Development and application of fiber-optic sensors in environmental

D E P A R T M E N T O F B I O L O G Y

F A C U L T Y O F S C I E N C E

U N I V E R S I T Y O F C O P E N H A G E N

PhD thesis

Lars F. Rickelt

Development and application of fiber-optic

sensors in environmental and life sciences

Academic advisor: Michael Kühl

Submitted: 01/09/2015

Development and application of fiber-optic sensors in environmental and life sciences

2

Section: Marine Biological Section Name of department: Department of Biology Author: Lars Fledelius Rickelt Titel: Udvikling af fiberoptiske sensorer og deres anvendelse i miljø- og

biovidenskab Title: Development and application of fiber-optic sensors in environmental

and life science Cover photo: Photograph of the spherical light collecting tip of a scalar irradiance

microprobe positioned in a culture dish next to a mouse embryo (from Ch.8; photo by M. Kühl)

Academic advisor: Michael Kühl Submitted: 01. September 2015 Grade: PhD


3

Development and application of

fiber-optic sensors in

environmental and life sciences

Ph. D. thesis by

Lars F. Rickelt

Supervisor: Professor Michael Kühl


4

Abstract

The light guiding properties of optical fibers are the fundament for fiber-optic sensors. The

composition of the fiber materials as well as the fabrication methods for both glass optical fibers

and plastic optical fibers (POF) are useful knowledge for improvements of the sensor design. A

majority of sensing materials includes imbedded luminescent dyes and all O2 fiber-optic sensors are

based on O2 quenching of a luminophore. The mechanisms of luminescence and O2 quenching are

described. A new procedure for etching a recess in the tip of multimode graded index optical glass

fibers was used to improve the mechanical stability of O2 optodes after the sensor chemistry was

immobilized into the recess. The recess improved focusing of the excitation light. An O2 optode

sensor array was constructed for long-term measurements in soil and sediments. The construction

and measuring characteristics of the sensor array system are presented along with a novel approach

for temperature compensation of O2 optodes. A method to measure the O2 tension in vivo in the

maxillary sinus of cystic fibrosis patients is described. The O2 tension is compared to the bacteria

found in the respective sinuses. O2 sensor spots placed inside vials with polymorphonuclear

leukocytes revealed strong O2 consumption. The O2 level was measured from outside the vials with

a POF. A new method for producing fiber-optic microprobes for measuring scalar irradiance is

presented along with an experimental setup for measuring the isotropic response in air and water.

The light collecting properties of differently sized scalar irradiance probes (30 μm – 470 µm)

produced by the new method were compared to probes produced with previously published

methods. A scalar irradiance microprobe was applied to assess the irradiance and total energy dose

from different microscopes during the in vitro embryonic development in mouse and pig and the

effect on the development was investigated. A highly luminescent cyclometalated iridium(III)

coumarin complex was compared to a ruthenium(II) polypyridyl complex or a platinium(II)

porphyrin complex. They were all three imbedded in polystyrene spin-coated on cover slips in a 1


5

µm thick layer. The combination of microscopic spatio-temporal O2 dynamics at the base of

heterotrophic biofilms and confocal imaging of biomass and structure demonstrated a complex

interaction between biomass distribution, mass transfer and flow. A simple ratiometric intensity

based O2 imaging protocol was developed using a conventional digital camera and the O2

distribution images were compared to life-time images obtained using a monochrome fast gate-able

CCD camera. The method was applied to a biofilm growth incubator incubated with bacteria

occurring in drinking water systems.


6

Abstrakt

Optiske fibres lysleder egenskaber er fundamentet for fiberoptiske sensorer. Fiber materialernes

sammensætning samt deres fremstillingsmetoder for både glas og plastik optiske fibre er nyttig

viden til forbedring af sensorenes design. Hovedparten af de fibre optiske sensorer benytter

luminiserende farvestoffer indlejret i en matrix og alle O2-fiberoptiske sensorer er baseret på

quenching af disse. Mekanismerne for luminescensens og O2-quenching er beskrevet. En ny metode

til at ætse hulrum i spidsen af multimode graded index fibre blev brugt til at forbedre styrken af O2-

optoder efter sensorkemien var immobiliseret i hulrummet. Det ætsede hulrum bevirkede en

forbedret fokusering af lysgangen. Et O2-optode spyd til langtidsmålinger i jord og sedimenter er

blevet fremstillet. Konstruktionen og målekarakteristikken af spyddet præsenteres sammen med en

ny måde at temperaturkompensere O2-optoder. En ny metode til at måle O2-trykket i bihulerne på

cystisk fibrose in vivo er beskrevet, og O2-trykket er sammenholdt med de bakterier, der blev fundet

i de respektive bihuler. O2-sensorspots anbragt indvendigt i små glasbeholdere indholdene

polymorphonukleare leukocytter afslørede disses kraftige O2-forbrug. O2-niveauet blev målet udefra

med optiske plastikfibre. En ny måde til fremstilling fiberoptiske mikroprober til måling af scalar

irradiance præsensters sammen med en eksperimentel opstilling til bestemmelse af deres isotropiske

respons i luft og vand. De lysopsamlende egenskaber af forskellige størrelser scalar irradiance

prober (30 μm – 470 µm) fremstillet efter den ny metode blev sammenlignet med prober fremstillet

efter tidligere publicerede metoder. En scalar irradiance mikroprobe blev benyttet til at bestemme

lysbestrålingen og den totale energi dosis fra forskellige mikroskoper under in vitro foster

udviklingen af mus og svin, og effekten på udviklingen blev undersøgt. Et kraftigt lysende

cyclometalated iridium(III) coumarin kompleks blev sammenlignet med et ruthenium(II)

polypyridyl kompleks og et platin(II) polypyridyl kompleks. De var alle tre indlejret i polystyren

spin-coatede dækglas i et 1 µm tykt lag. Kombinationen af den mikroskopiske spatio-temporale O2-


7

dynamik ved basis af heterotrofe biofilm og confocal afbildning af biomasse og struktur afslørede et

kompleks samspil mellem biomasse fordeling, bevægelse og flow. En enkel på ratiometrisk

intensitets byggende O2-billed protokol blev udviklet med brug af et konventionelt digitalt kamera

og O2-billederne blev sammenlignet med livstids billeder optaget med et sort-hvidt hurtigt skiftende

CCD-kamera. Metoden anvendtes på en biofilm-vækstinkubator inkuberet med bakterier, som

findes i drikkevand.


8

Preface

Ålsgårde, August 27, 2015

When I for two years ago sat linked to a wheelchair after a cerebral thrombosis I didn’t

dare thinking about ever finishing my PhD-thesis and now I am sitting and writing the last words. I

am extremely grateful to everyone who has been involved in this project. Whiteout their help it

would not have been possible.

Foremost I am grateful to my mentor and supervisor, Michael Kühl, for his excellent

assistance and guidance. Also former and current members of the Microenvironmental Ecology

Group at Marine Biological Section I owe great thanks.

Special thanks go to Egil Nielsen, master of the workshop, for help making small and big

devices and to our secretary Anne Drøscher for her help and positive attitude.

Sincere thanks!

Lars F. Rickelt


9

List of included papers (Ch.2) L.F. Rickelt, L.D. Ottosen, and M. Kühl:

Etching of multimode optical glass fibers: A new method for shaping the measuring tip and

immobilization of indicator dyes in recessed fiber-optic microprobes

Sensors and Actuators B 211: 462–468 (2015)

DOI: http://dx.doi.org/10.1016/j.snb.2015.01.091

(Ch.3) L.F. Rickelt, L. Askaer, E. Walpersdorf, B. Elberling, R.N. Glud, and M. Kühl:

An optode sensor array for long term in situ measurements of O2 in soil and sediment

Journal of Environmental Quality 42: 1267-1273 (2013) + Errata 43:1–1 (2014)

DOI: 10.2134/jeq2012.0334 + DOI:10.2134/jeq2010.0163er

(Ch.4) B. Elberling, R.N. Glud, C.J. Jørgensen, L. Askaer, L.F. Rickelt, H.P. Joensen, M. Larsen, and L.

Liengaard: Methods to assess high-resolution subsurface gas concentrations and gas fluxes in

wetland ecosystems

In: R.D. DeLaune, K.R. Reddy, C.J. Richardson, and J.P. Megonigal (Eds.), Methods in

Biogeochemistry of Wetlands. Soil Science Society of America Inc., Madison, (2013) 949-966

DOI: 10.2136/sssabookser10.c49

(Ch.5) K. Aaness, L.F. Rickelt, H.K. Johansen, C. von Buchwald, T. Pressler, N. Høiby, and P.Ø. Jensen:

Decreased mucosal oxygen tension in the maxillary sinuses in patients with cystic fibrosis

Journal of Cystic Fibrosis 10: 114–120 (2011)

DOI: 10.1016/j.jcf.2010.12.002

(Ch.6) K.N. Kragh, M. Alhede, P.Ø. Jensen, C. Moser, C.S. Jacobsen, S. Seier, S. Eickhardt, H. Trøstrup,

L. Christoffersen, H.-P. Hougen, L.F. Rickelt, M. Kühl, N. Høiby, and T. Bjarnsholt:

Polymorphonuclear leukocytes restrict the growth of Pseudomonas aeruginosa in lungs of cystic

fibrosis patients

Infection and Immunity 82: 4477-4486 (2014)

DOI: doi:10.1128/IAI.01969-14

(Ch.7) L.F. Rickelt, M. Lichtenberg, E.C.L. Trampe, and M. Kühl:

Fiber-optic probes for small scale measurements of scalar irradiance.

Submitted to Photochemistry & Photobiology (September 2015)

(Ch.8) R. Li, K.S. Pedersen, Y. Liu, H.S. Pedersen, M. Lægdsmand, L.F. Rickelt, M. Kühl, and H.

Callesen:

Effect of red light on development and quality of mammalian embryos.

Journal of Assisted Reproduction and Genetics 31: 795-801 (2014)

DOI: 10.1007/s10815-014-0247-7

(Ch.9) M. Staal, S. Borisov, L.F. Rickelt, I. Klimant, and M. Kühl:

Ultrabright planar optodes for luminescence life-time based microscopic imaging of O2 dynamics

in biofilms.

Journal of Microbiological Methods 85: 67–74 (2011)

DOI: 10.1016/j.mimet.2011.01.021


10

(Ch.10) M. Staal, E. Prest, H. Vrouwenvelder, L.F. Rickelt, and M. Kühl:

A simple optode based imaging technique to measure O2 distribution and dynamics in tap water

biofilms

Water Research 45: 5027-5037 (2011)

DOI: 10.1016/j.watres.2011.07.007


11

Contents

Abstract (English) 4

Abstrakt (dansk) 6

Preface 8

List of included papers 9

1. Introduction 12

2. Etching of multimode optical glass fibers 34

3. An optode sensor array

4. Methods in wetland ecosystems

5. Oxygen tension in the maxillary sinus

6. Polymorphonuclear leukocytes in cystic fibrosis

7. Fiber-optic scalar irradiance probes

8. Effect of red light on mammalian embryos

9. Ultrabright planar optodes

10. A simple optode for tap water biofilms

11. Appendix - Posters


12

1. Introduction

Optical Fibers

Optical fibers are flexible, transparent fibers made by drawing glass or plastic to a small diameter.

They usually consist of a core surrounded by a cladding with a lower refractive index. The fiber acts

as a waveguide due to total internal reflection. Multi-mode fibers support many propagation paths,

while single-mode fibers only support a single path. Single-mode fibers are well suited for a wide

variety of optical sensors measuring e.g. temperature, pressure, and chemicals [1]. Due to the

dielectric nature of the materials they are insensitive to electrical and magnetic noise, they are easy

to build and inexpensive [2]. The small core diameter of single-mode fibers (3-10 µm) makes

optical alignment crucial and effective light transmission is best accomplished with coherent light

sources as lasers or edge emitting diodes [3, 4]. The core diameter of multi-mode fibers is wider

(50-400 µm for glass) and they are better suited for optical sensors, where light intensity is detected,

or where the light path is split e.g. in optode instruments [3]. All fiber sensors in this thesis are

based on multi-mode fibers.

When a light ray comes from a media with higher refractive index and enters a media with lower

refractive index, total reflection occurs if a certain critical angel is exceed. From Snell’s law of

refraction ( ) the critical angle of the core-cladding interface of an optical

fiber can be calculated as:

defined by the refractive index of the core, , and of the cladding, . If a ray enters the optical

fiber (fig. 1.1) with an angel below a certain angel, the acceptance angle ( ), it will arrive at the

core-cladding interface at an angle and it will be totally reflected and can be transmitted


13

with relative little loss in energy over long distances. In contrast, light that enters the optical fiber at

an angle larger than will hit the core-cladding interface at an angle, e.g. , and will not be

totally reflected. At each reflection some of the light will dissipate in the cladding until total

dissipation by absorption in the cladding and scattering out of the cladding [5, 6]. The range of

angles over which wave guiding occurs within an optical fiber defines the numerical aperture (NA)

of the fiber and can be calculated from the refractive indices of the core ( ), the cladding ( ), and

the surrounding media ( ).

In air = 1 and in water [3, 5].

In multi-mode step-index optical fibers the light propagates in different modes or path. The

condition, that the angel of reflection being larger than is just a necessary condition for guiding

light in an optical fiber. Solution of Maxwell’s wave equations shows that the propagating of light

must also satisfy phase conditions at the boundary between core and cladding. That is, the phase

shift of the light wave between successive reflections must be an integer multiple of 2π. That

Figure 1.1. Schematic drawing of the guiding of light through an optical multi-mode step

index fiber. The core and the cladding have different refractive indices, and ,

respectively. If the light hits the core-cladding interface at an angle larger than the critical

angel e.g. it will be guide along the fiber by total internal reflection. If the angle is smaller

than the critical angle e.g. it will not be totally reflected, it will lose energy at each

reflection, and will eventually dissipate by light scattering out of the cladding. All light

entering the fiber at an angle smaller than , i.e. inside the acceptance cone, will be guided

along the fiber.


14

requirement results in a basic characteristic of a guided wave: the light beams can only propagate in

the fiber with discrete angles. The light with the smallest angel to the fiber axis is called the

fundamental mode, and others high-order modes. When the core diameter and the index difference

are small enough, only the fundamental mode can propagate inside the fiber. Such a fiber is called

single-mode fibers. In single-mode fibers up to 30% of the wave field, called the evanescent field,

propagates in the cladding at the core-cladding interface. The mode characteristics depend also on

wavelength. A fiber can be single-mode for longer wavelengths, but becomes multi-mode for

shorter wavelengths [1, 6].

Looking at the situation in multi-mode fibers from a ray optical view, the light rays that is reflected

at a steeper angles travel a much longer path than the light rays that are reflected at a shallow angel

(Fig. 1.1 and 1.2). This leads to differences in propagation time and thus, to pulse spreading called

intermodal dispersion. The intermodal dispersion can be minimized with graded-index fibers. They

have a parabolic shaped index profile with a maximum at the core axis (Fig. 1.2).

Figure 1.2. Overview of the three main types of optical fibers: multi-mode step index, multi-mode

graded-index, and single-mode fibers. 1st row is typical cross sections of the fibers; 2

nd row is the

corresponding refractive index profiles; 3rd

row is an example of an input pulse; 4th

row is a

longitudinal section of the fibers included some possible path; 5th

row is the resultant pulse of the

input pulse after passing a fiber length. (From Optical Fiber, Wikipedia)


15

Polymer Optical fibers

As mentioned, optical fibers can also be made out of plastic. PMMA is often the core material and

fluorinated polymer is the cladding material. A high refractive index difference between core and

cladding causing a high NA is common. Even with diameters much large than glass fibers they are

flexible and easy to handle. They can have diameters up to 2 mm where 96% of the cross section is

core. The price of polymer or plastic optical fibers (POF) is equal to glass optical fibers, but optical

links, connectors and installation is inexpensive. Due to the attenuation and distortion

characteristics of PMMA fibers are used for low-speed, short-distance (up to 100 m) applications.

There have been huge developments in materials for POFs in recent years with less attenuation loss

and higher bandwidth as result. The most common POF is multi-mode step-index, but graded-

index fibers are also available. It is possible to produce single mode POFs, but the advantage with

the low-cost installation equipment is lost with the small core diameter [1, 6-9].

Fabrication of optical glass fibers

The modern preparation of optical fibers includes two major stages. The first stage is to produce a

cylindrical preform of silica composition about 10-20 cm in diameter and 50-100 cm long. The

preform has a cladding with an inner core with the same refractive index profile and other optical

characteristics as the desired optical fiber. It is just much larger. There are developed several

methods for this procedure but the most common is based on modified chemical vapor deposition

(MCVD) technology invented by MacChesney et al. [10]. Al types of fiber preforms from single-

mode to step-index and graded-index multimode can be produced after this method.


16

Figure 1.3 shows a schematic diagram of the MCVD processing of a Ge-doped silica preform. A

tube of pure silica is placed in a lathe and rotated over an oxygen-hydrogen burner traversing along

the length. The reactive gases flowing through the tube form small particles (~0.1µm) called soot

that deposit on the tube wall downstream from the hot zone of the burner. The gas stream consists

of oxygen and extremely pure chlorides of silicon and germanium or other dopants. The main cost

by producing optical fibers is the production of these pure substances. The vapor phase reactions

produce mixtures of SiO2 and GeO2:

SiCl4 + O2 → SiO2 + 2Cl2

GeCl4 + O2 → GeO2 + 2Cl2

The deposition is made by application of one layer atop one another and the exact amount of dopant

added to each layer can be controlled by the flow of the gasses, thus controlling the refractive-index

profile. After the necessary layers are built up, additional heat is applied to the tube, causing it to

collapse into a solid rod or preform ready for the second stage [1, 2, 8, 10].

Figure 1.3. Schematic diagram of the MCVD process for Ge-doped silica fiber

preform. A pure silica tube is rotated over a traversing O2/H2-burner. The reactive

gases SiCl4, GeCl4, and O2 enter the tube from the end and forms small particles at

the back-side of the burner. The concentration of the reactive gases can be adjusted

controlling the reactive index in thin layers (From Fang et al. (2012) [1])


17

The second stage of optical glass fiber fabrication is the drawing process. The major steps of a

typical drawing procedure are shown in figure 1.4. The preform is put in a draw furnace and the

bottom tip is heated to melting allowing a fiber to be formed. This fiber is pulled through diameter-

monitoring equipment. It controls the fiber diameter by changing the drawing rate executed by the

pulling tractor assembly. A coating applicator applies a polymer coating or jacket over the cladding.

This process is controlled by concentricity-monitoring equipment. The polymer coating is cured by

UV-lamps or heat and the fiber is wound up on ready-to-ship reels. The drawing rate is critical to

Figure 1.4. Schematic diagram of typical drawing process.

The fiber preform is placed in the draw furnace and drawn

by the pulling tractor. The fiber diameter is monitored by

the diameter monitoring equipment controlling the drawing

rate and temperature. Fiber can also be drawn from a

double extrusion process where the cladding is put on the

core in an additional drawing process (from Mynbaev &

Scheiner 2006 [8]).


18

the process and is depended on the draw furnace temperature. The slower the drawing is, the better

the quality can be controlled. But a fast drawing rate is more productive. The drawing rate can run

from 200 m/min to 2000 m/min. To reach such a speed all rotating parts must be manufactured to

extremely tight tolerances and the process must be controlled with a high degree of accuracy.

This is a very simplified description and it took many years of intensive, costly research and

development efforts to make this process commercially possible [2, 8].

Dopants in optical glass fibers

Dopants are added to the pure silica (SiO2) to change the refractive index. Many different dopants

are used including oxides of aluminum and rare earth metals, but the most common employed are

germanium oxide (GeO2) and phosphorous(III) oxide (P2O5) to increase the refractive index and

boron(III) oxide (B2O3) and fluorine (F) to decrease the refractive index. They show a close to

linear relationship between the refractive index and the mole concentration in SiO2 (Fig. 1.5)

Fluorine is commonly used to lower the refractive index in the cladding of step-index fibers and

GeO2 is used in graded-index fibers to create parabolic shaped profiles with larger index in the fiber

center [1, 8, 11, 12].

Figure 1.5. Refractive index as a function of dopant materials and

their mole concentration in SiO2. (From Keiser 1991 [12])


19

HF etching of graded index optical fibers

The hydrofluoric acid (HF) etching rate of SiO2 doped with GeO2 increases with increasing dopant

content [13, 14]. This characteristic was applied to etch parabolic shaped recesses in the tip of flat-

cut tapered or untapered graded index optical fibers. The etched tips showed improved focusing of

light coupled into the opposite end, and very efficient excitation of thin layers of optical indicators

immobilized in recessed fiber tips. This allowed the construction of O2 microoptodes with improved

mechanical stability that can measure repeatedly even in very cohesive biofilms, tissue and soil

(Ch.2).

Fabrication of polymer optical fibers

For the core material high transparency is required and the polymer should be amorphous.

Poly(methyl methacrylate) (PMMA) and polystyrene (PS) are normally used as a core because

these polymers are easily purified at the monomer level by distillation and membrane particle

separation. No condensation reaction is involved in obtaining transparent polymer by radical

polymerization with initiators and chain transfer agents [15, 16]. POFs can be produced in a two

stage process similar to glass optical fibers, but step-index POFs can be manufactured efficient by

means of a continuous double extrusion process: the monomer, the initiator, and the chain transfer

agent are continuously fed into a reactor and the fiber is continuously drawn from the die. Cladding

is put on immediately after in an additional extruding process. The cladding material should have a

refractive index 2-5% less than the core material. PMMA and fluorinated polymers are used as

claddings onto PS and PMMA core, respectively [16].

Graded index POFs are produced similar to optical glass fibers in a two stage process. The first

stage is the creation of a preform and the second stage is the drawing process. The graded-index

preform can be made using interfacial gel-polymerization technique. A tube prepared out of the


20

cladding material e.g. PMMA is filled with methyl methacrylate (MMA) monomer, the

polymerization initiator, the chain transfer agent, and a dopant, which is another organic compound

with a higher refractive index than that of PMMA. Bromobenzen, benzyl butyl phthalate, benzyl

benzoate, and diphenyl phthalate have the desired properties which includes a higher molecular

volume than the monomer MMA. The mixture filled PMMA tube is placed horizontally in an oven

at 95⁰C and rotated on its axis at ~50 r.p.m. for 24 hours. A gel phase is formed on the inner tube

wall where the polymerization is accelerated due to the well-known so-called gel-effect. The

thickness of the polymer phase will gradually increase from the tube wall until it reaches the center

of the tube. During this polymerization process a radial graded-index distribution of the dopant is

formed. The diffusing of the MMA monomer into the polymer phase is faster than the aromatic

dopant due to the dopants higher molecular volume. Therefore, the dopant molecules are gradually

concentrated in the center region as the polymer phase increases [17].

Another approach for creation of graded-index POF preforms is the rod-in-tube method. A polymer

tube without dopants and a polymer rod with an even distribution of dopants with a higher

refractive index than the polymer are prepared separately. The polymer rod is inserted in the

polymer tube and placed in an oven at e.g. 150°C for 24 hours. During the heat treatment, the rod

and the tube will adhere to each other and the dopant diffuses from the rod into the tube. After this

heat diffusion process a preform with a graded index profile is obtained [18].

In the second stage the POF preform is drawn into a fiber much the same way as silica glass

preforms are drawn (see fig. 1.4)


21

Luminescence processes

All fiber optical O2 sensors (O2 optodes) are based on O2 collisional quenching of luminescence

(fluorescence or phosphorescence) [19].

Luminescence

Luminescence is the emission of light from any substance, and occurs from electronically excited

states. It is best explained by a

simplified Jablonski diagram (Fig.

2.1). The singlet ground, first, and

second electronic states are denoted

S0, S1, and S2, respectively. The

corresponding vibrational energy

levels are denoted 0, 1, 2, etc. When

absorbing light, a fluorophore is

excited to a higher vibrational level of

either S1 or S2 and will rapidly relax by

internal conversion to the lowest

vibrational leve of S1. The fluorophore

will then relax to some vibrational level

of S0 emitting light as fluorescence.

The emitted light will have less energy than the absorbed light, hence longer wavelength.

Molecules in the S1 state can also undergo a spin conversion, called intersystem crossing, to the first

triplet state T1. Emission from T1 is termed phosphorescence and is usually shifted to longer

wavelengths with lower energy relative to the fluorescence. Transition from T1 to S0 is forbidden

and as a resutlt the rate constants for triplet emission are several orders of magnitude smaller than

Figure 2.1. Jablonski diagram of electronic energy levels

showing absorption at ground singlet state So (hνA) and

emission of fluorescence from first singlet state S1 (hνF) or

emission of phosphorescence from first triplet state T1

(hνP) via intersystem crossing from S1 to T1.

(Adapted from photochemistry .wordpress.com)


22

Figure 2.2. Presentation of the dynamic process of collisional

O2 quenching of luminescence. When absorbing light the

luminophore is excited to a higher vibrational level. In the

absence of O2 the luminophore will fall back to the ground state

emitting light at a larger wavelength. In the presence of O2 the

luminophore can collide with O2 and fall back to the ground

state transferring the excitation energy to O2.

those for fluorescence. The phosporecent lifetime is correspondingly lager than the fluorecent

lifetime. The luminescent lifetime (τ) is defined as the average time the molecule spends in the

excited state prior to return to the ground state. For a single exponential decay 63% of the molecules

have decayed prior to t = τ and 37% decay at t > τ. Another impotant characteristc of luminophores

are the quatum yield. The luminoscence quatum yield is the ratio of the number of photons emitted

to the number absorbed [20].

Luminescence quenching by O2

The intensity of luminescence can be

decreased by collisional quenching.

Molecular oxygen (O2) has an

electronic triplet configuration in the

ground state and is a good quencher of

fluorescence and phosphorescence [21,

22]. The mechanism differs a little but

the over-all scheme can be depicted as

in figure 2.2. The excited luminophore

(S1 or T1 state) will in the absence of

O2 fall back to its ground state (S0)

under emission of light. In the presence

of O2 collisional quenching can take

place. It is thought to happen by transfer

of electronic excitation energy from the

triplet state of the luminophore to O2


23

resulting in a ground-state luminophore and an electronically excited singlet-state O2. The excited

single-state O2 will rapidly fall back to the triplet ground-state emitting the excess energy as heat

[23-25].

For collisional quenching the decrease in intensity or lifetime is described by the Stern-Volmer

equation:

In this expression and are the luminescent intensities at zero oxygen and a given O2

concentration, , respectively. and are the corresponding lifetimes. The over-all quenching

constant quantifies the quenching efficiency [26]. can be determined by a two point

calibration. The first point must be under anoxic conditions and the other point can be at any known

O2 concentration e.g. 100% airsaturation.

This simple two parameter equation is valid only for ideal systems such as diluted solutions of

luminophores in liquid solvents. For luminophores imbedded in a solid matrix e.g. polystyrene (PS)

it is necessary to modify the simple two-parameter equation by introducing another parameter ,

which is the fraction of the luminophore that is assumed not to be significantly quenched by O2 [27-

29]:

For this equation a three point calibration is necessary for determination of . For most sensor

matrices is temperature independent and constant over the dynamic range so it is only required to

define once and hereafter a two point calibration is sufficient [3].


24

Fig. 2.3 Electro optical set-up of an oxygen

measuring system: fcop – 2x2 fiber coupler; ex –

excitation LED; ref – reference LED; OF – optical

filters; ST – fiber connectors (from Holst et al.

2000 [30])

Figure 2.4. Sinusoidal modulated light for

measuring the phase angle shift Φ. – reference

light; - anoxic response; - response at O2 content c

Optical measuring techniques

Luminophores enable detection at a different

wavelength from excitation as described. This

improves the efficiency of measurements,

because both are guided by the same fiber.

Separation of the beams is achieved by a fiber

coupler and optical filters (Fig. 2.3). The first set-

ups for optical measurements of O2 with fibers

were based on detection of changes in intensity and

an effective optical insulation of the measuring tip

was necessary [29]. Modern O2 optode instruments

are based on life-time measurements. The largest

manufacturers of such instruments are Presens

(www.presens.de), PyroScience (www.pyro-

science.com), Centec (www.centec.de), Ocean Optics Inc. (www.oceanoptics.com), Oxysens Inc.

(www.oxysense.com), Finesse Inc. (www.finesse.com), and Hach-Lange GmbH (www.hach-

lange.de). In the medical field, OptiMedical Systems Inc. (www.optimedical.com) probably is the

largest [19]. The luminescent decay life-time is independent of indicator concentration, optical

environment or the light path including fiber bending and refraction index change [20]. The life-

time ( ) is usually measured by a phase modulation technique. The excitation light is sinusoidal

intensity modulated at a frequency ( ) and can be calculated from the phase angle shift ( )

between the excitation and emission signals:

http://www.presens.de/

http://www.centec.de/

http://www.oceanoptics.com/

http://www.oxysense.com/

http://www.finesse.com/

http://www.hach-lange.de/

http://www.hach-lange.de/

http://www.optimedical.com/


25

In practice it is done by detecting the delay or difference between a reference light and the emission

light (Fig. 2.3 and fig. 2.4). The reference light is modulated as the excitation light and the reference

LED has a wavelength close to the emission wavelength. If the two lifetimes and , that

describe the boundaries of the measuring range, are know the optimum modulation frequency

can be calculated from [20]:

Temperature compensation

Luminescence is dependent on temperature. Due to an increasing relaxation rate the luminescent

life-time decreases with increasing temperature. Most instruments for O2-optode measurements

enable temperature compensation, but it is only reasonable accurate within a few degrees from the

calibration temperature (Roland Thar, personal communication). For long-term in situ

environmental applications this is insufficient. The luminescent life-time of Ru(II) (diphenyl

phenanthroline)3 imbedded in an ultra-thin polystyrene layer (~1μm) showed a linear decline with

increasing temperature either at anoxic conditions and at 100% air saturation [31]. A corresponding

relationship was found for Pt(II) meso-tetra(pentafluorophenyl)porphine imbedded in ~10µm

polystyrene. This information was applied to develop temperature compensation valid over a range

of 0°-25°C. The compensation method was used on data collected from a spear-array with ten POF

O2 optodes fixed together with thermocouples. The spear-array was placed for several months in a

wetland peat bog and the POFs were connected to a multichannel fiberoptic meter. A data logger

acquired all signals from the multichannel meter and the thermocouples. The collected data showed

pronounced variations in O2 distribution after marked shifts in water level and anoxic conditions

below the water level but also diel variations in O2 concentrations in the upper layer presumably

due to rhizospheric oxidation by the main vegetation Phalaris arundinacea (Ch.3a). Later I realized


26

that a systematic error was committed during the calibration of the spear array and an erratum was

made (Ch.3b). A device for measuring in situ O2 profiles in gravely streambeds applied a similar

method for temperature compensation. The device had to be manually operated to obtain profiles

[32]. Fiber optode and planar optode O2 detection have also along with other dissolved gas

measuring methods in wetlands been presented in a book chapter (Ch.4)

Measuring pO2 in the maxillary sinus

A 2 mm POF was inserted in a nose catheter and equipped with a 2 mm disk planar optode for

measurements of pO2 in the maxillary sinus of cystic fibrosis and non-cystic fibrosis patients. The

catheter optodes were calibrated in air at 37°C in a thermostatically controlled climate room for the

100% air saturation value and in an anoxic bench for the zero O2 value. The O2 optode catheters

were sterilized prior to use in a plasma oven using hydrogen peroxide enabling sterilization at low

temperatures without damage or change in calibration values. The cystic fibrosis patients had a

significant lower pO2 on the mucosa but not in the sinus lumen as compared with the control group

of non-cystic fibrosis patients. Anoxic conditions were found in 7/39 (18%) of the sinuses from

where P. aeruginosa, Stenotrophomonas maltophilia and/or coagulase negative staphylococci were

cultured (Ch.5).

O2 consumption of polymorphonuclear leucocytes

Sensor spots mounted inside two airtight vials separated by a membrane were applied to measure

the O2 consumption of polymorphonuclear leucocytes (PMN). O2 levels in the lower, airtight

chamber containing P. aeruginosa and in the upper chamber containing the PMNs were monitored

from the outside with a 2 mm POF connected to a fiber-optic O2 meter. After 0, 2, and 4 h the

bacterial growth was checked. The growth rates of P. aeruginosa were significantly lower in the


27

presence of PMNs. To evaluate whether the PMNs restricted the growth of P. aeruginosa by O2

limitation, the medium was supplemented with an alternative electron acceptor, , which is not

affected by PMN metabolism. The growth rate of P. aeruginosa in the presence of PMNs and

was comparable to that observed without PMNs at both 2 h and 4 h (Ch.6).


28

Microprobes for light field measurements

Optical fibers take advantage of the inherent light-guiding capability and the fibers can relatively

simple be change to a light field probe. Figure 3.1 presents an overview of the three fundamental

optical parameters for characterizing light field in environmental microbiology [3].

A tapered and flat cut optical fiber coated on the tapered sides is an ideal microprobe for measuring

the field irradiance, , defined as the radiant flux per unit area and solid angel from a certain

direction specified by the zenith, , and azimuth, , angels in a spherical coordinate system.

Figure 3.1. The three fundamental optical parameters for characterizing light fields in environmental

microbiology. The field radiance, , is the energy or radiant flux, Φ, from a direction, specified

by the zenith, , and the azimuth, , angles in a spherical coordinate system, within the solid angle,

, through an area, , perpendicular to the flux. The downwelling irradiance, , is the radiance

per unit horizontal surface, , integrated over the upper hemisphere (2π solid angel and the radiance

is weighted with cosine to the incident zenith angel, ). The scalar irradiance, , is the radiant flux

incident from all directions about a point and can be express as the field irradiance integrated over the

hole sphere of a 4π solid angel. (adapted from Kühl and Jørgensen 1994 [33])


29

Also untapered flat cut optical fibers can be used as field irradiance probes [5, 33-35]. Another type

of field irradiance probe has been constructed from an irradiance sensor inside a concentric,

moveable light-absorbing sheath [36].

So-called cosine collectors can be applied for measurements of the downwelling irradiance, , i.e.

the radiance per unit horizontal surface integrated over the upper hemisphere. The radiance is

weighted with cosine to the incident zenith angel, . They can be made from tapered optical fibers

by casting a hemisphere of a light scattering material e.g. TiO2 in PMMA. The dry probe tip is

covered with opaque black paint and the hemisphere is grounded flat [37].

Scalar irradiance is the radiant flux from all directions about a point and can be measured by probes

with a scattering sphere upon tapered or untapered flat cut fibers (Ch.7). Such probes were applied

to assess the effect of irradiance from microscopes during the in vitro embryonic development cycle

in mouse and pig (Ch.8)

Applications of O2 planar optodes

A comparative study of cover slips spin-coated with three different luminescent O2 indicators

embedded in polystyrene established that Ir(III) acetylacetonato-bis(3-(benzothiazol-2-yl)-7-

(diethylamino)-coumarin) (IrC) had superior performance compared to Pt(II) meso-

tetra(pentafluorophenyl)porphyrin (PtTPFP) and Ru(II)-tris-4,7-diphenyl-1,10-phenanthroline

Ru(dpp)3. The combination of microscopic spatio-temporal O2 dynamics at the base of

heterotrophic biofilms and confocal imaging of biomass and structure demonstrated a complex

interaction between biomass distribution, mass transfer and flow (Ch.9). The spin-coating method

was developed earlier [31].

A simple ratiometric intensity based O2 imaging protocol was developed using a conventional

digital camera and the O2 distribution images were compared to life-time images obtained using a


30

monochrome fast gate-able CCD camera (PCO). The method was applied to a biofilm growth

incubator incubated with bacteria occurring in drinking water systems (Ch.10).

Future application of results

Although the tip configuration of fiber-optic microsensors plays an important role for their

performance, not much attention has been given to improve the design of the measuring tip and the

immobilization of the indicator with respect to improved mechanical and optical properties. The

investigation of scalar irradiance probes revealed a distinctive difference in response of the probes

whether measured in air or water if the optical insulation of the fiber was insufficient (Ch.7). A

large drop in intensity is also seen when optical non-insulated O2 microoptodes are inserted in water

from air. It is conventional explained by loss of light out of the sensing layer into the water due to

the difference in the refractive index between air and water. It could be that some light escape out of

the taper near the tip and an optical insulation of the taper will reduce the loss. In combination with

better focusing of the light by recess etching (Ch.2) only a very thin layer of sensing material will

be necessary resulting in ultra-fast responding optodes.

Spin coated cover slips with a reference and an O2 sensitive dye was thought to imagine O2 on a

confocal microscope using the microscopes laser for excitation and the microscopes image

processing system for data handling. It was thought to be a straight forward task, but preliminary

results showed severely damage to the dyes caused by the laser. I think it will be possible to do this

with the right settings, but additional work has to be done.


31

References

[1] Z. Fang, K.K. Chin, R. Qu, and H. Cai (2012): Fundamentals of Optical Fiber Sensors. 1st

ed. John Wiley & Sons, Inc. Hoboken, NJ, USA.

[2] E. Udd and W.B. Spillman, JR. (2011): Fiber Optic Sensors. An Introduction for Engineers

and Scientists. 2nd

ed. John Wiley & Sons, ISBN: 978-0-470-12684-4

[3] M. Kühl (2005): Optical microsensors for analysis of microbial communities.

MethodsEnzymol. 397 166–199. doi:10.1016/S0076-6879(05)97010-9

[4] M. Ettenberg, G. Olsen, and F. Hawrylo (1984): On the reliability of 1.3-µm InGaAsP/InP

edge-emitting LED’s for optical-fiber communication. J. Lightwave Technol. 2 (6) 1016-

1023

[5] T.C. Vogelmann, G. Martin, G. Chen, and D. Buttry (1991): Fibre optic microprobes and

measurements of the light microenvironment within plant tissue. Advances in Botanical

Research 18, 255-295

[6] D.A. Nolan, P.E. Blaszyk, and E. Udd (2011): Optical Fibers, in Fiber Optic Sensors. An

Introduction for Engineers and Scientists. ( ed.s E. Udd and W.B. Spillman) 2nd

ed. John

Wiley & Sons pp. 9-33

[7] J.M. Senior and M.Y. Jamro (2008): Optical Fiber Communications, Principles and Practice.

Prentice Hall

[8] D.K. Mynbaev and L.L. Scheiner (2006): Fiber-Optic Communications Technology.

Prentice Hall

[9] K. Peters (2011): Polymer optical sensors-a review. Smart Mater. Struct. 20 17pp

doi:10.1088/0964-1726/20/1/013002

[10] J.B. MacChesney, P.B. Connor, and H.M. Presby (1974): A New Technique for the

Preparation of Low-Loss and Graded-Index Optical Fibers. Proc. IEEE 62, 1280-1281

[11] J.W. Fleming and D.L.Wood (1983): Refractive dispersion and related properties in fluorine

doped silica. Applid Optics 22 (19) 3102-3104

[12] Gerd Keiser (1991): Optical Fiber Communications. 2nd

ed., McGraw-Hill, New York

[13] S. Neffe and M. Szustakowski (1986): Characterization of the microstructure of optical

fibers. Proc. SPIE 670, 100-103


32

[14] P. Pace, S.T. Huntington, K. Lyvtiänen, A. Roberts, and J.D. Love (2004): Refractive index

profiles of Ge-doped fibers with nanometer spatial resolution using atomic force

microscopy. Opt. Express 12 (7) 1452-1457

[15] Toshikuni Kaino (1996): Plastic optical fibers, in Polymeric Materials Encyclopedia, J.C.

Salamone (Ed.), pp.5288-5300 CRC Press: Boca Raton, FL. ISBN 0-8493-2470-X.

[16] J. Zubia and J. Arrue (2001): Plastic Optical Fibers: An Introduction to Their Technological

Processes and Applications. Opt. Fiber Technol. 7, 101-140

[17] Y. Koike, T. Ishigure, and E. Nihei (1995): High-Bandwidth Graded-Index Polymer Optical

Fiber. J. Light. Tecnol. 13, 1475-1489

[18] K. Makino, Y. Akimoto, K. Koike, A. Kondo, A. Inoue, and Y. Koike (2013): Low Loss

and High Bandwith Polystyrene-Based Graded Index Polymer Optical Fiber. J. Light.

Technol. 31, 2407-2412

[19] X.D.. Wang and O.S. Wolfbeis (2013): Fiber-optic Chemical Sensors and Biosensors (2008-

2012). Anal. Chem. 85, 487-508

[20] J.R Lakowicz (2006): Principles of Fluorescence Spectroscopy. 3rd ed. Springer, NY, USA

ISBN-10: 0-387-31278-1

[21] C. S. Parmenter and J. D. Rau (1969): Fluorescence Quenching in Aromatic Hydrocarbons

by Oxygen. J. Chem. Phys. 51 (5) 2242-2246 doi: 10.1063/1.1672322

[22] K. Kikuchi, C. Sato, M. Watabe, H. Ikeda, Y. Takahashi, T. Miyashi (1993): New aspects of

fluorescence quenching by molecular oxygen J. Am. Chem. Soc., 115 (12) 5180–5184 DOI:

10.1021/ja00065a033

[23] D.F. Evans (1957): Perturbation of singlet-triplet transitions of aromatic molecules by

oxygen under pressure. J. Chem. Soc. 1957 1351-1357 DOI: 10.1039/JR9570001351

[24] K. Kawaoka, A. U. Khan and David R. Kearns (1967): Role of Singlet Excited States of

Molecular Oxygen in the Quenching of Organic Triplet States. J. Chem. Phys. 46 (5) 1842-

1853 doi: 10.1063/1.1840943

[25] David R. Kearns and A. J. Stone (1971): Excited‐State Intermolecular Interactions Involving

Paramagnetic Molecules: Effect of Spin—Spin and Spin—Orbit Interactions on the

Quenching of Triplets. J. Chem. Phys. 55, 3383-3389 doi: 10.1063/1.1676589

[26] O. Stern and M. Volmer (1919): Über die Abklingzeit der Fluoreszenz. Phys. Z. 20, 183-188

[27] E. R. Carraway, J. N. Demas, B. A. DeGraff (1991): Luminescence quenching mechanism

for microheterogeneous systems. Anal. Chem. 63 (4) 332–336 DOI: 10.1021/ac00004a006


33

[28] E. R. Carraway, J. N. Demas, B. A. DeGraff, J. R. Bacon (1991): Photophysics and

photochemistry of oxygen sensors based on luminescent transition-metal complexes Anal.

Chem. 63 (4) 337–342 DOI: 10.1021/ac00004a007

[29] I. Klimant, V. Meyer, and M. Kühl (1995): Fiber-optic oxygen microsensors, a new tool in

aquatic biology. Limnol. Oceanogr. 40 (6) 1159-1165

[30] G. Holst, I. Klimant, O. Kohls, and M. Kühl (2000): Optical microsensors and microprobes.

In ‘‘Chemical Sensors in Oceanography’’ (M. Varney, ed.), pp. 143–188. Gordon & Breach.

2000ISBN: 9056992554

[31] M. Kühl, L.F. Rickelt, and R. Thar (2007): Combined imaging of bacteria and oxygen in

biofilms. Appl. Environ. Microbiol. 73 (19) 6289-6295 DOI: 10.1128/AEM.01574-07

[32] M. Vieweg, N. Trauth, J.H. Fleckenstein, and C. Schmidt (2013): Robust Optode-Based

Method for Measuring in Situ Oxygen Profiles in Gravelly Streambeds. Environ. Sci.

Technol.47 9858-9865 dx.doi.org/10.1021/es401040w

[33] M. Kühl and B.B. Jørgensen (1994): The light field of microbenthic communities: Radiance

distribution and microscale optics of sandy coastal sediments. Limnol. Oceanogr. 39 (6)

1368-1398

[34] M. Kühl and B.B. Jørgensen (1992): Spectral light measurements in microbenthic

phototrophic communities with a fiber-optic microprobe coupled to a sensitive diode array

detector. Limnol. Oceanogr. 37 1813-1823

[35] B.B. Jørgensen and D.J. Des Marais (1986): A simple fiber-optic microprobe for high

resolution light measurements. Limnol. Oceanogr. 31, 1376-1383

[36] T.E. Murphy, S. Pilorz, L. Prufert-Bebout, and B. Bebout (2015): A Novel Microsensor for

Measuring Angular Distribution of Radiative Intensity. Photochem. Photobiol. 91 862–868

[37] C. Lassen and B.B. Jørgensen (1994): A fiber-optic irradiance microsensor (cosine

collector): application for in situ measurements of absorption coefficients in sediments and

microbial mats. FEMS Microbiol. Ecol. 15 321-336


34

2. Etching of multimode optical glass fibers

L.F. Rickelt, L.D. Ottosen, and M. Kühl:

Etching of multimode optical glass fibers: A new method for shaping the measuring tip

and immobilization of indicator dyes in recessed fiber-optic microprobes

Sensors and Actuators B 211: 462–468 (2015)

DOI: http://dx.doi.org/10.1016/j.snb.2015.01.091

Etfi

La

b

a

ARRAA

KOTEIOM

1

sdatcdivooo(tdc

ah

h0

Sensors and Actuators B 211 (2015) 462–468

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

jo u r nal homep age: www.elsev ier .com/ locate /snb

tching of multimode optical glass fibers: A new method for shapinghe measuring tip and immobilization of indicator dyes in recessedber-optic microprobes

ars Fledelius Rickelta,∗, Lars D.M. Ottosenb, Michael Kühla

Marine Biological Section, Department of Biology, University of Copenhagen, Strandpromenaden 5, DK-3000 Helsingør, DenmarkBiological and Chemical Engineering Section, Department of Engineering, Aarhus University, Blichers Allé 20, DK-8830 Tjele, Denmark

r t i c l e i n f o

rticle history:eceived 31 October 2014eceived in revised form 7 January 2015ccepted 23 January 2015vailable online 31 January 2015

a b s t r a c t

We describe a new procedure for making recessed tips on multimode optical glass fibers. The method isbased on etching fiber tips in 40% hydrofluoric acid for defined immersion times. As the etching velocitydecreases radially from the core center in multimode graded index fibers, a recess can be formed in thetip of flat-cut tapered or untapered fibers. Etched fiber tips showed improved focussing of excitation lightcoupled into the fiber at the opposite end, and very efficient excitation of thin layers of optical indicatorsimmobilized into the recess. The sensor chemistry is well protected when immobilized in recessed fiber

eywords:ptical fiberaperingtchingndicator immobilizationxygen microoptode

tips and allows the construction of O2 microoptodes with improved mechanical stability that can measurerepeatedly even in very cohesive biofilms, tissue and dry soil.

© 2015 Elsevier B.V. All rights reserved.

echanical stability

. Introduction

Fiber-optic chemical microsensors (microoptodes) allow mea-urements at high spatio-temporal resolution and have beeneveloped for various analytes [1,2]. Such microsensors measure

chemical (e.g. O2, pH, CO2, salinity) or physical (e.g. tempera-ure, refractive index) variable via an analyte-dependent reversiblehange in the optical properties of an indicator, which is embed-ed in a polymer matrix immobilized onto the fiber tip. The

ndicator chemistry has mostly been applied to the fiber tipia dip coating or by mechanical deposition of a small dropletnto the end of the fiber tip. The first microoptodes were devel-ped for microscale measurements of O2 [3] and were basedn the dye, ruthenium(II)-tris-4,7-diphenyl-1,10-phenanthrolineRu(dpp)3) immobilized in polystyrene, but several other combina-ions of O2 sensitive dyes and immobilization matrices have beenescribed in recent years [4–9], and microoptodes are commer-ially available (www.pyro-science.com; www.presens.com).

Although the tip configuration of fiber-optic microsensors playsn important role for their performance, not much attentionas been given to improve the design of the measuring tip and

∗ Corresponding author. Tel.: +45 3532 1954; fax: +45 35321951.E-mail address: [email protected] (L.F. Rickelt).

ttp://dx.doi.org/10.1016/j.snb.2015.01.091925-4005/© 2015 Elsevier B.V. All rights reserved.

the immobilization of the indicator with respect to improvedmechanical and optical properties. Various fiber taper geome-tries and their influence on the performance of e.g. biosensorsand lensed fibers [10–13] have mainly involved use of singlemode fibers, and it was shown that tapered fibers have a supe-rior performance in collecting and transmitting light as comparedto untapered fibers [11,12]. Furthermore, it was shown that fibertips with relatively steep and conical tapers collect/focus lightmore efficiently than fiber tips with long and slender tapers[14].

Tapering of optical glass fibers can be done either by etchingthe fiber tip in hydrofluoric acid (HF) [11,13,15,16] or by pullingthe fiber in an IR laser-beam, in an electric arc [17] or in a smallflame from a micro torch (e.g. [2,18]). A constant tension duringthe melting process can be kept by a capillary puller [4,12] orby the force of gravity (as described here). The size of the flame,the pulling strength, and the timing all influence the final taperdimensions. While most work on chemical etching of opticalfibers has been done on single mode fibers, we found that thecladding of fused silica multimode graded-index optical fibers ismore resistant to hydrofluoric acid than the core and, therefore, a
concave recess can be etched into the tip. In this study, we describea simple method for etching recesses in tapered and untaperedmultimode optical fibers, we describe the optical performanceof such etched fibers and explore whether immobilization of an
dx.doi.org/10.1016/j.snb.2015.01.091

http://www.sciencedirect.com/science/journal/09254005

http://www.elsevier.com/locate/snb

http://crossmark.crossref.org/dialog/?doi=10.1016/j.snb.2015.01.091&domain=pdf

http://www.pyro-science.com;/

http://www.presens.com/

mailto:[email protected]

dx.doi.org/10.1016/j.snb.2015.01.091

d Actuators B 211 (2015) 462–468 463

oi

2

2

wsmioionnPwsto

flRcktuaFwJa

2

h

dpaiwmmtwet

wpfioa

dtetfitcf

Fig. 1. Schematic diagram of the setup for etching optical fiber tips with hydrofluoric

L.F. Rickelt et al. / Sensors an

ptical O2 indicator in the recess yields O2 microprobes withmproved mechanical stability.

. Materials and methods

.1. Fabrication of tapered fiber tips

We used fused-silica multimode graded index optical fibersith a 100/140 �m core/cladding diameter ratio. A 5 m long single

trand optical fiber patchcord (Radiall Fiber-Optic GmbH, Röder-ark, Germany) with a standard ST-connector at each end was cut

n two. The protective PVC coating and Kevlar fibers were removedver a length of 5–10 cm, and the Tefzel® polymer jacket enclos-ng the fiber was removed mechanically over several cm’s by usef a fiber stripper (Micro-Strip®, Thomas & Betts, Memphis, Ten-essee). For better handling, the fiber was fixed in a hypodermiceedle mounted on a syringe [2,18] or, alternatively, in a taperedasteur pipette. The fiber was secured with epoxy resin in such aay, that the exposed fiber was free of the needle or pipette tip. The

yringe or the pipette was mounted vertically in a micromanipula-or (MM33, Märtzhäuser, Wetzlar, Germany) with a small weightf 3.75 g attached to the bare fiber end.

A taper was made by heating the fiber with a small O2/propaneame from a miniature brazing and welding set (Roxy-Kit®,othenberger, Frankfurt a. M., Germany). Thereafter, the taper wasut back manually under a dissection microscope with a ceramicnife and a sharpened forceps to the desired diameter of the taperedip. The length of the taper and the tip diameter were measuredsing a calibrated compound microscope. Typical taper lengthsnd tip diameters were 300–800 �m and 20–40 �m, respectively.inally, the tip was cleaned in hexane. Untapered fibers were cutith an optical fiber cleaving tool (Thomas & Betts, Raritan, New

ersey, USA) to obtain a straight and flat-cut fiber tip before etchingnd subsequent rinsing.

.2. Etching of fiber tips

A recess in the fiber tip was made by etching a cavity with 40%ydrofluoric acid as follows:

A small volume (0.1 ml) of the HF was placed in an Eppen-orf tube and carefully covered with 1 ml paraffin oil (Fig. 1). Thearaffin oil prevented HF evaporation, the formation of aerosols,nd removed adherent HF from the fiber tip when withdrawingt from the etching bath. The fiber was mounted vertically and

as introduced into the etching bath with a computer-controlledotorized micromanipulator (Unisense A/S, Denmark). The micro-anipulator software (Profix, Unisense A/S, Denmark) controlled

he time the tip was immersed in the HF and the velocity withhich the fiber was withdrawn from the etching solution. After

tching, the fiber tip was cleaned by successive immersion in des-illed water, acetone (99%), and xylene (98%).

For material etching rate experiments, only untapered fibersith straight and flat cut tips were used. Several 2–3 cm long fiberieces were made from the same fiber cable and each piece wasxed with plasticine on the tip of a glass Pasteur pipette. The effectf etching on the fiber dimensions was observed and measured on

calibrated optical microscope.For untapered fibers, the dimensions of the recess only

epended on the time the tip was immersed in the HF, and theotal depth of the recess could therefore be calculated from thetching rate. The actual recess depth was confirmed by observa-ion of etched tips on a calibrated optical microscope. For tapered
ber tips, the shape of the recess also depended on the tip diame-er and geometry, due to differences in the relative thickness of theladding and core material in the tapered region after pulling. Thusor very thin and long tapers, the etching process became more
acid. The same setup was used for testing the mechanical stability of O2 microop-todes. For this, the Eppendorf tube was replaced with a glass beaker containing thetest media and a microoptode was connected to a fiber-optic O2 meter.

undefined, but a central cavity was always formed in the fiber tipduring etching for <15 min. By combining the etching procedurewith sealing off parts of the fiber tip with polystyrene, it was alsopossible to create different shaped tips, e.g. conical tips.

2.3. Characterization of recessed fibers

The light emission from bare fiber tips was investigated underan optical microscope. For this, the fibers were coupled to eithera fiber-optic fluorometer [19] or a fiber-optic O2 meter (MICROX1, Presense GmbH, Regensburg, Germany) from which light from ablue LED was coupled into the optical fiber. The light emitting fibertip was placed into a flat glass capillary (internal dimensions 8 by0.8 by 40 mm; VitroCom Inc., Mt.Lks., NJ, USA) filled with dilutedmilk. The milky suspension enabled visualization of the emittedlight field from the fiber tip via scattering. The milky solution wasreplaced by an aqueous solution of ruthenium(II) tris(4,7-diphenyl-1,10-phenanthroline 4′,4′′-disulfonic acid) dichloride, i.e., a water-soluble O2 indicator. The indicator was synthesized according to Linet al. [20] from potassium penta-chloro-aquoruthenate(III), whichwas changed from RuCl3 (Fluka Chemie, Buchs, Switzerland) [21],and 4,7-diphenyl-1,10-phenanthroline 4′,4′′-disulfonic acid (FlukaChemie, Buchs, Switzerland). The emitted light field was monitored
via the induced luminescence of the indicator around the fiber tip.Photographs of the fiber tips and the emitted light field were takenin a dark room with a Leica camera equipped with a 42 cm bellowsand a light sensitive film Fujichrome Provia Daylight 400 F, RHP III

4 d Actu

1a

2

t(mcstottv

swtasfiwe

F2

64 L.F. Rickelt et al. / Sensors an

35 (Fuji Photo Film Co., Ltd., Tokyo, Japan) using a fixed aperturend an exposure time of 30 s.

.4. Immobilization of sensor chemistry in recessed fibers

An O2 sensitive indicator was immobilized to the fiberips as a filtered polymer solution of 4% (w/v) polystyreneGoodfellow, Cambridge, UK) in chloroform with 5 mmol Pt(II)

eso-tetra(pentafluorophenyl)porphine per kg polymer. The indi-ator/polymer mixture was applied to the fiber tip with a smallpatula under a dissection microscope. The spatula was dipped intohe polymer solution and was moved to the fiber tip until the dropn the spatula touched it. A small fraction of the drop adhered tohe tip. It was necessary to wait a few seconds for letting some ofhe CHCl3 evaporate to make the indicator/polymer mixture moreiscous and adhesive, while touching the fiber tip.

For fiber tips with a deep recess, it was difficult to get the sen-or solution into the bottom of the recess without enclosure of air,hen CHCl3 was used as a solvent. To prolong the evaporating of

he solvent it was thus necessary to use a less volatile solvent suchs 1,1,2-trichloroethane. After the spatula was dipped into the sen-
or solution, the drop on the spatula was moved until it touched theber tip and some of the solution went into the recess. The spatulaas removed awaiting the air in the bottom of the recess to pen-
trate to the surface of the solution. The recess was then refilled

ig. 2. Photographs of etched fiber tips. A flat-cut optical fiber tip etched in HF for40 s (A). A tapered optical fiber tip etched for 90 s (B).

ators B 211 (2015) 462–468

with the polymer solution. To avoid detachment of deposited lay-ers, this procedure was done repeatedly until a small meniscusof the polymer mixture just protruded out of the recess after thesolvent evaporated.

2.5. Characterization of microoptodes

Microoptodes were connected to a fiber-optic O2 meter (Microx3, Presens GmbH, Regensburg, Germany) for characterization. Fortwo sets of straight cut sensors (8 without recess and 8 with∼25 �m recess), the O2 dependent phase angle and the fluores-cence intensity (amplitude) were measured in air-saturated water,and in an aqueous solution of 1% sodium sulfite (zero oxygen). Theresponse time was measured as the time before the signal reached95% of the full response when the optode was rapidly moved fromair-saturated water to the sodium sulphite solution.

2.6. Mechanical stability of microoptodes

Recessed optodes were tested for mechanical stability mea-suring O2 concentration profiles in different media. The sensorswere mounted in the micromanipulator and connected to the O2meter. After testing, the fiber tips were examined under an optical
microscope. Two sets of sensors with tapered tips were tested: 12sensors without recess and 11 sensors with recess. A sensor wasplaced in the micromanipulator and four or more profiles weredone in a 2% agarose gel. The agar was then substituted with a very
0 50 0 100 0 150 0 20 000

10

20

30

40

50

60

70

80

0 20 0 400 600 800 100 0 120 0 14000

20

40

60

80

100

120

140

160

180

radi

us (µ

m)

A

core

cladd ing

B

rem

oved

fibe

r (µm

)

etching t ime (second s)

recess depth 22oC removed cladding 22oC removed cladding 23oC

Fig. 3. Fiber radius of untapered optical fibers as a function of the etching time at22 ◦C and 23 ◦C. The position of the core and cladding is indicated (A). The depthof the recess (22 ◦C) and the amount of cladding material removed as a function ofetching time (22 ◦C and 23 ◦C) (B).

d Actu

ds

tsea

3

3

sa

zero in the micromanipulator software. It was not possible to mea-

Fpt


ense and cohesive bacterial biofilm, i.e., a microbial mat from aolar saltern [22].

In addition to these short-term tests of mechanical stability inhe laboratory, the recessed sensors were also tested for long termtability in soil. Recessed sensors were applied in the soil over anxtended period of 12 days to measure the development of anoxiand the reintroduction of O2 following liquid manure injection.

. Results and discussion

.1. Etching rates

Examples of an etched straight cut fiber and a tapered fiber arehown in Fig. 2. For etching times <10 min, a central cavity waslways formed in the fiber tip. The depth of this recess at 22 ◦C

ig. 4. Images of the light emission from different types of optical fiber tips. The light soanels A, C, E, G) and in a solution of the water soluble O2 indicator Ru(dpp(SO3Na)2)3 (rigips were etched and an additional set of pictures was taken (C, D and G, H).

ators B 211 (2015) 462–468 465

and the fiber radius at 22 ◦C and 23 ◦C were determined as a func-tion of etching time (Fig. 3). While the cladding (70–50 �m) wasetched with a constant velocity, the etching rate increased throughthe core with the highest etching rate in the center of the core.As the etching rates were constant through both the cladding andin the center of the core glass material, they could be calculatedfrom the slopes of the two lines. At 22 ◦C, the etching rates for thecladding and the center of the core glass material were found to be∼0.014 �m s−1 and ∼0.28 �m s−1, respectively (Fig. 3B). The ethingrate for the cladding was found to be ∼0.016 �m s−1 at 23 ◦C. Themeasurement at 0 s was performed by setting the etching time to

sure any change in the diameter of the fiber, but a small recess∼3 �m was etched at the tip. As the etching rate was highest in thecenter of the core, a conical parabolic shaped cavity was formed.

urce was a blue LED. The fiber tips were inserted in a dilute milk suspension (leftht panels B, D, F, H). After pictures were taken of the flat-cut tips (A, B and E, F), the

466 L.F. Rickelt et al. / Sensors and Actuators B 211 (2015) 462–468

Table 1Comparison of response time and signal amplitude of flat cut fiber O2 optodes with and without recess. Numbers indicate means ± standard deviation (n = 8).

Response time 100–5% air saturation Amplitude 0% (anoxic) a.u. Amplitude 100% (air saturated) a.u.

Straight cut sensors without recess 29.3 ± 8.8 s 17,715 ± 6388 7609 ± 248215,100 ± 7897 7331 ± 4090

3

scctawu

DcLne

3

wt1lTf

3

tTcww

ssw(Sms3>auido

3

rrap

0 5 10 15 20 25 30 35 40 45 50 55 60 65

0

50

100

% a

irsat

urat

ion

tim e (secon ds)

whitout recess recess

Fig. 5. Response time curves for 16 flat cut sensors; 8 without recess (red) and 8with a ∼25 �m deep recess (blue). Each fiber was moved from air saturated water to

pure SiO2 or it can be doped with B2O3 or F; both will lower therefractive index [23–25]. Annealed SiO2 doped with B2O3 shows alower HF etch rate than pure glass [26].

0 50 10 0 15 0 20 0 25 00

10

20

30

40

50

60

70

time ( hours)

O2 (

% a

ir sa

tura

tion)

0

5000

10000

15000

20000

nor

mal

ized

am

plitu

de (a

.u.)

Straight cut sensors with ∼25 �m recess 11.7 ± 4.7 s

.2. Light emission

Light emission from a flat cut untapered tip (Fig. 4A and B)howed strongest light closest to the fiber surface within a nearlyylinder-shaped beam with a diameter corresponding to the fiberore over a distance of 1–2 times the fiber diameter. From a taperedip without recess (Fig. 4E and F), the emitted beam was broad, butlso more concentrated close to the 37 �m wide tip. This agreesith the fact that tapered optodes produce a stronger signal thanntapered, due to focussing of light in the tapered region [14].

Untapered fiber tips with a recess depth of 50 �m (Fig. 4C and) showed that the recess apparently acts as a parabolic reflectoroncentrating the light beam in the recess before it is spread out.ight emitted from tapered tips with a recess also showed a pro-ounced focusing of the light within the recess leading to enhancedxcitation of the O2 indicator (Fig. 4G and H).

.3. Response time

Response time signal curves of fiber-optic O2 sensors with andithout recess are shown in Fig. 5. Average response times for the

wo sensor sets were calculated to 29.3 ± 8.8 s (without recess) and1.7 ± 4.7 s (with recess) with no significant difference between the

uminescence amplitude between the two sensor types (Table 1).he signal changes were fully reversible and no hysteresis wasound.

.4. Mechanical stability

All sensors survived the agar test. In the cohesive mat, malfunc-ion occurred when the sensor tip was pulled back from the mat.he sensors without recess lost their entire signal and the sensorhemistry was completely torn off without damaging the tip itself,hereas recessed sensors still showed good signals albeit someere a bit damaged at the edge of the recess.

It is normally not possible to avoid some mechanical stress toensor tips during prolonged insertion in soils, primarily due tohrinking or expansion of the soil as a consequence of changingater contents. Consequently, O2 recordings with normal optodes

without recess) in similar experiments have hitherto often failed.ince extended deployment in wet environments (in this case aix of soil and liquid manure) can result in a softening of the sen-

or coating, it is easily lost and the experiment must be aborted. In experiments however, recessed sensors maintained signal over10 days in soil. A plot of the O2 measurement together with themplitude of the luminescence signal normalized to the amplitudender anoxic conditions is shown in Fig. 6. The experiment was

nterrupted after 12 days. The recessed sensor was still in good con-ition and the normalized amplitude plot indicates no detachmentf the sensor chemistry.

.5. Refractive index and etching rate correlation

The etching of multimode fibers in HF showed an apparent cor-
elation between refractive index of the glass material and etchingate. The core was etched at faster rates relative to the cladding,nd a parabolic recess was formed. The core refractive indexrofile in multimode graded-index fibers is parabolic with the
anoxic water (1% Na2SO3) at time 0 s. (For interpretation of the references to colourin this figure legend, the reader is referred to the web version of the article.)

index decreasing from the center of the core to the core–claddinginterface, while the refractive index in the cladding is constant[23]. The concave etched recess is consistent with the use of GeO2in multimode optical fibers for variation of the refractive index inthe core with decreasing concentration of GeO2 from the centerto the core resulting in a gradual increase in the refractive index.The HF etch rate is found to exhibit a monotonic dependency ofthe germanium concentration in SiO2 [24] and the etch rate thusincreased with increasing GeO2 contents. The cladding is usually

Fig. 6. Development of anoxia and reintroduction of O2 to an agricultural soil fol-lowing injection of liquid manure as measured with a recessed etched microoptode.Curves show the calibrated O2 measurments (−) as well as the luminescence ampli-tude signal normalized to the amplitude under anoxic conditions (−).

d Actu

dd

S

ccbie

4

tiflimnrtrasoflmctIwett

omlppasfi

A

ftdt

R

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

[

[


These effects of dopants on etching rate are well described. Theissolution of vitreous SiO2 into an aqueous HF solution can beescribed by the simplified overall reaction:

iO2 + 6HF → H2SiF6 + 2H2O

Vitreous SiO2 consist of tetragonal Si units connected at all fourorners covalently with silioxane bonds. For a specific HF con-entration the rate-determining step for dissolution of SiO2 is thereakage of the siloxane bond. The breakage of the equivalent bond

n the presence of Ge is faster and more GeO2 thus means fastertching-rate [23–26].

. Conclusions

Recessed fibers showed a stronger focusing of light at the fiberip in comparison to normal flat cut fibers. This has importantmplications for manufacturing fiber-optic microsensors, whereuorescent indicator dyes are immobilized onto the tip of fibers,

.e., microoptodes. The focussing of excitation light in the recesseant that it was possible to manufacture sensors with thin-

er layers of fluorescent indicator chemistry and therefore fasteresponse times. Another important feature for microoptodes isheir mechanical stability. In cohesive biofilms both flat-cut andecessed sensors lasted longer when having a larger tip diameternd therefore more chemistry attached to the tip. However, theensing layer was considerably thinner for the recessed sensorsbtaining the same signal and better mechanical stability as for theat-cut sensors with the same tip diameter. When used in cohesiveaterials, the sensor chemistry was easily dragged off when dip-

oated sensors were withdrawn from the measuring object, whilehe sensor material was better protected inside recessed sensors.mmobilizing the dye inside a recess thus yielded microoptodes

ith a better mechanical stability and faster response times. Tiptching and immobilization of indicator material in recessed fibersherefore represents an important improvement in the construc-ion of the microoptodes.

Recessed fiber tips may also allow easier construction ofther types of optical microsensors such as fiber-optic irradianceicroprobes for quantifying light intensity at high spatial reso-

ution. Such probes currently require a complex manufacturingrocedure, where a miniature disk of a TiO2–methacrylate com-ound is fixed to the fiber tip and polished [27,28] and such probeslso exhibit a limited mechanical stability when profiling in cohe-ive media. Immobilization of the scattering matrix into a recessedber tip may resolve these limitations.

cknowledgements

This study was financed by grants from the Danish Councilor Independent Research Natural Sciences, (1323-00065B) andhe Danish Council for Independent Research Technology and Pro-uction Sciences (0602-00618B). Birgit Thorell and Anni Glud arehanked for excellent technical assistance.

eferences

[1] G. Holst, I. Klimant, M. Kühl, O. Kohls, Optical microsensors and microprobes,in: M.S. Varney (Ed.), Ocean Science and Technology, vol. 1, Chemical Sensorsin Oceanography, Gordon and Breach, Amsterdam, 2000, pp. 143–188.

[2] M. Kühl, Optical microsensors for analysis of microbial communities, in: J.R.Leadbetter (Ed.), Methods in Enzymology, vol. 397, Elsevier, Amsterdam, 2005,pp. 166–199 (Chapter 10).

[3] I. Klimant, V. Meyer, M. Kühl, Fiber-optic microsensors, a new tool in aquaticbiology, Limnol. Oceanogr. 40 (6) (1995) 1159–1165.

[4] I. Klimant, F. Ruckruh, G. Liebsch, A. Stangelmayer, O.S. Wolfbeis, Fast responseoxygen micro-optodes based on novel soluble ormosil glasses, Mikrochim. Acta131 (1999) 35–46.

ators B 211 (2015) 462–468 467

[5] I. Klimant, M. Kühl, R.N. Glud, G. Holst, Optical measurement of oxygen and tem-perature in microscale: strategies and biological applications, Sens. ActuatorsB: Chem. 38 (1997) 29–37.

[6] A.M. Morin, W. Xu, J.N. Demas, B.A. DeGraff, Oxygen sensors based on quenchingof tris-(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) in fluorinated poly-mers, J. Fluoresc. 10 (2000) 5–12.

[7] Y. Amao, T. Miyashita, T. Okura, Novel optical sensing material: platinumoctaethylporphyrin immobilized in a copolymer film of isobutyl methacrylateand tetrafluoropropyl methacrylate, React. Funct. Polym. 47 (2001) 49–54.

[8] X.-D. Wang, O.S. Wolfbeis, Optical methods for sensing an imaging oxy-gen: materials, spectroscopies and applications, Chem. Soc. Rev. 43 (2014)3666–3761.

[9] C. Ast, E. Schmälzlin, H.-G. Löhmannaröben, J.T. van Dongen, Optical oxygenmicro- and nanosensors for plant applications, Sensors 12 (2012) 7015–7032.

10] J.P. Golden, G.P. Anderson, S.Y. Rabbany, F.S. Ligler, An evanescent wavebiosensor-Part II: fluorescent signal acquisition from tapered fiber optic probes,IEEE Trans. Biomed. Eng. 41 (1994) 585–591.

11] G.P. Anderson, J.P. Golden, F.S. Ligler, A fiber optic biosensor: combinationtapered fibers designed for improved signal acquisition, Biosens. Bioelectr. 8(1993) 249–256.

12] G.A. Valaskovic, M. Holton, G.H. Morrison, Parameter control, characterization,and optimization in the fabrication of the optical fiber near-field probes, Appl.Opt. 7 (1995) 1215–1228.

13] H.H. Gao, Z. Chen, J. Kumar, S.K. Tripathy, D.L. Kaplan, Tapered fiber tips for fiberoptic biosensors, Opt. Eng. 34 (1995) 3465–3470.

14] O. Kohls, G. Holst, M. Kühl, Micro-optodes: the role of fibre tip geometry forsensor performance, SPIE Proc. 3483 (1998) 106–108.

15] D.R. Turner, Etch procedure for optical fibers, US Patent 4,469,554 (1984).16] T. Alder, A. Stöhr, R. Heinzelmann, D. Jäger, High-efficiency fiber-to-chip cou-

pling using low-loss tapered single-mode fiber, IEEE Photon. Technol. Lett. 12(2000) 1016–1018.

17] B. Grunwald, G. Holst, Fibre optic refractive index microsensor based on white-light SPR excitation, Sens. Actuators B: Chem. 113 (2004) 174–180.

18] M. Kühl, B.B. Jørgensen, Spectral light measurement in microbentic pho-totrophic communities with a fiber-optic microprobe coupled to a sensitivediode array detector, Limnol. Oceanogr. 37 (8) (1992) 1813–1823.

19] R. Thar, M. Kühl, G. Holst, Fiber-optic fluorometer for microscale mapping ofphotosynthetic pigments in microbial communities, Appl. Environ. Microbiol.67 (6) (2001) 2823–2828.

20] C.-T. Lin, W. Böttcher, M. Chou, C. Creutz, M. Sutin, Mechanism of the quenchingof the emission of substituted polypyridineruthenium (II) complexes by iron(III), chronium (III), and europium (III) ions, J. Am. Chem. Soc. 98 (21) (1976)6536–6544.

21] E.E. Mercer, R.R. Buckley, Hexaaquoruthenium (II), Inorg. Chem. 4 (12) (1963)1692–1695.

22] A. Wieland, J. Zopfi, M. Benthien, M. Kühl, Biogeochemistry of an iron-richhypersaline microbial mat (Camargue France): oxygen, sulfur and carboncycling, Microb. Ecol. 49 (2005) 34–49.

23] P. Pace, S.T. Huntington, K. Lyvtikänen, A. Roberts, J.D. Love, Refractive indexprofiles of Ge-doped fibers with nanometer spatial resolution using atomicforce microscopy, Opt. Express 12 (7) (2004) 1452–1457.

24] S. Neffe, M. Szustakowski, Charicterization of the microstructure of opticalfibers, Proc. SPIE 670 (1986) 100–103.

25] G.A.C.M. Spierings, Wet chemical etching of silicate glasses in hydrofluoric acidbased solutions, J. Mater. Sci. 28 (1993) 6263–6273.

26] A.S. Tenney, M. Ghezzo, Etch rates of doped oxides in solutions of buffered HF,J. Electrochem. Soc. 120 (1973) 1091–1095.

27] M. Kühl, C. Lassen, B.B. Jørgensen, Light penetration and light intensity in sandysediments measured with irradiance and scalar irradiance fiber-optic micro-probes, Mar. Ecol. Prog. Ser. 105 (1994) 139–148.

28] C. Lassen, B.B. Jørgensen, A fiber optic irradiance microsensor (cosine collector):application for in situ measurements of absorption coefficients in sedimentsand microbial mats, FEMS Microbiol. Ecol. 15 (1994) 321–336.

Biographies

Lars Fledelius Rickelt was born on December 8, 1954. He received his M.Sc. inchemical engineering (1983) at the Technical University of Denmark, where hedid research in natural products chemistry at the Institute of Organic Chemistry(1988–1993). After employment in the industry, he became a member of a diabetesresearch group at the Institute of Medical Physiology, University of Copenhagen(1996–1999). Since 2000, he is a member of the Microenviromental Ecology researchgroup at the Marine Biological Section, Department of Biology, University of Copen-hagen (Denmark), where he is developing fiber-optic microsensors and advancedimaging techniques for environmental analysis.

Lars D.M. Ottesen was born in April 1971. He received a PhD in Microbial Ecologyin 2000. Following a degree in Economics, also in 2000, he started worked in the
industry, with strong focus on research and development. In 2007–2009 he workedas assistant professor at the microbiology department at AU before returning toindustry until 2013, where he became associate professor and head of the biologicaland chemical engineering department at AU Engineering. Lars D.M. Ottosens work,both in industry and academia, has focused on applied microbiology and chemistry.
http://refhub.elsevier.com/S0925-4005(15)00117-3/sbref0145


















































































































































































































































































































































































































































































































































































































































































































































4 d Actu

Ip

MaFte

68 L.F. Rickelt et al. / Sensors an

n addition to industrial R&D insight, he has more than 20 scientific papers andatents.

ichael Kühl was born on June 16, 1964. He received his M.Sc. in biology (1988)nd Ph.D. in microbiology (1992) from the University of Aarhus, Aarhus (Denmark).rom 1992–1998 he established and headed the microsensor research group athe Max-Planck-Institute for Marine Microbiology, Bremen (Germany) developinglectrochemical and fiber-optic microsensors and advanced imaging techniques for

ators B 211 (2015) 462–468

environmental analysis. Since 1998 he has continued this research at the Marine Bio-logical Section, Department of Biology, University of Copenhagen (Denmark) wherehe is full professor in microbial ecology and heads the Microenvironmental Ecol-
ogy research group. He is also adjunct professor at the University of TechnologySydney, Australia and a visiting professor at the Nanyang Technological Univer-sity, Singapore. He is a member of the Royal Danish Academy of Sciences and Letters,associate editor of Marine Biology, Aquatic Microbiology, Environmental Biology, andFaculty of 1000/Environmental Microbiology.

3. An optode sensor array

L.F. Rickelt, L. Askaer, E. Walpersdorf, B. Elberling, R.N. Glud, and M. Kühl:

An optode sensor array for long term in situ measurements of O2 in soil and sediment

Journal of Environmental Quality 42: 1267-1273 (2013) + Errata 43:1–1 (2014)

DOI: 10.2134/jeq2012.0334 + DOI:10.2134/jeq2010.0163er

TECHNICAL REPORTS

1267

Long-term measurements of molecular oxygen (O2) dynamics in wetlands are highly relevant for understanding the effects of water level changes on net greenhouse gas budgets in these ecosystems. However, such measurements have been limited due to a lack of suitable measuring equipment. We constructed an O2 optode sensor array for long-term in situ measurements in soil and sediment. The new device consists of a 1.3-m-long, cylindrical, spear-shaped rod equipped with 10 sensor spots along the shaft. Each spot contains a thermocouple fixed with a robust fiberoptic O2 optode made by immobilizing a layer of Pt(II) meso-tetra(pentafluorophenyl)porphine in polystyrene at the end of a 2-mm polymethyl methacrylate plastic fiber. Temperature and O2 optode readings are collected continuously by a data logger and a multichannel fiberoptic O2 meter. The construction and measuring characteristics of the sensor array system are presented along with a novel approach for temperature compensation of O2 optodes. During in situ application over several months in a peat bog, we used the new device to document pronounced variations in O2 distribution after marked shifts in water level. The measurements showed anoxic conditions below the water level but also diel variations in O2 concentrations in the upper layer presumably due to rhizospheric oxidation by the main vegetation Phalaris arundinacea. The new field instrument thus enables new and more detailed insights to the in situ O2 dynamics in wetlands.

An Optode Sensor Array for Long-Term In Situ Oxygen Measurements in Soil and Sediment

L. F. Rickelt,* L. Askaer, E. Walpersdorf, B. Elberling, R. N. Glud, and M. Kühl

Molecular oxygen (O2) is a key environmental parameter in most ecosystems, where its concentration and dynamics not only outline the activity and distri-

bution of aerobic processes but also provide a proxy for overall biogeochemical carbon fixation and mineralization (Glud, 2008). Measurements of O2 transport and dynamics in organic soils and sediments are central for estimating the consequences of spatio-temporal water level fluctuations because decomposition of sub-surface carbon pools and the resulting emissions of the greenhouse gases CO2, CH4, and N2O are strongly affected by the O2 avail-ability (Askaer et al., 2010; Jørgensen et al., 2012; Liengaard et al., 2013). For example, in peat soil, O2 is the primary factor control-ling subsurface CH4 dynamics (Megonigal et al., 2003). A deeper understanding of how O2 supply is related to water level regimes is therefore of primary interest when assessing CH4 emissions to the atmosphere, especially in relation to the possible consequences of hydrological changes in a global change perspective, be it sea level rise affecting shallow coastlines and estuaries, increased rain-fall intensity causing flooding events, or increased draught periods causing water level draw down (IPCC, 2007).

The water level is a primary control of subsurface O2 availability due to low O2 solubility and an approximately 104 times slower O2 diffusion in water as compared with air. Spatio-temporal changes in O2 concentration and dynamics are not well understood, and there is a need for a continuous high-resolution monitoring of O2 dynamics in situ. Numerous studies have investigated water table effects on CH4 emissions (e.g., Macdonald et al., 1998; Daulat and Clymo, 1998; Hargreaves and Fowler, 1998; Kettunen, 2003; Schäfer et al., 2012). Also, the complex interactions of subsurface O2 concentrations with CH4 production and consumption have been demonstrated in a laboratory study (Askaer et al., 2010). Besides a field study linking subsurface O2 and CH4 concentrations (Elberling et al., 2011), to our knowledge

Abbreviations: POF, polymethyl methacrylate optical fiber.

L.F. Rickelt and M. Kühl, Marine Biological Section, Dep. of Biology, Univ. of Copenhagen, Strandpromenaden 5, DK-3000 Helsingør, Denmark; L. Askaer, E. Walpersdorf, and B. Elberling, Dep. of Geosciences and Natural Resource Management, Univ. of Copenhagen, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark; B. Elberling, Center for Permafrost (CENPERM), Univ. of Copenhagen, Øster Voldgade 10, DK-1350 Copenhagen K., Denmark; R.N. Glud, Univ. of Southern Denmark & Nordic Center for Earth Evolution–NordCEE, Odense, Denmark, Scottish Association for Marine Science (SAMS), Marine Laboratory Dunstaffnage, Oban, Argyll PA37 1QA, Scotland, UK, and Greenland Climate Research Center, Greenland Institute of Natural Resources, Nuuk Greenland; M. Kühl, Plant Functional Biology and Climate Change Cluster, Univ. of Technology Sydney, P.O. Box 123, Ultimo Sydney NSW 2007, Australia, and Singapore Centre on Environmental Life Sciences Engineering, School of Biological Sciences, Nanyang Technological Univ., Singapore. Assigned to Associate Editor Søren O. Petersen.

Copyright © American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. 5585 Guilford Rd., Madison, WI 53711 USA. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. J. Environ. Qual. 42:1267–1273 (2013) doi:10.2134/jeq2012.0334 Received 2 Sept. 2012. *Corresponding author ([email protected]).

Journal of Environmental Quality SHORT COMMUNICATIONS

Published April 30, 2013

1268 Journal of Environmental Quality

continuous O2 measurements in situ have not been available to support the interpretation of CH4 emissions.

Optical oxygen sensors (i.e., O2 optodes) are well suited for environmental monitoring due to their mechanical robustness and good long-term stability of measuring signals (Kühl, 2005; Holst et al., 2000). Fiberoptic O2 sensor arrays have previously been described for microscale measurements of O2 distribution (Holst et al., 1997; Glud et al., 1999; Fischer and Koop-Jakobsen, 2012) and for measurements in soil and sediment (Hecht and Kölling, 2001; Kölling et al., 2002). These systems did not include temperature compensation. Here, we present a robust optical O2 sensor array enabling long-term in situ deployment and continuous monitoring of O2 penetration and temperature on a time scale of several months. The construction, measuring characteristics, and temperature compensation of the new instrument are described, together with an example of in situ application in a peat soil.

Materials and MethodsConstruction of the Sensor Array

The sensor array was built into a cylindrical spear (Fig. 1) made of polyethylene (1200 mm length, 50 mm outer diameter, and a wall thickness of 5 mm). Along the cylinder shaft, 10 holes (8 mm in diameter) were drilled at distances consistent with measuring depths of 2, 5, 10, 20, 30, 40, 50, 60, 70, and 80 cm when the spear is placed at an angle of 30° relative to a vertical position and with 200 mm of the spear above the sediment/soil surface. A 4-mm-deep countersink with a 20-mm-diameter was cut around each hole for the placement of a bolt to fix the sensors (see below). A hole (20 mm diameter) was drilled 50 mm from the top and used for joint sensor connection to a data logger. Each of the 10 combined O2 and temperature sensors was constructed and mounted in the following way.

Two holes (2.1 mm diameter) were drilled in each of 10 steel screw bolts (8 × 20 mm) lengthwise from the head. One hole was drilled through, and the other stopped 0.5 mm from the bottom. The bolts were then fixed with a sealant (Loctite, Henkel Norden) to a stainless steel bar (3 mm × 16 mm × 905 mm) with 10 8-mm holes corresponding to the holes in the cylinder shaft, and 10-mm silicone disks were placed on each bolt thread. The soft silicone made it possible to tighten the outer nuts flush with the bolt ends and the shaft surface when the system was assembled. The disks also function as spacer between the steel bar and the shaft.

A welded-tip thermocouple (Type T, PTFE Insulated, Z2-T-2M, Labfacility Ltd.) and the sensing end of a 5-m-long plastic polymethyl methacrylate optical fiber (POF) (step index, 2 mm diameter, with a polymethyl methacrylate core and a fluorinated polymer cladding; Laser Components GmbH) were mounted in each steel screw with two-component epoxy resin (Fig. 1). The thermocouple and fiber were held together by 3.2-mm heat-shrinkable tubing (Low Shrink Temperature polyolefin tubing; Farnell), leaving 3 cm free at both ends.

The steel bar was pushed into the shaft, and the bolts were tightly fixed with 5-mm nuts and polyurethane rubber (Sikaflex-11FC, Sika A/S) through the side holes of the spear. The detector ends of each thermocouple and fiber pair was pushed out through the 20-mm hole at the top of the sensor array shaft. After polishing, each POF fiber end was mounted in a SMA-connector (Laser Components GmbH) enabling connection to a fiberoptic O2 meter.

A disk of a transparent and O2–impermeable carrier foil coated with O2 sensor chemistry (diameter, 2.5 mm; thickness, 0.125 mm) (sMylar, Goodfellow Cambridge Ltd.) was glued with UV curing resin (Dymax) onto the measuring end of the POF at the outer surface of the steel screw. Gluing the sensor foils to the measuring end of the POF ensured a durable and efficient optical coupling of the sensor layer luminescence into the POF. The disk was prepared with a drop of a 3.3% (w/w) solution of an O2 sensor chemistry cocktail consisting of 25 mg Pt(II) meso-tetra(pentafluorophenyl)porphine (Frontier Scientific, Inc.), 1 g polystyrene (Goodfellow Cambridge Ltd.), and 0.5 g TiO2 in chloroform. After evaporation of the sensor cocktail solvent, each sensor spot was knife-coated by a thin layer of highly O2 permeable black silicone (Elastosil N189, Wacker-Chemie GmbH) to seal it flush with the shaft of the array. The spatial resolution of the O2 sensors is estimated to be approximately 2.5 mm (the fibers are 2 mm), and the spatial resolution of the temperature sensors is estimated to be approximately 10 mm (the steel bolts are 8 mm).

The lower-most end of the sensor array spear was closed with a 100-mm conical tip, and the top of the cylinder was closed with a 20-mm cylinder, both made of solid PVC and equipped with a 40 mm plug. The sensor array was filled with clean sand. In the field, all sensor signals were acquired by a data logger (CR1000, Campbell Scientific Ltd.). Thermocouples were connected directly to the data logger, whereas the O2 optodes were connected to a multichannel fiberoptic meter (OXY-10, Presens GmbH) interfaced to the data logger. The instruments were powered by

Fig. 1. Schematic drawing of the O2 and temperature sensor array spear and the connection of sensors to the data logger.

www.agronomy.org • www.crops.org • www.soils.org 1269

220V in the laboratory and during in situ field measurements. We evaluated the O2 optodes with respect to performance, temperature response, and response time. These measurements were done with the OXY-10 meter or with a single channel version with similar measuring characteristics (FIBOX-3, Presens GmbH).

Performance of the Oxygen OptodeOxygen optode measurements can be based on measuring

O2–dependent changes in the intensity or decay time (also called lifetime) of specific luminescent indicator dyes that are quenched by O2 (Kühl 2005). Luminescence lifetime-based measurements are more robust and less prone to optical artifacts than luminescence intensity-based O2 measurements (Holst et al., 1995, 2000). The fiberoptic O2 meter used in this study measures the luminescence lifetime with a phase-modulation technique (Holst et al., 1995). When the luminescent O2 indicator is excited with sinusoidally intensity-modulated light (at frequency, fmod), the lifetime of the indicator causes a time delay of the emitted red-shifted luminescence; this delay is called the “phase angle” (F) between the excitation and the emitted signal. The lifetime, t, is calculated from the measured phase angle as:

tan(F) = 2p ⋅ fmod ⋅ t [1]

As shown in earlier work, reviewed by Holst et al. (2000) and by Kühl (2005), the response of POF-based O2 optodes follows a modified Stern-Volmer relation that includes a nonquenchable fraction, a:

c c c

0 0 0 SV

tan( ) 1tan( ) 1

II K c

F t -a= = = +a

F t + × [2]

where F0, I0, and t0 are the phase angle, luminescence intensity, and lifetime, respectively, of the indicator in the absence of O2, and Fc, Ic, and tc are the phase angle, luminescence intensity, and lifetime of the indicator at a given O2 concentration, c. KSV is a characteristic quenching coefficient of the immobilized indicator. For a given mixture of indicator and matrix material, a is usually temperature independent and constant over the dynamic range of the optode (Kühl, 2005), and Eq. [2] accurately describes the nonlinear behavior of I or t vs. O2 content. The nonquenchable fraction, a, and KSV can be determined from measurements of luminescence intensity or lifetime under at least three defined O2 contents (e.g., 0, c1 = 100% air saturation, and c2 = 20% air saturation) (see details in Kühl [2005]). Once a has been determined, the calibration of the optical O2 sensor can be done by a simple two-point calibration (e.g., measurements at zero O2 and at a known O2 content), typically water at atmospheric saturation at experimental temperature and salinity.

We had originally placed the O2 sensing material directly on the end of the POF, which yielded a mechanically robust sensor as only 1 sensor out of 10 lost its signal over a 7-mo pilot deployment period. However, with this type of direct immobilization, the chloroform in the O2 sensor chemistry cocktail dissolved some of the POF material, causing a relatively high nonquenchable fraction (i.e., a = 0.415 ± 0.002 [mean ± SD]; n = 3). Instead, the O2 indicator layer was cast onto a transparent carrier foil, which could be attached onto the POF using UV-curing glue. This type of immobilization yielded a values typical of planar (2D) optodes (i.e., a ?0.11) (Kühl and

Polerecky, 2008), and it was simpler to manufacture such sensors reproducibly while retaining good mechanical robustness. This immobilization method was therefore used in the final design of the sensor array spear.

Calibration and Temperature Response of Oxygen Optodes

We quantified the temperature dependency of all 10 sensors in the array. The luminescence lifetime, and with it F and KSV, depend on temperature. Optode sensors show a linear decrease in tan(F) with increasing temperature, T, which in the relevant temperature interval (0–25°C) can be described by:

T 0tan( ) tan( ) d TF = F + × [3]

where d is the slope and tan(F)0 is the intercept of the curve with the y axis at 0°C, and (F)T is the phase angle at temperature T. Using Eq. [3], the temperature dependency of KSV can be described as:

SV11 0 1

0 0 0

1 1( ) 1

tan( )tan( )

K Tcd T

d T

ì üï ïï ïï ïï ïï ï-aï ï= - ×í ýé ùï ïF + ×ï ïê úï ï-aï ïê úF + ×ï ïë ûï ïî þ

[4]

The anoxic calibration and the corresponding measurement of temperature response of the optodes were performed in the following way. The spear was placed in a water-filled cylindrical container (length, 115 cm; diameter, 19 cm) in a temperature-controlled room at approximately 4°C. For removal of O2, 200 g sodium dithionite (technical grade; Struers) was dissolved to a final concentration of 0.75%. The solution was kept anoxic by continuous flushing with N2 gas (using a pumice stone) for 2 d, and then the system, including the N2 flushing, was transferred to a 21°C room. The sensors were immediately connected to the OXY-10 fiberoptic O2 meter, and the O2–dependent phase angles of each sensor were recorded every 15 s during the warm-up period. The water temperature was measured continuously with an Omnitherm Pt 100 Digital Thermometer (RS components). The experiment was interrupted when the water temperature reached 16.7°C due to depletion of the N2 supply.

Calibration and temperature responses in 100% air saturated water were determined in a similar way. The container was filled with water and flushed with ambient air. The system was left for 2 d to equilibrate. After a first measurement at ~4°C in the cold room, the system was moved to the 21°C room, while still flushing with air. The water temperature reached 20.5°C after 24 h, and the experiment was stopped.

In principle, increasing hydrostatic pressure in the calibration container could affect sensor readings due to the combined pressure effect on sensor materials and the O2 solubility in the water (Uchida et al., 2008). However, detailed measurements with a single optode done at 10-cm intervals throughout the well mixed calibration chamber did not reveal any difference in phase angle. Apparently, the difference in signal caused by hydrostatic pressure differences was below the sensitivity level of these sensors. Hydrostatic pressure effects in situ were negligible due to the shallow O2 penetration in the peat soil.


Response Time of Oxygen OptodesResponse time measurements of the sensor array were

performed at 20°C. A cylindrical container (length, 120 cm; diameter, 12 cm) equipped with a pumice stone at the bottom was used in this case. The container was filled with water and flushed with air to obtain atmospheric saturation, and sodium sulfite (Merck) was added to the water (final concentration, 1%) together with continuous flushing with N2 to induce anoxic conditions while continuously logging the O2–sensor response at 15-s intervals. Water temperature was monitored continuously. Separate measurements with a fast-responding micro-optode (Klimant et al., 1995) were included for reference.

In Situ ApplicationIn situ O2 measurements were initiated in November 2008 in

a peat soil within the wetland area Maglemosen located 20 km north of Copenhagen, Denmark (55°51¢ N; 12°32¢ E). The peat soil in this area is subject to fluctuations in water level within the upper meter but maintains moist surface conditions year-round. Water level increases are induced by precipitation events resulting in water percolation from above and lateral flow from elevated surroundings. Water level was measured at 10-min intervals with a barometer-compensated pressure transducer (PCR 1830 series, Druck, ThermX) ( Jørgensen et al., 2012). The transducer wire was attached to a horizontal bar, and the transducer head was submerged in a 2-m-long perforated plastic tube placed in a sand cast drill hole. The horizontal bar was mounted with 3-m-long stainless steel rods inserted into the underlying mineral soil to avoid potential errors caused by seasonal displacement of the surface after shrinkage and swelling of the peat soil.

The spear was inserted into a predrilled hole (48 mm diameter) and with the measuring spots pointing downward to reduce any effects of lateral preferential flow. The O2 content was measured continuously at 10-min intervals at all 10 depths. The O2, temperature, and water level measurements were logged with the CR1000 data logger. For illustration, we present soil O2 contents at the 10 depths during a 3-wk period in spring (12 May to 1 June 2009).

Results and DiscussionEvaluation of Optode Sensor Array Performance

Our measurements showed a linear change in tan(F) (and thus the luminescence lifetime) vs. increasing temperature T (in °C) in accordance with Eq. [3], both under anoxic (Fig. 2A) and air saturated conditions (Fig. 2B) and with similar slopes among 10 sensors (R2 > 0.99 [slope = −0.00424 ± 0.00016; n = 10] and R2 > 0.99 [slope = -0.00338 ± 0.00017; n = 10] for anoxic and air saturation, respectively). The variation of KSV with temperature was calculated by Eq. 4 from the temperature calibrations of the 10 sensors (Fig. 2C). Thus, it is possible with a good approximation to transform the two calibration values for each sensor done at a given temperature to calibration values at other temperatures using the slopes derived from the anoxic and the air-saturated system. Using such calculated calibration values, it becomes in principle possible to calculate the O2 content from phase angle data measured at temperatures of 0 to 25°C using Eq. [4]. Unpublished experimental results with similar sensors have shown that the relationship between tan(F) and temperature

is not completely linear from 0 to 80°C, although it is safe to assume linearity in 25°C intervals. The thermocouples were calibrated at 4 and 21°C ( Jørgensen et al., 2012).

Separate measurements with a fast-responding micro-optode showed that anoxic conditions had developed within 30 s. The average 90% response time of the O2 sensors of the array was 255 ± 58 s (Fig. 3; n = 10). This relatively slow operational response time was due to the use of a rather thick indicator layer combined with a relatively thick silicone coating conferring strong mechanical stability. Response times of a few seconds could be obtained by reducing the thickness of the indicator and the silicone coating, which would potentially compromise the mechanical stability. However, for the long-term in situ measurements reported here, where diel to seasonal changes of soil O2 levels were to be monitored, a response time of several minutes was adequate.

Monitoring Site CharacteristicsEarlier studies have reported marked spatial variability

in peat characteristics over the depth range, where the water table typically fluctuates (Askaer et al., 2010). These peat characteristics, in combination with plant-mediated subsurface O2 transport, regulate the actual net transport and depth distribution of oxygen in the root zone (Elberling et al., 2011; Askaer et al., 2011). Although the water level typically varied from 0 cm to several cm above the soil surface during winter to approximately 50 cm below in the driest summer period, the water level ranged from 12 to 41 cm below the soil surface in the

Fig. 2. Temperature response of 10 optical O2 sensors, quantified as the O2–dependent luminescence lifetime proportional tan(F) with sensors in anoxic water (A) and in water at 100% atmospheric saturation (B). Stern-Volmer quenching constant KSV (see Eq. [3]) vs. temperature for each optical O2 sensor (C).


selected 3-wk measurement period. During this time, the water level declined, although precipitation events during the period resulted in transient rises in water level (Fig. 4).

At the monitoring site, the organic material in the top 0.6 m of the peat soil has been deposited in a freshwater lake, turning the lake into a moist wetland (Askaer et al., 2010). Soil auger cores (0–120 cm) showed that peat depth was fairly uniform throughout the wetland independent of surface vegetation. Three contrasting layers were observed. The top approximately 25 cm consisted of recognizable plant parts and fresh plant litter. At 26 to 60 cm depth, the peat became increasingly humified and darker in color, indicative of silty detritus deposited in a fresh water environment. At approximately 60 cm depth, there was an abrupt change in the sediment from peat to carbonate-rich organic silt. This layer extended to approximately 80 cm below the surface, with a few thin dark layers of peat. Bulk density increased from 0.2 to 0.3 g cm-3 in the upper 10 cm to 0.4 g cm-3 at 40 to 50 cm depth, coinciding with a decrease in porosity and apparent diffusivity. This indicated decreasing hydraulic conductivity and increasing water retention with depth (Boelter, 1965), in particular below the main root zone (0–25 cm depth).

Oxygen Conditions in the Peat Soil ProfileBelow the active root depth of the dominant vegetation

type (Phalaris arundinaceae), O2 supply is limited by molecular diffusion under saturated conditions (Askaer et al., 2010, 2011; Jørgensen et al., 2012). In general, O2 distribution in the peat soil was strongly affected by water level. When the water level reached below a specific sensor depth, the local O2 content increased, and vice versa. The sensors at 50 cm depth and below showed anoxia (<1% air saturation) throughout the measuring period. At 40 cm depth, the O2 content was <1% air saturation until the water level dropped below this depth by the end of May, resulting in a maximum O2 concentration of 11% air saturation on 1 June (Fig. 4A, lower right hand corner). At 30 cm depth, the O2 content was constant at <1% air saturation until 22 May. Subsequently, it momentarily increased to 5% air saturation in response to a rapid water level decline reaching the sensor measuring depth in the horizon. On 25 May, the water level again dropped below the 30 cm sensor, resulting in a rapid increase in O2 availability in the range of 50 to 65% air saturation. On 29 May, the water level rose to 10 cm above the sensor for a 2-d period, resulting in a decrease in O2 content to <1% air saturation before rapidly rising to 55 to 65% air saturation. At 20 cm depth, the water level declined below the sensor on 15 May for a 2-d period, resulting in an increase in O2 content up to 43% air saturation, although it rapidly decreased to <1% air saturation when the water level increased. As the water level decreased for a longer period of time, around 20 May, the O2 content increased steadily until reaching stability around 28 May at 80% air saturation.

The water level did not increase above the O2 sensor at 10 cm depth. Still, an unexpected decrease in O2 content to 53% air saturation was observed on 18 May in response to precipitation events on 16 May (3.4 mm), 17 May (0.7 mm), and 18 May (1.0 mm) after a period without precipitation. It took roughly 5 d to re-establish stable O2 contents around 83% air saturation. This is consistent with the lack of reaction of the 20-cm sensor during 18 to 21 May, where the water level measurements showed a level below this sensor.

The O2 contents at the 2- and 5-cm depths were very similar. At these depths, the soil was continuously oxic, with O2 contents of 87

Fig. 3. The response time of the 10 calibrated fiberoptic minisensors used in the sensor array spear upon a shift from air saturated to anoxic water at 20°C.

Fig. 4. In situ measurements in a peat soil (Maglemosen, Denmark) during a 3-wk period in 2009. (A) Oxygen and water level variations. The dashed lines mark the depths of the respective O2 sensors in relation to the water level. (B) The measured temperature at the same depths.


to 105% (mean 93.9%) air saturation at 2 cm depth and 88 to 98% (mean, 91.9%) air saturation at 5 cm depth. The high O2 content was due to the fact that the water level did not reach above these probes during the monitoring period reported here. Apparently, O2 diffusion from the atmosphere was effective at replenishing microbial O2 consumption. These near-surface sensors exhibited diel O2 variations, with O2 contents above 100%. Rhizospheric oxidation by the dominating wetland plant Phalaris arundinacea could be a possible explanation for the diel variations in O2 content of the uppermost soil layers (Askaer et al., 2011; Jørgensen et al., 2012). We have obtained preliminary data on O2 dynamics in the Phalaris rhizosphere as monitored in 2D with planar optodes (Liengaard et al., unpublished data), but a more detailed discussion and comparison of the new spear system for O2 measurements with planar optodes is outside the scope of this paper.

The high water storage capacity of fresh peat and its ability to shrink and swell in accordance with available moisture minimize water level fluctuations and maintain the water table below (but close to) the surface (Ingram, 1983). Such shrinking and expansion can affect other parameters, such as soil bulk density, water retention, hydraulic conductivity, and, consequently, gas diffusion and O2 distribution (Elberling et al., 2011). However, we can only speculate on the importance of such effects on our field data because we did not monitor the shrinking and swelling of peat at the field site.

Variability of Oxygen DistributionThe O2 conditions in wetland soils under varying water

saturation levels are complex, and few studies have attempted to document O2 dynamics in such systems at high spatiotemporal resolution. A study of the effect of small-scale peat soil structure on O2 availability (Askaer et al., 2010) found vast soil heterogeneity and the presence of oxic zones below the water level in periods up to days after water saturation. This was presumably due to wetting inhibition combined with air inclusion, the formation of water-repelling film, and alterations of pore structure due to shrinkage during the drained period (Schwärzel et al., 2002). Anoxic zones were also observed above the water level (Askaer et al., 2010). These observations suggest a discontinuous soil pore system. During diel temperature variations, discontinuous pores could experience pressure changes potentially affecting the partial pressure and thus the measured O2 content. With the diel temperature variations of <3°C, this possible error is estimated to be insignificant.

Two-dimensional O2 imaging with planar O2 optodes in peat from the same wetland area has previously demonstrated a large spatial variation at the mm scale (92 ± 5% air saturation when fully waterlogged with oxygenated water) within the surface layers of the soil profile (Askaer et al., 2010). The sensor array described here measures exclusively in one point at each depth, and although pronounced cyclic variations were seen, the highest diurnal variations were within the same range as the spatial variation (e.g., 94.1 ± 5.6% air saturation on 13 May).

Temperature and Oxygen DynamicsFigure 5 shows detailed O2 and temperature readings for the

sensors at the 2-, 5-, and 10-cm depths over the first 3 d of the monitoring period. Soil temperature showed an increase from the morning until sunset and a subsequent decrease during the night, whereas the O2 content had a maximum shortly after

noon, when the photosynthetic activity was presumably highest, and then decreased until the next morning.

For the upper 10 cm, diel variations of the measured temperature and O2 levels did not coincide. The O2 levels showed a very regular and constant diel variation, reaching a maximum close to noon, which was 11 ± 4% higher (at 2 cm) than the minimum O2 level measured 0 to 3 h after midnight. The daily average temperature increased over the same 3-wk period, and the diel temperature variations in the top 2 cm were less regular, with a minimum between 07:00 and 10:00 and a maximum between 19:30 and 22:30 (Fig. 4). At 2 cm depth, the difference between the daily minimum and maximum temperatures was 1.4 ± 0.6°C. At 100% air saturation, a 1°C temperature difference typically gives a change in O2 optode signal equivalent to <2% air saturation. The variations of in situ temperature were thus far too small to have a significant effect on the O2 measuring signals even if the sensors were not temperature corrected. Because the sensors were temperature corrected and the temperature variations were out of phase with the observed O2 fluctuations, we rule out the possibility that the observed O2 fluctuations in the upper soil layers were due to temperature fluctuations.

Methodological IssuesThe O2 optode sensor array has proven to have remarkable long-

term stability. A spear array was inserted in the peat soil in November 2008 and was retrieved in December 2012; the 3-wk in situ data set reported here was obtained from the same spear deployment. From November 2008 to May 2009, the water level was at ground level, resulting in measurements showing anoxia at all 10 sensor depths. In December 2012, the spear array was removed from the measuring site. The retrieved sensor array was brought back to the laboratory, washed with water, and put in the water-filled calibration container. After more than 4 yr of deployment, 6 out of 10 sensors were still fully functional. The sensors at 20, 40, 50, 60, 70, and 80 cm were intact except for the black silicone coating, which was probably lost

Fig. 5. A magnified representation of the dynamics of O2 (A) and temperature (B) at depths of 2, 5, and 10 cm as measured over the first 3 d of the measuring period shown in Fig. 4. The water level decreased from 12 to 19 cm below ground level over the same period, and there was no precipitation.


during removal of the spear from the soil. The fact that the upper sensors of the array were lost is probably due to the much higher mechanical impact on these sensors during retrieval because they were tightly enclosed by the rhizomes of the vegetation. When recalibrated at 100% air saturation and 14.5°C, the phase angle of the six functional sensors had increased by 1.11 ± 0.45° relative to the original calibration value, corresponding to an apparent decrease in the measured O2 content from 100 to 93.6%. The anoxic recalibration was performed at 15.4°C and showed an average decrease in the phase angle (relative to the original calibration before installation of the spear) of 1.18 ± 0.19°. This corresponds to an apparent increase in measured O2 from 0 to 1.9% air saturation when referring to the original calibration.

In this study, we monitored the 10 optodes mounted in the spear with a 10-channel fiberoptic O2 meter. Another possibility for using multiple sensors is to image the end of a fiber bundle with the O2–sensing ends distributed in the sample (Fischer and Koop-Jakobsen, 2012); these authors used an expensive life-time imaging system for sensor read-out. Both the bundle approach and the spear approach presented here can be combined with ratiometric O2 sensor readout using the inexpensive camera system of Larsen et al. (2011), and such work is in progress (Larsen et al., unpublished).

ConclusionsWe present a new robust tool for spatially resolved in situ

measurements of O2 and temperature in soil and sediment undergoing changes in water level. The new sensor array device allows for simultaneous O2 and temperature measurements at 10 depths in the soil horizon using a combination of fiberoptic O2 sensors and thermocouples. Based on the linear response of O2–dependent sensor luminescence to temperature, it is possible to correct sensor signals for fluctuating temperature. The long-term stability of optical O2 sensors enabled in situ measurement in a peat soil for several months in combination with monitoring of water levels, temperature, and a range of other environmental data. This application illustrated the value of mapping O2 conditions in wetlands at high spatiotemporal resolution in relation to water level fluctuations. The new device thus allows long-term in situ studies of O2 dynamics, which is highly relevant for understanding the effects of water level changes on net greenhouse gas budgets in wetlands.

Future tests of the sensor array should include a more detailed analysis of sensor sensitivity with respect to total pressure variations, coating of secondary phases on the sensors, soil compaction around the array after installation, preferential root growth, and water flow along the spear.

AcknowledgmentsThis study was supported by grants from the Danish Natural Science Research Council (BE, RNG, MK), the Danish Research Council for Technology and Production (MK), and the Danish National Advanced Technology Foundation (MK). RNG was financially supported by the National Environmental Research Council (NERC)– NE/F018612/1; NE/F0122991/1, the Commission for Scientific Research in Greenland; KVUG; GCRC6507, EU; HYPOX-226213; and ERC advanced grant 2010-AdG_20100224. The authors thank Egil Nielsen with the mechanical construction of the sensor array and Christian Juncher Jørgensen for quality testing and managing the data base.

ReferencesAskaer, L., B. Elberling, R.N. Glud, M. Kühl, F.R. Lauritsen, and H.P. Joensen. 2010.

Soil heterogeneity effects on O2 distribution and CH4 emissions from wetlands:

In situ and mesocosm studies with planar O2 optodes and membrane inlet mass spectrometry. Soil Biol. Biochem. 42:2254–2265. doi:10.1016/j.soilbio.2010.08.026

Askaer, L., B. Elberling, T. Friborg, C.J. Jørgensen, and B.U. Hansen. 2011. Plant-mediated CH4 transport and C gas dynamics quantified in-situ in a Phalaris arundinacea-dominant wetland. Plant Soil 343:287–301. doi:10.1007/s11104-011-0718-x

Boelter, D.H. 1965. Hydraulic conductivity of peats. Soil Sci. 100:227–231. doi:10.1097/00010694-196510000-00001

Daulat, W.E., and R.S. Clymo. 1998. Effects of temperature and water table on the efflux of methane from peat land surface cores. Atmos. Environ. 32:3207–3218. doi:10.1016/S1352-2310(98)00078-8

Elberling, B., L. Askaer, C.J. Jørgensen, H.P. Joensen, M. Kühl, R.N. Glud, and F.R. Lauritsen. 2011. Linking soil O2, and CH4 concentrations in a wetland soil: Implications for CO2 and CH4 fluxes. Environ. Sci. Technol. 45:3393–3399. doi:10.1021/es103540k

Fischer, J.P., and K. Koop-Jakobsen. 2012. The multi fiber optode (MufO): A novel system for simultaneous analysis of multiple fiber optic oxygen sensors. Sens. Actuators B 168:354–359. doi:10.1016/j.snb.2012.04.034

Glud, R.N. 2008. Oxygen dynamics of marine sediments. Mar. Biol. Res. 4:243–289. doi:10.1080/17451000801888726

Glud, R.N., I. Klimant, G. Holst, O. Kohls, V. Meyer, M. Kühl, and J.K. Gundersen. 1999. Adaptation, test and in situ measurements with O2 microopt(r)odes on benthic landers. Deep-Sea Res. 46:171–183. doi:10.1016/S0967-0637(98)00068-5

Hargreaves, K.J., and D. Fowler. 1998. Quantifying the effects of water table and soil temperature on the emission of methane from peat wetland at the field scale. Atmos. Environ. 32:3275–3282. doi:10.1016/S1352-2310(98)00082-X

Hecht, H., and M. Kölling. 2001. A low-cost optode array measuring system based on 1 mm plastic optical fibers: New technique for in situ detection and quantification of pyrite weathering processes. Sens. Actuators B 81:76–82. doi:10.1016/S0925-4005(01)00934-0

Holst, G., M. Kühl, and I. Klimant. 1995. A novel measuring system for oxygen microoptodes based on a phase modulation technique. Proc. SPIE 2508:387–398. doi:10.1117/12.221754

Holst, G., R.N. Glud, M. Kühl, and I. Klimant. 1997. A microoptode array for fine scale measurements of oxygen distribution. Sens. Actuators B 38:122–129. doi:10.1016/S0925-4005(97)80181-5

Holst, G., I. Klimant, O. Kohls, and M. Kühl. 2000. Optical microsensors and microprobes. In: M. Varney, editor, Chemical sensors in oceanography. Gordon & Breach, New York. p. 143–188.

Ingram, H.A.P. 1983. Hydrology. In: A.J.P. Gore, editor, Mires: Swamp, bog, fen and moor. General studies. Ecosystems of the World. Vol. 4A. Elsevier, New York.

IPCC. 2007. IPCC climate change 2007: Synthesis report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. R.K. Pachauri. and A. Reisinger, editors. IPCC, Geneva, Switzerland.

Jørgensen, C.J., S. Struwe, and B. Elberling. 2012. Temporal trends in N2O flux dynamics in a Danish wetland: Effects of plant-mediated gas transport of N2O and O2 following changes in water level and soil mineral-N availability. Glob. Change Biol. 18:210–222. doi:10.1111/j.1365-2486.2011.02485.x

Kettunen, A. 2003. Connecting methane fluxes to vegetation cover and water table fluctuations at microsite level: A modelling study. Global Biogeochem. Cycles 17:1051–1070. doi:10.1029/2002GB001958

Klimant, I., V. Meyer, and M. Kühl. 1995. Fiber-optic oxygen microsensors, a new tool in aquatic biology. Limnol. Oceanogr. 40:1159–1165. doi:10.4319/lo.1995.40.6.1159

Kölling, M., H. Hecht, and G. Holst. 2002. Simple plastic fiber based optode array for the in-situ measurement of the ground air oxygen concentrations. In: T. Vo-Dinh and S. Büttgenbach, editors, Advanced environmental sensing technology II. SPIE Proceedings 4576:75–86.

Kühl, M. 2005. Optical microsensors for analysis of microbial communities. Methods Enzymol. 397:166–199. doi:10.1016/S0076-6879(05)97010-9

Kühl, M., and L. Polerecky. 2008. Functional and structural imaging of phototrophic microbial communities and symbioses. Aquat. Microb. Ecol. 53:99–118. doi:10.3354/ame01224

Larsen, M., S.M. Borisov, B. Grünwald, I. Klimant, and R.N. Glud. 2011. A simple and inexpensive high resolution color ratiometric planar optode imaging approach: Application to oxygen and pH sensing. Limnol. Oceanogr. Methods 9:348–360. doi:10.4319/lom.2011.9.348

Liengaard, L., L.P. Nielsen, N.P. Revsbech, B. Elberling, A. Priemé, A.E. Prast, and M. Kühl. 2013. Extreme emission of N2O from tropical wetland soil (Pantanal, South America). Front. Microbiol. 3:433. doi:10.3389/fmicb.2012.00433

Macdonald, J.A., D. Fowler, K.J. Hargreaves, U. Skiba, L.D. Leith, and M.B. Murray. 1998. Methane emission rates from a northern wetland; Response to temperature, water table and transport. Atmos. Environ. 32:3219–3227. doi:10.1016/S1352-2310(97)00464-0

Megonigal, J.P., M.E. Hines, and P.T. Visscher. 2003. Anaerobic metabolism: Linkages to trace gases and aerobic processes. In: H.D. Holand and K.K. Turekian, editors, Treatise on geochemistry. Vol. 8. Elsevier, New York. p. 317–424.

Schäfer, C.-M., L. Elsgaard, C.C. Hoffmann, and S.O. Petersen. 2012. Seasonal methane dynamics in three temperate grasslands on peat. Plant Soil 357:339–353. doi:10.1007/s11104-012-1168-9

Schwärzel, K., M. Renger, R. Sauerbrey, and G. Wessolek. 2002. Soil physical characteristics of peat soils. J. Plant Nutr. Soil Sci. 165:479–486. doi:10.1002/1522-2624(200208)165:4<479::AID-JPLN479>3.0.CO;2-8

Uchida, H., T. Kawano, I. Kaneko, and M. Fukasawa. 2008. In situ calibration of optode-based oxygen sensors. J. Atmos. Ocean. Technol. 25:2271–2281. doi:10.1175/2008JTECHO549.1

TECHNICAL REPORTS

1

There was an error in the calibration value used for “air saturated” water in the original article. The calibration value for air saturated water was 105%, not 100%. This means that the calculated O2 concentrations of the in situ deployment data are ~5% too low, especially concentrations close to 100% air saturation. This error arose because it was assumed that the water in the calibration chamber was homogeneously kept at 100% air saturation. The water column in the calibration container (115 cm length; 19 cm diam.) caused a buildup of hydrostatic pressure, and the air flushing was performed using a pumice stone ~100 cm from the

An Optode Sensor Array for Long-Term In Situ Oxygen Measurements in Soil and Sediment

L. F. Rickelt,* L. Askaer, E. Walpersdorf, B. Elberling, R. N. Glud, and M. Kühl Journal of Environmental Quality 2013 42:1267–1273

water surface near the bottom of the container to ensure a good mixing throughout the water column. As stated in the article, good mixing was checked with a single O2 optode at 10-cm intervals, showing no difference in the O2–dependent phase angle signal. For air bubbles to penetrate the pumice stone out in the water column at 100-cm depth, the pressure inside the bubble must be more than 1.1 times the atmospheric pressure. It will continuously decrease until atmospheric pressure at the surface. Because of the mixing of the water column, this resulted in an averaging of the O2 concentration to a value of 105% air saturation in the present study. An additional experiment in a tube (150 cm length; 3 cm diam.) flushed less vigorously with air from the bottom showed a continuous increase in O2 concentration from 100% air saturation at the surface to 112% at 120-cm depth. Because of the small diameter, there was almost no mixing of water in the vertical direction in this case.

The recalibration has no implications for the conclusions of the manuscript, but the oxygen concentrations in Fig. 4 and 5 have been recalculated to take the calibration artifact into account. The revised versions of Fig. 4 and Fig. 5 are given here.

Copyright © American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. 5585 Guilford Rd., Madison, WI 53711 USA. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. J. Environ. Qual. 43:1–1 (2014) Received 8 July 2014. doi:10.2134/jeq2010.0163er *Corresponding author ([email protected]).

Journal of Environmental Quality ERRATA

Fig. 4. In situ measurements in a peat soil (Maglemosen, Denmark) during a 3-wk period in 2009. (A) Oxygen and water level variations. The dashed lines mark the depths of the respective O2 sensors in relation to the water level. (B) The measured temperature at the same depths.

Fig. 5. A magnified representation of the dynamics of O2 (A) and temperature (B) at depths of 2, 5, and 10 cm as measured over the first 3 d of the measuring period shown in Fig. 4. The water level decreased from 12 to 19 cm below ground level over the same period, and there was no precipitation.

4. Methods in wetland ecosystems

B. Elberling, R.N. Glud, C.J. Jørgensen, L. Askaer, L.F. Rickelt, H.P. Joensen, M. Larsen, and L.

Liengaard

Methods to assess high-resolution subsurface gas concentrations and gas fluxes in wetland

ecosystems

In: R.D. DeLaune, K.R. Reddy, C.J. Richardson, and J.P. Megonigal (Eds.), Methods in

Biogeochemistry of Wetlands. Soil Science Society of America Inc., Madison, (2013) 949-966

DOI: 10.2136/sssabookser10.c49

433

Chapter 49

B. Elberling, C.J. Jørgensen, L. Askaer, H.P. Joensen, and L. Liengaard, Center for Permafrost, Dep. of Geosciences and Natural Resource Management, Univ. of Copenhagen, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark. *Corresponding author ([email protected]). M. Kühl and L.F. Rickelt, Marine Biological Section, Dep. of Biology, Univ. of Copenhagen, Strandpromenaden 5, DK-3000 Helsingør, Denmark. M. Kühl, also Plant Functional Biology and Climate Change Cluster, Dep. of Environmental Science, Univ. of Technology, P.O. Box 123 Broadway, Sydney, NSW 2007, Australia. R.N. Glud and M. Larsen, Scottish Association for Marine Science, Dunstaffnage Marine Lab., Oban, Argyll, PA37 1QA, UK, Dep. of Biology and Nordic Center for Earth Evolution, Univ. of Southern Denmark, Odense M, Denmark, and Greenland Climate Research Centre, Kivioq 2, Box 570, 3900 Nuuk, Greenland.

doi:10.2136/sssabookser10.c49

Copyright © 2013 Soil Science Society of America, 5585 Guilford Road, Madison, WI 53711-5801, USA. Methods in Biogeochemistry of Wetlands. R.D. DeLaune, K.R. Reddy, C.J. Richardson, and J.P. Megonigal, editors. SSSA Book Series, no. 10.

Methods to Assess High-Resolution Subsurface Gas Concentrations and Gas Fluxes in Wetland Ecosystems

Bo Elberling,* Michael Kühl, Ronnie N. Glud, Christian Juncher Jørgensen, Louise Askaer, Lars F. Rickelt,

Hans P. Joensen, Morten Larsen, and Lars Liengaard

AbstractThe need for measurements of soil gas concentrations and surface fluxes of greenhouse gases at high temporal and spatial resolution in wetland eco-system has lead to the introduction of several new analytical techniques and methods. In addition to the automated flux chamber methodology for high-resolution estimates of greenhouse gas fluxes across the soil–atmo-sphere interface, these high-resolution methods include microsensors for quantification of spatiotemporal concentration dynamics in O2 and N2O at micrometer scales, fiber-optic optodes for long-term continuous point mea-surements of O2 concentrations, and planar optodes for high-resolution two-dimensional measurements of O2 and pH. This chapter summarizes the principles behind the methods and shows examples of new insights obtained using combinations of these different methods in a Danish fresh-water wetland system. The results highlight that wetland and peat soils are highly heterogeneous, containing a mosaic of dynamic macropore systems created by both macrofauna and flora leading to distinct spatial and temporal variations in gas concentration on a scale of millimeters and minutes. Applications of these new methodologies allow measurements of greenhouse gas dynamics in wetlands on a scale at which the underlying processes are regulated by microenvironmental controls.

434 Elberling, Kühl, Glud, Jørgensen, Askaer, Rickelt, Joensen, Larsen & Liengaard

Knowledge of the links between spatiotemporal variations in subsurface gas concentration dynamics and surface fluxes is crucial to describe large-scale ecosystem fluxes based on underlying biogeochemical processes. Most of

the existing studies that have reported on greenhouse gas dynamics and exchange in wetlands have focused quite specifically either on the drivers or microbiolo-gy of greenhouse gas production and consumption in the soil or on net fluxes of greenhouse gases across the soil–atmosphere interface. A fundamental challenge therefore exists in linking the knowledge of smaller scale processes from soil iso-lates to field-scale ecosystem fluxes and determining how the knowledge from these different approaches can be upscaled in a meaningful way. This becomes particularly important when models based on fitting equations and/or statistical analyses are used to simulate ecosystem fluxes, which are subsequently scaled up to regions. In terms of understanding the pitfalls and limits of such upscaling, and in relation to the calibration and validation of models, data measurements at high spatiotemporal resolution are needed. This chapter provides an overview of recently developed methods that can provide such high-resolution data sets and thereby help to bridge the current gap in our knowledge of wetland gas fluxes and their regulation.

SUBSURFACE GAS MEASUREMENTSDepth-Specific Gas Sampling: Silicone Probes

Soil air samples from specified depths in a soil profile can be obtained using bur-ied sampling probes where an air volume is connected to a sampling port at the surface, allowing nondestructive repeated measurements over time. Under water-saturated or near-water-saturated conditions, however, such probes become water filled and do not allow the extraction of a representative air sample. To overcome this problem, the use of water-tight silicone tubing has been proposed in which air samples from water-saturated soil conditions are obtained via gas diffusion through the silicone membrane (Jacinthe and Dick, 1996; Kammann et al., 2001). This type of silicone probe has been tested and used for repeated depth-specific gas sampling in water-saturated peat soils over a number of years (Askaer et al., 2010, 2011; Jørgensen et al., 2012; Jørgensen and Elberling, 2012), where 1.3-m-long silicone tubes (10-mm i.d., 3-mm wall thickness, 100-mL total probe volume) closed with rubber septa at both ends were rolled into a coil. A 0.92-mm-i.d. stain-less steel tube was inserted through the outer septa of the probe to connect the silicone probe in the soil with the soil surface (1 mL of dead volume per meter steel tube) and closed with a three-way stopcock at the end of the steel tube. The probes were inserted in precut cavities in the wall of a soil pit, which were subsequently refilled with soil, horizon by horizon, allowing the soil to collapse around the probe after insertion.

After installation of such silicone probes, nondestructive soil gas samples can be taken on a regular basis, taking into account the total time to reach steady-state conditions in the tube after sampling, which is dependent on the dimensions and wall thickness of the silicone tube. Careful testing of the installed tubes revealed

Methods to Assess High-Resolution Subsurface Gas Concentrations and Gas Fluxes 435

that steady-state conditions of the probe internal gas volume were obtained within less than 3 d under field conditions, allowing a potential temporal sampling fre-quency on a subweekly timescale. The spatial resolution of the silicone probes are limited by the physical size of the tube and the installation pattern, which in com-bination in this case facilitated a depth-specific resolution of 5 cm (Jørgensen et al., 2012; Jørgensen and Elberling, 2012). Askaer et al. (2010, 2011) took depth-specific gas samples of approximately 40 mL from each probe on a weekly basis, which were transferred to evacuated glass vials via a needle and syringe before analysis in the laboratory. Subsurface gas concentration profiles of CO2 and CH4 using sili-cone tubes under contrasting environmental conditions and soil moisture contents are shown in Fig. 49-1.

Oxygen and Nitrous Oxide Microsensors for ProfilingMicrosensors are widely used in biogeochemistry and many fields of fundamen-tal and applied research. A large number of sensor designs have been developed

Fig. 49-1. Field-observed subsurface gas concentrations under near-steady-state conditions: (a) mea-surements during cold and flooded conditions and (b) measurements during warm conditions with the water level below 50 cm; under flooded conditions, O2 was also measured using microsensors (diamonds) (modified from Elberling et al., 2011).


for measuring the concentrations of key elements of the C, N, and S cycles, as well as important physical properties such as diffusivity, temperature, and light penetration (reviewed in Kühl and Revsbech, 2001; Revsbech, 2005; Kühl, 2005). Microsensors can be applied in both aquatic ecosystems and in soil, where they have revealed important characteristics of the soil microenvironment and dynam-ics of biogeochemical processes (e.g., Tiedje et al., 1984; Højberg et al., 1994; Elberling and Damgaard, 2001; Elberling et al., 2010; Jørgensen and Elberling, 2012). While many variables can be measured with microsensors, the focus is here on the application of O2 and N2O microsensors for assessing subsurface gas con-centrations in water, soil slurries, and partially as well as fully saturated peat soils.

Amperometric Oxygen and Nitrous Oxide MicrosensorsWhile the electrode materials and electrolytes are different, electrochemical microsensors for O2 and N2O both rely on an amperometric measuring principles based on a Clark-type electrode configuration, where the partial pressure of O2 or N2O is measured. The analyte gas is reduced at a negatively polarized measur-ing electrode positioned together with reference and guard electrodes inside an electrolyte-filled glass casing that is sealed with a thin silicone membrane allow-ing only gases to enter (Revsbech, 2005). The measuring electrode is positioned a few microns behind a silicone membrane at the sensor tip, which is also just a few micrometers wide and thick. To avoid O2 interference on N2O measurements, the N2O microsensors are fitted with an O2 trap, i.e., an extra casing with a thin layer of strongly reducing alkaline ascorbate solution in front of the N2O measuring tip (Andersen et al., 2001). A typical N2O microsensor is therefore slightly larger and exhibits a longer response time than typical O2 microsensors.

The small tip dimensions ensure that only a small amount of O2 or N2O is con-sumed by the microsensor in the measuring process; typical sensors consume less than a single living microbe. Furthermore, the thin membrane and short distance between the membrane and measuring electrode ensure a fast and stable response due to efficient diffusive supply to the sensor across such short distances. The O2 and N2O microsensors can therefore be used in stagnant media without affecting the local microenvironment outside the microsensor tip. In this way, microsensors exhibit low stirring sensitivity, with typical variations of only a few percentage points between stagnant and stirred conditions (Gundersen et al., 1998). Further-more, the outer casing of such sensors can be customized for use in various media, e.g., a thin and slender casing for measurements in fine-grained sediment or bio-films or a thicker and more robust casing for measuring in coarse sediments or even dry soil (Wenzhöfer et al., 2000). The casing can even be put inside a hypo-dermic needle for even greater protection against physical damage or breakage while retaining the high-resolution measuring characteristics of the sensors. While the construction of these microsensors requires special skills and is time consum-ing, both the electrochemical O2 and N2O microsensors are commercially available in various designs and tip sizes in the range of <5 mm to >1 mm (Unisense A/S).


Optical Oxygen Microsensors (Microoptodes)Fiber-optic O2 microsensors, i.e., microoptodes, have been developed as an alter-native to amperometric O2 microsensors (Klimant et al., 1995, 1997a; Kühl, 2005), and both optodes and detectors are commercially available for a variety of envi-ronmental and biogeochemical uses (Pyro-Science GmbH; Presens GmbH). The measuring principle is based on dynamic collisional quenching of the lumines-cence of an O2–sensitive indicator dye, which is immobilized in a polymer matrix at the tip of a tapered optical fiber. When the indicator dye is exposed to an exci-tational light source, it emits a red-shifted luminescence (either fluorescence or phosphorescence) at the fiber tip, and this luminescence is guided via the fiber to a sensitive detector. When O2 molecules collide with the excited indicator dye, energy is dissipated through generation of elemental O that dissipates quickly back to O2. In this way, the higher O2 concentration at the sensor tip results in a lower luminescence level (for more details on specific O2 indicators and measur-ing procedures, see Kühl, 2005). Such optical O2 measurement can also be used in larger fiber-optic sensors and in planar optodes for imaging O2 concentration (see section on planar optodes below). Generally, optical O2 sensors exhibit better long-term stability of their measuring characteristics than electrochemical O2 sensors. Many different combinations of indicators, matrices, and excitation lights—each with different properties—have been tailored to specific research questions. A variety of luminescent O2 indicators with different quenching characteristics and spectral properties can be used in combination with various immobilization matri-ces (reviewed by Amao, 2003; Borisov and Wolfbeis, 2008), and such sensors can be tailored to specific research questions (e.g., systems optimized for low O2 sens-ing, non-consuming measuring principle, long term stability, fast response time).

Application of MicrosensorsThe strength of microsensor analyses is that they enable measurements of con-centration gradients across micrometer to millimeter distances in the soil matrix with minimal disturbance to the investigated system, both in terms of mechanical disturbance due to their small dimensions and in terms of chemical disturbance due to their negligible O2 consumption. While they give very detailed informa-tion on the local concentration and dynamics of O2 and N2O, it is important to realize that microsensors only provide point measurements in a heterogeneous matrix (see also planar optodes below). Figure 49-1 shows a typical example of the application of a Clark-type O2 microsensor in a wetland soil (Elberling et al., 2011). The O2–dependent microsensor current was measured by a picoampere meter (PA2000, Unisense), which simultaneously kept the measuring cathode polarized against the internal reference. The sensors typically have tip diameters between 20 and 250 mm, a stirring sensitivity of <1 to 2%, and a 90% response time of 0.2 to 3 s (Gundersen et al., 1998). With increasing tip diameter and silicone mem-brane thickness, the robustness of the sensor increases markedly, with a trade-off in increased response times. The linearity of the sensor response was confirmed by recording the output current in picoamperes in water sparged with N2, air, and pure O2. The sensor needs to be calibrated at the experimental temperature, e.g.,


by a two-point calibration from measurements in (i) the overlying water with known O2 content (either at atmospheric saturation or determined independently by titration or calibrated macrosensors) and (ii) a strongly reduced solution (e.g., sodium dithionite or an alkaline 0.1 mol L−1 sodium ascorbate solution).

High-resolution O2 concentration profiles can be made with O2 microsensors across small depth intervals where the actual depths can be controlled either man-ually or by an automated profiling system. Depth intervals typically range from 50 to 500 mm and should be considered with respect to the sensor tip diameter. The reference depth for vertical positioning is often the water–sediment interface, which can be determined by approaching the sediment surface with the sensor tip in small steps and moving the sensor horizontally, e.g., by gentle tapping on the micromanipulator after each step. The highest position at which the sensor is able to make surface particles move can be defined as the sediment surface. Sub-sequently, the reference depth can be confirmed by a change in the slope of the O2 gradient at the sediment surface (Fig. 49-2). Observation of the change in slope can be the only method of identification of the surface position during in situ measure-ments (Glud et al., 1995), where visual detection is often not possible; however, combining the microsensor tip with a tapered fiber-optic reflectance probe can also enable surface determination without visual inspection (see details in Kli-mant et al., 1997a).

The details of submerged substrates that can be obtained under field and laboratory conditions using microsensors are illustrated in Fig. 49-1b. Such mea-surements can, for example, resolve concentration gradients across the diffusion

Fig. 49-2. Schematic presentation of the O2 concentration profile through a flooded wetland ecosystem, where hdbl is the height of the diffusive boundary layer, dp is the O2 penetration depth, C0 is the O2 con-centration at the tailing surface, Csat is the O2 concentration in the saturated water phase, dCw/dx(0) is the O2 concentration gradient in the diffusive boundary layer at the tailing surface, and dCs/dx(0) is the O2 concentration gradient in the tailings at the surface (modified from Elberling and Damgaard, 2001). The reference depth can be confirmed by the change in the slope of the O2 gradient at the sediment surface.


boundary layer between wetland soils and the overlying water, and the O2 con-sumption rates can be related directly to O2 fluxes across the sediment–water interface (Fig. 49-2).

Nitrous oxide microsensors (Revsbech et al., 2005), which are O2 insensitive (see above), have also been used successfully to measure N2O concentrations in wetland soils using a similar experimental setup as described above for O2 micro-sensors (Elberling et al., 2010; Liengaard et al., 2011; Jørgensen and Elberling, 2012). Calibration of N2O microsensors typically involves measurements in solutions of known N2O concentration at the experimental temperature, e.g., by repeated addi-tion of known volumes from a N2O-saturated stock solution.

The limitations of all glass microsensors discussed above are particularly related to the lack of robustness when sensors are inserted directly in soils with roots and stones; however, more robust versions have been constructed and applied successfully (e.g., Markfoged et al., 2011). Drift in sensor response and therefore the need for recalibration on a regular basis when measuring over longer times makes such amperometric sensors less suitable for long-term measurements. Needle probes are only thin at the measuring tip; they are not suitable for repeated measurements in the exact same measuring holes because the top layers become influenced by the penetration of the thicker part of the sensor.

Fiber-Optic Oxygen Optodes for Continuous In Situ MeasurementsFiber-optic O2 optodes are very useful for long-term in situ measurements due to the lack of gas consumption and minimal sensor drift combined with a good mechanical strength, especially when depth-specific gas concentrations need to be measured repeatedly on an hourly to daily basis in the field and in different depth horizons. Commercial optodes are available for pH, CO2, and O2, but so far only O2 has been used for longer term measurements in wetlands (Elberling et al., 2011; Jørgensen et al., 2012; Jørgensen and Elberling, 2012). In these studies, long-term installation has demonstrated the use of optodes for high-resolution assessment of in situ depth-specific O2 concentrations. These investigations were based on one of the most successful O2 types consisting of a polymer optical fiber with a polished distal tip, which is coated with a small patch of a transparent foil with immobi-lized O2 indicator in a polystyrene matrix. The optical fiber can be mounted in a steel tube to protect both the sensor material and the fiber itself and arranged in an sensor array inserted into the wetland soil to varying depths, enabling moni-toring of O2 concentrations in particular soil horizons over longer times. While these relatively large and robust O2 optodes are excellent tools for long-term in situ monitoring of O2, they are less suitable for repeated profiling in the soil and therefore cannot quantify the O2 dynamics at the same spatial resolution as O2 microsensors. Figures 49-3 and 49-4 illustrate in situ measurements with an O2 optode array inserted in a peat-soil profile at depths between 5 and 120 cm. Each sensor consisted of a robust fiber-optic O2 optode mounted together with a type K mini thermocouple enabling temperature-compensated measurements. This is essential because many optode types are highly temperature sensitive. The O2 optodes were connected to a multichannel fiber-optic O2 meter (FIBOX-4 or FIBOX


Fig. 49-3. Example of changes in O2 concentrations and transport via plants over 10 d: O2 concentra-tions in the upper 5-cm depth (red circles), soil temperature (black line), and photosynthetically active radiation (PAR) (brown circles). A significant correlation (P < 0.001) between diurnal variations in PAR (W m−2) indicates plant transport of O2 into the root zone as a result of active photosynthesis. The timing of the daily maximum soil temperature (at 5 cm) is, on average, delayed about 3 to 5 h from the maximum incoming radiation. In the first half of the period, O2 concentration varies from full oxygenation at a depth of 5 cm to fully O2 depleted at a depth of 10 cm below terrain (modified from Jørgensen et al., 2012). ???PERMISSION ISSUE???

Fig. 49-4. Example of changes in O2 concentrations and transport over months during marked changes in water level (black line in bottom graph) and water content controlled by precipitation (modified from Elberling et al., 2011).

Author: Please forward the release from the publisher or replace or delete the figure.


10, Presens GmbH) and the thermocouples were connected to a thermocouple thermometer (RS 206-3722, RS Components Ltd.). All sensors were calibrated by a two-point temperature and O2 calibration. Sensors measured with a standard deviation £5% and an average 90% response time across the measuring range (0–100% air saturation) of ?15 min. Recalibration after 1 yr showed long-term sen-sor stability within 5%.

Figure 49-3 illustrates high-resolution temporal measurements of the near-surface O2 concentrations during periods with a stable soil water table, where O2 variations were closely related to plant activity as controlled by irradiance and the associated release of O2 from the roots during the daytime (Fig. 49-3). Based on the same type of data, Fig. 49-4 shows high-resolution temporal-spatial mea-surements at longer time scales, where seasonal changes in the water table cause marked changes in the subsurface O2 availability.

Arrays of mini optodes mounted at specific depths in the soil are well suited for simultaneous measurement of O2 at several depths but are not suitable for profiling in soils because the repeated pushing in and out of the soil matrix can damage the sensor layer at the measuring tip of the optode due to physical abra-sion. An alternative use of fiber-optic minisensors has been developed, however, for high-resolution measurements of O2 and pH distributions in soil and rhizo-spheres (Blossfeld et al., 2011). In that study, the O2– or pH-sensitive optical sensor chemistry was cast onto a foil that was inserted into the soil, where two-dimen-sional measurements of O2 or pH were taken by moving a so-called readout fiber over the sensor foil in a grid of measuring points. Such a measurement approach has so far not been applied in wetland soils and is inherently limited to laboratory setups, where such spatial readouts can be realized with the use of specialized plant growth chambers, where roots are forced to grow along the optode foil sur-face. Even higher spatial resolution, two-dimensional mapping of O2 and pH in soils can be obtained by planar optode technology in combination with various imaging setups (see below).

Membrane Inlet Mass SpectrometryDepth-specific measurements of dissolved gas concentrations in wetland soil can be made using membrane inlet mass spectrometry (MIMS). A detailed descrip-tion of the method and the MIMS probe itself was provided by Sheppard and Lloyd (2002). The method has been developed and applied for use in peat meso-cosms under laboratory conditions (Benstead and Lloyd, 1994; Thomas and Lloyd, 1995; Lloyd et al., 1996). An example of field use and in situ measurements using the MIMS probe in a Danish peatland can be seen in Fig. 49-5 (unpublished data, 2008), from a study in which a quadrupole mass spectrometer (Balzers QMA 125) was mounted in a temperature-controlled unit and used with stainless steel gas sampling probes, allowing detailed gas profiles over several weeks (see details in Askaer et al., 2010). The sampling probe consisted of a 50- to 125-cm-long stain-less steel probe (3.2-mm o.d, 1.6-mm i.d.) fitted with a 20-mm-long narrow tip (1.6-mm o.d., 1.0-mm i.d.) in one end and attached to the mass spectrometer at the other end via a 50-cm-long flexible metal bellows. The narrow tip was closed at


the sampling end and had a 0.5-mm orifice 10 mm from the end. The orifice was covered with a 50-mm-thick microporous polypropylene membrane (Celgard 2502, Hoechst Celanese) with an effective pore size of 0.075 mm and a porosity of 45%. In contrast to the silicone rubber membranes used by Benstead and Lloyd (1994), Thomas and Lloyd (1995), and Lloyd et al. (1996), this type of membrane does not result in a highly preferential transport of gases compared with water through the membrane; instead, all compounds pass the membrane at predictable rates. The data shown in Fig. 49-5 demonstrate how this setup can provide a much higher spatial resolution of CO2 concentrations in the soil than measurements with pre-installed silicone probes.

Planar Optodes for Two-Dimensional Oxygen MeasurementsWhile fiber-optic O2 optodes are highly suitable for long-term monitoring of depth-specific O2 concentrations at specific points in soil profiles, they do not provide information on small or microscale variations in the O2 distribution and O2 dynamics in response to the soil heterogeneity and/or temporal changes in environmental conditions within the soil system. To map the two-dimensional O2 concentration in the soil matrix at a very high temporal resolution, planar O2 optodes can be used (Fig. 49-6). Such optodes use the same sensor materials as used in fiber-based O2 sensors, but an O2–sensitive indicator is immobilized in an extended homogeneous polymer layer fixed to a transparent support foil. This foil can, e.g., be mounted on inverted periscopes for in situ application or inside a transparent plant root chamber in which the indicator foil can be excited and the distribution of O2–dependent luminescence can be recorded with a digital camera equipped with a long-pass filter (Glud et al., 2005). Several imaging systems exist, each optimized for different settings and environmental conditions (Glud et al., 1996; Oguri et al., 2006; Yanzhen et al., 2011). A common feature for these systems is the use of various O2–sensitive light characteristics to quantify the O2 distribu-tion; some use simple intensity while others use the lifetime of the luminescence, which again can be quantified by a number of approaches. In general, lifetime-based measurement systems provide a better signal quality and use transparent

Fig. 49-5. Deep CO2 concentration profiles measured using both silicone probes (solid symbols) and mem-brane inlet mass spectrometry (open symbols) in the upper 50 cm of a wetland (unpublished data, 2008).


optodes but require also faster response times as well as sophisticated and expen-sive equipment. An alternative strategy is to use radiometric approaches, where an O2–sensitive intensity signal is normalized to an O2–insensitive signal in each image pixel, followed by relative simple post-processing in traditional image pro-cessing software. Recently, an inexpensive and simple camera system analyzing the red, green, blue (RGB) signal in each image pixel was developed and applied for pH and O2 imaging (Larsen et al., 2011). This system provides high-quality data while being both robust and relatively inexpensive.

The spatial resolution of planar O2 optodes can be optimized for a given investigation by regulating the camera optics of the setup, but optical smearing as the luminescent signal passes the light-guiding support foil and the chamber wall can compromise the spatial resolution. For microscale investigations, fiber optic phase plates allowing a maximum pixel resolution on the order of ?10 mm is rec-ommended (Fischer and Wenzhöfer, 2010).

Planar optodes have been only recently introduced in wetland studies (Askaer et al., 2010; Elberling et al., 2011). In these pioneering works, the O2–quenchable indicator Ru(II)-tris-4,7-diphenyl-1,10-phenanthroline (Ru-dpp) was immobilized onto a 0.125-mm-thick transparent polyethylene foil (Mylar). The luminescent indicator has an excitation maximum at 460 nm and a red luminescence emission maximum at 610 nm, which is quenched in the presence of O2. Two planar optode foils (100 by 250 mm) were taped together (100 by 500 mm) and mounted on the inside of a Plexiglas sheet glued onto a cut polyvinyl chloride cylinder. In the experimental setup, excitation light was supplied by an array of blue light-emitting diodes (wavelength = 470 nm) illuminating the sensor foil through the Plexiglas and the sensor support foil (Fig. 49-6). Images of the O2–dependent luminescence

Fig. 49-6. Schematic drawing of the experimental setup for monitoring O2 dynamics in peat soil meso-cosms by a planar optode imaging system (not to scale). The vacuum pump allows a fast regulation of the water table while the air flow on the overlying water phase ensures homogenous distribution of O2 above the soil during flooding (modified from Askaer et al., 2010).


lifetime were quantified by a thermoelectrically cooled gateable charge coupled device camera (SensiCam Sensimod, PCO Computer Optics) equipped with a 25-mm/1.4 Nikon wide-angle lens. All images covering an area of 70 by 50 mm were converted into O2 images using a modified Stern–Volmer equation, and sig-nals fixed areas of the foil exposed to known O2 concentrations (i.e., 0 and 100% air saturation) as described by Wenzhöfer and Glud (2004). Image systems similar to that described above have been used to resolve O2 dynamics in other quite differ-ent aquatic environments (Frederiksen and Glud, 2006; Precht et al., 2004; Cook et al., 2007; Volkenborn et al., 2010).

Figure 49-7a shows a planar optode black and white image of a mesocosm peat profile with the water table located 4 cm below the surface. Figure 49-7b shows the same image after interpretation of the O2 distribution. In combination, the two images allow detailed inspection of the O2 distribution in relation to soil structures such as dead roots, animal pellets, gas bubbles, and preferential travel paths.

An active wetland rhizosphere exhibits a very heterogeneous O2 distribution and steep concentration gradients. To evaluate the O2 distributions around grow-ing Phalaris arundinacea L. roots, a peat core was extracted in early spring and placed in a transparent mesocosm equipped with O2–sensitive planar optodes as described above. The mesocosm was placed under controlled light (16 h light, 8 h dark) and temperature conditions (18°C light temperature, 12°C dark tempera-ture). The well-developed biomass clearly mediated a significant transport of O2 to deeper soil layers (Fig. 49-8). It can also be seen that not all roots released similar amount of O2. This is presumably related to root age because time series record-ings reflected how growing roots released the most O2.

Another important driver for micro- to mesoscale heterogeneity in wetlands is macrofaunal activity. Burrow structures help ventilate the soil, but the mucus lining along burrows of earthworms (Lumbricus terrestris) can also induce micro-bial hot spots depleted with respect to O2 in otherwise well-aerated soil (Fig. 49-9A and 49-9B). The intensified microbial activity associated with such sites may also induce local pH minima (Fig. 49-9C).

Time series of O2 images can be especially useful for quantifying gas dynam-ics linked to variations in water level as well as root and faunal activity in wetland soil. An example is displayed in Fig. 49-10, which shows images recorded over 2 d in a controlled experiment where the water table gradually decreased from 5 cm above the soil surface to 40 cm below the soil surface, with a following overall increase in O2 availability in the topsoil. Anoxic waterlogged patches remained, however, leaving a mosaic-like pattern of anoxic microsites within the aerated soil matrix (Fig. 49-10). Animations of this type of high-resolution O2 dynamics in the soil can be found in the online supplementary material for Wenzhöfer and Glud (2004) and Askaer et al. (2010).

Limitations and SuggestionsThe need for combining methods is evident from the discussion above; no single method is suitable for providing multiple gas measurements at high temporal and spatial resolution. Planar optodes are the state-of-the-art method, but for wetland


Fig. 49-9. (A and B) Two paired examples of soil and O2 images around a mucus-enriched earthworm burrow in an otherwise well-aerated soil, depicted by a transparent planar optode using a lifetime-based imaging system (O2 scale bar reflects air saturation, %), and (C) the pH distribution around a similar bur-row using a nontransparent pH optode (scale bar reflect pH units) and a red, green, blue (RGB)-based imaging system (unpublished data, 2011).

Fig. 49-7. (a) Black and white image of a mesocosm peat profile as recorded through a transparent planar optode, (b) the corresponding O2 image, and (c) schematic image of soil heterogeneity, O2 distribution, and the potential fate of CH4 produced in the anoxic zone.

Fig. 49-8. Black and white image of a Phalaris arundinacea root system as obtained through a semitrans-parent planar optode and the corresponding O2 image documenting the leakage of O2 to the surrounding water-saturated soil, scaled to ensure sufficient resolution of the O2 distribution in the O2–depleted soil. The color scale bar indicates the air saturation (%) (unpublished data, 2007).


studies they have so far only been used for imaging O2 and pH. The approach has been applied in situ in aquatic settings, but this step still needs to be adapted to wetland ecosystems. An interesting development is the combination of optic-fiber bundles equipped with sensing chemistry in one end fixed to an imaging system and an excitation light source in the other. In this system, numerous fibers

Fig. 49-10. Observed O2 (air saturation, %) distribution in the peat column during a drainage experi-ment; T is time (min), WTD is water table depth. The image size is 6 by 8 cm. At T0 the column was fully saturated with 5 cm of aerated water above the soil surface and was drained to −40 cm (modified from Askaer et al., 2010).


that can each be positioned at selected sites can be interrogated by the camera in real time. Such work is in progress in more laboratories and will in future allow true multichannel measurements without the use of expensive measuring devices. More sophisticated methods, such as in situ profiling using mass spectrometry, have been successfully applied in the field on a short-term basis. To provide high-resolution data on the key greenhouse gases, however, these methods need further development before they can be applied as standards under contrasting environ-mental conditions.

NET SURFACE GAS FLUX MEASUREMENTSHigh-Resolution Flux Chamber Methodology

Increased awareness of the importance of wetlands with respect to net CO2 and CH4 emissions and more recently reported pulses of N2O associated with water table fluctuations (Jørgensen et al., 2012; Jørgensen and Elberling, 2012) has high-lighted the importance of high-resolution and near-continuous measurements of greenhouse gas fluxes across the soil–atmosphere interface. Conventional, low-frequency flux measurements using manual closed static chambers are associated with a number of problems. For instance, physical disturbance of wetland soils due to walking in the area before and during measurements can have a major effect on the release of greenhouse gases via ebullition (which is in particular impor-tant for CH4). Also, the relatively few number of samples taken per chamber over time may not necessarily allow optimized time intervals for the various gases of interest. The faster increase in CO2 concentrations should preferably be measured over 50 to 100 s, while the typically slower concentration buildup and low signal strength for N2O requires measurement periods of typically 15 to 45 min. Fur-thermore, the relatively few sample points obtained by manual sampling may not allow accurate flux estimate calculations, especially if the measured greenhouse gas concentrations are close to the detection limit or if the gas emissions occur via ebullition. Also, because the transport of both CH4 and N2O from the soil to the atmosphere may occur via plant-mediated gas transport in aerenchymous plant tissue generic to many wetland plants or in short-lived emission burst events (Jør-gensen and Elberling, 2012), the weekly sampling frequency often encountered with manual flux chamber measurements will not necessarily capture the in situ gas exchange dynamics and thus allow a meaningful integration of seasonal or annual budgets.

Figures 49-11 and 49-12 illustrate the differences between applying manual closed static chambers without plants and automated closed static chambers with plants. Once installed, the automated closed static chambers provide flux mea-surements on a temporal resolution of 15 to 45 min. Depending on the number of chamber replicates, hourly mean values may be calculated. The data presented (Fig. 49-12) are based on automated flux chambers (60 cm deep by 60 cm wide by 60 cm high ± a 50-cm extender) made of transparent 6-mm polycarbonate sheets installed in steel base frames inserted 10 cm into the soil to prevent horizontal gas flow. Transmission of radiation in the photosynthetically active radiation (PAR)


Fig. 49-11. Schematic illustration of the difference between total ecosystem flux and soil ecosystem flux measured by automated closed static (ACS) cham-bers and manual closed static (MCS) chambers, re-spectively (modified from Askaer et al., 2011).

Fig. 49-12. (a) Daily average CH4 fluxes measured by automatic chambers (black circles) and static chambers (gray circles) calculated by linear regression, (b) groundwater level (WL) and contour map of depth-specific soil CH4 concentrations constructed from weekly measurements at the 5-, 10-, 20-, and 50-cm depths (measurement frequency is shown as white dots), (c) soil temperatures at the 2- and 30-cm depths, and (d) daily precipitation and soil surface moisture content (SMC). Error bars show 1 SD from the mean. Roman numerals and gray bars signify high temporal resolution measurement periods (modified from Askaer et al., 2011).


spectrum through the chamber walls was measured to be >80% of incoming PAR. The chambers were fitted with both inlet and outlet tube connectors. During mea-surements, air from the chamber headspace was circulated through RS 293-2000 tubing (0.5-mm i.d., 0.9-mm o.d.) from the chamber to the gas analyzers in a closed and pressure-tight loop at approximately 2.5 L min−1. During measurement, the air volume inside the closed chamber headspace was circulated using a 12-V fan to prevent the buildup of concentration gradients of the measured gases within the chambers. Real-time concentrations of CO2, CH4, N2O, and water vapor (H2O) were determined using both a nondispersive infrared gas analyzer (LiCor LI-840), a high-accuracy CH4 analyzer (DLT-100, Los Gatos Research), and an in-line pho-toacoustic trace gas analyzer (INNOVA 1312, LumaSense Technology) similar to other automated flux measurement studies (Ambus and Robertson, 1998; Yamulki and Jarvis, 1999). Simultaneous measurements of CO2 and H2O concentrations were performed by both the LI-840 and the INNOVA 1312 to achieve a 30-s tempo-ral resolution of CO2 concentrations by the LI-840 and to provide an independent CO2 control on the status of the low-concentration measurements (nL L−1 region) of N2O by the photoacoustic gas analyzer. To achieve an acceptable signal strength for N2O, the flux chamber closing period was 30 min. To stabilize the water vapor pressure in the measurement cell, the sample gas was dried before analysis using a noninterfering Nafion dryer (PermaPure MD110) with continuous purging of dry air.

The importance of automatic, high-resolution flux ecosystem chambers for capturing in situ greenhouse gas exchange dynamics is illustrated for CH4 (Fig. 49-13), where distinct variations can be observed on both a daily and a seasonal basis. The need for flux measurements at a high temporal resolution for captur-ing actual field-scale greenhouse gas dynamics is further illustrated in Fig. 49-14, where a significant but short-lived N2O burst can be observed in response to a rapid shift in water level (see also Jørgensen and Elberling, 2012).

Limitations and SuggestionGreenhouse gas transport through aerenchymous tissue in a variety of wet-land macrophytes is a topic of ongoing research. Several plant–soil–atmosphere interactions could be highly important for modifying net gas fluxes across the soil–atmosphere interface and thereby influencing the net annual ecosystem greenhouse gas budget for a particular wetland. While the use of automated, high-resolution closed static chambers is often an appropriate method for quantifying in situ flux dynamics, this approach has several important limitations. These limitations are mainly related to the fact that once closed for measurements, flux chambers cre-ate their own isolated environment and microclimate in the chamber headspace, which can influence both microbial processes in the soil and plant physiological parameters related to the release or uptake of greenhouse gases. These changes in headspace properties are dependent on the overall site-specific conditions: night vs. day measurements, low or high rates, active or inactive plants, direct or indi-rect light, the size of the chamber vs. actual rates. The next step in automated, closed static chambers is to create as close to natural ambient conditions in the


chamber headspace during measurement as possible for the major environmental parameters that could influence both microbial and plant-related processes. That means essentially control of the headspace temperature, relative humidity, incom-ing light, and auxiliary gas concentrations other than the gas compound being quantified, particularly rapid changes in CO2 concentrations.

Fig. 49-13. Hourly average CO2 and CH4 fluxes for a 3-d period during different growth stages of Phalaris arundinacea corresponding to the high temporal resolution measurement periods in Fig. 12: (i) dormancy, (ii) sprouting, (iii) early growth season (high water level), and (iv) mid growth season (low water level). All measurements were made by automatic closed chambers. Dates on the x axis represent 0000 h. Error bars show 1 SD from the mean, n = 3 (modified from Askaer et al., 2011).

Fig. 49-14. Hourly average N2O fluxes for a 5-d period during a natural flooding event. Measurements were made by automatic closed chambers (modified from Jørgensen and Elberling, 2012).


CONCLUSIONSHigh spatial and temporal variability in subsoil gas distribution and surface flux patterns have been documented and illustrated mainly from one specific wetland site in Denmark; however, the variability and complexity of interacting biogeochemical processes is considered representative for a wide range of wetland systems. The results shown highlight the need for high-resolution measurements of appropriate sample frequency in future wetland studies. The results also suggest that new findings with respect to understanding the linkages between subsurface greenhouse gas processes and actual gas flux across the soil and plant surfaces probably will require the combination of different high-resolution technologies. In the investigated Danish wetland, the combination of various high-resolution methods has proven an effective approach for quantifying the role of plant-induced gas transport, the importance of living roots in releasing O2 into anaerobic layers, and the micro- to mesoscale heterogeneity in wetlands controlled by macrofaunal activity, including burrow structures, which help ventilate the soil, and microbial hot spots depleted with respect to O2 in an otherwise well-aerated soil.

The improvement of methods achieved during the last 20 yr in other sciences have more recently been applied in wetland systems, which represent one of the most dynamic natural ecosystems with major global impacts in terms of water improvement, biomass production, and greenhouse gas budgets. In all of these cases, linking subsurface gas dynamics as a function of water table, fluxes of nutri-ents, and climate to net ecosystem fluxes is a key issue and the application of the methods described in this chapter may grossly improve our process understand-ing and current quantitative assessments of gas dynamics in wetland systems.

ACKNOWLEDGMENTSThis work was conducted within the framework of the projects “Oxygen

availability controlling the dynamics of buried organic carbon pools and green-house gas emissions” and “Nitrous oxide dynamics: The missing links between controls on subsurface N2O production/consumption and net atmospheric emis-sions” financed by the Danish Natural Science Research Council (PI: B. Elberling); M. Larsen and R.N. Glud were financially supported by the National Environ-mental Research Council NE/F018612/1 and NE/F0122991/1, the Commission for Scientific Research in Greenland, KVUG, GCRC6507, and ERC Advanced Grant ERC-2010-AdG20100224. Thanks to J. Santner and E. Oburger for assistance with pH images.

REFERENCESAmbus, P., and G.P. Robertson. 1998. Automated near-continuous measurement of carbon

dioxide and nitrous oxide fluxes from soil. Soil Sci. Soc. Am. J. 62:394–400. doi:10.2136/sssaj1998.03615995006200020015x

Andersen, K., T. Kjær, and N.P. Revsbech. 2001. An oxygen insensitive microsensor for ni-trous oxide. Sens. Actuators B 81:42–48. doi:10.1016/S0925-4005(01)00924-8

Amao, Y. 2003. Probes and polymers for optical sensing of oxygen. Microchim. Acta 143:1–12. doi:10.1007/s00604-003-0037-x


Askaer, L., B. Elberling, T. Friborg, C.J. Jørgensen, and B.U. Hansen. 2011. Carbon dioxide and methane fluxes linked to oxygen availability in a graminoid dominant temperate northern peatland. Plant Soil 343:287–301. doi:10.1007/s11104-011-0718-x

Askaer, L., B. Elberling, R.N. Glud, M. Kühl, F. Lauritsen, and H.P. Joensen. 2010. Soil het-erogeneity control on the distribution of oxygen and resultant methane emissions from wetlands: Assessed by 2-D planar optodes and membrane inlet mass spectrometry. Soil Biol. Biochem. 42:2254–2265. doi:10.1016/j.soilbio.2010.08.026

Benstead, J., and D. Lloyd. 1994. Direct mass spectrometric measurement of gases in peat cores. FEMS Microbiol. Ecol. 13:233–240. doi:10.1111/j.1574-6941.1994.tb00070.x

Blossfeld, S., D. Gansert, B. Thiele, A.J. Kuhn, and R. Lösch. 2011. The dynamics of oxygen concentration, pH value, and organic acids in the rhizosphere of Juncus spp. Soil Biol. Biochem. 43:1186–1197. doi:10.1016/j.soilbio.2011.02.007

Borisov, S.M., and O.S. Wolfbeis. 2008. Optical biosensors. Chem. Rev. 108:423–461. doi:10.1021/cr068105t

Cook, P.L.M., F. Wenzhöfer, R.N. Glud, F. Janssen, and M. Huettel. 2007. Benthic solute ex-change and carbon mineralization in two shallow subtidal sandy sediments: Impact of advective porewater exchange. Limnol. Oceanogr. 52:1943–1963. doi:10.4319/lo.2007.52.5.1943

Elberling, B., L. Askaer, C.J. Jørgensen, M. Kühl, H.P. Joensen, R.N. Glud, and B.U. Hansen. 2011. Linking soil O2, CO2, and CH4 concentrations in a wetland soil: Implications for CO2 and CH4 fluxes. Environ. Sci. Technol. 45:3393–3399. doi:10.1021/es103540k

Elberling, B., H.H. Christiansen, and B.U. Hansen. 2010. High nitrous oxide production from thawing permafrost. Nat. Geosci. 3:332–335. doi:10.1038/ngeo803

Elberling, B., and L.R. Damgaard. 2001. Micro-scale measurements of oxygen diffusion and consumption in subaqueous sulfide tailings. Geochim. Cosmochim. Acta 65:1897–1905. doi:10.1016/S0016-7037(01)00574-9

Fan, Y., Q. Zhu, R.C. Aller, and D.C. Rhoads. 2011. An in situ multispectral imaging system for planar optodes in sediments: Examples of high-resolution seasonal patterns of pH. Aquat. Geochem. 17:457–471. doi:10.1007/s10498-011-9124-5

Fischer, J.P., and F. Wenzhöfer. 2010. A novel planar optode set-up for concurrent oxygen and light field imaging: Application to a benthic phototrophic community. Limnol. Oceanogr. Methods 8:254–268. doi:10.4319/lom.2010.8.254

Frederiksen, M.S., and R.N. Glud. 2006. Oxygen dynamics in the rhizosphere of Zostera marina: A two-dimensional planar optode study. Limnol. Oceanogr. 51:1072–1083. doi:10.4319/lo.2006.51.2.1072

Glud, R.N., K. Jensen, and N.P. Revsbech. 1995. Diffusivity in surficial sediments and benthic mats determined by use of a combined N2O–O2 microsensor. Geochim. Cosmochim. Acta 59:231–237. doi:10.1016/0016-7037(94)00321-C

Glud, R.N., N.B. Ramsing, J.K. Gundersen, and I. Klimant. 1996. A new tool for fine scale measurements of two-dimensional O2 distribution in benthic communities. Mar. Ecol. Prog. Ser. 140:217–226. doi:10.3354/meps140217

Glud, R.N., F. Wenzhöfer, A. Tengberg, M. Middelboe, K. Oguri, and H. Kitazato. 2005. Distribution of oxygen in surface sediments from central Sagami Bay, Japan: In situ measurements by microelectrodes and planar optodes. Deep Sea Res., Part I 52:1974–1987. doi:10.1016/j.dsr.2005.05.004

Gundersen, J.K., N.B. Ramsing, and R.N. Glud. 1998. Predicting the signal of oxygen micro-sensors from physical dimensions, temperature, salinity, and oxygen concentrations. Limnol. Oceanogr. 43:1932–1937.

Højberg, O., N.P. Revsbech, and J.M. Tiedje. 1994. Denitrification and nitrous oxide accumula-tion in soil aggregates as measured with a combined microsensor for oxygen and nitrous oxide. Soil Sci. Soc. Am. J. 58:1691–1698. doi:10.2136/sssaj1994.03615995005800060016x

Jacinthe, P.A., and W.A. Dick. 1996. Use of silicone tubing to sample nitrous oxide in the soil atmosphere. Soil Biol. Biochem. 28:721–726. doi:10.1016/0038-0717(95)00176-X


Jørgensen, C.J., and B. Elberling. 2012. Effects of flooding-induced N2O production, con-sumption and emission dynamics on the annual N2O emission budget in wetland soil. Soil Biol. Biochem. 53:9–17. doi:10.1016/j.soilbio.2012.05.005

Jørgensen, C.J., S. Struwe, and B. Elberling. 2012. Temporal trends in N2O flux dynamics in a Danish wetland: Effects of plant-mediated gas transport of N2O and O2 following chang-es in water level. Global Change Biol. 18:210–222. doi:10.1111/j.1365-2486.2011.02485.x

Kammann, C., L. Grunhage, and H.-J. Jager. 2001. A new sampling technique to monitor con-centrations of CH4, N2O and CO2 in air at well-defined depths in soils with varied water potential. Eur. J. Soil Sci. 52:297–303. doi:10.1046/j.1365-2389.2001.00380.x

Klimant, I., G. Holst, and M. Kühl. 1997a. A simple fiber-optic sensor to detect the penetra-tion of microsensors into sediments and other biological materials. Limnol. Oceanogr. 42:1638–1643. doi:10.4319/lo.1997.42.7.1638

Klimant, I., V. Meyer, and M. Kühl. 1995. Fiber-optic oxygen microsensors, a new tool in aquatic biology. Limnol. Oceanogr. 40:1159–1165. doi:10.4319/lo.1995.40.6.1159

Kühl, M. 2005. Optical microsensors for analysis of microbial communities. Methods Enzymol. 397:166–199. doi:10.1016/S0076-6879(05)97010-9

Kühl, M., and N.P. Revsbech. 2001. Biogeochemical microsensors for boundary layer studies. In: B.P. Boudreau and B.B. Jørgensen, editors, The benthic boundary layer. Oxford Univ. Press, Oxford, UK. p. 180–210.

Larsen, M., S.M. Borisov, B. Gunwald, I. Klimant, and R.N. Glud. 2011. A simple and inexpen-sive, high-resolution color-ratiometric planar optode imaging approach: Application to oxygen and pH sensing. Limnol. Oceanogr. Methods 9:348–360. doi:10.4319/lom.2011.9.348

Liengaard, L., L.P. Nielsen, N.P. Revsbech, B. Elberling, A. Prieme, A. Prast, and M. Kühl. 2011. Effects of flooding cycles in the Pantanal on the turnover of soil nitrogen pools and emission of N2O. Biogeosci. Discuss. 8:5991–6030. doi:10.5194/bgd-8-5991-2011

Lloyd, D., K. Thomas, D. Price, B. O’Neil, K. Oliver, and T.N. Williams. 1996. A membrane-inlet mass spectrometer miniprobe for the direct simultaneous measurement of mul-tiple gas species with spatial resolution of 1 mm. J. Microbiol. Methods 25:145–151. doi:10.1016/0167-7012(96)00011-5

Markfoged, R., L.P. Nielsen, T. Nyord, L.D.M. Ottosen, and N.P. Revsbech. 2011. Transient N2O accumulation and emission caused by O2 depletion in soil after liquid manure injection. Eur. J. Soil Sci. 62:541–550. doi:10.1111/j.1365-2389.2010.01345.x

Oguri, K., H. Kitazato, and R.N. Glud. 2006. Platinum octaetylporphyrin based planar op-todes combined with UV-LED excitation light source: An ideal tool for high resolu-tion O2 imaging in O2 depleted environments. Mar. Chem. 100:95–107. doi:10.1016/j.marchem.2005.11.005

Precht, E., U. Franke, L. Polerecky, and M. Huettel. 2004. Oxygen dynamics in permeable sediments with a wave-driven pore water exchange. Limnol. Oceanogr. 49:693–705. doi:10.4319/lo.2004.49.3.0693

Revsbech, N.P. 2005. Analysis of microbial communities with electrochemical microsen-sors and microscale biosensors. Methods Enzymol. 397:147–166. doi:10.1016/S0076-6879(05)97009-2

Revsbech, N.P., J.P. Jacobsen, and L.P. Nielsen. 2005. Nitrogen transformations in microen-vironments of river beds and riparian zones. Ecol. Eng. 24:447–455. doi:10.1016/j.eco-leng.2005.02.002

Sheppard, S.K., and D. Lloyd. 2002. Direct mass spectrometric measurement of gases in soil monoliths. J. Microbiol. Methods 50:175–188. doi:10.1016/S0167-7012(02)00025-8

Thomas, K.L., and D. Lloyd. 1995. Measurement of denitrification in estuarine sedi-ment using membrane inlet mass spectrometry. FEMS Microbiol. Ecol. 16:103–114. doi:10.1111/j.1574-6941.1995.tb00274.x

Tiedje, J.M., A.J. Sexstone, T.B. Parkin, and N.P. Revsbech. 1984. Anaerobic processes in soil. Plant Soil 76:197–212. doi:10.1007/BF02205580

Volkenborn, N., L. Polerecky, D.S. Wethey, and S.A. Woodin. 2010. Oscillatory porewater bioadvection in marine sediments induced by hydraulic activities of Arenicola marina. Limnol. Oceanogr. 55:1231–1247. doi:10.4319/lo.2010.55.3.1231


Wenzhöfer, F., and R.N. Glud. 2004. Small-scale spatial and temporal variability in benthic O2 dynamics of coastal sediments: Effects of fauna activity. Limnol. Oceanogr. 49:1471–1481. doi:10.4319/lo.2004.49.5.1471

Wenzhöfer, F., O. Holby, R.N. Glud, H.K. Nielsen, and J.K. Gundersen. 2000. In situ micro-sensor studies of a shallow water hydrothermal vent at Milos (Greece). Mar. Chem. 69:43–54.

Yamulki, S., and S.C. Jarvis. 1999. Automated chamber technique for gaseous flux mea-surements: Evaluation of a photoacoustic infrared spectrometer–trace gas analyzer. J. Geophys. Res. 104(D5):5463–5469. doi:10.1029/1998JD100082

5. Oxygen tension in the maxillary sinus

K. Aaness, L.F. Rickelt, H.K. Johansen, C. von Buchwald, T. Pressler, N. Høiby, and

P.Ø. Jensen:

Decreased mucosal oxygen tension in the maxillary sinuses in patient s with cystic

fibrosis

Journal of Cystic Fibrosis 10: 114–120 (2011)

DOI: 10.1016/j.jcf.2010.12.002

Journal of Cystic Fibrosis 10 (2011) 114–120www.elsevier.com/locate/jcf

Original Article

Decreased mucosal oxygen tension in the maxillary sinuses in patients withcystic fibrosis

Kasper Aanaesa,⁎, Lars Fledelius Rickeltb, Helle Krogh Johansenc, Christian von Buchwalda,Tacjana Presslerd, Niels Høibyc, Peter Østrup Jensenc

a Department of Otolaryngology – Head & Neck Surgery, Rigshospitalet, Denmark. Blegdamsvej 9, DK-2100 Copenhagen, Denmarkb Marine Biological Laboratory, Department of Biology, University of Copenhagen, Strandpromenaden 5, DK-3000 Helsingør, Denmark

c Department of Clinical Microbiology, Rigshospitalet, Denmark. Blegdamsvej 9, DK-2100 Copenhagen, Denmarkd Copenhagen CF Centre, Pediatric Pulmonary Service, Department of Pediatrics, Rigshospitalet, Denmark. Blegdamsvej 9, DK-2100 Copenhagen, Denmark

Received 4 October 2010; received in revised form 29 November 2010; accepted 1 December 2010

Abstract

Background: Pseudomonas aeruginosa in the sinuses plays a role in the lungs in cystic fibrosis (CF) patients, but little is known about the sinusenvironment where the bacteria adapt. Anoxic areas are found in the lower respiratory airways but it is unknown if the same conditions exist in thesinuses.Methods: The oxygen tension (pO2) was measured, using a novel in vivo method, in the maxillary sinus in a group of 20 CF patients.Results: The CF patients had a significant lower pO2 on the mucosa but not in the sinus lumen as compared with a control group of non-CFpatients. Anoxic conditions were found in 7/39 (18%) of the sinuses from where we cultured P. aeruginosa, Stenotrophomonas maltophilia and/orcoagulase negative staphylococci.Conclusion: These findings support our hypothesis that P. aeruginosa can adapt or acclimate to the environment in the lungs, during growth inanoxic parts of the paranasal sinuses.© 2010 European Cystic Fibrosis Society. Published by Elsevier B.V. All rights reserved.

Keywords: Pseudomonas aeruginosa; Cystic fibrosis; Maxillary sinuses; Oxygen tension; Sinus surgery; Catheter optode

1. Background

Cystic fibrosis (CF) is a genetic disease caused by mutationsin the gene for the CF transmembrane conductance regulator(CFTR) protein, resulting in altered chloride transport, in mul-tiple organs, which comprises the mucociliary function andrenders the mucus viscosity and the mucosa more susceptibleto infections [1]. Nasal and sinus inflammation is a frequentcondition in patients with CF, commonly leading to findingsas congestion, mucopurulent material in the nose cavity, pol-

⁎ Corresponding author. Department of Otolaryngology – Head & NeckSurgery, Rigshospitalet, F2071, Copenhagen University Hospital, Blegdamsvej9, DK-2100 Copenhagen, Denmark. Tel.: +45 35 45 8691; fax: +45 35 45 26 90.

E-mail address: [email protected] (K. Aanaes).

1569-1993/$ - see front matter © 2010 European Cystic Fibrosis Society. Publishedoi:10.1016/j.jcf.2010.12.002

yposis, abnormalities of the lateral nasal wall, mucoceles, andhypoplasia of the paranasal sinuses [2]. Such conditions mightaffect the O2 exchange and the O2 content in the sinuses andthus the microenvironment of its bacterial community.

The gas exchange in the maxillary sinus takes place via theostium and the mucosa that absorbs and consumes O2. Thediffusion through the ostium obeys simple physical laws anddepends on the patency of the ostium, the volume of the sinusand the respiratory work in the nose.

The absorption of O2 from the sinuses as well as the changein the O2 content in the paranasal sinuses after obstruction of theostium, are thought to be important factors in the pathogenesisof sinusitis. There is a significant decrease in the luminal O2

content in persons with acute sinusitis or allergic rhinitis com-pared with patients without symptoms [3,4]. Aust and Drettner[4] also found decreased O2 tension (pO2) in a group of patients

d by Elsevier B.V. All rights reserved.

http://dx.doi.org/10.1016/j.jcf.2010.12.002





115K. Aanaes et al. / Journal of Cystic Fibrosis 10 (2011) 114–120

with recurrent sinusitis and in a group of patients withobstructed ostia (all non-CF). They did not find pO2 to berelated to the presence or absence of antral pus or mucus.Carenfelt and Lundberg [5] reported a pO2 close to zero in somepurulent sinus secretions with Streptococcus pneumoniae orHaemophilus influenza compared with 96 mmHg in non-purulent secretions. They conclude that the gas compositionin the sinuses influences the bacterial growth, as well as thebactericidal function of the granulocytes, and that the O2

levels also might be of importance to the mucociliary activity.Purulent sinusitis should thus be treated by drainage of the sinuscavity, not only to reduce the debris, but also to improve thecondition for the local host defense mechanisms of the sinus.

The limited research that has been published regarding pO2

in the sinus deals with non-CF patients. CF patients have adifferent inflammatory response in their sinus mucosa as com-pared with non-CF related rhinosinusitis [6,7], and the O2

conditions and their importance for rhinosinusitis in thesepatients are unknown. However, it has been shown that hypoxiacontributes to a reduction of cell surface CFTR [8], which mighthave an additional negative impact in the sinuses of CF patients.

Chronic Pseudomonas aeruginosa lung infection developsin most patients with CF. Once the bacteria have establisheda chronic infection in the lungs, they cannot be eradicated.P. aeruginosa is a facultative anaerobe, which can proliferateand adapt to anaerobic environments, when thick layers ofmucoid exopolysaccharide surround the bacteria (biofilm).Such biofilms are known to exist in the sinuses [6] and inthe lungs, and the presence of such biofilms limit the diffusivesupply of O2 which then can lead to total O2 depletion inthe sputum [9]. In addition, numerous of polymorphonuclearleucocytes (PMNs) in the infected bronchi exhibit strong con-sumption of O2 for production of reactive oxygen species(ROS) [9–13].

Under such anoxic conditions P. aeruginosa can achieveanaerobic growth either based on denitrification using nitrateas a terminal electron acceptor or by fermentation of arginine[9–12,14].

It has been indicated that P. aeruginosa responds to hypoxicmucus with an upregulation of alginate production, which maydecrease the susceptibility to some antibiotics. Also, noveltherapies for CF include removal of hypoxic mucus plaques andthe use of antibiotics effective against P. aeruginosa adaptedto anaerobic environments [10].

In the mucus with low pO2, P. aeruginosa can makealterations due to mutations caused by the ROS or conversion inphenotype, e.g. becoming mucoid and developing antibioticresistance, in order to adapt to different focal niches [15,16].Studies of P. aeruginosa suggest a correlation between nutrientlimitation, growth rates and conversion to mucoidy [17], andwe speculate that the same applies for anoxic conditions.

The upper airways are shown to be a gateway for acquisi-tion of opportunistic bacteria like P. aeruginosa, where theparanasal sinuses can act as a reservoir. Concordant genotypeshave been found in the sinuses and in the lungs [18]. Ourhypothesis is that P. aeruginosa adapts to the environment inthe paranasal sinuses where some bacteria mutate or converse

their phenotype [19]. This results in bacterial strains that are fitfor spreading to the lungs, where they can maintain an ongoingdeleterious infection.

Our present knowledge about P. aeruginosa is primarilyrelated to the lungs. The pattern of inflammation differs inthe sinus from the findings in the lower airway specimensof chronically infected patients with CF [20]. There is a Th2dominated response in the lungs, while there is a significantlyreduced PMN response in the sinuses, probably due to thehigher concentration of IgA in the sinuses then in the lungs[21]. The latter statement combined with the fact that anti-biotics more difficultly penetrates and achieves therapeuticlevels in the sinus cavity than in the lungs, are some of thereasons why the immune response in the sinuses is less chal-lenging than in the lungs. Based on the above mentionedknowledge, it is important to determine whether P. aeruginosacan adapt to anaerobic environments in the sinuses. In thisstudy we determined the pO2 in the maxillary sinuses in CFpatients never infected, intermittently infected and chronicallyinfected with P. aeruginosa in their lungs as a first step towardsdetermining under which conditions P.aeruginosa adapts inthe sinuses.

2. Materials and methods

The patients were recruited at the CF Centre in Copenhagen.The CF-diagnosis was based on characteristic clinical features,abnormal sweat electrolytes and the genotype. CF patientsplanned for sinus surgery were invited to participate in ourstudy. As a control group, we asked non-CF patients whounderwent surgery under general anesthesia because their nasalseptum needed correction. Patients suffering of acute or chronicrhinosinusitis were excluded from the control group. All invitedCF-patients accepted to be included in the study, while twopatients who were invited to join the control group denied.

The CF-patients follows a routine with monthly medicalexaminations including lung function tests and cultures takenfrom the lower airways. At least every third month bloodsamples are taken for measurements including antibodiesagainst P. aeruginosa (precipitating antibodies).

No standardized guidelines comprising criteria and motiva-tions for sinus surgery in CF patients exist [22]. At our insti-tution we select patients based on the following criteria indescending order:

1. Patients with declining lung function despite intensiveantibiotic-chemotherapy and/or increasing antibodies againstGram-negative bacteria despite negative bacteriology in theirsputum samples. Especially patients with unknown focus andincreasing antibodies against P. aeruginosa, Achromobacterxylosoxidans or Burkholderia multivorans are given priority.The majority of patients has been operated due to criteria 1,but may also fulfill criteria 2 or 3 as well.

2. Patients who have undergone lung transplantation within thelast year.

3. Patients with severe symptoms of rhinosinusitis according toEPOS guidelines [22].

116 K. Aanaes et al. / Journal of Cystic Fibrosis 10 (2011) 114–120

2.1. Ethics

All the measurements were done during anesthesia, whichthe patients underwent for other reasons. The measurementswere done through the natural ostia to the maxillary sinus, so nopermanent damage was done to the control group, and the CFpatients all had their sinuses opened during surgery afterwards.The study was approved by the local ethics committee (H-A-2008-141), and all patients gave informed consent. In patientsb18 years of age, consent was also obtained from their parents.

2.2. Preparation of optodes

The pO2 in the sinus was measured with a new type ofcatheter O2 optode. See [23] for a recent review of fiber-opticO2 sensor technology. The optode was manufactured from anEndonasal Suction Tip (EST)(Medioplast, Malmö, Sweden,3.0×150 mm 11 G) and a 3 m length of a plastic PMMA opticalfiber (POF, step index, 2 mm diameter with a polymethylmethacrylate core and a fluorinated polymer cladding; LaserComponents GmbH, Olching, Germany). The end of thePOF was glued inside the EST with two components SuperEpoxy (Plastic Padding, Henkel Technologies) with ~1 mmprotruding from the tip of the EST. The other part was coveredwith 2.4 mm shrink tubing (Low Shrink Temperature (LSTT)polyolefin tubing; RS Components, Denmark) and a SMA-connector (Laser Components GmbH, Germany, SMA-B2100)was glued at the end. Both ends were subsequently polisheddown to the metal. At the EST end, a disc (2 mm diameter) ofa 0.125 mm thick transparent Mylar®-foil (Goodfellow, UK)was glued on the fiber tip with 1:1 diluted contact glue (BostikKontaktlim A3). Subsequently, the foil was covered with athin layer of a luminescent O2 indicator layer composed of asolution of 1 g polystyrene (Goodfellow), 500 mg TiO2, and25 mg Pt(II) meso Tetra(pentafluorophenyl)porphine (PtTFPP)(Frontier Scientific, USA), in 29 g CHCl3.

3. Calibration and measurements

Prior to use, the catheter O2 optodes were sterilized in aplasma oven (Sterrad 100 S), which enables sterilization at lowtemperatures using hydrogen peroxide. Calibration and mea-surements were done with the catheter O2 optodes connectedto a fiber-optic O2 meter (Fibox 3, Minisensor Oxygen Meter,Presens Precision Sensing GmbH, Regensburg, Germany)[24].The optode meters used in this study measures the luminescencelifetime with a phase-modulation technique [22], where the O2

dpendent luminescence lifetime of the PtTFPP indicator, τ, canbe calculated from the measured phase angle shift, Φ, betweenthe sinusoidal intensity modulated (at frequency, fmod) exci-tation and emission signals:tan(Φ)=2π ⋅ fmod ⋅τ

The sensor response can be described by a modified Stern-Volmer equation (x):

tanðΦÞ = tanðΦ0Þ 1−α1þKSV O2½ � + α

� �ð1Þ

where, Ф0 is the phase angle of the indicator in the absenceof O2, Φ is the phase angle of the indicator at a given O2

concentration, KSV is a characteristic temperature dependentquenching coefficient of the immobilized indicator, and α is thenon-quenchable fraction of the indicator (α=0.11 in this study)in the carrier foil. For a given mixture of indicator and matrixmaterial, α is usually constant over the dynamic range (x). TheStern-Volmer constant (Ksv) was calculated from calibrationvalues measured in anoxic and atmospheric air. The Φ0 wasmeasured in an anoxic bench and the ΦSat was measured in athermostated climate room with atmospheric air temperature of37 °C. The values were not corrected for air pressure:

KSV =1−α

tanðΦSatÞtanðΦ0Þ −α

−1

0@

1A 1

O2½ �Satð2Þ

The pO2 for a given experimental phase angle measurementwas calculated according to:

O2½ � = 1−αtanðΦÞtanðΦ0Þ−α

−1

0@

1A 1

KSVð3Þ

3.1. Measurement

The patients (CF and non-CF) were anaesthetized using anoral tracheal tube and intravenous solutions. The patients werepre-oxygenated with 100% O2 before intubation. We waitedat least 15 min before onset of the O2-measurements. Duringanesthesia, all patients were given ~40% O2 of supplementaryO2. At the beginning of the general anesthesia but prior tosurgery the catheter was introduced, through the nasal cavity,into the maxillary sinus through the natural maxillary ostium(Figs. 1 and 2). The pO2 in the maxillary sinuses was con-tinuously recorded, first in the lumen of the sinus, and thenwith the catheter touching the mucosa in the bottom of thesinus. It was noted if the patient had an occluded or a patentmaxillary ostium. Temperature, pH and the volume of thesinuses were not recorded. After the pO2 was measured, aregularly sinus surgery was performed, where each maxillarysinus was cultured separately.

All CF patients were CT scanned within 6 weeks prior tosurgery and staged according to the Lund McKay stagingsystem, where every sinus scores from 0 (no opacification) to 2(total opacified).

3.2. Statistical methods and definitions

Statistical significance was evaluated by unpaired t-test forobservations with parametric data. Catagorical data wereanalysed by Fisher's exact test. A p-valueb0.05 was consideredstatistically significant. The tests were performed with Prism4.0c (GraphPad Software, La Jolla, California, USA).

We define anoxia as pO2 valuesb0.8 mmHg.

Fig. 1. The catheter with the incorporated optode.


4. Results

4.1. Patients

The pO2 was measured in 20 CF patients, representing 39maxillary sinuses (one measurement failed). The mean ageof the CF patients was 18 years (range 6–39 years.), 12 of thepatients were under 18 years of age. One patient had previouslyundergone sinus surgery but his ostia to the maxillary sinuseswere totally obstructed.

Sixteen of the 20 CF patients were categorized as intermit-tently infected [25], two were chronically infected and two werelung transplanted (former chronically P.aeruginosa infected).

Four out of 40 CF-sinuses had a Lund McKay score [26] of 0(no opacification) seen on the CT scan. The average score of allsinuses was 1.2. The control group contained 11 patientsrepresenting 22 sinuses. The mean age was 31 years ( range 22–66 years.)

Fig. 2. The catheter on its way t

4.2. Microbiology

In nine CF-patients representing 13 sinuses, mucoid and/ornon-mucoid P. aeruginosa were found. In only one maxillarysinus no bacteria were detected. Besides P. aeruginosa, theresults from the cultivation were as follows: 3 sinuses containedStenotrophomonas maltophilia, 3 A. xylosoxidans, 23 coagulasenegative staphylococci (CNS), 3 Staphylococcus aureus, 3Haemophilus influenzae, 1 Enterobacter cloacae and 3 sinuscontained Candida albicans. (Several sinuses had growth ofmore than one microorganism).

4.3. pO2 measurements

A significantly (pb0.03) lower pO2 on the maxillary mucosawas found in CF patients, as compared with the control group(Fig. 3). There was no difference between the patients aboveand under 18 years of age within the CF group.

o the right maxillary sinus.

image of Fig.�2

Fig. 3. PO2 on the mucosa of all maxillary sinuses (t-test) Pb0.0263.

Fig. 4. A typical coronal CT scan of a CF patient. A right total opacifiedmaxillary sinus is seen. The left maxillary sinus shows a little air (black)medially in the top between the natural ostium and the middle turbinate.


In contrary to what we expected, there were only twomeasurements in the lumen of the sinuses below 20 mmHg O2

(10.0 and 13.8 mmHg) and none with an anoxic lumen. Therewas no correlation between the types of bacteria cultured fromthe sinuses, the presence of pus or visible enlarged ostia, orthe pO2 content in the maxillary lumen. The pO2 was notsignificantly lower in the CF patients' lumen compared to thecontrols either.

We found anoxic conditions on the mucosa in five CF-patients, comprising three patients with unilateral anoxia andtwo patients with bilateral anoxia. Thus, anoxia was found at ahigher frequency in the CF patients than in non-CF patients(pb0.02). Four of the sinuses harbored P. aeruginosa and/or S.maltophilia but we also diagnosed total anoxia in three sinusesthat only harbored CNS.

In eight sinuses in five patients no macroscopically pus wasfound. These sinuses were not anoxic. However, we also foundhigh pO2 in sinuses with pus and P. aeruginosa.

In three sinuses from three patients a large visible naturalostium to the maxillary sinus existed in accordance with highpO2 values in the lumen (average of 128.3 mmHg). Unexpect-edly, anoxia on the mucosa was demonstrated in one ofthese patients. No significant difference was found between theluminal pO2.

5. Discussion

Little is known about how O2 influences the immune systemand the bacterial community in the sinuses, and to our knowl-edge no research has previously been done in relation to eitherthe pO2 lumen or the pO2 mucosa in the sinuses of CF patients.We found a significantly lower pO2 on the mucosa in CFpatients than in the healthy control group, and also found some

CF patients with anoxic conditions. However, we suspect thatour study has a tendency to overestimate the pO2 values. Thereasons are: 1: If the sensitive optode comes in contact with a lotof blood it will measure the pO2 in the blood; 2: The fact that allpatients were given supplementary O2 which could increase thepO2 in the sinuses; 3: We only measured a little area in thebottom of the sinus mucosa, from which we cannot exclude thepossibility that other areas of the mucosa had lower or higherpO2. Our measurements were done in the maxillary sinuses,but we expect our findings to be representative for the otherparanasal sinuses as well.

No follow up study was made, but additional measurementsin patients with anoxic conditions could have determinedwhether the pO2 increased after surgery. In principle, it ispossible to re-measure some patients only using local anes-thetics, as long as the sinus ostium exceeds 4 mm and the patientis capable of lying still for about two minutes while the optodeis in contact with the mucosa. Due to unfortunate circumstancesas severe nasal septum deviation, young age, and two patientsdropping out of our follow up study, we have not, so far, hadpatients for re-measuring their pO2.

The CF-sinus-anatomy varies, probably due to the associatedchronic rhinosinusitis when the sinus development takes place.Consequently the natural maxillary ostium is often obstructed.However, among the minority a very enlarged ostium is ob-served. Opacified (blurred) sinuses and viscous mucus compli-cates the O2 diffusion in the sinuses (Fig. 4). CF patients oftenhave opacified sinuses diagnosed by a CT-scan. In physicalexamination or during sinus surgery, the nose and sinusesoften present clinical signs of inflammation and infection, butthis is less commonly related to rhinosinusitis symptoms thanin non-CF patients. We have shown somemaxillary sinuses with

image of Fig.�3

image of Fig.�4


aerobic areas and some with anaerobic areas. We hypothesizethat some sinuses may contain aerobic as well as anaerobic areas,probably hosting two different niches of P.aeruginosa, as seenin CF-sputum [9], which could enter the lower airways. In theanoxic areas, we suspect a lower growth rate and an adaption tothe environment as seen in the endobronchial mucus [9]. Ourbacterial findings and the observed O2 depletion confirms that P.aeruginosa, as well as other bacteria, can adapt to and live underanaerobic conditions in the maxillary sinuses. We suspect thatthe sinus epithelium, the bacteria and the PMNs along with otherinflammatory mechanisms competes to use the limited amountof O2. These results support the hypothesis that P. aeruginosacan adapt to and change phenotype in the sinuses where there isO2 depletion, a lover antibiotic concentration and a less activeimmune system/PMN response than in the lungs.

No significant difference was found in the lumen pO2, whichwe suspect is due to O2 consumption primarily taking placein the mucus/mucosa, and due to the fact that O2 in the lumenrepresent a large O2 reservoir compared with the mucus,wherein O2 is less soluble as in air. Full depletion of lumen O2 isthus unlikely.

No correlation between the pO2 values on the left and rightsinuses was shown. This compares with the fact that we oftenfound different microbiological flora in the two maxillarysinuses, and that the anatomic findings in the two sides oftendiffer. This supports the theory that bacteria, the size of thesinus, and the size of the ostium influence the pO2.

Since the function of the CFTR gene is reduced by hypoxia[8] and because it has been suggested that the immune system isless functional in O2 depleted areas [5], we speculate whetheranoxia itself should be an indication for sinus surgery, and ifharmless bacteria, such as CNS, can induce unfavorable con-ditions with anoxic pus. We have not been able to show asignificant difference between the pO2 in the maxillary lumen,but we hypothesize that surgery gives the possibility for sinusirrigations with saline and antibiotics and hence a better pO2

besides the removal of pus and bacteria.

6. Conclusion

We present a new method for determining the pO2in themaxillary sinuses. In this first study, it was only used duringsinus surgery, but the new method can, in some cases, be usedafter surgery without anesthesia. We show that patients with CFhave a lower pO2 on the mucosa as compared to a control group.Some CF patients have anoxic conditions, probably due to themucosa's and the bacteria's O2 consumption. In contrast, therewas no anoxia in the sinus lumen. Both P. aeruginosa andCNS can thrive under anoxic conditions in the sinuses. Thesefindings support our hypothesis that P. aeruginosa can adapt tothe environment, mutate or converse phenotype in the paranasalsinuses. In addition, hypoxia influences the regulation of theCFTR protein, may facilitate biofilm formation and may in-fluence the immune system. It is not known, whether surgeryalone improves the anoxia in the mucosa, but surgery is oftennecessary if one should attempt to remove the anoxic secretionsbuy suction and nasal irrigations.

Acknowledgements

We would like to thank Professor Michael Kühl for helpfulsuggestions and corrections. The work was thankfullysupported by Candys Foundation.

References

[1] Gysin C, Alothman GA, Papsin BC. Sinonasal disease in cystic fibrosis:clinical characteristics, diagnosis, and management 1. Pediatr Pulmonol2000 Dec;30(6):481–9.

[2] Mak GK, Henig NR. Sinus disease in cystic fibrosis 1. Clin Rev AllergyImmunol 2001 Aug;21(1):51–63.

[3] Aust R, Stierna P, Drettner B. Basic experimental studies of ostial patencyand local metabolic environment of the maxillary sinus. Acta OtolaryngolSuppl 1994;515:7–10.

[4] Aust R, Drettner B. Oxygen tension in the human maxillary sinus undernormal and pathological conditions 1. Acta Otolaryngol 1974 Sep;78(3–4):264–9.

[5] Carenfelt C, Lundberg C. Purulent and non-purulent maxillary sinussecretions with respect to pO2, pCO2 and pH 1. Acta Otolaryngol 1977Jul;84(1–2):138–44.

[6] Hekiert AM, Kofonow JM, Doghramji L, Kennedy DW, Chiu AG, PalmerJN, et al. Biofilms correlate with TH1 inflammation in the sinonasal tissueof patients with chronic rhinosinusitis 1. Otolaryngol Head Neck Surg2009 Oct;141(4):448–53.

[7] Sobol SE, Christodoulopoulos P, Manoukian JJ, Hauber HP, Frenkiel S,Desrosiers M, et al. Cytokine profile of chronic sinusitis in patients withcystic fibrosis 1. Arch Otolaryngol Head Neck Surg 2002 Nov;128(11):1295–8.

[8] Guimbellot JS, Fortenberry JA, Siegal GP, Moore B, Wen H, Venglarik C,et al. Role of oxygen availability in CFTR expression and function 1. Am JRespir Cell Mol Biol 2008 Nov;39(5):514–21.

[9] Bjarnsholt T, Jensen PO, Fiandaca MJ, Pedersen J, Hansen CR, AndersenCB, et al. Pseudomonas aeruginosa biofilms in the respiratory tract ofcystic fibrosis patients. Pediatr Pulmonol 2009 Jun;44(6):547–58.

[10] Worlitzsch D, Tarran R, Ulrich M, Schwab U, Cekici A, Meyer KC, et al.Effects of reduced mucus oxygen concentration in airway Pseudomonasinfections of cystic fibrosis patients 1. J Clin Invest 2002 Feb;109(3):317–25.

[11] Koch C, Hoiby N. Pathogenesis of cystic fibrosis 1. Lancet 1993 Apr24;341(8852):1065–9.

[12] YangL,Haagensen JA, Jelsbak L, JohansenHK, Sternberg C,HoibyN, et al.In situ growth rates and biofilm development of Pseudomonas aeruginosapopulations in chronic lung infections 1. J Bacteriol 2008 Apr;190(8):2767–76.

[13] Kolpen M, Hansen CR, Bjarnsholt T, Moser C, Christensen LD, van GM.Polymorphonuclear leucocytes consume oxygen in sputum from chronicPseudomonas aeruginosa pneumonia in cystic fibrosis 1. Thorax 2010Jan;65(1):57–62.

[14] Vander WC, Pierard A, Kley-Raymann M, Haas D. Pseudomonasaeruginosa mutants affected in anaerobic growth on arginine: evidence fora four-gene cluster encoding the arginine deiminase pathway. J Bacteriol1984 Dec;160(3):928–34.

[15] HoffmannN, Rasmussen TB, Jensen PO, Stub C, HentzerM,Molin S, et al.Novel mouse model of chronic Pseudomonas aeruginosa lung infectionmimicking cystic fibrosis. Infect Immun 2005 Apr;73(4):2504–14.

[16] Ciofu O, Mandsberg LF, Bjarnsholt T, Wassermann T, Hoiby N. Geneticadaptation of Pseudomonas aeruginosa during chronic lung infection ofpatients with cystic fibrosis: strong and weak mutators with heterogeneousgenetic backgrounds emerge in mucA and/or lasR mutants. Microbiology2010 Apr;156(Pt 4):1108–19.

[17] Terry JM, Pina SE, Mattingly SJ. Role of energy metabolism in conversionof nonmucoid Pseudomonas aeruginosa to the mucoid phenotype. InfectImmun 1992 Apr;60(4):1329–35.


[18] Mainz JG, Naehrlich L, Schien M, Kading M, Schiller I, Mayr S, et al.Concordant genotype of upper and lower airways P aeruginosa and Saureus isolates in cystic fibrosis. Thorax 2009 Jun;64(6):535–40.

[19] Hansen SK, Johansen HK, von Buchwald C, Høiby N, Molin SZ.Diversification and evolution of Pseudomonas aeruginosa in the upperairways of intermittently colonized CF childrenPoster at the XII InternationalConference, Hanover, Germany, August 13–17 2009; 2009.

[20] Hartl D, Griese M, Kappler M, Zissel G, Reinhardt D, Rebhan C, et al.Pulmonary T(H)2 response in Pseudomonas aeruginosa-infectedpatients with cystic fibrosis 1. J Allergy Clin Immunol 2006 Jan;117(1):204–11.

[21] Johansen HK, Aanaes K, Pressler T, Hansen SK, Skov M, Nielsen KG,et al. P. aeruginosa (PA) sinusitis in CF — is IgA beneficial or harmful?Pediatr Pulmonol 2009(Suppl 32) Article no. Meeting Abstract 307:320.

[22] Thomas M, Yawn BP, Price D, Lund V, Mullol J, Fokkens W. EPOSPrimary Care Guidelines: European Position Paper on the Primary CareDiagnosis and Management of Rhinosinusitis and Nasal Polyps 2 1. PrimCare Respir J 2008 Jun;17(2):79–89.

[23] Kuhl M. Optical microsensors for analysis of microbial communities 1.Meth Enzymol 2005;397:166–99.

[24] Holst GA, Kühl M, Klimant I. Novel measuring system for oxygen micro-optodes based on a phase modulation technique. Proc SPIE 2005;2508:387–98, doi:10.1117/12.2217548 387 (1995).

[25] Johansen HK, Hoiby N. Seasonal onset of initial colonisation and chronicinfection with Pseudomonas aeruginosa in patients with cystic fibrosis inDenmark 1. Thorax Feb 1992;47(2):109–11.

[26] Lund VJ, Mackay IS. Staging in rhinosinusitus 35. Rhinology 1993Dec;31(4):183–4.

http://dx.doi.org/10.1117/12.2217548

6. Polymorphonuclear leukocytes in cystic fibrosis

K.N. Kragh, M. Alhede, P.Ø. Jensen, C. Moser, C.S. Jacobsen, S. Seier, S. Eickhardt, H.

Trøstrup, L. Christoffersen, H.-P. Hougen, L.F. Rickelt, M. Kühl, N. Høiby, and T.

Bjarnsholt:

Polymorphonuclear leukocytes restrict the growth of Pseudomonas aeruginosa in lungs

of cystic fibrosis patients

Infection and Immunity 82: 4477-4486 (2014)

DOI: doi:10.1128/IAI.01969-14

Polymorphonuclear Leukocytes Restrict Growth of Pseudomonasaeruginosa in the Lungs of Cystic Fibrosis Patients

Kasper N. Kragh,a Morten Alhede,a,b Peter Ø. Jensen,b Claus Moser,b Thomas Scheike,c Carsten S. Jacobsen,d Steen Seier Poulsen,f

Steffen Robert Eickhardt-Sørensen,a Hannah Trøstrup,b Lars Christoffersen,b Hans-Petter Hougen,g Lars F. Rickelt,e Michael Kühl,e,h,i

Niels Høiby,a,b Thomas Bjarnsholta,b

Department of International Health, Immunology and Microbiology, University of Copenhagen, Copenhagen, Denmarka; Department of Clinical Microbiology,Rigshospitalet, Copenhagen, Denmarkb; Department of Public Health, Section of Biostatistics, University of Copenhagen, Copenhagen, Denmarkc; Geological Survey ofDenmark and Greenland, Copenhagen, Denmarkd; Marine Biological Section, Department of Biology, University of Copenhagen, Helsingør, Denmarke; Department ofBiomedical Science, University of Copenhagen, Copenhagen, Denmarkf; Department of Forensic Medicine, University of Copenhagen, Copenhagen, Denmarkg; PlantFunctional Biology and Climate Change Cluster, University of Technology Sydney, New South Wales, Sydney, Australiah; Singapore Centre on Environmental Life ScienceEngineering, School of Biological Science, Nanyang Technological University, Singaporei

Cystic fibrosis (CF) patients have increased susceptibility to chronic lung infections by Pseudomonas aeruginosa, but theecophysiology within the CF lung during infections is poorly understood. The aim of this study was to elucidate the in vivogrowth physiology of P. aeruginosa within lungs of chronically infected CF patients. A novel, quantitative peptide nucleic acid(PNA) fluorescence in situ hybridization (PNA-FISH)-based method was used to estimate the in vivo growth rates of P. aerugi-nosa directly in lung tissue samples from CF patients and the growth rates of P. aeruginosa in infected lungs in a mouse model.The growth rate of P. aeruginosa within CF lungs did not correlate with the dimensions of bacterial aggregates but showed aninverse correlation to the concentration of polymorphonuclear leukocytes (PMNs) surrounding the bacteria. A growth-limitingeffect on P. aeruginosa by PMNs was also observed in vitro, where this limitation was alleviated in the presence of the alternativeelectron acceptor nitrate. The finding that P. aeruginosa growth patterns correlate with the number of surrounding PMNspoints to a bacteriostatic effect by PMNs via their strong O2 consumption, which slows the growth of P. aeruginosa in infectedCF lungs. In support of this, the growth of P. aeruginosa was significantly higher in the respiratory airways than in the conduct-ing airways of mice. These results indicate a complex host-pathogen interaction in chronic P. aeruginosa infection of the CF lungwhereby PMNs slow the growth of the bacteria and render them less susceptible to antibiotic treatment while enabling them topersist by anaerobic respiration.

Patients with the genetic disorder cystic fibrosis (CF) havehighly viscous endobronchial mucus and decreased mucocili-

ary clearance of the airways, which render them susceptible tochronic bacterial lung infections. Severe chronic Pseudomonasaeruginosa lung infections are the most common cause of morbid-ity and mortality in CF patients (1, 2). Lungs of CF patients withchronic P. aeruginosa infections are characterized by intrabron-chial mucus-imbedded aggregates of bacterial cells (biofilms) sur-rounded by high numbers of polymorphonuclear leukocytes(PMNs) (3, 4). Such PMN-surrounded biofilms can persist overthe lifetime of CF patients, despite an extensive inflammatory re-sponse and aggressive antibiotic treatment (5).

Slow growth within bacterial biofilms is recognized as a majorcontributor to high antibiotic tolerance because the effectivenessof the majority of antibiotics in clinical use decreases with lowbacterial metabolism (6, 7). Limited molecular oxygen (O2) canfurther increase the tolerance of P. aeruginosa biofilms for antibi-otics in vitro (8). Mucus in the conducting airways of chronicallyinfected CF patients is characterized by steep O2 concentrationgradients ranging from normoxic to anoxic conditions, and thecombination of slow diffusive transport and intense O2 consump-tion within the mucus leads to anoxia (9). This is accompanied byongoing denitrification, as evidenced by N2O production (10), thedenitrification biomarker OprF in sputum (11), antibodiesagainst nitrate reductase in serum (12), and upregulation of genesfor denitrification (13, 14). O2 gradients in the endobronchialsecretions are primarily a result of the O2 consumed by PMNs for

the formation of reactive oxygen species (15, 16) during respira-tory burst (15, 16) and for the production of nitric oxide (10) bynitric oxide synthase (54, 55). In addition, the fraction of O2 con-sumed by PMNs for aerobic respiration in endobronchial secre-tions from chronically infected CF patients is negligible (15, 16).The growth rate of P. aeruginosa is diminished by the low avail-ability of O2 (17); therefore, depletion of O2 within the mucus ofCF patients could serve as a limiting factor for the growth of P.aeruginosa and may contribute to the slow growth of P. aeruginosain the sputum of chronically infected CF patients (18). Alterna-tively, it has been suggested that isolates from chronically infectedCF patients may develop genetic adaptations that reduce thegrowth rate of the bacteria (19). Under these conditions, theabove-mentioned denitrification indicators point to anaerobicrespiration using nitrate as an alternative metabolic mode of P.

Received 24 April 2014 Returned for modification 11 July 2014Accepted 5 August 2014

Published ahead of print 11 August 2014

Editor: B. A. McCormick

Address correspondence to Thomas Bjarnsholt, [email protected].

Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.01969-14.

Copyright © 2014, American Society for Microbiology. All Rights Reserved.

doi:10.1128/IAI.01969-14

November 2014 Volume 82 Number 11 Infection and Immunity p. 4477– 4486 iai.asm.org 4477

on May 6, 2015 by K

B, K

øbenhavns U

niversitetsbibliotekhttp://iai.asm

.org/D

ownloaded from

http://dx.doi.org/10.1128/IAI.01969-14



http://iai.asm.org

http://iai.asm.org/

aeruginosa, resulting in lower energy yields and possibly lowergrowth rates in biofilms in the CF lung. However, only the in vitrogrowth of bacteria isolated from sputum samples has been studied(18), and the actual growth rates of P. aeruginosa within CF lungshave neither been mapped nor correlated to growth limitation invivo.

In this study, we developed a new quantitative peptide nucleicacid fluorescence in situ hybridization (PNA-FISH)-basedmethod that enabled mapping of the in vivo growth rates of P.aeruginosa for the first time. This method was used to investigatethe growth of P. aeruginosa in chronically infected CF lungs and inthe conducting and respiratory airways of P. aeruginosa-infectedmice. A significant negative correlation was observed between thegrowth rate and the abundance of PMNs surrounding the bacte-rial biofilm aggregates. A strong PMN-induced O2 limitation on P.aeruginosa growth was confirmed in vitro, while the bacterialgrowth limitation was alleviated in the presence of an alternativeelectron acceptor (nitrate) that enabled denitrification.

MATERIALS AND METHODSBacterial strains. The P. aeruginosa PAO1 wild-type strain used in all invitro experiments was obtained from the Pseudomonas Genetic StockCenter (strain PAO0001 [http://www.pseudomonas.med.ecu.edu]). TheEscherichia coli laboratory strain MG1655 was used for production ofspike-in DNA (20).

Ex vivo CF patient samples. Samples were obtained from explantedlungs of three CF patients chronically infected with P. aeruginosa (onemale and two females ranging from 30 to 42 years old). Tissue was col-lected following approval (KF-01278432) from the Danish Scientific Eth-ical Board. All three patients had undergone double-sided lung transplan-tation at the Copenhagen University Hospital, Rigshospitalet. Lung tissuesamples (n � 6 to 7 from each patient) were removed immediately afterextraction. Samples (n � 20) were fixed in phosphate-buffered salinecontaining 4% paraformaldehyde and embedded in paraffin. Sections (4�m thick) were cut using a standard microtome and fixed on glass slides.The slides were stored at 4°C until further analysis. In total, 59 bacterialbiofilms were analyzed.

Mouse model. To examine differences in bacterial growth rate as afunction of O2 partial pressure, we used a recently described model basedon the instillation of bacteria immobilized on small or large alginate beadsinto the respiratory or conducting zone of the lungs (21). Briefly, the P.aeruginosa strain PAO579 was propagated overnight at 37°C in ox broth(Statens Serum Institute, Denmark). The overnight culture was centri-fuged at 4°C and 4,400 � g and resuspended in 5 ml of serum bouillon(Department of Clinical Microbiology, Herlev Hospital, Denmark). Alg-inate (Protanal LF 10/60; FMC BioPolymer, Norway) was dissolved in0.9% NaCl to a concentration of 1% and sterile filtered. The bacterialculture was diluted 1:20 in the alginate solution. The solution was trans-ferred to a 10-ml syringe and placed in a syringe pump (model 3100;Graseby, United Kingdom), which fed the alginate to the encapsulationunit (Var J30; Nisco Engineering, Zurich, Switzerland). The alginate waspumped into a gelling solution of 0.1 M CaCl2 prepared in 0.1 M Tris-HClbuffer (pH 7.0), which was agitated with a magnetic stirrer (RCT Basic;IKA, Germany). The beads stabilized under continuous stirring in thegelling bath for 1 h and were then washed twice with 0.9% NaCl contain-ing 0.1 M CaCl2 before being transferred to 20 ml of 0.9% NaCl contain-ing 0.1 M CaCl2. Five milliliters of alginate beads was prepared, with meanbead diameters of 136 �m (range, 74 to 205 �m; n � 72) and 40 �m(range, 15 to 85 �m; n � 72) for large and small beads, respectively.

The number of bacteria in the alginate beads was determined by dis-solving the beads in 0.1 M citric acid buffer (pH 5.0) and plating thesupernatant for CFU counts.

BALB/c female mice (11 weeks old; Taconic Europe A/S, Denmark)were allowed to acclimatize for 1 week before use. The mice had free access

to chow and water and were handled by trained personnel. All animalexperiments were authorized by the National Animal Ethics Committee,Denmark.

Mice were anesthetized subcutaneously with a 1:1 mixture of etomi-date (Janssen, Denmark) and midazolam (Roche, Switzerland) (10 ml/kgbody weight) and then tracheotomized. Alginate beads embedded with P.aeruginosa PAO579 were installed in the left lung of BALB/c mice using abead-tipped needle. All mice received similar amounts of alginate beadsand P. aeruginosa cells (1 � 108 CFU/ml for both groups).

Eight mice (four with each size of aggregate beads) were examinedeach day; two mice from both groups were euthanized at 0, 1, 3, and 5 daysafter bacterial inoculation. The left lungs were fixed in a 4% (wt/vol)formaldehyde solution (VWR, Denmark). Bacterial growth was measuredin 34 aggregates (14 aggregates from respiratory airways and 20 aggregatesfrom conducting airways).

Quantitative PNA-FISH. Paraffin-embedded samples were deparaf-finized by treatment with xylene (twice for 5 min), 99.9% ethanol (EtOH;twice for 3 min), and 96% EtOH (twice for 3 min) and were then washedin MilliQ water three times for 3 min. A drop of a Texas Red-conjugatedPNA-FISH probe specific for P. aeruginosa 16S rRNA (AdvanDx, USA)was applied to the tissue section and then covered with a coverslip (22).Samples were incubated for 90 min at 55°C (AdvanDx Workstation, Ad-vanDx, USA). The coverslip was removed, and the slides were washed inwarm washing buffer at 55°C (AdvanDx, USA) for 30 min and then airdried in the dark. A drop of Vectashield mounting medium with 4=,6=-diamidino-2-phenylindole (DAPI; Vector, USA) was placed on top of theslide, which was then covered with a coverslip (Menzel-Glaser, Germany)and air dried for 15 min.

Mounted slides were scanned using a confocal laser scanning micro-scope (CLSM) (Axio Imager.Z2, LSM710 CLSM; Zeiss) and the accom-panying software (Zen 2010, version 6.0; Zeiss, Germany). Extremely highresolution and color depth are required for precise quantification. There-fore, fluorescence images were recorded at an emission wavelength of 615nm with a resolution of 6,144 by 6,144 pixels and at a color depth of 16 bitswith a 63�/1.4 (numerical aperture) oil objective using laser excitation at594 nm. Each pixel was scanned twice. Images were stored in 16-bit TIFformat. Fluorescence in individual cells was quantified using the freewareprogram ImageJ (National Institutes of Health, Bethesda, MD, USA). Thebackground signal was defined by a threshold value using the automatedMultiThresholder macro for ImageJ (K. Baler, G. Landini, and W. Ras-band, NIH, Bethesda, MD). For quantification, the ImageJ function “an-alyze particles” was used. The fluorescence intensity was calculated influorescence units (FU) as the mean of gray-scale units over a range from0 to 65,535. The correlation between FU and growth rate was used toestimate the growth rate (see Fig. 2; see also Supplemental Materials andMethods in the supplemental material). Using a correlation can result inthe prediction of a negative growth rate. Growth rates cannot be less thanzero; therefore, in these cases, the growth rate was considered to be slow.

The length, width, and cross-sectional area of biofilm aggregates inlung tissue samples from CF patients, as well as the distances from indi-vidual biofilms to the edge of their mucus clumps in the lung tissue, weremeasured using Zeiss Zen 2010, version 6.0. A proxy for the level of in-flammation and PMN activity around each biofilm aggregate was ob-tained by counting all PMNs stained with DAPI within a distance of 20�m from the edge of the biofilm using Zeiss Zen 2010, version 6.0.

PMNs on P. aeruginosa in vitro. One hundred milliliters of Krebs-Ringer buffer (KRB) (Panum Institute, Denmark) supplemented with 1%glucose was inoculated with 100 �l of PAO1 and incubated overnight at37°C in a shaker. When the culture reached an optical density at 450 nm(OD450) of 0.2, it was diluted to an OD450 of 0.1 using KRB containingglucose at 37°C, after which 500 �l was added to the airtight lower cham-ber of a 0.2-�m single-step filter vial (Thomson, USA) (see Fig. 7). Hu-man blood was collected from healthy volunteers with the approval (H-3-2011-117) of the Danish Scientific Ethical Board. PMNs were isolated asdescribed elsewhere (23). Extracted PMNs were resuspended in KRB con-

Kragh et al.

4478 iai.asm.org Infection and Immunity

on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://www.pseudomonas.med.ecu.edu

http://iai.asm.org

http://iai.asm.org/

taining glucose at 37°C to a final density of 2.5 � 107 PMNs/ml. In total,200 �l of the PMN suspension was added to the chamber above the filter,while the chamber below the filter received 200 �l of KRB containingglucose. Half of both the PMN-treated and untreated vials was supple-mented with 10 mM KNO3, and the vials were incubated at 37°C. After 0,2, and 4 h, 20 �l of bacterial suspension from the airtight chamber wasfixed on Super Frost Plus slides (Thermo Scientific, USA) with GN Fixa-tion Solution (AdvanDx, USA) at 65°C for 20 min. Slides were analyzed byquantitative PNA-FISH as described above.

O2 levels in the lower, airtight chamber containing the bacteria and inthe upper chamber containing the PMNs were measured using O2-sensi-tive sensor spots mounted inside the vials and monitored with a fiber-optic O2 meter (Fibox 3; PreSens, Germany) equipped with a 2-mm fiber-optic cable (24, 25). PMN activation was induced by 10 �M phorbol12-myristate 13-acetate (PMA) (Sigma-Aldrich, USA).

Statistical analysis. Statistical significance was evaluated using aMann-Whitney test. Multiple regressions were used to evaluate multifac-tor models of data. To evaluate relationships without parametric assump-tions, Spearman’s rank correlation was used. P values of �0.05 were con-sidered to be significant. All tests were performed using GraphPad Prism,version 5 (GraphPad Software, USA) and InStat, version 3 (GraphPadSoftware).

RESULTS

Schaechter et al. defined a proportional relationship between therate of growth and the ribosomal content in Salmonella enterica

serovar Typhimurium cells (26), enabling estimates of the bacte-rial growth rate from the number of ribosomes. Fluorescentlyconjugated PNA was hybridized to the RNA of intact ribosomes inP. aeruginosa cells, and the fluorescent signal was correlated to thegrowth rate of the bacteria. Based on the use of quantitative PNA-FISH and real-time PCR (RT-PCR) specific for P. aeruginosa 16SrRNA, the ribosomal content of in vitro pure culture samplestaken at different growth phases was determined. The specificgrowth rate was calculated at the exact sampling time based onoptical density (OD) measurements. For each sample, the averagenumber of fluorescence intensity units (FU) emitted by the PNA-FISH-treated cells was quantified. Between 10 and 200 cells weremeasured at each time point, and the number of rRNA moleculesper rRNA gene molecule (i.e., the number of ribosomes per ribo-some/protein-encoding gene, or the number of ribosomes) wasquantified by RT-PCR. Fluorescence microscopy showed a cleardifference in the fluorescence intensity of cells sampled at differentgrowth rates (Fig. 1).

Correlating fluorescence to growth rate. The mean number ofFU emitted in pixels, within a threshold that discriminates back-ground fluorescence, was plotted as a function of the number ofribosomes in each sample. Samples were taken to represent cul-tures in exponential growth, decreasing growth, and stationaryphase (labeled green, yellow, and red, respectively, in Fig. 2).

FIG 1 Pseudomonas aeruginosa at different growth rates. The cells were treated with PNA-FISH probes targeting P. aeruginosa 16S rRNA. The specific growthrates, in divisions (div) per hour, are indicated on the panels.

FIG 2 Correlations between fluorescence intensity and rRNA or growth rate. (A) Correlation between the average fluorescence intensity in P. aeruginosa cells andthe number of rRNA molecules per rRNA gene molecule, as measured by RT-PCR. The black line shows the calculated correlation, and the two dotted lines showthe 95% confidence interval. The correlation has an R2 fit at 0.7222. The relationship is described by y � 0.0447 � x � 46.3. (B) Correlation between the averagemeasured fluorescence intensity (FU) in P. aeruginosa and the growth rate determined from OD measurements of bacterial culture samples as a function of time.The black line shows the calculated correlation, and the two dotted lines show the 95% confidence interval. The correlation has an R2 fit at 0.7627. Therelationship is described by y � (1.146 � 10�4) � x � 1.031.

PMNs Restrict P. aeruginosa Growth in CF Lungs

November 2014 Volume 82 Number 11 iai.asm.org 4479

on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://iai.asm.org

http://iai.asm.org/

There was a significant (P � 0.0001) linear correlation betweenFU values and rRNA: number of rRNA molecules per rRNA genemolecule � 0.0447 � FU � 46.3 (R2 � 0.722) (Fig. 2A). The lackof a linear relationship and normally distributed residuals for lowlevels of FU does not invalidate our conclusion of a significantlypositive relationship, as shown by the computed Spearman’s cor-relation (� � 0.7936, P � 0.0001). As the specific growth rateswere known for each sample, the fluorescence intensity and therRNA content could be expressed as a function of the growth rateand vice versa. There was also a significant (P � 0.0001) linear corre-lation between FU and the specific growth rate: growth rate �(1.146 � 10�4) � FU � 1.031 (R2 � 0.763) (Fig. 2B). Theserelationships enabled us to estimate the growth rate based on flu-orescence intensity measurements. The few outliers that prevent anormal distribution of the data did not alter the statistically signifi-cant positive relationship (Spearman’s correlation, � � 0.9104; P �0.0001).

Environmental regulation of ribosomal activity. Quantita-tive PNA-FISH was used in two starvation experiments to inves-tigate ribosomal content in P. aeruginosa in response to suddencarbon/nitrate starvation and O2 depletion. When the exponen-tially growing culture was deprived of all carbon or nitrogensources, a decline in the FU value was observed that could bedescribed as a mono-exponential decay (see Fig. S1A in the sup-plemental material) (R2 � 0.93), with an FU decay constant of58.9% h�1, which reached an asymptotic value of 8,558 FU after 6to 7 h. This correlated well with findings from the growth phasestudy showing a baseline of 8,800 932 FU (mean standarddeviation [SD]) in cells from a very late stationary phase (�24 hafter inoculation). When cells were exposed to a sudden shift fromoxic to anoxic conditions, resulting in O2 depletion, a similar ex-ponential decay of FU (R2 � 0.91 and FU decay constant of 51.4%

h�1) was observed, which reached an asymptotic level of 8,267 FUafter 7 h of anoxia (see Fig. S1B).

Growth rates in clinical samples. To directly estimate thegrowth of P. aeruginosa in the lungs of end-stage CF patients, thequantitative PNA-FISH method was used on explanted lungsfrom three CF patients. Tissue samples (n � 20) were collected torepresent all regions of the infected lungs. Many biofilm aggre-gates were embedded in mucus in the conducting zone and weresurrounded by PMNs, consistent with earlier observations (4)(Fig. 3). Using quantitative PNA-FISH, the mean specific growthrate was estimated in each of the biofilm aggregates (n � 59). Ahigh variability in growth rate was found among the samples fromall three patients, ranging from 0 to 0.90 divisions per hour (Fig.4). Similar heterogeneity was also observed within each tissue sec-tion. In a 1-cm2 section, growth rates ranging from 0 to 0.65 divi-sions per hour were observed. Interestingly, bacteria isolated justprior to lung transplantation were not growth limited in vitro asthe median growth rate was estimated to be 1.2 divisions per hour(range, 1.15 to 1.60), which is significantly (P � 0.0058) higherthan the in situ growth rate of 0.217 divisions per hour (range,�0.10 to 0.67).

Effects of biofilm aggregate size, depth, and diameter ongrowth rate. The size, depth within the mucus, and diameter ofeach biofilm aggregate in the CF lung samples were measured, andpossible synergistic correlations to the in vivo growth rates of P.aeruginosa were investigated by multiple regression analysis. Nosignificant synergistic interactions were found that could explainthe observed heterogeneity of the bacterial growths rates, such assize, depth, and diameter (P � 0.3665), size and depth (P �0.2413), size and diameter (P � 0.4513), or depth and diameter(P � 0.6841). The average size of the biofilm aggregates was 520

FIG 3 Micrograph of P. aeruginosa-infected lung tissue. Light and fluorescence microscopy images (magnification, �170) of periodic acid-Schiff- and hema-toxylin-stained sections (A and B) and PNA-FISH-stained sections (C and D) containing luminal and mucosal accumulations of inflammatory cells. The P.aeruginosa-positive areas are seen as well-defined lobulated clarifications surrounded by inflammatory cells. Red arrows indicate PMNs, and green arrowsindicate P. aeruginosa biofilm aggregates.

Kragh et al.


on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://iai.asm.org

http://iai.asm.org/

�m2 (range, 4 to 3,227 �m2). The lack of correlation betweengrowth rate and size is depicted in Fig. 5.

PMN counts and effects on P. aeruginosa growth. While ag-gregate size and location do not affect growth rate in the CF lung,an alternative explanation is that slower-growing aggregates maybe limited for an important nutrient. Previous observations showthat PMNs increase their O2 consumption upon contact with bac-teria in vitro (11, 27); we hypothesized that slow-growing aggre-gates may be surrounded by significantly higher levels of PMNsthan aggregates with a higher growth rate. The number of PMNswas counted within 20 �m around each biofilm aggregate, and asignificant inverse correlation was found (� � �0.4471, P �

0.0004) between the PMN count and the in vivo growth rate of P.aeruginosa (Fig. 6).

In vitro confirmation of a biostatic function of PMNs. To testwhether PMNs can limit the growth of P. aeruginosa, bacterial

FIG 4 Growth rates measured in lung tissue and peak growth rate achieved by isolates. The specific growth rate of Pseudomonas aeruginosa was estimated byquantitative PNA-FISH in 59 biofilm aggregates in 20 tissue sections from explanted lungs from three CF patients (solid symbols). The highest exponential (exp)growth measurements from isolates are shown as open symbols. The horizontal line represents the median rate in each patient. Dates of sampling are indicated.pr, patient; **, P � 0.01; ***, P � 0.001.

FIG 5 Growth rates versus biofilm aggregate size. Growth rates measured inbiofilm clusters in ex vivo CF lung tissue are shown as a function of size. Therewas no significant correlation between size and growth (P � 0.1891).

FIG 6 Growth rates measured in lung tissue as a function of the number ofsurrounding PMNs. The specific in vivo growth rate of Pseudomonas aerugi-nosa was estimated by quantitative PNA-FISH as a function of the total num-ber of PMNs within a 20-�m radius from the edge of the 59 measured biofilmaggregates in 20 tissue sections of explanted lungs from three CF patients.There was a significant negative Spearmen’s correlation (� � �0.4471, P �0.0004) between the specific growth rate and the number of PMNs in thesurrounding mucus.



on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://iai.asm.org

http://iai.asm.org/

growth was examined in vials with filters that physically separatedPMNs and bacteria within the same chemical environment.Growth rates in the absence or presence of PMNs were deter-mined after 0, 2, and 4 h. O2 was rapidly depleted in the vials whenPMNs were stimulated by PMA (Fig. 7). Quantitative FISH anal-ysis showed significantly lower growth rates of P. aeruginosa in thepresence of PMNs than in the absence of PMNs after 2 h (P �0.0001) and 4 h (P � 0.0001) (Fig. 8A and B). To evaluate whetherthe PMNs restricted the growth of P. aeruginosa by O2 limitation,the medium was supplemented with an alternative electron accep-tor, NO3

�, which is not affected by PMN metabolism. The growthrate of P. aeruginosa in the presence of PMNs and NO3

� wascomparable to that observed without PMNs at both 2 h and 4 h(Fig. 4). The addition of NO3

� to P. aeruginosa in the absence ofPMNs did not have any effect on the growth rate.

Growth of P. aeruginosa in infected mouse lungs. A pulmo-nary infection mouse model (21, 28) was used to further elucidatethe effect of O2 partial pressure on the P. aeruginosa growth rate.By controlling the size of alginate-embedded P. aeruginosa micro-

colonies, installation of P. aeruginosa was predominantly directedto either the conducting zone (large alginate beads) or the respi-ratory zone (small alginate beads) of mouse lungs. As is knownfrom human physiology, the respiratory zone is oxygenated due tocontinuous O2 supply from the venous blood passing alveoli (28).Conversely, the infectious mucus in the bronchi of the conductingzone is anoxic (9). Mice infected with small aggregates showedinfection in both the respiratory and conducting airways. In con-trast, mice infected with large aggregates showed infection only inthe conducting airways. Quantitative PNA-FISH was applied to 34single aggregates in eight mouse lungs. The P. aeruginosa growthrate was significantly higher (P � 0.008) in the respiratory airwaysthan in the conducting airways (Fig. 9). To further characterize theeffect of aggregate size on the growth rate of immobilized P.aeruginosa, the size of each aggregate was measured and correlatedto the estimated growth rate. There was no correlation (P � 0.6)between the size of the aggregate and the bacterial growth rate.

DISCUSSION

Bacterial biofilm aggregates persist in the conducting zone of thelungs of CF patients despite high-dose antibiotic treatment (4). Ithas long been speculated that the low growth rates of bacterial cellswithin chronically infected lungs of CF patients contribute to theirtolerance toward antibiotic treatment (29–31), but experimentalevidence has been lacking.

In the present study, we demonstrated extremely slow in vivogrowth of P. aeruginosa in the mucus of chronically infected CFlungs. These findings are consistent with the slow growth of P.aeruginosa in expectorated CF sputum (18). For the first time, thein vivo growth rate of P. aeruginosa was directly estimated in a largenumber of mucus-embedded biofilm aggregates in the lumens ofthe conducting airways of explanted lungs from three CF patients.These biofilm aggregates had a diameter of 10 to 80 �m and weremostly surrounded by PMNs, in accordance with earlier observa-tions (32). The growth rates among biofilm aggregates in thesesamples were highly variable, and we could not identify any sub-populations of differentiated growth patterns within the individ-ual biofilm aggregates. Such differential growth patterns of P.aeruginosa were previously observed in vitro (33–35). Our findingof an undifferentiated growth rate of P. aeruginosa aggregates invivo is in accordance with a previous report that calculated a mod-est (25%) decrease in the P. aeruginosa growth rate for biofilmdepths of 40 �m using a continuous-flow biofilm reactor (36).The majority of biofilm aggregates observed in the examined lungsamples rarely exceeded 40 �m in diameter. Furthermore, no cor-relation was found between the size or position of biofilm clusterswithin the endobronchial mucus in the CF lung tissue and the P.aeruginosa growth rate.

The apparent lack of a limiting effect of biofilm aggregate sizeon P. aeruginosa growth in vivo, an effect that is typically seen invitro, may be explained by the limited amount of O2 reaching P.aeruginosa biofilm aggregates within mucus plugs, where rapiddepletion of O2 limits aerobic respiration. Worlitzsch et al. (9)demonstrated anoxic conditions in the endobronchial mucus ofinfected CF patients, and O2 consumption during the respiratoryburst of activated PMNs was proposed as the main cause of theaccelerated O2 depletion in infected mucus (15).

P. aeruginosa growth rates in the biofilm aggregates in vivo werecompared to the numbers of PMNs surrounding the biofilms.Interestingly, the growth rates in the biofilm aggregates decreased

FIG 7 Oxygen consumption by PMNs measured in filter vial. The schematicdrawing shows an in vitro experiment with PMNs in a chamber separated by amembrane from an airtight chamber containing P. aeruginosa (upper panel).O2 concentration (con) measurements taken in the chamber above the mem-brane or in the airtight chamber below the membrane were plotted versus time(lower panel). All measurements were done with PMNs (either unstimulatedor stimulated with PMA) in the chamber above the membrane.

Kragh et al.


on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://iai.asm.org

http://iai.asm.org/

with an increase in the number of PMNs surrounding the biofilmaggregates. The reduction in growth rate may be owing to PMNconsumption of O2, which reduces aerobic respiration in P.aeruginosa biofilms (37). Growth rates of P. aeruginosa are de-

creased under low-O2 conditions (17), possibly because O2 is anessential electron acceptor for ATP generation during aerobic res-piration. By supplying P. aeruginosa with NO3

� as an alternativeelectron acceptor for anaerobic respiration (38), the limitation ofaerobic respiratory metabolism was alleviated, and P. aeruginosagrowth rates increased significantly, even in the presence ofPMNs. We speculate that such conditions are representative of theanoxic environment of P. aeruginosa in the endobronchial mucusof CF lungs, where high concentrations of PMNs surround thebiofilm aggregates and where the concentration of NO3

� is suffi-ciently high to support growth by anaerobic respiration (10, 39).

The PMN-induced inhibition of P. aeruginosa growth wasgreatest in biofilms surrounded by approximately 30 PMNs. Thegrowth-inhibitory effect was lower in biofilms surrounded by ahigher number of PMNs, perhaps because there is a critical massof PMNs at which the strong hypoxia approaches complete an-oxia. At this point, the absence of O2 would presumably preventany further O2-dependent growth delay by the additional accu-mulation of PMNs. The slight increase in the P. aeruginosa growthrate observed at very high PMN concentrations could be owing tothe increased availability of NO2 and NO3

�. PMNs can produceboth nitrite and nitrate in the CF lung (40), and this may enablethe growth of P. aeruginosa through anaerobic respiration bydenitrification (39). It was also recently shown that PMN produc-tion of nitric oxide, which is an intermediate produced duringdenitrification in infected CF sputum, leads to oxygen consump-tion (10).

FIG 8 Effects of PMNs on P. aeruginosa growth in vitro. Changes in the growth rate of P. aeruginosa in the absence (Bac alone) or presence (Bac � PMN) ofPMNs, untreated or supplemented with NO3

�, were determined. (A) Growth after 2 h of treatment. (B) Growth after 4 h of treatment. The growth rate was lowerin P. aeruginosa samples with PMNs than in samples of P. aeruginosa alone. The effect of the PMNs could be alleviated by the addition of NO3

�. The horizontalline represents the median growth rate. ***, P � 0.001.

FIG 9 Growth rates of P. aeruginosa observed in mouse model. The specific invivo growth rate of Pseudomonas aeruginosa was measured in the conductingor respiratory airways of seven mice by quantitative PNA-FISH (n � 34 singleobservations). The growth rate was significantly higher (P � 0.0077) in respi-ratory airways than in conducting airways. The horizontal line represents themedian growth rate. **, P � 0.01.



on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://iai.asm.org

http://iai.asm.org/

The observed relationship between the P. aeruginosa growthrate and the assumed availability of O2 in CF lungs was confirmedby our findings in a mouse model, where immobilized P. aerugi-nosa grew faster in respiratory airways than in conducting airways.The biofilm aggregates in the alveoli are provided with a continu-ous supply of O2 due to their close proximity to the arterial bloodsupply, in contrast to biofilm aggregates in the conducting air-ways, where the O2 availability is lower (28, 41).

Therefore, O2 availability is a key factor regulating P. aerugi-nosa growth in the lungs of CF patients, and PMNs in infectedmucus can inhibit the growth of P. aeruginosa by O2 depletionthrough their respiratory burst. The aerobic in vitro growth of P.aeruginosa isolates from CF patients is 2- to 3-fold slower than thatof laboratory reference strains, probably owing to genetic adapta-tion to low-oxygen conditions in vivo (19). Besides such adapta-tions, our data point to an active growth-limiting effect caused bythe persisting metabolic state of inflammatory cells that respondto the biofilm aggregates. Therefore, in addition to the decreasedlung function associated with endobronchial PMN accumulation(42), bacterial proliferation is also dependent on PMN activity ininfected CF lungs.

The increasing growth rate of P. aeruginosa in response toNO3

� supplementation demonstrates that P. aeruginosa may beable to continue growing under PMN-induced O2 depletion byemploying anaerobic respiration by denitrification in the infectedendobronchial mucus in CF patients. Denitrification by P. aerugi-nosa during chronic lung infection in CF is evidenced by a deni-trification biomarker, the porin OprF, in CF sputum (11) and theupregulation of genes for denitrification in CF isolates (13, 14).Furthermore, we recently demonstrated active denitrification infreshly expectorated sputum from CF patients with chronic P.aeruginosa infection via measurements of N2O production (a keyintermediate in denitrification) and simultaneous nitrate deple-tion followed by nitrite depletion (10). Denitrification duringchronic lung infection in CF may also be used by other CF patho-gens because we recently demonstrated the genetic setup for deni-trification as well as anaerobic N2O production in Achromobacterxylosoxidans (43), an emerging CF pathogen that induces an in-flammatory response resembling the response induced by P.aeruginosa (44). Consumption of glucose by activated PMNs (45,46) may lead to a reduction in available glucose, which may alsolimit the growth of P. aeruginosa (47, 48). However, the high levelsof glucose (2 to 4 mM) in CF airway fluids (49) suggest that bac-terial growth in infected mucus is not limited by available glucose.The extent to which the ability of P. aeruginosa to grow anaerobi-cally on L-arginine, by substrate-level phosphorylation (50), is af-fected by the consumption of L-arginine by activated PMNs (51)in CF sputum remains to be firmly established. However, the re-portedly high levels (approximately 300 mM) of L-arginine in in-fected CF sputum (52) do not support growth limitation by L-ar-ginine depletion.

The bactericidal effects of most antibiotics rely on metaboli-cally active, growing bacteria (53). Therefore, our results offer anew possible explanation for why antibiotic treatment can clearbacteria in the respiratory zone but not in the conducting zone ofCF lungs (4). Higher P. aeruginosa growth rates observed in therespiratory airways of mice with higher O2 availability may corre-spond to an infected untreated CF patient. However, when pa-tients are undergoing intravenous administration of antibiotics,the higher growth rate in the respiratory airways, combined with

proximity to alveolar capillaries, would result in the improvedclearance of P. aeruginosa infecting the respiratory zone. In con-trast, the low growth rate in the conducting airways protectsbacterial cells from antibiotic treatment, and these areas can serveas a reservoir for future recolonization of respiratory airways (2,4, 18).

In conclusion, our results suggest that PMNs control thegrowth of P. aeruginosa in chronically infected CF lungs. Ourdemonstration of PMN-related restriction of bacterial growthsuggests, for the first time, that a bacteriostatic effect due to inten-sive O2 depletion contributes to the antibacterial activity of acti-vated PMNs. However, P. aeruginosa seems well adapted to lifeunder constant O2 depletion by PMNs because the bacteria in thebiofilm aggregates can employ anaerobic respiration by denitrifi-cation to exploit alternative electron acceptors such as NO3

�, al-beit with a lower energy yield. Such long-term persistence of P.aeruginosa under low, but stable, energy conditions may explainits low susceptibility to antibiotic treatment and enable it to adaptto the biofilms of chronically infected CF lungs.

ACKNOWLEDGMENTS

We thank Anne Kirstine Nielsen for excellent technical assistance, PiaBach Jakobsen for help with DNA/RNA purification, Heidi Marie Paulsenand Lise Strange for cutting CF lung tissue samples, and Bodil Bojsen, SysJørgensen, and Anett Wollesen for cutting mouse lung samples. AdvanDxgenerously supplied the PNA-FISH probes.

Financial support was provided by grants from the Gerda og AageHaenschs Fond to T.B., the Human Frontier Science Program to T.B., theLundbeck Foundation to T.B., and the Danish Council for IndependentResearch to M.K.

We declare that we have no conflicts of interests.

REFERENCES1. Koch C, Hoiby N. 2000. Diagnosis and treatment of cystic fibrosis. Res-

piration 67:239 –247. http://dx.doi.org/10.1159/000029503.2. Hoiby N. 2011. Recent advances in the treatment of Pseudomonas aerugi-

nosa infections in cystic fibrosis. BMC Med. 9:32. http://dx.doi.org/10.1186/1741-7015-9-32.

3. Alhede M, Bjarnsholt T, Jensen PO, Phipps RK, Moser C, ChristophersenL, Christensen LD, van Gennip M, Parsek M, Hoiby N, Rasmussen TB,Givskov M. 2009. Pseudomonas aeruginosa recognizes and responds ag-gressively to the presence of polymorphonuclear leukocytes. Microbiol-ogy 155:3500 –3508. http://dx.doi.org/10.1099/mic.0.031443-0.

4. Bjarnsholt T, Jensen PO, Fiandaca MJ, Pedersen J, Hansen CR, Ander-sen CB, Pressler T, Givskov M, Hoiby N. 2009. Pseudomonas aeruginosabiofilms in the respiratory tract of cystic fibrosis patients. Pediatr. Pul-monol. 44:547–558. http://dx.doi.org/10.1002/ppul.21011.

5. Høiby N, Ciofu O, Johansen HK, Song ZJ, Moser C, Jensen PO, MolinS, Givskov M, Tolker-Nielsen T, Bjarnsholt T. 2011. The clinical impactof bacterial biofilms. Int. J. Oral Sci. 3:55– 65. http://dx.doi.org/10.4248/IJOS11026.

6. O’Toole GA. 2003. To build a biofilm. J. Bacteriol. 185:2687–2689. http://dx.doi.org/10.1128/JB.185.9.2687-2689.2003.

7. Evans DJ, Allison DG, Brown MR, Gilbert P. 1991. Susceptibility ofPseudomonas aeruginosa and Escherichia coli biofilms towards ciprofloxa-cin: effect of specific growth rate. J. Antimicrob. Chemother. 27:177–184.http://dx.doi.org/10.1093/jac/27.2.177.

8. Borriello G, Werner E, Roe F, Kim AM, Ehrlich GD, Stewart PS. 2004.Oxygen limitation contributes to antibiotic tolerance of Pseudomonasaeruginosa in biofilms. Antimicrob. Agents Chemother. 48:2659 –2664.http://dx.doi.org/10.1128/AAC.48.7.2659-2664.2004.

9. Worlitzsch D, Tarran R, Ulrich M, Schwab U, Cekici A, Meyer KC, BirrerP, Bellon G, Berger J, Weiss T, Botzenhart K, Yankaskas JR, Randell S,Boucher RC, Doring G. 2002. Effects of reduced mucus oxygen concentra-tion in airway Pseudomonas infections of cystic fibrosis patients. J. Clin.Invest. 109:317–325. http://dx.doi.org/10.1172/JCI13870.

10. Kolpen M, Kuhl M, Bjarnsholt T, Moser C, Hansen CR, Liengaard L,

Kragh et al.


on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://dx.doi.org/10.1159/000029503

http://dx.doi.org/10.1186/1741-7015-9-32

http://dx.doi.org/10.1186/1741-7015-9-32

http://dx.doi.org/10.1099/mic.0.031443-0

http://dx.doi.org/10.1002/ppul.21011

http://dx.doi.org/10.4248/IJOS11026

http://dx.doi.org/10.4248/IJOS11026

http://dx.doi.org/10.1128/JB.185.9.2687-2689.2003

http://dx.doi.org/10.1128/JB.185.9.2687-2689.2003

http://dx.doi.org/10.1093/jac/27.2.177

http://dx.doi.org/10.1128/AAC.48.7.2659-2664.2004

http://dx.doi.org/10.1172/JCI13870

http://iai.asm.org

http://iai.asm.org/

Kharazmi A, Pressler T, Hoiby N, Jensen PO. 2014. Nitrous oxideproduction in sputum from cystic fibrosis patients with chronic Pseu-domonas aeruginosa lung infection. PLoS One 9:e84353. http://dx.doi.org/10.1371/journal.pone.0084353.

11. Yoon SS, Hennigan RF, Hilliard GM, Ochsner UA, Parvatiyar K,Kamani MC, Allen HL, DeKievit TR, Gardner PR, Schwab U, Rowe JJ,Iglewski BH, McDermott TR, Mason RP, Wozniak DJ, Hancock RE,Parsek MR, Noah TL, Boucher RC, Hassett DJ. 2002. Pseudomonasaeruginosa anaerobic respiration in biofilms: relationships to cystic fibro-sis pathogenesis. Dev. Cell 3:593– 603. http://dx.doi.org/10.1016/S1534-5807(02)00295-2.

12. Beckmann C, Brittnacher M, Ernst R, Mayer-Hamblett N, Miller SI,Burns JL. 2005. Use of phage display to identify potential Pseudomonasaeruginosa gene products relevant to early cystic fibrosis airway infections.Infect. Immun. 73:444 – 452. http://dx.doi.org/10.1128/IAI.73.1.444-452.2005.

13. Son MS, Matthews WJ, Jr, Kang Y, Nguyen DT, Hoang TT. 2007. In vivoevidence of Pseudomonas aeruginosa nutrient acquisition and pathogene-sis in the lungs of cystic fibrosis patients. Infect. Immun. 75:5313–5324.http://dx.doi.org/10.1128/IAI.01807-06.

14. Lee B, Schjerling CK, Kirkby N, Hoffmann N, Borup R, Molin S, HoibyN, Ciofu O. 2011. Mucoid Pseudomonas aeruginosa isolates maintain thebiofilm formation capacity and the gene expression profiles during thechronic lung infection of CF patients. APMIS 119:263–274. http://dx.doi.org/10.1111/j.1600-0463.2011.02726.x.

15. Kolpen M, Hansen CR, Bjarnsholt T, Moser C, Christensen LD, vanGennip M, Ciofu O, Mandsberg L, Kharazmi A, Doring G, Givskov M,Hoiby N, Jensen PO. 2010. Polymorphonuclear leucocytes consume oxygenin sputum from chronic Pseudomonas aeruginosa pneumonia in cystic fi-brosis. Thorax 65:57– 62. http://dx.doi.org/10.1136/thx.2009.114512.

16. Babior BM, Curnutte JT, McMurrich BJ. 1976. The particulate super-oxide-forming system from human neutrophils. Properties of the systemand further evidence supporting its participation in the respiratory burst.J. Clin. Invest. 58:989 –996.

17. Alvarez-Ortega C, Harwood CS. 2007. Responses of Pseudomonas aerugi-nosa to low oxygen indicate that growth in the cystic fibrosis lung is byaerobic respiration. Mol. Microbiol. 65:153–165. http://dx.doi.org/10.1111/j.1365-2958.2007.05772.x.

18. Yang L, Haagensen JA, Jelsbak L, Johansen HK, Sternberg C, Hoiby N,Molin S. 2008. In situ growth rates and biofilm development of Pseu-domonas aeruginosa populations in chronic lung infections. J. Bacteriol.190:2767–2776. http://dx.doi.org/10.1128/JB.01581-07.

19. Rau MH, Hansen SK, Johansen HK, Thomsen LE, Workman CT,Nielsen KF, Jelsbak L, Hoiby N, Yang L, Molin S. 2010. Early adaptivedevelopments of Pseudomonas aeruginosa after the transition from life inthe environment to persistent colonization in the airways of human cysticfibrosis hosts. Environ. Microbiol. 12:1643–1658. http://dx.doi.org/10.1111/j.1462-2920.2010.02211.x.

20. Bachmann B. 1996. Derivations and genotypes of some mutant deriva-tives of Escherichia coli K-12, p 2460 –2488. In Neidhardt FC, Curtiss R, III,Ingraham JL, Lin ECC, Low KB, Magasanik B, Reznikoff WS, Riley M,Schaechter M, Umbarger HE (ed), Escherichia coli and Salmonella: cellularand molecular biology, 2nd ed. ASM Press, Washington, DC.

21. Christophersen LJ, Trostrup H, Malling Damlund DS, Bjarnsholt T,Thomsen K, Jensen PO, Hougen HP, Hoiby N, Moser C. 2012. Bead-size directed distribution of Pseudomonas aeruginosa results in distinctinflammatory response in a mouse model of chronic lung infection. Clin.Exp. Immunol. 170:222–230. http://dx.doi.org/10.1111/j.1365-2249.2012.04652.x.

22. Perry-O’Keefe H, Rigby S, Oliveira K, Sorensen D, Stender H, Coull J,Hyldig-Nielsen JJ. 2001. Identification of indicator microorganisms usinga standardized PNA FISH method. J. Microbiol. Methods 47:281–292.http://dx.doi.org/10.1016/S0167-7012(01)00303-7.

23. Bjarnsholt T, Jensen PO, Burmolle M, Hentzer M, Haagensen JA, HougenHP, Calum H, Madsen KG, Moser C, Molin S, Hoiby N, Givskov M. 2005.Pseudomonas aeruginosa tolerance to tobramycin, hydrogen peroxide andpolymorphonuclear leukocytes is quorum-sensing dependent. Microbiol-ogy 151:373–383. http://dx.doi.org/10.1099/mic.0.27463-0.

24. Kuhl M. 2005. Optical microsensors for analysis of microbial communi-ties. Methods Enzymol. 397:166 –199. http://dx.doi.org/10.1016/S0076-6879(05)97010-9.

25. Rickelt LF, Askaer L, Walpersdorf E, Elberling B, Glud RN, Kühl M.2013. An optode sensor array for long term in situ measurements of O2 in

soil and sediment. J. Environ. Qual. 42:1267–1273. http://dx.doi.org/10.2134/jeq2012.0334.

26. Schaechter M, Maaloe O, Kjeldgaard NO. 1958. Dependency on me-dium and temperature of cell size and chemical composition during bal-anced grown of Salmonella typhimurium. J. Gen. Microbiol. 19:592– 606.http://dx.doi.org/10.1099/00221287-19-3-592.

27. Baldridge CW, Gerard RW. 1932. The extra respiration of phagocytosis.Am. J. Physiol. 103:235–236.

28. West. 2000. Respiratory physiology: the essentials, 6th ed. Lippincott,Williams, and Wilkins, Baltimore, MD.

29. Keren I, Kaldalu N, Spoering A, Wang Y, Lewis K. 2004. Persister cellsand tolerance to antimicrobials. FEMS Microbiol. Lett. 230:13–18. http://dx.doi.org/10.1016/S0378-1097(03)00856-5.

30. Costerton JW, Lewandowski Z, Caldwell DE, Korber DR, Lappin-ScottHM. 1995. Microbial biofilms. Annu. Rev. Microbiol. 49:711–745. http://dx.doi.org/10.1146/annurev.mi.49.100195.003431.

31. Hoiby N, Bjarnsholt T, Givskov M, Molin S, Ciofu O. 2010. Antibioticresistance of bacterial biofilms. Int. J. Antimicrob. Agents 35:322–332.http://dx.doi.org/10.1016/j.ijantimicag.2009.12.011.

32. Bjarnsholt T, Alhede M, Eickhardt-Sorensen SR, Moser C, Kuhl M,Jensen PO, Hoiby N. 2013. The in vivo biofilm. Trends Microbiol. 21:466 – 474. http://dx.doi.org/10.1016/j.tim.2013.06.002.

33. Wentland EJ, Stewart PS, Huang CT, McFeters GA. 1996. Spatial vari-ations in growth rate within Klebsiella pneumoniae colonies and biofilm.Biotechnol. Prog. 12:316 –321. http://dx.doi.org/10.1021/bp9600243.

34. Williamson KS, Richards LA, Perez-Osorio AC, Pitts B, McInnerney K,Stewart PS, Franklin MJ. 2012. Heterogeneity in Pseudomonas aeruginosabiofilms includes expression of ribosome hibernation factors in the anti-biotic-tolerant subpopulation and hypoxia-induced stress response in themetabolically active population. J. Bacteriol. 194:2062–2073. http://dx.doi.org/10.1128/JB.00022-12.

35. Stewart PS, Franklin MJ. 2008. Physiological heterogeneity in bio-films. Nat. Rev. Microbiol. 6:199 –210. http://dx.doi.org/10.1038/nrmicro1838.

36. Roberts ME, Stewart PS. 2005. Modelling protection from antimicrobialagents in biofilms through the formation of persister cells. Microbiology151:75– 80. http://dx.doi.org/10.1099/mic.0.27385-0.

37. Jesaitis AJ, Franklin MJ, Berglund D, Sasaki M, Lord CI, Bleazard JB,Duffy JE, Beyenal H, Lewandowski Z. 2003. Compromised host defenseon Pseudomonas aeruginosa biofilms: characterization of neutrophil andbiofilm interactions. J. Immunol. 171:4329 – 4339. http://dx.doi.org/10.4049/jimmunol.171.8.4329.

38. Zumft WG. 1997. Cell biology and molecular basis of denitrification.Microbiol. Mol. Biol. Rev. 61:533– 616.

39. Palmer KL, Brown SA, Whiteley M. 2007. Membrane-bound nitratereductase is required for anaerobic growth in cystic fibrosis sputum. J.Bacteriol. 189:4449 – 4455. http://dx.doi.org/10.1128/JB.00162-07.

40. Francoeur C, Denis M. 1995. Nitric oxide and interleukin-8 as inflam-matory components of cystic fibrosis. Inflammation 19:587–598. http://dx.doi.org/10.1007/BF01539138.

41. Hogg JC. 2004. Pathophysiology of airflow limitation in chronic obstruc-tive pulmonary disease. Lancet 364:709 –721. http://dx.doi.org/10.1016/S0140-6736(04)16900-6.

42. Mayer-Hamblett N, Aitken ML, Accurso FJ, Kronmal RA, KonstanMW, Burns JL, Sagel SD, Ramsey BW. 2007. Association between pul-monary function and sputum biomarkers in cystic fibrosis. Am. J. Respir.Crit. Care Med. 175:822– 828. http://dx.doi.org/10.1164/rccm.200609-1354OC.

43. Jakobsen TH, Hansen MA, Jensen PO, Hansen L, Riber L, Cockburn A,Kolpen M, Ronne Hansen C, Ridderberg W, Eickhardt S, Hansen M,Kerpedjiev P, Alhede M, Qvortrup K, Burmolle M, Moser C, Kuhl M,Ciofu O, Givskov M, Sorensen SJ, Hoiby N, Bjarnsholt T. 2013. Completegenome sequence of the cystic fibrosis pathogen Achromobacter xylosoxidansNH44784 to 1996 complies with important pathogenic phenotypes. PLoSOne 8:e68484. http://dx.doi.org/10.1371/journal.pone.0068484.

44. Hansen CR, Pressler T, Nielsen KG, Jensen PO, Bjarnsholt T, Hoiby N.2010. Inflammation in Achromobacter xylosoxidans infected cystic fibrosispatients. J. Cyst. Fibros. 9:51–58. http://dx.doi.org/10.1016/j.jcf.2009.10.005.

45. Borregaard N, Herlin T. 1982. Energy metabolism of human neutrophilsduring phagocytosis. J. Clin. Invest. 70:550 –557. http://dx.doi.org/10.1172/JCI110647.



on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://dx.doi.org/10.1371/journal.pone.0084353


http://dx.doi.org/10.1016/S1534-5807(02)00295-2

http://dx.doi.org/10.1016/S1534-5807(02)00295-2

http://dx.doi.org/10.1128/IAI.73.1.444-452.2005

http://dx.doi.org/10.1128/IAI.73.1.444-452.2005


http://dx.doi.org/10.1111/j.1600-0463.2011.02726.x

http://dx.doi.org/10.1111/j.1600-0463.2011.02726.x

http://dx.doi.org/10.1136/thx.2009.114512

http://dx.doi.org/10.1111/j.1365-2958.2007.05772.x

http://dx.doi.org/10.1111/j.1365-2958.2007.05772.x

http://dx.doi.org/10.1128/JB.01581-07

http://dx.doi.org/10.1111/j.1462-2920.2010.02211.x

http://dx.doi.org/10.1111/j.1462-2920.2010.02211.x

http://dx.doi.org/10.1111/j.1365-2249.2012.04652.x

http://dx.doi.org/10.1111/j.1365-2249.2012.04652.x

http://dx.doi.org/10.1016/S0167-7012(01)00303-7


http://dx.doi.org/10.1016/S0076-6879(05)97010-9

http://dx.doi.org/10.1016/S0076-6879(05)97010-9

http://dx.doi.org/10.2134/jeq2012.0334

http://dx.doi.org/10.2134/jeq2012.0334

http://dx.doi.org/10.1099/00221287-19-3-592

http://dx.doi.org/10.1016/S0378-1097(03)00856-5

http://dx.doi.org/10.1016/S0378-1097(03)00856-5

http://dx.doi.org/10.1146/annurev.mi.49.100195.003431

http://dx.doi.org/10.1146/annurev.mi.49.100195.003431

http://dx.doi.org/10.1016/j.ijantimicag.2009.12.011

http://dx.doi.org/10.1016/j.tim.2013.06.002

http://dx.doi.org/10.1021/bp9600243

http://dx.doi.org/10.1128/JB.00022-12

http://dx.doi.org/10.1128/JB.00022-12

http://dx.doi.org/10.1038/nrmicro1838

http://dx.doi.org/10.1038/nrmicro1838


http://dx.doi.org/10.4049/jimmunol.171.8.4329

http://dx.doi.org/10.4049/jimmunol.171.8.4329

http://dx.doi.org/10.1128/JB.00162-07

http://dx.doi.org/10.1007/BF01539138

http://dx.doi.org/10.1007/BF01539138

http://dx.doi.org/10.1016/S0140-6736(04)16900-6

http://dx.doi.org/10.1016/S0140-6736(04)16900-6

http://dx.doi.org/10.1164/rccm.200609-1354OC

http://dx.doi.org/10.1164/rccm.200609-1354OC






http://iai.asm.org

http://iai.asm.org/

46. Laval J, Touhami J, Herzenberg LA, Conrad C, Taylor N, Battini JL, SitbonM, Tirouvanziam R. 2013. Metabolic adaptation of neutrophils in cysticfibrosis airways involves distinct shifts in nutrient transporter expression. J.Immunol. 190:6043–6050. http://dx.doi.org/10.4049/jimmunol.1201755.

47. Palmer KL, Aye LM, Whiteley M. 2007. Nutritional cues control Pseu-domonas aeruginosa multicellular behavior in cystic fibrosis sputum. J.Bacteriol. 189:8079 – 8087. http://dx.doi.org/10.1128/JB.01138-07.

48. Brennan AL, Gyi KM, Wood DM, Johnson J, Holliman R, BainesDL, Philips BJ, Geddes DM, Hodson ME, Baker EH. 2007. Airwayglucose concentrations and effect on growth of respiratory pathogensin cystic fibrosis. J. Cyst. Fibros. 6:101–109. http://dx.doi.org/10.1016/j.jcf.2006.03.009.

49. Baker EH, Clark N, Brennan AL, Fisher DA, Gyi KM, Hodson ME,Philips BJ, Baines DL, Wood DM. 2007. Hyperglycemia and cysticfibrosis alter respiratory fluid glucose concentrations estimated by breathcondensate analysis. J. Appl. Physiol. 102:1969 –1975. http://dx.doi.org/10.1152/japplphysiol.01425.2006.

50. Vander Wauven C, Pierard A, Kley-Raymann M, Haas D. 1984. Pseu-domonas aeruginosa mutants affected in anaerobic growth on arginine:

evidence for a four-gene cluster encoding the arginine deiminase pathway.J. Bacteriol. 160:928 –934.

51. Chen LY, Mehta JL. 1996. Variable effects of L-arginine analogs on L-argi-nine-nitric oxide pathway in human neutrophils and platelets may relateto different nitric oxide synthase isoforms. J. Pharmacol. Exp. Ther. 276:253–257.

52. Grasemann H, Tullis E, Ratjen F. 2013. A randomized controlled trial ofinhaled L-arginine in patients with cystic fibrosis. J. Cyst. Fibros. 12:468 –474. http://dx.doi.org/10.1016/j.jcf.2012.12.008.

53. Eng RH, Padberg FT, Smith SM, Tan EN, Cherubin CE. 1991. Bacte-ricidal effects of antibiotics on slowly growing and nongrowing bacteria.Antimicrob. Agents Chemother. 35:1824 –1828. http://dx.doi.org/10.1128/AAC.35.9.1824.

54. Carreras MC, Pargament GA, Catz SD, Poderoso JJ, Boveris A. 1994.Kinetics of nitric oxide and hydrogen peroxide production and formationof peroxynitrite during the respiratory burst of human neutrophils. FEBSLett. 341:65– 68. http://dx.doi.org/10.1016/0014-5793(94)80241-6.

55. Alderton WK, Cooper CE, Knowles RG. 2001. Nitric oxide synthases:structure, function and inhibition. Biochem. J. 357(Pt 3):593– 615.

Kragh et al.


on May 6, 2015 by K

B, K

øbenhavns U


.org/D

ownloaded from

http://dx.doi.org/10.4049/jimmunol.1201755

http://dx.doi.org/10.1128/JB.01138-07



http://dx.doi.org/10.1152/japplphysiol.01425.2006

http://dx.doi.org/10.1152/japplphysiol.01425.2006


http://dx.doi.org/10.1128/AAC.35.9.1824

http://dx.doi.org/10.1128/AAC.35.9.1824

http://dx.doi.org/10.1016/0014-5793(94)80241-6

http://iai.asm.org

http://iai.asm.org/

7. Fiber-optic scalar irradiance probes

L.F. Rickelt, M. Lichtenberg, E.C.L. Trampe, and M. Kühl:

Fiber-optic probes for small scale measurements of scalar irradiance

Submitted to Photochemistry & Photobiology (September 2015)

For submission to PHOTOCHEMISTRY AND PHOTOBIOLOGY

1

Fiber-optic probes for small scale measurements of scalar irradiance

Lars Fledelius Rickelt1,

*, Mads Lichtenberg1, Erik Christian Løvbjerg Trampe

1, Michael Kühl

1, 2

1Marine Biological Section, Department of Biology, University of Copenhagen,

Strandpromenaden 5, DK-3000 Helsingør, Denmark

2Plant Functional Biology and Climate Change Cluster, University of Technology Sydney

PO Box 123, Ultimo Sydney NSW 2007, Australia.

Running title: Scalar irradiance microprobes

Keywords: optical fiber; scalar irradiance; fluence rate; light; fiber-optics;

*Corresponding author, e-mail: [email protected] Phone: +45 3532 1954



2

Abstract

A new method for producing fiber-optic microprobes for scalar irradiance (=fluence rate)

measurements is described. Such fine scale measurements are important in many photobiological

disciplines. With the new method, it is possible to cast spherical 30-600 µm wide light integrating

sensor tips onto tapered or untapered optical fibers. The sensor tip is constructed by first casting a

clear poly-methyl methacrylate (PMMA) sphere (~80 % of the size of the final probe tip diameter)

onto the optical fiber via dip-coating. Subsequently, the clear sphere is covered with light diffusing

layers of PMMA mixed with TiO2 until the fiber probe exhibits a satisfactory isotropic response

(typically ±5-10%). We also present an experimental setup for measuring the isotropic response of

fiber-optic scalar irradiance probes in air and water. The fiber probes can be mounted in a syringe

equipped with a needle, facilitating retraction of the spherical fiber tip. This makes it e.g. possible to

cut a hole in cohesive tissue with the needle before inserting the probe. The light collecting

properties of differently sized scalar irradiance probes (30 μm, 40 μm, 100 μm, 300 μm, and 470

μm) produced by this new method were compared to probes produced with previously published

methods. The new scalar irradiance probes showed both higher throughput of light, especially for

blue light, as well as a better isotropic light collection over a wide spectral range. The new method

also allowed manufacturing of significantly smaller scalar irradiance microprobes (down to 30 µm

tip diameters) than hitherto possible, and such sensors allow minimally invasive high resolution

scalar irradiance measurements in thin biofilms, leaves and animal tissues.


3

1 Introduction

Light is essential for life on Earth and is an important environmental parameter in biology, but also

in medicine, where light is used for diagnosis and treatment e.g. in photodynamic therapy (PDT). It

is a challenge to measure light in dense media such as sediments, biofilms and tissues, where

intense scattering and absorption results in strong light attenuation and steep light gradients, and

where phenomena such as photon trapping and path length enhancement come from multiple

scattering and internal reflections at optical boundaries with variations in refractive indices [1-3]. It

is thus essential in both biomedical dosimetry and photosynthetic studies in plant physiology and

microbiology to determine the total light the cells receive [1, 4, 5]. This involves measuring the

integral quantum flux from all directions about a point; this parameter, (𝐸0) is often denoted as the

photon scalar irradiance (µmol photons m-2

s-1

) in environmental research or the radiant energy

fluence rate (in W m-2

) in biomedical research.

The light field in a given medium can be described by detailed measurements of field radiance (𝐿)

with flat-cut optical fibers that have well-defined light acceptance characteristics [6, 7]. The field

radiance, 𝐿(𝜃, 𝜑), from a given direction specified by the zenith and azimuth angles (𝜃, 𝜑) in a

spherical coordinate system is defined as:

𝐿(𝜃, 𝜑) =𝑑2𝛷

𝑑𝐴𝑑𝜔

where 𝛷 is the radiant flux from that direction per unit solid angle, 𝑑𝜔, per unit area perpendicular

to the direction of light propagation, 𝑑𝐴.

The scalar irradiance (𝐸0) or fluence rate at a given point can be expressed as the field radiance

integrated over the whole sphere of 4π solid angle [1, 4, 5]:

𝐸0 = ∫ 𝐿(𝜃, 𝜑)𝑑𝜔

4𝜋


4

Scalar irradiance probes are usually built as spherical light collectors exhibiting an isotropic angular

response to incident light, i.e., light from all directions is captured by the probe tip and channeled to

the detector with equal probability. Scalar irradiance probes can be manufactured by fixing a light

diffusing sphere with isotropic light collection properties at the end of an optical fiber. The size of

the diffusing sphere, its angular isotropic response, the transmittance at different wavelengths, and

the mechanical stability are important parameters for the choice of probe for a particular

application. Ideal scalar irradiance probes should have a very good isotropic response, i.e., a

standard deviation <10% of the mean detector response for different incident light angles as well as

a small blind angle, where the optical fiber is in contact with the spherical tip. All wavelengths

should be transmitted identically. Ultraviolet (UV) radiation is usually not transmitted well due to

strong absorption in the probe material, and special optical fibers with high OH-content for

transmission in the UV region are needed. To resolve the steep light gradients in scattering media, a

small spherical tip is crucial for measurements in sediments, biofilms and tissues. A small tip

diameter also minimized local impact at the insertion point, especially in cohesive media where it

can be necessary to precut a hole with a needle. However, untapered fibers with larger spheres are

less fragile than tapered fibers with small spheres.

Several different methods for preparing fiber-optic scalar irradiance probes with spherical tip

diameters <1000 μm have been described for application in biomedical and environmental sciences.

Marijnissen and Star [8, 9] developed scalar irradiance probes for applications in tissue optics and

PDT. The sensors were constructed from a light diffusing plastic, Arnite® (polyethylene

terephthalate) sphere with ~800 μm diameter that was machined on a lathe, and glued to a 200 μm

wide flat-cut optical fiber. The probes exhibited a good isotropic response of ±10% and good

mechanical strength; however their size is quite large for most biological applications. They can be


5

used in clinical medicine, and are commercially available (MedLight SA, Switzerland; PDT

Systems, Buellton, CA, USA).

Henderson [9, 10] developed an alternative manufacturing method based on a light-cured polymer

probe using a white dental fissure sealant (Helioseal®, Ivoclar Vivadent) to form<800 μm wide

sphere on the tip of an untampered optical fiber. These sensors exhibited an angular light collection

isotropy of ±5% at 488 nm, and ±7% at 632 nm. A similar method was used to cure a resin mixed

with TiO2 forming a 50 µm wide light collecting sphere on a tapered fiber, albeit with a less ideal

isotropic response [11].

Using a fundamentally different approach, Lilge et al. [12] developed two different types of scalar

irradiance probes based on the use of a dye-doped measuring tip fixed at a distance from the light

collecting optical fiber by a transparent material. Type I was made from polymethyl methacrylate

(PMMA) and different fluorescent dyes with measuring tip diameters of 265-615 μm and shows a

light collection isotropy of ±10% in water. In type II, the fiber cladding was removed by HF

etching, and inserted in a capillary tube with a “fluorescent dye doped UV-curing glue” at the end.

The result was a cylindrical isotropic probe with diameter of 170 μm, and a sensing length of 200-

400 μm showing an isotropy of ±20% in air. The responsivity of these designs was, however, two

orders of magnitude lower than for probes made according to Marijnissen and Star [8, 9, 12]. Their

angular response was detected by rotating the light source around the probes. For measurements in

water, the probes were held in a round container filled with water. The material of the container is

not explained, and no comparison of probes measured in both air and water were given.

Dodds [13] fixed a drop of Titanium White acrylic paint at the end of a 125 μm step-index optical

fiber that dried to a scattering sphere of ~250 µm in diameter. Such a probe was used to measure

scalar irradiance in sediments and microbial mats. The angular response was less ideal and to make


6

up for that, four measurements were done turning the sensor 90° along its axis and integrating the

results, when a light profile was done.

Instead of paint, Lassen et al. [14] used poly-methyl, and poly-butyl methacrylate (PMMA and

PBMA) dissolved in xylene (Plexisol, PM 560 and PM 709) and mixed with TiO2 powder to cast a

70-100 µm wide scattering sphere on the end of a tapered fiber. These microprobes show a good

isotropic angular response of ±10% at 450 nm, 650 nm, and 850 nm both in air and water and they

have been widely used for visible and near-infrared light measurements in aquatic photosynthesis

studies [6].

To alleviate the bad transmission in the UV region of the methacrylate-based scalar irradiance

microprobes, Garcia-Pichel [15] produced an UV transmitting ~100 µm wide vitro-ceramic

spherical tip by back melting of a long optical fiber taper coated with MgO crystals. Such sensors

enable UV scalar irradiance measurements down to 250 nm with an isotropic angular response of

±15%.

The mentioned scalar irradiance probes have been applied in a variety of biomedical, and ecological

applications, where the small probe size has allowed new insights to the light microenvironment,

and optical properties of biofilms, sediments, and tissues [5, 7, 16-19]. Some recent examples

include the investigation on the effect of light on in vitro cultivated blastocysts in mouse embryos

[20], detailed measurements of vertical, and lateral light gradients within and across coral tissues in

different scleractinian species [21], characterization of the biophotonic properties of iridocytes in

photosymbiotic giant clams [11], and measurements of radiative energy budgets in different

photosynthetic microbial mats [22].

A comparison of different types of fiber-optic scalar irradiance probes in terms of their measuring

characteristics and material properties has to our knowledge not been reported in the literature. In

this study, we compare four different types of scalar irradiance probes, and present a further


7

development of the type invented by Lassen et al. [14], along with details on a setup for measuring

the isotropic performance of scalar irradiance probes in air and water. The new manufacturing

method enables construction of ultra-small scalar irradiance microprobes with 30-150 µm wide

spherical light collectors cast onto tapered fibers, or 220-600 µm wide spherical collectors cast onto

untapered fibers. These sensors show ~5 times less light attenuation, and good isotropic responses

at different wavelengths. The optical fiber can be is fixed within a syringe equipped with a needle,

facilitating a retractable probe. This makes the microprobe easier to handle, which is crucial in

various applications, such as for measuring in cohesive microbial mats, leaves and various tissues.

The needle with the retracted probe can cut a hole in the object before the probe is inserted.

2 Materials and methods

2.1 Fixation in syringe, cutting, and fabrication of tapered fiber tips

A 5 m long single strand fused-silica multimode optical fiber patch cord with standard SMA-

connectors was used for manufacturing all scalar irradiance probes in this study. The optical fiber

was a step index fiber with 105/125 µm core/cladding diameter ratio, and a numerical aperture in air

of NA = 0.22 (FG105LCA, Thorlabs, USA). The patch cord was cut in two, and the protective PVC

coating and Kevlar fibers were removed over a length of ~15 cm from one end. The underlying

Tefzel® polymer jacket enclosing the fiber was removed over ~7 cm with a mechanical fiber

stripping tool (Micro-Strip®, Micro Electronics Inc., USA). For better handling, the fiber was fixed

to the piston in a 1 ml syringe painted with an opaque black paint (Conductive Carbon Paint, SPI

Supplies, West Chester, PA, USA); this also made the fiber retractable (see Fig. 1, where

components are indicated with letters):

A 5 mm hole was drilled in the piston head (A). The black rubber gasket (B) was removed from the

piston (C), and about a third of the gasket holder side was cut off with a Stanley knife. The gasket


8

Figure 1. Overall design of the new scalar irradiance microprobe. The upper panel shows a

complete probe with a protective cap on the needle and syringe containing the optical fiber. The

lower panel shows a scalar irradiance microprobe before assembly comprised of the piston head

(A), a rubber gasket (B) on the piston (C) with the protective metal tube (D) and the black-painted

optical fiber (E) protruding out of the metal tube. The piston with the attached fiber is mounted

inside a black-painted plastic syringe (F) and a hypodermic needle (G) equipped with a protective

cap (H) when the light-collecting sensor tip is retracted into the needle.

C

FE

B EF

G

D

H

A


9

was put back, and a hypodermic needle with removed Luer connector (D) (Sterican 21G, 0.80 x 80

mm, B. Braun Melsungen AG, Germany) was pushed through the gasket. Then the fiber (E) was

first put through the hole of the piston head (A), and then through the needle tube (D). The sharp

end of the tube was pushed into the fiber protection, and the fiber protection was fixed to the piston

(C) with black tape (Vinyl Electric Tape, Scotch® Super 33+, 3M Electrical Products Division,

USA). The piston including the fixed fiber was put back into the black painted syringe (F) with the

fiber end (E) protruding from the opening. A hypodermic needle (G) (23G, 0.6 x 25 mm, Fine-

Ject®, Henke-Sass Wolf GmbH, Germany) was attached to the syringe with the bare fiber pushed

through the needle. The syringe was then mounted vertically in a micromanipulator (MM33,

Märtzhäuser, Wetzlar, Germany) with a small weight (3.75 g) attached to the bare fiber end. A taper

was made by heating the fiber with a small oxygen/propane flame from a miniature brazing and

welding set (Roxy-Kit®, Rothenberger, Frankfurt a. M., Germany). Thereafter, the syringe was

mounted horizontally in a micromanipulator with the fiber tip placed under a dissection microscope.

The fiber outside the needle was painted with opaque black paint diluted 1:1 with isopropanol, and

the taper was cut back manually with a sharpened forceps to the desired diameter (10 - 15 μm) of

the tapered tip. Untapered fibers were cut with an optical fiber cleaving tool (Thomas & Betts,

Raritan, New Jersey) to obtain a straight and flat-cut tip before it was put through a hypodermic

needle (23G, 0.6 x 25 mm, Fine-Ject®). The outermost 2-3 cm of the fiber was coated with an

opaque layer of the black paint diluted 1:1 with isopropanol. Finally, the plastic Luer connector of

the needle was painted with undiluted black paint.

2.2 Casting of light collecting spheres onto fiber tips

2.2.1 PMMA/TiO2 –based probes


10

For the construction of the light collecting spheres, two stock solutions of 25% w/w of poly-methyl

methacrylate (PMMA) (Goodfellow Cambridge Ltd., UK; refractive index n=1.49), were made in

chlorobenzene: Solution A consisted of 5 g PMMA in 15 g chlorobenzene, and solution B consisted

of 3 g PMMA 1 g TiO2, 9 g chlorobenzene (25 % w/w TiO2 in the dry PMMA). From solutions A

and B, three other solutions were made with a final content of TiO2 in dry PMMA of 12.5%

(solution C), 6.25% (solution D), and 3.15% (solution E), respectively. Chlorobenzene was chosen

as the solvent, because it dissolves PMMA and evaporates with an adequate speed. It can be

exchanged with xylene without problems.

The syringe with the fiber protruding out of the needle was mounted in a micromanipulator, and the

tip was observed under a dissection microscope. Any black paint was carefully removed from the

flat-cut fiber end, and a clear sphere was cast by dip coating the fiber tip in solution A aiming after

placing the center of this sphere at the tip of the fiber. For this, a drop of the solution was placed on

a small spatula, and was moved to the fiber until the drop on the spatula touched it. After retraction

from the polymer droplet, the surface tension of the adhered material formed a small sphere on the

fiber tip. After drying for 3-5 minutes, the process continued until the desired sphere size was

obtained. The final sphere size usually was ~20-30% wider than the diameter of this clear sphere.

A schematic drawing of the final result for a 470 µm probe with an inner 360 µm clear sphere is

shown in Figure 2. The casting process was then continued with mixtures containing different

concentrations of TiO2. The isotropic response was quickly checked after each adherence of a new

layer (see section 2.3.2) and the process was stopped when a sufficient isotropic response was

obtained. If no satisfactory result could be obtained, the sphere was removed completely with

CHCl3, and the process repeated.

The critical angel for light reflected out of PMMA was calculated to 42.9° in air and 64.8° in water.

For spheres on both tapered and untapered fibers, the best result was obtained using mixture C for


11

Figure 2. Schematic drawing of a scalar irradiance probe produced with the new method outlined in

the text. The fiber is a multimode step-index fiber with a core/cladding ratio of 105/125 µm and

painted with an opaque black paint. The inner sphere is made out of pure PMMA, while the outer

sphere (470 µm) is made of PMMA with TiO2; both materials are applied to the fiber tip by dip

coating.


12

the first layers, and D or E for finer adjustments. Using this approach, scalar irradiance probes with

5 different sizes were used in this study: 3 supported by tapered fibers (30 µm, 40 µm, and 100 µm)

and 2 supported by untapered fibers (300 µm and 470 µm).

We also manufactured a large scalar irradiance probe (500 µm), with an untapered fiber and using

only solution C for the sphere, following the casting method of Lassen et al. [14].

2.2.2 Helioseal® curing-based scalar irradiance probes

We manufactured scalar irradiance probes according to Henderson ([10] pp. 83-93, [9]) in the

following way: After removal of the black paint at the fiber tip, the distal end of the fiber was

connected to a UV-light source (UV-glue lamp, Dymax, Germany), and the fiber tip was placed into

a small amount of an UV-light curing white dental fissure sealant (Helioseal®, Ivoclar Vivadent,

Liechtenstein; refractive index 𝑛025 = 1.51). The curing light was applied for 10 seconds, where after

the fiber was pulled out of the sealant solution creating a 500 µm wide sphere of Helioseal® at the

tip. The surface of the sphere was not completely cured, but was finished by turning the sphere

directly in front of the UV-source for some minutes. Helioseal® is cured with 400-500 nm light.

The cured sphere was then washed in ethanol, and the fiber immediately behind the spherical tip

was painted with diluted black paint. It was not possible to produce isotropic spheres on tapered

fibers with this method. The critical angle for light reflected out of Helioseal® was calculated to

41.4° in air and 61.6° in water.

2.2.3 Scalar irradiance probes based on gluing a diffusing sphere to fibers.

A scalar irradiance probe with a diffusing sphere glued to a step-index optical silica glass fiber was

compared to the other types of probes. We used a commercially available version of a probe

developed by Marijnissen and Star [8] (Medlight isotropic probe model IP850 with a gold


13

radiomarker, tip diameter 850 μm, isotropy of ±10% in air 40°-360° , wavelength range 480-800

nm; silica, low OH, 400 μm core, NA = 0.37; SMA-connector, Medlight S.A., Switzerland). After

the sphere was machined on a lathe out of Arnite™, a blind hole was drilled towards the sphere

center and the fiber was glued into the hole with a transparent UV-adhesive [9]. The diameter of the

fiber was measured between the sphere and the gold radiomarker to 450 µm. The refractive index of

Arnite™ was 1.51 [23]. The critical angel for light reflected out of Arnite™ was calculated to 41.4°

in air, and 61.6° in water.

The probe was mounted within a hypodermic needle (17G, 1.5 x 50 mm) mounted on a 1 ml black

painted syringe for easier handling during measurements. It was also necessary to paint the fiber

between the needle and the sphere including the part between the radiomarker and the sphere for

correct measurements. Due to the good mechanical stability of the sphere, stains of paint on the

probe could be removed by carefully scraping with a small dissection knife.

2.3 Probe characterization

2.3.1 Detectors and light sources

A custom-built light meter, with a relatively flat spectral quantum responsivity for 400 to 700 nm

light, developed for microscale measurements of photosynthetically active radiation (PAR, 400-700

nm) [24], was used for characterizing the fiber optic scalar irradiance probes. The light meter signal

was recorded on a strip-chart recorder (BD25, Kipp & Zonen, Netherlands).

A fiber-optic spectrometer (USB2000 operated with the Spectra Suite software, Ocean Optics,

Dunedin, USA) was used to record the spectral response of the probes and bare fibers. All spectra

were recorded as an average of 10 scans, using a boxcar smoothing width of 4, and with the

spectrometers nonlinearity and stray light correction enabled. The integration time was set as high

as possible, and the corresponding dark noise was automatically subtracted for each recorded


14

spectrum. A light meter (Universal Light Meter, ULM-500, Walz, Germany) equipped with a

calibrated quantum irradiance sensor (LI-190, Li-Cor, USA) was used for measurements of the

absolute photon irradiance of incident PAR (400-700 nm) from the collimated light source in units

of µmol photons m-2

s-1

.

Collimating optics was used for all sensor characterization measurements (Ovio Collimated Source,

Ovio Optics, France) together with different light emitting diodes (LED). The LED’s were

connected to a trigger box controlled by the software Look@RGB (both available from www.fish-

n-chips.de), which enabled PC-controlled adjustment of the LED intensity [25]. Three different

LEDs were used in connection with the collimation optics: Two different white LEDs, one for low

light intensities (Oslon 1 LED; ILH-OW01-STWH-SC211-WIR200; RS-Components, UK), and

one for high light intensities (Oslon 4 LEDs; ILH-OO04-ULWH-SC211-WIR200; RS-Components,

UK), and a 405 nm LED (Oslon 4 LEDs; ILH-OW04-UVBL-SC211; RS-Components, UK). The

opening diameter of the Ovio optics was 32 mm, and at the backside of the box (at a distance of

54.5 cm) the beam diameter was 55 mm in air. The beam divergence for the collimated light was

calculated to be <2.5° in air. The value was similar in water after a glass plate was placed in front of

the collimation lens.

2.3.2 Measurement of isotropic response

For characterizing the angular light-collecting properties of fiber-optic scalar irradiance probes in

air and water, a device was constructed from 1 cm thick black PVC (parts are identified by letters in

Fig. 3): It consisted of a 60 x 60 x 10 cm box with a 10 cm disk (A) glued to the bottom in the

middle of the box. A needle was put vertical through the center of the disk. Another 23.5 cm disk

(B) with a 10 cm hole in the center exactly fitting the first disk was placed so it could be revolved

freely. A holder (C) for the sensors was fixed on the outer disk with a metal rod so it could be

http://www.fish-n-chips.de/

http://www.fish-n-chips.de/


15

Figure 3. Setup for measuring the light collecting isotropy of scalar irradiance probes comprised of

a flat black box with a fixed disk (A) placed in the center with an angular scale and a needle, a

rotational disk (B) on which a scalar irradiance probe is mounted in a holder (C) with the probe tip

placed over the center of the fixed disk, a read out pointer (D), and the collimated light source

mounted water-tight in a holder (E). During probe readout, the box is covered with a black lid, and

the box can also be filled with water.


16

rotated around its own axis. A plastic screw was used to lock the holder after centering of the sensor

sphere. A laminated print of a graduated circle was glued to the central disk (A). A pointer (D) was

placed in front of the sensor holder for easy reading of the measurement angle.

The collimation optics was mounted water-tight on one side of the box in a holder (E) that was

sealed with an O-ring (Simmerring®). A glass plate was placed in front of the collimator lens to

prevent water from leaking, dew on the backside of the lens, and to keep the light collimation

similar with air and water in the measuring chamber. The collimated light source and probe holders

were constructed in a way, making the spherical probe tip position adjustable, as to be put in the

center of the light beam precisely over the needle in the center of the disk. The chamber could be

closed with a light-tight lid to prevent ambient light from disturbing the measurements.

For measurements, the scalar irradiance probe was attached to the holder on the revolving table with

the sphere exactly over the guiding needle placed at the center of the disk (Fig. 3). The distal end of

the fiber-optic scalar irradiance probe was connected to a light meter. The directional response of

the probe was determined by rotating the revolving table from +160° to -160° in angular steps of

10° in a beam of collimated light. At 0° the probe fiber and the light source were aligned. To check

for possible variations in the measuring setup such as light source fluctuations, fiber bending effects

or e.g. air bubbles in the water, measurements at -90°, 0°, 90°, and 160° were measured several

times for each probe.

Four series of measurements were recorded with the PAR-meter, and two series with the scalar

irradiance probes connected to the spectrometer. All probes were first measured in air, where after

the chamber was filled with water and measurements continued. After a measurement series, the

probe was turned 90° around its longitudinal axis in the sensor holder, and the angular light

collection was measured again. The water was then drained from the chamber and the probe

properties was measured once more in air.


17

2.3.3 Calibration of scalar irradiance probes for photon irradiance measurements

A 7 x 7 x 7 cm black PVC box with a lid and two aligned holes in opposite walls was used to

calibrate the response of scalar irradiance sensors in absolute units of µmol photons m-2

s-1

. The

collimated light source and a quantum irradiance sensor (LI190, LiCOR, USA; Universal Light

Meter, ULM-500, Walz, Germany) were placed in opposite holes. The photon irradiance (µmol

photons s-1

m-2

) was measured at 8 different LED currents increasing in 100mA intervals, (100mA-

800mA), for each of the white, blue and red LED’s as adjusted by the LED power supply.

Subsequently, the irradiance sensor was exchanged with a fiber-optic scalar irradiance probe with

its measuring tip placed at exactly the same spot and distance in the light field relative to the

collimator, and the sensor response was measured in mV with the PAR-meter. Measurements with

the scalar irradiance probe could then be converted to µmol photons m-2

s-1

, from the linear

correlation between mV and µmol photons m-2

s-1

.

2.3.4 Light attenuation in scalar irradiance probes

We compared the light attenuation in each of the four different types of scalar irradiance probes,

based on different sphere types fixed on the same untapered fiber. The sensors were mounted in the

light calibration box (see section 2.3.3), and light spectra were recorded with the spectrometer

system. The percentage of light passing from the light scattering sphere at the tip into the probe

fiber was calculated by dividing the probe spectra with spectra recorded with bare fibers placed at

similar distance and position in the light field (Fig. 4). A 400 µm core step index fiber (Laser

Components, Germany) was used corresponding to the Medlight scalar irradiance probe. All probes

were supported by untapered fibers, and the results were corrected for different cross-sectional areas

due to different sphere diameters. Besides measuring probe spectra with a white LED light source,


18

380 390 400 410 420 430 440 450

0.00

0.05

0.10

0.15

450 500 550 600 650 700

0.0

0.2

0.4

0.6

Rela

tive

In

ten

sity (

%)

Wavelength (nm)

A

This study

Marijnissen & Star (1987)

Henderson (1990)

Lassen et al. (1992)

B

Figure 4. A comparison of spectral light attenuation in scalar irradiance probes made of 4 different

types of light collecting sphere materials , all supported by the same type of untapered fibers. A.

Measurements using a 405 nm LED as light source. B. Measurements using a white LED light

source. The light intensity is expressed relative to the measuring signal obtained with bare

untapered fibers in the light path.


19

additional spectra were recorded with a 405 nm LED to obtain sufficient signal in the 380-450 nm

region.

2.4 Photography of sensors

The different types of scalar irradiance probes were photographed with a commercially available

digital SLR camera (Canon EOS 7D MkII, Canon Europe Ltd., Middlesex, UK), connected via a

0.5x photo-adapter (IMAG-AX, Heinz Walz GmbH, Effeltrich, Germany) to an epifluorescence

microscope (Axiostar Plus FL, Carl Zeiss GmbH, Germany), fitted with either a 10x or 20x plan-

Apochromate objective (Carl Zeiss GmbH, Germany). Illumination of the probes was achieved by

means of the light source of the microscope and a fiber-optic halogen lamp (KL-1500, Schott AG,

Mainz, Germany) fitted with a double light guide for side illumination.

2.5 Application of new scalar irradiance microsensors

Spectral scalar irradiance was measured using a micro-profiling setup with the sensor mounted in a

motorized micromanipulator (MU-1, PyroScience GmbH, Germany) and controlled by PC-software

(Profix, PyroScience GmbH, Germany); this allowed advances in 100 µm vertical steps through the

sample. Spectra of the scalar irradiance were measured with the probe fiber connected to a fiber-

optic spectrometer (USB2000, Ocean Optics, FL, USA) that was interfaced to a PC running

dedicated spectral acquisition software (SpectraSuite, Ocean Optics, FL, USA). Incident light was

provided either with a fiber optic tungsten halogen lamp with a collimating lens (Schott

KL2500LCD, Schott AG, Germany) or from a LED ring (Walz GmbH, Germany). Spectra of scalar

irradiance can be integrated in the spectral region of interest, e.g. the PAR region (400-700nm), and

normalized to the incident downwelling irradiance measured over a black non-reflective light-well,

to construct scalar irradiance profiles through tissues and characterize the amount of


20

photosynthetically available radiation (PAR) in a certain tissue depth. For a detailed description of

the collection and analysis of acquired spectra, see [26].

3 Results

3.1 Light attenuation in the different scalar irradiance probes

A comparison of light attenuation in the four different probe materials showed that wavelengths

<410 nm were strongly absorbed in probe tips made of PMMA or Helioseal, whereas the machined

sphere probe made according to Marijnissen and Star [8] exhibited a much better performance in

the UV-region. At wavelengths >415 nm, probes manufactured according to Henderson [10] and

Lassen et al. [14] exhibited similar light attenuation, while probes made according to Marijnissen

and Star [8] and probes made with the new procedure described in this study showed about 5 times

less light attenuation in the probe material (Fig. 4).

3.2 Angular light collecting properties of scalar irradiance probes

We manufactured different types of scalar irradiance sensors with spherical tip diameters of 30 -

850 µm (Fig. 5). The isotropic response of the different scalar irradiance probes revealed relatively

large differences in isotropy between different probes when measured in air and water and for

different wavelength ranges (Table 1; Fig. 6-9). Sensors manufactured according to Marijnissen and

Star [8] and Lassen et al. [14] showed similar isotropy for blue, green and red light, whereas the

Helioseal-based sensor manufactured according to Henderson [10] showed a large color-

dependence, where the isotropy for green light was 3-5 times smaller than for blue and red light

(Fig. 8C). Except for some minor peaks in water, for some sensors, in the region -120° to -160°, and

120° to 160°, the sensors manufactured with the new method presented here generally exhibited a

good isotropic light collection in both air and water that was similar or better than the other types of


21

Figure 5. Photographs of different fiber-optic scalar irradiance probes. Sensors made with the new

manufacturing procedure are shown in panels A, B (40 µm sphere diameter) at two different

magnifications, where B was illuminated through the fiber, C (100 µm), and D (470 µm). Panel E

shows a probe made according to Lassen et al. (1992) (500 µm). Panel F shows a probe (500 µm)

made according to Henderson (1990). Panels G and H show a commercial probe (850 µm) made

according to Marijnissen & Star (1987) with and without a black overcoat of the fiber, respectively.


22

Probe No. 1 2 3 4 5 6 7 8

Type This study

This study

This study

This study

This study

Marijnissen and Star (1989)

Lassen et al. (1992)

Henderson (1990)

Diameter (µm) n.d./30 35/40 80/100 n.d./300 360/470 850 500 500

Fiber tip geometry tapered un-tapered

Air 9.3 4.6 7.2 5.5 6.8 17.2 8.9 7.3

Air 90° 6.3 5.4 7.3 4.5 9.3 17.8 12.3 5.1

Water 9.2 14.7 7.6 6.3 6.6 18.5 8.3 8.5

Water 90° 7.5 15.0 7.3 8.3 7.3 18.5 14.7 7.8

Air blue (450nm) 13.2 9.2 7.8 6.9 8.0 18.7 13.1 27.4

Air green (550nm) 7.5 5.6 6.6 6.8 7.0 19.5 13.5 5.5

Air red (650nm) 7.6 7.2 8.7 7.7 9.3 20.2 13.8 23.5

Water blue (450nm) 18.9 21.1 11.0 9.4 9.4 21.4 8.9 24.5

Water green (550nm) 12.1 15.7 6.5 6.8 7.5 22.6 8.8 11.4

Water red (650nm) 12.3 15.0 4.6 5.1 7.8 23.3 9.9 34.0

Table 1. Light collecting properties of different types of scalar irradiance probes. Probe isotropy

was measured in the setup shown in Figure 3 in air as well as in water with the sensor turned 90°

around the longitudinal fiber axes between measurements. The isotropy was quantified as the

standard deviation of probe signal normalized to the highest angular signal reading. The probes

were manufactured with 4 different methods: Probe no. 1-5 (different diameters) were made with

the new fabrication method based on casting first an inner sphere of PMMA and then a diffusing

shell of PMMA doped with TiO2 onto the fiber tip (the first number is the size of the clear sphere

and the last number is the final probe diameter); Probe no. 6, was obtained with a 850 µm wide

machined diffusing plastic sphere fixed to an optical fiber according to Marijnissen and Star (1987);

Probe no. 7 was made by casting 500 µm wide sphere of PMMA doped with TiO2 onto the fiber tip

according to the method of Lassen et al. (1992); Probe no. 8 was made by curing a 500 µm wide

sphere of Helioseal® on the fiber tip according to Henderson (1990).


23

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

DC

BThis study

Air

90° Air

Water

90° Water

Rela

tive R

esponse (

%)

A Lassen et al. (1992)

Henderson (1990)

Angle between Fiber and Light (Degrees)

Marijnissen & Star (1987)

Figure 6. The isotropic light collecting properties of 4 different types of scalar irradiance probes.

The relative response was measured with the probes connected to a PMT light meter with a flat

quantum response over 400-700 nm and illuminated with a collimated white LED light source.

Each probe was mounted in the angular calibration setup shown in Figure 3 and the probe response

was measured as a function of incident light angle in 10° steps. Measurements were done in air and

water. After each measurement, the sensor was turned 90° around its own axis and another set was

recorded to check for spherical homogeneity in light collection.


24

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

A 30 m

Air

90° Air

Water

90° Water

B 40 m

C 100 m

Rela

tive

Re

sp

on

se

(%

)

Angle Between Fiber and Light (Degrees)

D 300 m

Figure 7. The isotropic light collecting properties of 4 different sizes of scalar irradiance probes

manufactured with the new method. The relative response was measured with the probes connected

to a PMT light meter with a flat quantum response over 400-700 nm and with a collimated white

LED light source. Probes were mounted in the setup shown in Figure 3 and the angular response

was measured as a function of incident light angle in 10° steps. Measurements were done in air and

water. After each measurement, the probe was turned 90° around its own axis and another set was

recorded to check for spherical homogeneity in light collection.


25

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

160

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

160

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

160

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

160

PAR

Blue

Green

Red

This study (water)A Lassen et al. (1992)B

Re

lative

Re

sp

on

se

(%

)


Henderson (1990)C Marijnissen & Star (1987)D

Figure 8. The spectral dependence of the isotropic light collecting properties of 4 different types of

scalar irradiance probes. The relative response was measured with the probes connected to a

spectroradiometer and illuminated with a collimated white LED light source. Probes were mounted

in the setup shown in Figure 3 and the probe response was measured as a function of incident light

angle in 10° steps. Measurements were done in air except for A.


26

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180

0

20

40

60

80

100

120

140

PAR

Blue

Green

Red

30 mA 40 mB

Rela

tive

Re

sp

on

se

(%

)


100 mC

300 mD

Figure 9. The spectral dependence of the isotropic light collecting properties of 4 different sizes of

scalar irradiance probes manufactured with the new method. The relative response was measured

with the probes connected to a spectroradiometer and illuminated with a collimated white LED light

source. Probes were mounted in the setup shown in Figure 3 and the angular response was

measured as a function of incident light angle in 10° steps. Measurements were done in air.


27

scalar irradiance probes (Fig.6, 8) over a wide range of tip diameters (Fig. 7, 9). The angular light

response of most probes was symmetrical about 0°. The probe manufactured according to

Marijnissen and Star [8] showed a characteristic parabolic shape, whereas the other types of scalar

irradiance probes exhibited a more periodical variation in angular response.

.

3.3 Application of scalar irradiance microprobes

The new manufacturing procedure presented in this study enabled production of small scalar

irradiance microprobes with tip diameters <50 µm, which can be used for measuring scalar

irradiance attenuation profiles in thin specimens. We successfully tested such probes in the tough

thallus of the brown macroalga Fucus serratus and in the cohesive tissue of the reef-building coral

Montastrea curta (Fig. 10).

In both specimens, it was possible to measure detailed light attenuation profiles of photon scalar

irradiance of photosynthetically active radiation (PAR, 400-700 nm). The light microprofiles

showed local enhancement of scalar irradiance near the surface of the tissue due to photon trapping,

created by multiple scattering and enhancement of the photon pathlength [6] followed by an

exponential attenuation of light in deeper tissue layers.

4 Discussion

We present the first comparison of the light collecting properties of different types of fiber-optic

scalar irradiance probes along with a new manufacturing method that enables fabrication of scalar

irradiance microprobes with a good isotropic response and significantly smaller tip diameters than

previously realized. Furthermore, the new scalar irradiance probes exhibited a higher throughput of

light in the visible to near-infrared spectral range, thus alleviating constraints on detector sensitivity.

The detailed characterization of probe performance, using a new set-up for measuring the angular


28

1.2

1.0

0.8

0.6

0.4

0.2

0.0

-0.2

10 100 10 100

1.2

1.0

0.8

0.6

0.4

0.2

0.0

-0.2

Photon scalar irradiance (% of incident irradiance)

Depth

(m

m)

A B

Figure 10. Two examples of scalar irradiance microsensor measurements in biological tissues. A. A

vertical profile of photon scalar irradiance (400-700 nm) in the living tissue of the scleractinian

coral Montastrea curta. B. Photon scalar irradiance measurements in a macroalgal thallus of the

brown alga Fucus serratus. Values are expressed in % of the incident downwelling photon

irradiance as measured with the fiber probe positioned over a black non-reflective light well.


29

light collection of scalar irradiance probes, revealed distinct difference between the different probe

types, as discussed below.

4.1. Characterizing the angular light collection of scalar irradiance probes

To our knowledge, a thorough comparison of different types of scalar irradiance sensors has not

previously been published. With the construction of a new device enabling angular response

measurements in air and water (Fig. 3), it was possible to compare the performance of different

scalar irradiance probes. If the probes and sensors are manufactured correctly, there is not a

significant difference in the behavior of the sensors, whether they are measured in air or water, but

such comparative measurements can effectively reveal any flaws in the production that can direct

optimization of probe construction.

As an example, it was necessary to paint the fiber all the way to the sphere on the probe

manufactured according to Marijnissen and Star [8], before a reasonable isotropic angular response

was obtained; while the standard deviations in the isotropy plot where not improved by the painting,

the shape of the plot became much different. Before painting, the plot in air was relatively flat from

90° to -90° with a decreasing angular response at the ends. In water, peaks appeared around 130°

and -130°. After painting, the plots were similar in air and water (Fig. 6D; data before painting not

shown). This underscores the importance of avoiding light entry or leakage via the fiber behind the

spherical probe tip or via the optical fiber strand.

When the first recording of the “Marijnissen and Star” sensor was done, no protective painting was

applied, not even on the syringe. This resulted in a large peak at -140° to -150° in both air and

water, but only in one position of the syringe. This was not seen when the syringe was turned 90° in

the holder. After painting of the syringe, this peak disappeared, but the overall plot still looked

strange although the overall variability in angular response was ok. We concluded that the minor


30

peaks in the angle response plots (Fig. 6, 7) are artefacts due to flaws in the black painting of the

fibers very close to the sphere where it is difficult to see (Fig. 5) and from which some light can

leak out. The peaks arise because more light is apparently leaked out upon direct illumination of the

front of the spherical tip and less when the light comes from the back. This may also explain the

observed difference in angular response found in air and water, where in the latter case more light

can be coupled out of the fiber more easily due to a lower difference in refractive index between

fiber and water as compared to fiber and air due to larger acceptance angle between the fibers in

water than in air. For a fiber with a 105/125 μm core/cladding ratio, 30% of the fiber cross-section

is cladding and the cladding refractive index is only a little smaller than the core refractive index.

This indicates that a little less than 30% of the light from the sphere will enter the cladding. In air

most of this light will soon rapidly be caught by the core, but in water the acceptance angle is ~20°

wider and more of the light will escape out if the protective painting is insufficient.

For tapered fibers, light coupling across the sides will be more pronounced than for untapered

fibers. This was confirmed by redoing the 100 µm probe (3) (Fig. 7C), and the 500 µm Lassen

probe (7) (Fig. 6B), where preliminary results showed large differences in isotropic response in air

and water (data not shown). When the new probes were produced, great attention was put into

ensuring a flawless painting of the fibers.

Lassen et al. [14] showed an isotropic plot of the same sensor in both air and water, but did not

describe how it was carried out [14]. Lilge et al. [12] measured some sensors in water and some in

air, but did not present any comparison of the same sensor in both air and water. The set-up was

described in some detail but it seems problematic as the probe was held in a cylindrical water bath

and the collimated light was moved around it. It cannot be assumed that the light will still be

collimated inside the water bath described due to focusing effects. The probes measured in air, was

excited with a 351 nm laser, however light attenuation in water was is to strong [12]. Van Staveren


31

et al. [9] used a similar set-up as Lilge et al. [12] but placed a lens in front of the water bath to

counteract the above mentioned focusing effects. They found differences in isotropy for probes

inserted in different media and gave some correction factors for absolute calibration depending on

the refractive index of the probe material and the surrounding medium. The probe fibers did not

have any protection against light collection or leakage [3, 9, 17]. They also compared isotropic

properties for probes acting either as a light detector or as a light source, and measured large

differences due to dissimilar light paths. A probe with an isotropic light emission (< ±15%) could

be anisotropic for light detection (> ±35%), and vice versa [9].

4.2 Optical throughput and spectral differences in isotropic response

The difference in the angular response to blue, green, and red light of the various scalar irradiance

probe types can largely be explained by absorption differences in the sphere materials. This affects

the light attenuation in the light collecting sphere due to scattering enhancing differences in the

spectral absorption characteristics of the probe sphere material. That is, the path length

enhancement for photons in the scattering sphere material enhances the probability of photon

absorption at wavelengths overlapping with the absorption maxima of the sphere material, while

this effect is much smaller for photons in spectral ranges outside characteristic absorption maxima.

In this way, both material properties, the homogeneity of the sphere in terms of material thickness,

and the way the sphere is fixed onto the fiber will affect the probability of photons collected and

channeled to the detector via the optical fiber.

The refractive index of the sphere material can also affect the light collecting properties of scalar

irradiance probes. The difference in refractive indices of air and water has no major significance for

the light penetrating the probes but for the radiation reflected out of the probe there is a difference

in the critical angel. For Helioseal® and Arnite™, the difference is ~20°, and for PMMA it is a little


32

larger, ~22°. Photons collected in the sphere material are thus more efficiently retained in the sphere

when measuring in air as compared to measurements in water. On the other side, backscatter of

incident light in the sphere surface will be higher in air than in water. Such immersion effects are

well known, also from other types of scalar irradiance probes, and for probes made according to

Marijnissen and Star [8], such effects can lead to minor uncertainties in the range of <5% when

measuring in turbid media [23, 27].

Probes manufactured according to Marijnissen and Star generally exhibit a small uniform

absorption of visible light [9], hence the observed difference in the three colors was very small (Fig.

8D). PMMA-based sensors [14] showed stronger absorption of blue light as compared to green and

red light. The spectral variation in the isotropic response of such sensors could be related to the

length of the path the light must travel inside the probe.

Such effect can be illustrated by looking at the construction of the new probes (Fig. 2), where the

fiber tip is not placed in the center of the probe, but in the center of the clear sphere so the path

through the TiO2 doped PMMA is longer for light coming from the front than from the rear. This

can give a somewhat stronger attenuation of the frontal light. In probes made according to Lassen et

al. [14], the fiber tip is placed more central in the PMMA sphere, and such differences in path

lengths are less pronounced. But as mentioned above, the solid PMMA+TiO2 sphere of these probes

causes a much higher overall attenuation of incident light as compared to probes made with the

novel method presented here.

Scalar irradiance probes made according to Henderson [10], displayed a much stronger absorbance

of blue light because the Helisoseal® matrix is produced to absorb in this wavelength range as it is

used for curing of the matrix [9, 10].

4.3 Fabrication of scalar irradiance microprobes


33

The variance in isotropic light collection of the different scalar irradiance probes is strongly affected

by the actual position of the fiber tip inside the spherical light collector casted on the fiber. This is

also relevant for the Marijnissen-Star-type probes[9], where the parabolic shape of their angular

light collecting function is caused by the placement of the sphere on the fiber, as well as a small

ratio between the sphere diameter (850 μm), and the fiber diameter (450 μm); in the original

publications the ratio was recommended to be 4-5 [9].

The insertion point of the fiber itself creates a blind angle for light collection by the sphere and will

in most cases set a limit for how efficient light is collected at incident angles >±120°. However,

during the fabrication of scalar irradiance probes made according to the new method, one can

partially compensate for such effects. The construction based on a clear inner sphere coated with a

scattering shell thus makes it possible to produce small probes down to 30 μm on tapered fibers

with a tip diameter of ~15 μm, and on untapered fibers down to 240 μm with good isotropic

response (Table 1).

It has not previously been possible to create probes made according to Lassen et al. [ref] on

untapered fibers. This is mainly due to the concentration of PMMA in xylene (40-45% w/w) of the

used polymer solutions, which leads to a high surface tension in the material and larger spheres will

not dry with a smooth surface [14]. For larger spheres, the idea behind the Henderson method is

elegant, but the optical properties of the probe material are not ideal for broad spectral light

measurements as their angular light collection varies strongly with wavelength [10].

With the new method presented here, it is both faster to produce scalar irradiance probes with good

isotropic performance and a higher optical throughput as compared to probes based on casting a

sphere of only PMMA+TiO2 [14]. The new method makes it easier to vary the tip dimensions of the

probe, and it is easier to predict the size of the final sphere during the fabrication process.


34

4.4 Mechanical stability of scalar irradiance probes.

It is difficult to give an exact evaluation of the mechanical stability of the probes without destroying

them, but the material, and diameter of the employed optical fiber, and how well the sphere material

sticks to the fiber are essential parameters. The fiber material is silica-glass for all probes compared

in this study. Naturally, tapered fibers are thinner and more fragile than untapered fibers. Van

Staveren et al. [9] found that the mechanical strength of sensors made according to Marijnissen and

Star [8], and Henderson [10] was similar. Sensors made with PMMA spheres cast on the fiber tip

are less sturdy, especially when measuring in very cohesive media, such as dense tissue, where

there can be a substantial drag on the sphere when retracting the probe, with a potential loss of the

spherical tip. Etching of the fiber tip [28] prior to casting the sphere results in better adhesion, and

thus more robust scalar irradiance probes. Furthermore, the smaller probes made with the new

method also appeared sturdier when used for measurements in plant and animal tissues (Fig. 10).

4.5 Application of new scalar irradiance probes

Collection of scalar irradiance profiles in thin specimens, e.g. in the tough thallus of the brown

macroalga Fucus serratus, and in the cohesive tissue of the reef-building coral Montastrea curta

(Fig. 10) is only possible using very small sensors. This is due to i) the thickness of the tissue under

investigation, which is often < 1 mm, and ii) the toughness of the outer tissue layers (epidermis in

corals and outer cortex in algae), which makes it difficult for larger objects to penetrate. In addition,

larger spheres are often actively pulled off the fiber tip by coral mesenterial filaments acting as a

defense mechanism to fiber insertion, but apparently such filament attachment are unable to pull off

the small spheres (<50 µm) (Daniel Wangpraseurt and Mads Lichtenberg, personal

communication).


35

4.6 Conclusion

The new fabrication method yields scalar irradiance probes with excellent isotropic light collecting

properties, being comparable or better than previously developed scalar irradiance probes, and

spherical tip diameters ranging from as small as 30 µm on tapered fibers up to ~600 μm on

untapered fibers. The presented experimental setup for measuring the isotropic light collection of

the scalar irradiance probes in air and any relevant liquid is easy to build and operate. If probes are

carefully produced, e.g. by an experienced technician, it is not necessary to test them both in air and

water. However, testing can effectively reveal flaws in the probe performance due to imperfect

coating/isolation of fiber tips or damage of the spherical tip. The possibility of making well-

functioning scalar irradiance microprobes with tip diameters <50 µm now enables light

measurements at higher spatial resolution in thin objects, such as terrestrial leaves and thin-tissued

corals, and aquatic macrophytes. These new probes thus give rise to many new fields of application

in photobiology.

Acknowledgments

This study was supported by grants from the Danish Council for Independent Research | Natural

Sciences (MK) and an Eliteforsk talent grant from the Danish Council for Independent Research

(ECLT). The authors thank Egil Nielsen for help with the mechanical construction of the isotropic

calibration box and Bjørn Jacobsen for assisting with photography.


36

References

[1] M. Kühl, C. Lassen, B.B. Jørgensen (1994): Optical properties of microbial mats: light

measurements with fiber-optic microprobes. In L.J. Stal & P. Caumette (Eds.) Microbial

Mats: Structure, Development and Environmental Significance, NATO Series G, Vol. 35,

Springer, Berlin pp. 149-167

[2] Flemming, H.-C., J. Wingender (2001): Relevance of microbial extracellular polymeric

substances (EPSs) – Part I: Structural and ecological aspects. Water Science and

Techynology 431-8

[3] W.M. Star, J.P.A. Marijnissen (1989): Diffusion theory of isotropic light dosimetry probe

response. Laser Med. Sci. 4 341-350

[4] L.O. Björn (2008): Principles and Nomenclature for the Quantification of Light

Photobiology. In: L.O. Björn (Ed.), The Science of Life and Light, pp. 41-49.

[5] M. Kühl, B.B. Jørgensen (1992): Spectral light measurements in microbenthic phototrophic

communities with a fiber-optic microprobe coupled to a sensitive diode array detector.

Limnol. Oceanogr. 37 (8) 1813-1823.

[6] M. Kühl (2005): Optical microsensors for analysis of microbial communities, Environ.

Microbiol. 397 166-199

[7] M. Kühl, C. Lassen, B.B. Jørgensen (1994): Light penetration and intensity in sandy marine

sediments measured with irradiance and scalar irradiance fiber optic microprobes. Mar.

Ecol. Prog. Ser. 105 (1994) 139-148

[8] J.P.A. Marijnissen, W.M. Star (1987): Quantitative Light Dosimetry in Vitro and Vivo.

Lasers Med. Sci. 2 235-242.

[9] H.J. van Staveren, J.P.A. Marijnissen, M.C.G. Aalders, W.M. Star (1995): Construction,

quality assurance and calibration of spherical isotropic fibre-optic light diffusers. Laser

Med. Sci. 10 137–47

[10] B. Henderson (1990): An Isotropic Dosimetry Probe for Monitoring Light in Tissue,

Theoretical and Experimental Assessment. Ph.D.-thesis, Medical Laser Unit, Heriot-Watt

University, Edinburgh, UK.


37

[11] A.L. Holt, S. Vahidinia, Y.L. Gagnon, D.E. Morse, A.M. Sweene (2014): Photosymbiotic

giant clams are transformers of solar flux. J. Royal Soc. Interface DOI: 10.1098/rsif.2014.0678 +

rsif.2014.0678supp1

[12] L. Lilge, T. Haw, B.C. Wilson(1993): Minaíature isotropic optical fibre probes for

quantitative light dosimetry in tissue. Phys. Med. Biol. 38 215-230.

[13] W.K. Dodds (1992): A modified fiber-optic light microprobe to measure spherically

integrated photosyntethic photon flux density: Chararaterization of periphyton

photosynthesis-irradiance patterns. Limnol. Oceanogr. 37(4) 871-878.

[14] C. Lassen, H. Plough, B.B. Jørgensen (1992): A fiber-optic scalar irradiance microsensor:

application for spectral light measurements in sediments. FEMS Microbiology Ecology 86

247-254

[15] F. Garcia-Pichel (1995): A scalar irradiance fiber-optic microprobe for measurement of

ultraviolet radiation at high spatial resolution, Photochem. Photobiol. 61 (3) 248-254

[16] M. Kühl, R.N. Glud, H. Plough, N.B. Ramsing (1996): Microenvironmental control of

photosynthesis and photosynthesis-coupled respiration in an epilithic cyanobacterial biofilm.

J. Phycol. 32 799-812

[17] Star, W.M., J.P.A. Marijnissen, J.C. van Gemert (1988): Light dosimetry in optical

phantoms and in tissues: 1. Multiple flux and transport theory. Phys. Med. Biol. 33 437-454

[18] B.M. Bebout, F. Garcia-Pichel (1995): UV B-Induced Vertical Migrations of Cyanobacteria

in a Microbial Mat. App. Environ. Microbiol. 4215–4222

[19] H. Plough, C. Lassen, B.B. Jørgensen (1993): Action spectra of microalgal photosynthesis

and depth distribution of spectral scalar irradiance in a costal marine sediment of

Limfjorden, Denmark. FEMS Microbiol. Ecol. 102 261-270

[20] R. Li, K.S. Pedersen, Y. Liu, H.S. Pedersen. M. Lægdsmand, L.F. Rickelt, M. Kühl, H.

Callesen (2014): Effect of red light on the development and quality of mammalian embryos.

J. Assist. Reprod. Genet. DOI: 10.1007/s10815-014-0247-7


38

[21] D. Wangpraseurt, A.W.D. Larkin, P.J. Ralph, M. Kühl (2012): Light gradients and optical

microniches in coral tissue. Front. Microbiol. DOI: 10.3389/fmicb.2012.00316 +

supplement

[22] M.A.A. Al-Najjar, D. de Beer, M. Kühl, L. Polerecky (2012): Light utilization in

photosynthetic microbial mats. Environ. Microbiol. 14 (2) 982-992

[23] J.P.A. Marijnissen, W.M. Star (2002): Performance of isotropic light dosimetry probes

based on scattering bulbs in turbid media. Phys. Med. Biol. 47 (12) 2049-2058

[24] Michael Kühl, Carsten Lassen, Niels Peter Revsbech (1997): A simple light meter for

measurements of PAR (400 to 700 nm) with fiber-optic microprobes: application for P vs

E0(PAR) measurements in a microbial mat. Aquatic Microbial Ecology 13 197-207

[25] M. Larsen, S.M. Borisov, B. Grunwald, I. Klimant, R.N. Glud (2011): A simple and

inexpensive high resolution color ratiometric planar optode imaging approach: application to

oxygen and pH sensing. Limnol. Oceanogr. Methods 9 348-360

[26] M. Lichtenberg, M. Kühl (2015): Pronounced gradients of light photosynthesis and O2

consumption in the tissue of the brown alga Fucus serratus. New Phytologist 7 DOI:

10.1111/nph.13396

[27] J.P.A. Marijnissen, W.M. Star (1996): Calibration of isotropic light dosimetry probes based

on scattering bulbs in clear media. Phys. Med. Biol. 41 (7) 1191-1208

[28] L.F. Rickelt, L.D.M. Ottosen, M. Kühl (2015): Etching of multimode glass fibers: A new

method for shaping the measuring tip and immobilization of indicator dyes in recessed fiber-

optic microprobes. Sen. Actuators B Chem. 211 462-468

8. Effect of red light on mammalian embryos

R. Li, K.S. Pedersen, Y. Liu, H.S. Pedersen, M. Lægdsmand, L.F. Rickelt, M. Kühl, and H.

Callesen:

Effect of red light on development and quality of mammalian embryos.

Journal of Assisted Reproduction and Genetics 31: 795-801 (2014)

DOI: 10.1007/s10815-014-0247-7

TECHNOLOGICAL INNOVATIONS

Effect of red light on the development and quality of mammalianembryos

Rong Li & Kamilla Sofie Pedersen & Ying Liu &

Hanne Skovsgaard Pedersen & Mette Lægdsmand &

Lars Fledelius Rickelt & Michael Kühl & Henrik Callesen

Received: 13 January 2014 /Accepted: 5 May 2014# Springer Science+Business Media New York 2014

AbstractPurpose To assess irradiance and total energy dose fromdifferent microscopes during the in-vitro embryonic develop-mental cycle in mouse and pig and to evaluate its effect onembryonic development and quality in pig.Method Spectral scalar irradiance (380–1050 nm) was mea-sured by a fiber-optic microsensor in the focal plane of adissection microscope, an inverted microscope and a time-lapse incubation system. Furthermore, the effect of three differ-ent red light levels was tested in the time-lapse system onmousezygotes for 5 days, and on porcine zona-intact and zona-freeparthenogenetically activated (PA) embryos for 6 days.Results The time-lapse system used red light centered at625 nm and with a lower irradiance level as compared to thewhite light irradiance levels on the dissection and invertedmicroscopes, which included more energetic radiation<550 nm. Even after 1000 times higher total energy dose ofred light exposure in the time-lapse system, no significant

difference was found neither in blastocyst development ofmouse zygotes nor in blastocyst rates and total cell numberof blastocysts of porcine PA embryos.Conclusions Our results indicate that red light (625 nm, 0.34W/m2) used in the time-lapse incubation system does not decreasethe development and quality of blastocysts in both mouse zy-gotes and porcine PA embryos (both zona-intact and zona-free).

Keywords Spectral irradiance .Microsensor . Dosimetry .

Mouse . Pig . Time-lapse observation

Introduction

Mammalian oocytes and embryos are typically exposed toseveral kinds of light exposure during their in-vitro process-ing, which all can affect the subsequent embryonic develop-ment and quality [1, 2]. This can encompass fluorescent roomlight in the lab, halogen light on the microscopes and evendaylight during ovary collection. Some types of light such assunlight are lethal, but can easily be avoided [3], whereassome light exposure for observation during in-vitro manipu-lations is inevitable. There is e.g. a need for frequent observa-tions to select the best in-vitro produced embryos for transfer,because better embryonic quality means higher chance forimplantation and ultimately pregnancy [4–6]. In practice, thenumber of observations used in clinical IVF on human andanimal embryos is therefore a compromise between a wish tokeep the embryos undisturbed (with respect to light, temper-ature and pH) and to have as many observations as possible tofacilitate optimal selection for transfer.

For decades, evaluation of embryonic quality hasbeen based on simple morphological observations per-formed either a couple of times during culture or at thetime of selection for transfer. The limitations in thisapproach have been known just as long, but there was

Capsule Our results indicate that red light (625 nm, 0.34 W/m2) used inthe time-lapse incubation system does not decrease the development andquality of blastocysts in both mouse zygotes and porcine PA embryos(both zona-intact and zona-free).

R. Li (*) :Y. Liu :H. S. Pedersen :H. CallesenDepartment of Animal Science, Aarhus University, Blichers Allé 20,DK-8830 Tjele, Denmarke-mail: [email protected]

K. S. Pedersen :M. LægdsmandUnisense FertiliTech A/S, Tueager 1, DK-8200 Aarhus N, Denmark

K. S. PedersenDivision of Life Science, Medical Biotechnology, DanishTechnological Institute, Kongsvang allé 29, DK-8000 Aarhus,Denmark

L. F. Rickelt :M. KühlMarine Biological Section, Department of Biology, University ofCopenhagen, Strandpromenaden 5, DK-3000 Helsingør, Denmark

J Assist Reprod GenetDOI 10.1007/s10815-014-0247-7

not any realistic alternative until the introduction oftime-lapse cinematography to observe mammalian em-bryos some 30 years ago [7, 8]. Since then, this type offrequent observation with images taken regularly forsubsequent analyses has become more or less routinein several IVF set-ups, especially in the human field[9–12]. In the animal field, such additional measureshave found less application except for experimentalpurposes where it is greatly valued [13, 14]. With therapidly spreading use of time-lapse observations, espe-cially on human embryos in fertility clinics, it is impor-tant to investigate, whether light exposure for imagerecording affects or even compromises the embryo’sdevelopmental competence. Different measures basedon studies of animal model embryos have already beenimplemented to minimize light exposure of visible [15]or longer wavelength light [2, 15, 16], and in combina-tion with sensitive camera systems it is possible toobtain high quality images for evaluation without disad-vantages on embryonic development and quality [17].Microscopy typically works with visible light (380–700 nm), where wavelengths <500-550 nm are regardedas harmful to the development and quality of mamma-lian embryos [2, 15]. Increasing exposure time leads tohigher total energy dose potentially causing more stressto the embryos. In mammalian cells, the function offibroblast cells was disrupted when the total dose ofblue light was increased to 10 kJ/m2 [18]. However,most studies only use light intensity to quantify illumi-nation effect on embryos [3, 15, 16]. Light at longerwavelength carries lower energy and red light is recom-mended as a safe illumination source for embryo obser-vation systems [15] and is now routinely applied in time-lapse incubation systems [11, 12, 19–21]. However, themaximum tolerance of embryos to red light is unknown,and no comparisons have yet been made of an eventualeffect of maximizing light exposures to obtain even moremorphological details over a full pre-implantation periodin-vitro, i.e. for 5, 6 or 7 days.

In the present study, embryos were therefore exposed tothree different levels of red light during the whole in-vitroculture period in a time-lapse system: no light, low exposurefor normal image recording for time-lapse observation (75 ms/15 min and 75 ms/20 min), or extra exposure (50 s/20 min,60 s/15 min and 100 s/20 min). Mouse embryos were chosento represent a robust model that is fairly tolerant to lightexposure [3], while porcine embryos were applied as a modelmore sensitive to in-vitro conditions [22, 23]. The specificpurposes of our study were: (1) to compare the total lightenergy dose under different illumination scenarios during anin-vitro mammalian embryonic developmental cycle, as quan-tified by microscale spectral irradiance measurements; (2) totest the developmental rates and quality of both mouse and pig

embryos after extra light exposure of red light in a time-lapseincubator.

Materials and methods

All chemicals were purchased from Sigma–Aldrich Corp.(St Louis, MO, USA) except otherwise indicated.

Light measurements

To quantify light exposure of embryos when observed on aninverted microscope, a dissection microscope and in a time-lapse incubation system (EmbryoScope®; UnisenseFertiliTech A/S, Aarhus, Denmark), spectral scalar irradiance(380–1050 nm) was measured by a fiber-optic microsensorconnected to a fiber-optic spectrometer calibrated for absoluteirradiance measurement (USB2000, Ocean Optics, USA) [24,25]. Briefly, a scalar irradiance micro-probe with a tip diam-eter of ~100 μm, i.e., similar dimensions as a mouse embryo,was used to measure the spectral light exposure (in units ofμW/cm2/nm) in a culture dish (shown in Fig. 1) mounted onthe two different microscope types and in the well of anEmbryoSlide® culture dish in the time-lapse incubator(Unisense FertiliTech A/S). Measurements were conductedin the focal plane on all three microscope platforms withtypical light settings used for embryo inspection as adjustedby a trained person. Measurements were recorded in 9 differ-ent positions in the light exposed field of view. Spectral datawere calculated as the means of these 9 values (Fig. 2a). Basedon spectral integration, the total scalar irradiance was calcu-lated for 380–1050 nm light, as well as for the harmfulwavelength range 380–550 nm only (Fig. 2b).

Fig. 1 Photograph of the spherical light collecting tip of a scalar irradi-ance microsensor positioned in a culture dish next to a mouse embryo

J Assist Reprod Genet

Preparation of culture dish

All 12 wells in an EmbryoSlide® culture dish (UnisenseFertiliTech A/S) were filled with 25 μl cultivation media(for mouse: Global culture medium (LifeGlobal, Guelph,Canada); for pig: porcine zygote medium-3 (PZM-3,[26])) at room temperature; any gas bubbles were careful-ly removed using a glass pipette. To cover the 12 wells,1.2 ml mineral oil was filled into the common reservoir.The culture dish was left to equilibrate for at least 20 heither at 37 °C and 5 % CO2 in air for mouse embryos orat 38.5 °C in an atmosphere of 5 % CO2, 5 % O2, and90 % N2 for porcine PA embryos.

Mouse embryo preparation

Cryopreserved 1-cell embryos from cross-bred mice wereobtained from Vitrolife (Vitrolife Inc, San Diego, USA). Thestraw containing the embryos was removed from the liquidnitrogen tank and left to thaw horizontally for 2 min in air atroom temperature. Subsequently, the holder was removed, andthe straw was equilibrated for 3 min in a 37 °C water bath andfor 2min in a 20-25 °Cwater bath. After thawing, the embryoswere washed twice in 4-well dishes (Nunc, Roskilde,Denmark) with GALW washing media (LifeGlobal) cov-ered with LiteOil (LifeGlobal) to prevent evaporation.After washing, the embryos were transferred to 4-welldishes with Global culture media supplemented with7.5 % LGPS (LifeGlobal,) and then transferred to theEmbryoSlide® culture dish. The mouse embryos fromboth light treatment and normal observation groups werecultured in Global culture media supplemented with7.5 % LGPS (LifeGlobal) in EmbryoSlide® culturedishes at 37 °C with 5 % CO2 in air.

Porcine embryo production

Cumulus–oocyte complexes (COCs) were aspirated from 2 to6 mm follicles in slaughterhouse-derived sow ovaries andmatured as described earlier [14]. Briefly, groups of 50COCs with compact and at least two layers of cumulus cellswere selected and cultured for 42–44 h in 4-well dishes in400 μl bicarbonate-buffered TCM-199 (supplemented with10 % (v/v) cattle serum (CS; Danish Veterinary Institute,Frederiksberg, Denmark), 10 % (v/v) sow follicular fluid,10 IU/ml pregnant mare serum gonadotrophin and 5 IU/mlhuman chorionic gonadotrophin (Suigonan Vet, Intervet,Boxmeer, Holland)) covered with 400 μl mineral oil at38.5 °C in 5 % CO2 with 100 % humidity.

After maturation, oocytes were parthenogenetically activat-ed (PA) as described earlier [14]. Briefly, oocytes were equil-ibrated for 10–15 s in drops of activation medium (0.3 Mmannitol, 0.1 mM MgSO4, 0.1 mM CaCl2 and 0.01 % poly-vinyl alcohol). Under a 0.12 kV/cm alternating current, oo-cytes were aligned to the wire of a fusion chamber (Microslide0.5-mm fusion chamber, model 450; BTX, San Diego, CA,USA). Meanwhile, a single direct current pulse (1.26 kV/cm,80 μs) was applied. The time of activation by electricity wasdefined as Day 0. After washing twice in drops of TCM-199(supplemented with 10 % CS), groups of 100 oocytes wereincubated for 4 h in 400 μl PZM-3 (supplemented with4 mg/ml bovine serum albumin, 5 μg/ml cytochalasin B and10 μg/ml cycloheximide) covered with 400 μl mineral oil at38.5 °C in an atmosphere of 5 % CO2, 5 % O2 and 90 % N2

with 100 % humidity. PA embryos were then washed threetimes in culture medium prior to culture. To acquire zona-freePA embryos, the zona pellucida (ZP) was removed as de-scribed earlier [14]. In brief, the embryos were placed in0.3 % (w/v) pronase for 30 s followed by immediate washing2–3 times in culture medium, and then the remaining ZP was

Fig. 2 Spectral composition and intensity of observation light on differ-ent microscopes. A) Spectral scalar irradiance. The harmful wavelengthrange (380–550 nm) is shown in the gradient area; B) Integral scalar

irradiance for total light (380–1050 nm, shown in outlined column) andlight <550 nm (harmful part of light, shown in color column) on thedifferent microscopes


removed mechanically by repeated pipetting by a glass pipette(diameter: 200–300 μm).

The porcine embryos from both extra light treatment andnormal observation groups were cultured in PZM-3 inEmbryoSlide® culture dishes at 38.5 °C in an atmosphere of5 % CO2, 5 % O2 and 90 % N2. As control, embryos werecultured in groups of 20–25 per well in 4-well dishes in 400 μlculture medium covered with 400 μl oil at 38.5 °C in anatmosphere of 5 % CO2, 5 % O2 and 90 % N2 with 100 %humidity.

Time-lapse observation of embryo development

The embryos were cultured in the EmbryoScope® time-lapseincubator in the EmbryoSlide® culture dishes. Mouse embry-os were cultured for 5 days at 37 °C in 5 % CO2 in air, andporcine PA embryos were cultured for 6 days at 38.5 °C in anatmosphere of 5 % CO2, 5 % O2 and 90 % N2. All embryosfrom the control groups were cultured without light in stan-dard incubator for 5 days (mouse) or 6 days (pig). Digitalimages of the cultured embryos were obtained at each record-ing interval (15/20 min) with a red light emitting diode (LED,R42182, Seoul Semiconductor, Korea) that was only turnedon during image acquisition at 75 ms intervals (15 ms for eachimage capture x 5 different focus planes=75 ms/15-20 min).The images were collected and analyzed using theEmbryoViewer® software (Unisense Fertilitech A/S).

The low light exposure treatment used the standard lightsettings for image recording on the instrument. For the extralight exposure treatment, a modified EmbryoScope® time-lapse system was used to keep the light on for 60 s/15 minor 100 s/20 min.

Evaluation of the embryonic development and quality

Embryonic development was checked on Day 5 for mouseand Day 6 for pig embryos. For mouse embryos, only blasto-cyst formation was assessed and defined as the formation of ablastocoel cavity. All porcine blastocysts were evaluated bydividing the quality into 4 grades according to their morphol-ogy as previously described [14]: (1) Excellent: Spherical,regular border, symmetrical with cells of uniform size, evendistribution of color and texture in trophectoderm; (2) Good:Fragmentations (<10 %), irregular shape of blastocoel cavity;(3) Fair: Fragmentations (10-30 %), vesiculation of cavity; (4)

Poor: Fragmentations (>30 %), varying sizes of cells, numer-ous vesicles.

The quality of porcine blastocysts was evaluated by thetotal cell number on Day 6. Briefly, the blastocysts werestained for 20 min with 1 μg/ml Hoechst 33342 (using 4 %paraformaldehyde as base medium) and were then mounted inglycerol on a microscope glass slide. Stained embryos wereexamined and images were taken on fluorescence microscopy(360±20 nm excitation, ebq 100 Filter, Leica, Germany).

Statistical analysis

A Chi-square test was used to analyze the development rate ofmouse embryos. For the porcine PA embryos, 3 replicateswere performed for each experiment, and one-way ANOVA,Tukey’s Honestly Significant Difference (TukeyHSD) testwas used to analyze the blastocyst rates and total cell number.All statistical analyses were performed with software of R(version 2.14.2). A value of P<0.05 was considered to bestatistically significant.

Experimental design

Experiment 1: Comparison of the spectral scalar irradiance inthree systems: inverted microscope (used for intracytoplasmicsperm injection), dissection microscope (used for embryohandling and checking) and EmbryoScope® time-lapsesystem.

Experiment 2: Testing the effect of the light source onmouse embryos.

Experiment 3: Comparison of the embryonic developmentand quality of porcine PA embryos (zona-intact) using threelevels of light exposure with red LED light.

Experiment 4: Comparison of the embryonic developmentand quality of zona-free porcine PA embryos under threelevels of light exposure with red LED light.

Results

Experiment 1: The spectral scalar irradiance in the threemicroscope systems (Fig. 2).

There was no light at <550 nm in the time-lapse system(Fig. 2a), and the system illuminated specimens with anintegrated scalar irradiance (380–1050 nm) of 0.34 W/m2

Table 1 Total energy dose in 3levels of light exposure for 5 daysobservation and the effect on theblastocyst rate of mouse embryosin the time-lapse system

Exposure time/interval 75 ms/20 min 50 s/20 min 100 s/20 min

Total dose (J/m2) 10 6120 12240

Blastocyst % (Blastocysts/zygotes) 96 (27/28) 100 (11/11) 95 (20/21)

Replicates 3 1 2


when the red LED light was turned on (Fig. 2b). In compar-ison, the two microscopes illuminated specimens with higherscalar irradiance, including the potentially harmful 380–550 nm range (Fig. 2b).

Experiment 2: The effect of red light in the time-lapsesystem on mouse embryos (Table 1).

In 60 incubated mouse zygotes, no significant differencewas found in the blastocyst rates, even when the total lightdose was increased up to 1000 times (>10 kJ/m2).

Experiment 3: The development and quality of zona-intactporcine PA embryos after 3 levels of red light exposure in thetime-lapse system (Table 2).

In 130 activated oocytes, there were no significant differ-ences in the blastocyst rates and quality between control andextra light exposure groups.

Experiment 4: The development and quality of zona-freeporcine PA embryos after 3 levels of red light exposure intime-lapse system (Table 3).

Of 267 activated oocytes, 131 were used in one experimentwith a light exposure of 60 s/15 min, while 136 were used inan experiment with a higher light exposure of 100 s/20 min.There was no decrease in the blastocyst rates and qualitybetween the control and extra light exposure groups.

Discussion

Light is a common stress factor for embryos during their in-vitro processes, especially when being handled and observedfor evaluation and selection. The type [2, 3, 15], intensity [15]

and exposure time [27] of light can affect the subsequentdevelopment of embryos. In the present study, the light sourcein the time-lapse systemwas a red LED light emitting within anarrow wavelength range peaking at 625 nm. The scalarirradiance and therefore light exposure in this system wasfound to be much lower than in dissection and invertedmicroscopes in both tested wavelength areas. Furthermore,the blastocyst rates and total cell numbers in both mouse andporcine embryos were not affected even after a considerablyincreased exposure to red light.

A total dose of 10 kJ/m2 blue light was found to damage thefunction ofmammalian fibroblast cells [18]. In the present study,we used a maximum energy dose of red light in the time-lapsesystem of 10–15 kJ/m2 after extra 100 s/20 min light exposureover 5 to 6 days. This caused no significant difference in theblastocyst rates and quality of incubated embryos from bothmouse and pig. Possible explanations for this result can be:

(i) The red light used in the EmbryoScope® time-lapsesystem is not harmful. Light exposure to the wavelength rangeof 380–550 nm has been shown to induce harmful effects onembryonic quality and survival under in-vitro manipulationand monitoring [2, 15]. This spectral range can increase heatshock proteins (Hsp70) gene expression and the formation ofreactive oxygen species, which result in more apoptotic cellsappearing at the blastocyst stage [2, 15].

(ii) The exposure to red LED light in the time-lapse systemwas low. Under visible light intensities up to 1200 lux, thedevelopment and quality of hamster embryos can be significant-ly decreased after only 30 min exposure [3]. Moreover, thedevelopment and quality of embryos will gradually decrease

Table 2 Effect of extra light exposure (60 s/15 min) on development and quality of zona intact porcine PA embryos

Exposuretime/interval

Total time (s) Total energy (J/m2) Activated oocytes(Replicates)

Blastocyst%* Grade1 & 2blastocyst%*

Total cellnumber**

60 s/15 min 34560 11750 36 (3) 74.7±8.3 (27) 55.7±5.6 (20) 55.2±2.7 (27/3)

75 ms/15 min 44 15 36 (3) 88.7±5.7 (32) 72.3±2.8 (26) 51.1±3.2 (32/3)

Control 0 0 58 (3) 77.7±2.7 (45) 62.3±3.6 (36) 57.6±3.1 (25/2)

*mean of replicates±SEM (No. of blastocysts), **mean of blastocyst±SEM (No. of blastocysts/replicates). There is no significant difference in each column

Table 3 Effect of extra light exposure (60 s/15 min and 100 s/20 min) on development and quality of zona free porcine PA embryos

Exposuretime/interval

Total time (s) Total energy (J/m2) Activated oocytes(Replicates)

Blastocyst%* Grade1 & 2blastocyst%*

Total cellnumber**

60 s/15 min 34560 11750 36 (3) 69.3±7.3 (25) 52.7±12.2 (19) 34.6±2.1 (15/2)

75 ms/15 min 44 15 36 (3) 75±4.6 (27) 52.7±5.3 (18) 41.4±3.2 (18/2)

Control 0 0 59 (3) 81.3±6.8 (49) 66±9.5 (38) 41.9±2.8 (29/2)

100 s/20 min 43200 14690 36 (3) 72.3±5.3 (26) 44.6±2.7 (16) 51.5±5.1 (8/1)

75 ms/20 min 33 10 36 (3) 69.4±10.0 (25) 47.3±2.7 (17) 46.8±2.2 (6/1)

Control 0 0 64 (3) 65.3±15.6 (43) 52.7±15.2 (35) 59.3±4.6 (8/1)

*mean of replicates±SEM (No. of blastocysts), **mean of blastocyst±SEM (No. of blastocysts/replicates). There is no significant difference in each column


with increasing light intensity when the intensity was over 200lux [15]. The light intensity in the EmbryoScope® time-lapsesystemwas much lower, and our results indicate that even undera dose of 10–15 kJ/m2 of red light, i.e., the maximal induciblelight level during observation in the EmbryoScope® time-lapsesystem, there was apparently no harmful effect on the develop-ment and quality of mammalian embryos.

Over a 5 to 7 day normal observation period in the time-lapse system, we showed that the total exposure time wasmaximally 50 s resulting in a total energy dose of 10–20 J/m2.This is a very short exposure time and represents a much lowertotal energy dose as compared to light exposure under differ-ent traditional manipulation steps such as a typical 10–15 mincommon evaluation of embryo quality under a dissectionmicroscope, or e.g. a typical time frame of 30 min forperforming an intracytoplasmic sperm injection on theinverted microscope. It is thus important to reduce the lightexposure during such in-vitro manipulations, either by de-creasing exposure time and irradiance [15, 27] or by avoidingharmful wavelengths by the use of filters or LED’s with amore narrow spectral emission [2, 28].

Conclusion

The red light (625 nm, 0.34 W/m2) applied in theEmbryoScope® time-lapse system does not compromise theblastocyst rates and quality in either mouse zygotes or porcinePA embryos (both zona-intact and zona-free).

Acknowledgments The authors thank Anette M. Pedersen, JanneAdamsen, Klaus Villemoes and Ruth Kristensen for excellent technicalassistance. Part of this work was supported by grants from the DanishNational Advanced Technology Foundation and the Danish Council ofIndependent Research.

References

1. Schultz RM. Of light and mouse embryos: Less is more. Proceedingsof the National Academy of Sciences of the United States of America2007; 104(37):14547–14548

2. Korhonen K, Sjovall S, Viitanen J, Ketoja E, Makarevich A, PeippoJ. Viability of bovine embryos following exposure to the greenfiltered or wider bandwidth light during in vitro embryo production.Human Reproduction. 2009;24(2):308–14.

3. Takenaka M, Horiuchi T, Yanagimachi R. Effects of light on devel-opment of mammalian zygotes. Proceedings of the NationalAcademy of Sciences of the United States of America.2007;104(36):14289–93.

4. Puissant F, Vanrysselberge M, Barlow P, Deweze J, Leroy F. EmbryoScoring As A Prognostic Tool in Ivf Treatment. HumanReproduction. 1987;2(8):705–8.

5. Fisch JD, Rodriguez H, Ross R, Overby G, Sher G. The GraduatedEmbryo Score (GES) predicts blastocyst formation and pregnancyrate from cleavage-stage embryos. Human Reproduction.2001;16(9):1970–5.

6. Vernon M, Stern JE, Ball GD, Wininger D, Mayer J, Racowsky C.Utility of the national embryo morphology data collection by theSociety for Assisted Reproductive Technologies (SART): correlationbetween day-3 morphology grade and live-birth outcome. Fertilityand Sterility. 2011;95(8):2761–3.

7. Massip A, Mulnard J. Time-Lapse Cinematographic Analysis ofHatching of Normal and Frozen-Thawed Cow Blastocysts. Journalof Reproduction and Fertility 1980; 58 (2):475-&.

8. Massip A, Mulnard J, Vanderzwalmen P, Hanzen C, Ectors F. TheBehavior of Cow Blastocyst Invitro - Cinematographic andMorphometric Analysis. Journal of Anatomy 1982; 134 (MAR):399–405.

9. Payne D, Flaherty SP, Barry MF, Matthews CD. Preliminary obser-vations on polar body extrusion and pronuclear formation in humanoocytes using time-lapse video cinematography. HumanReproduction. 1997;12(3):532–41.

10. Wong CC, Loewke KE, Bossert NL, Behr B, De Jonge CJ, Baer TM,et al. Non-invasive imaging of human embryos before embryonicgenome activation predicts development to the blastocyst stage.Nature Biotechnology. 2010;28(10):1115–99.

11. Kirkegaard K, Hindkjaer JJ, Grondahl ML, Kesmodel US, IngerslevHJ. A randomized clinical trial comparing embryo culture in aconventional incubator with a time-lapse incubator. Journal ofAssisted Reproduction and Genetics. 2012;29(6):565–72.

12. Meseguer M, Herrero J, Tejera A, Hilligsoe KM, Ramsing NB,Remohi J. The use of morphokinetics as a predictor of embryoimplantation. Human Reproduction. 2011;26(10):2658–71.

13. Holm P, Shukri NN, Vajta G, Booth P, Bendixen C, Callesen H.Developmental kinetics of the first cell cycles of bovine in vitroproduced embryos in relation to their in vitro viability and sex.Theriogenology. 1998;50(8):1285–99.

14. Li R, Liu Y, Pedersen H, Kragh P, Callesen H. Development andquality of porcine parthenogenetically activated embryos after re-moval of zona pellucida. Theriogenology. 2013;80(1):58–64.

15. Oh SJ, Gong SP, Lee ST, Lee EJ, Lim JM. Light intensity andwavelength during embryo manipulation are important factors formaintaining viability of preimplantation embryos in vitro. Fertilityand Sterility. 2007;88:1150–7.

16. Takahashi M, Saka N, Takahashi H, Kanai Y, Schultz RM, Okano A.Assessment of DNA damage in individual hamster embryos by cometassay. Molecular Reproduction and Development. 1999;54(1):1–7.

17. Nakahara T, Iwase A, Goto M, Harata T, Suzuki M, Ienaga M, et al.Evaluation of the safety of time-lapse observations for human em-bryos. Journal of Assisted Reproduction andGenetics. 2010;27(2–3):93–6.

18. Wataha JC, Lockwood PE, Lewis JB, Rueggeberg FA, Messer RLW.Biological effects of blue light from dental curing units. DentalMaterials. 2004;20(2):150–7.

19. CruzM,Garrido N, Herrero J, Perez-Cano I, MunozM,MeseguerM.Timing of cell division in human cleavage-stage embryos is linkedwith blastocyst formation and quality. Reproductive BiomedicineOnline. 2012;25(4):371–81.

20. Li R, Li J, Liu Y, Kragh PM, Hyttel P, Schmidt M, et al. Optimaldevelopmental stage for vitrification of parthenogenetically activatedporcine embryos. Cryobiology. 2012;64(1):60–4.

21. Campbell A, Fishel S, Bowman N, Duffy S, Sedler M, Thornton S.Retrospective analysis of outcomes after IVF using an aneuploidyrisk model derived from time-lapse imaging without PGS.Reproductive Biomedicine Online. 2013;27(2):140–6.

22. Niemann H. Sensitivity of Pig Morulae to Dmso/Pvp Or GlycerolTreatment and Cooling to 10-Degrees-C. Theriogenology.1985;23(1):213.

23. Wilmut I. Low-Temperature Preservation of Mammalian Embryos.Journal of Reproduction and Fertility. 1972;31(3):513–4.

24. KühlM. Optical microsensors for analysis of microbial communities.Methods in Enzymology. 2005;397:166–99.


25. Exposure to light during image acquisition in the EmbryoScope.http://www.fertilitech.com/Admin/Public/DWSDownload.aspx?File=%2FFiles%2FFiles%2Fdownloads%2FLight+exposure+in+EmbryoScope+v4.pdf. 2012. Ref Type: Online Source

26. Yoshioka K, Suzuki C, Tanaka A, Anas IMK, Iwamura S. Birth ofpiglets derived from porcine zygotes cultured in a chemically definedmedium. Biology of Reproduction. 2002;66(1):112–9.

27. Zhang JQ, Li XL, Peng YZ, Guo XR, Heng BC, Tong GQ.Reduction in exposure of human embryos outside the incubatorenhances embryo quality and blastulation rate. ReproductiveBiomedicine Online. 2010;20(4):510–5.

28. Ottosen LDM, Hindkjaer J, Ingerslev J. Light exposure of the ovumand preimplantation embryo during ART procedures. Journal ofAssisted Reproduction and Genetics. 2007;24(2–3):99–103.


http://www.fertilitech.com/Admin/Public/DWSDownload.aspx?File=%2FFiles%2FFiles%2Fdownloads%2FLight+exposure+in+EmbryoScope+v4.pdf



9. Ultrabright planar optodes

M. Staal, S. Borisov, L.F. Rickelt, I. Klimant, and M. Kühl:

Ultrabright planar optodes for luminescence life-time based microscopic imaging of O2 dynamics in

biofilms.

Journal of Microbiological Methods 85: 67–74 (2011)

DOI: 10.1016/j.mimet.2011.01.021

Journal of Microbiological Methods 85 (2011) 67–74

Contents lists available at ScienceDirect

Journal of Microbiological Methods

j ourna l homepage: www.e lsev ie r.com/ locate / jmicmeth

Ultrabright planar optodes for luminescence life-time based microscopic imagingof O2 dynamics in biofilms

M. Staal a, S.M. Borisov b, L.F. Rickelt a, I. Klimant b, M. Kühl a,c,⁎a Marine Biological Section, Department of Biology, University of Copenhagen, Strandpromenaden 5, DK-3000 Helsingør, Denmarkb Institute of Analytical Chemistry and Radiochemistry, Graz University of Technology, Stremayrgasse 16/III, 8010 Graz, Austriac Plant Functional Biology and Climate Change Cluster, University of Technology Sydney, PO Box 123, Ultimo Sydney NSW 2007, Australia

⁎ Corresponding author at: Strandpromenaden 5, DTel.: +45 35321956; fax: +45 35321951.

E-mail address: [email protected] (M. Kühl).

0167-7012/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.mimet.2011.01.021

a b s t r a c t
a r t i c l e i n f o
Article history:Received 7 December 2010Received in revised form 17 January 2011Accepted 18 January 2011Available online 26 January 2011

Keywords:BiofilmImagingLuminescenceMicroscopyOxygenOptode

New transparent optodes for life-time basedmicroscopic imaging of O2 were developed by spin-coating a μm-thin layer of a highly luminescent cyclometalated iridium(III) coumarin complex in polystyrene onto glasscover slips. Compared to similar thin-film O2 optodes based on a ruthenium(II) polypyridyl complex or aplatinum(II) porphyrin, the new planar sensors have i) higher brightness allowing for much shorter exposuretimes and thus higher time resolution, ii) more homogeneous and smaller pixel to pixel variation over thesensor area resulting in less noisy O2 images, and iii) a lower temperature dependency simplifying calibrationprocedures. We used the new optodes formicroscopic imaging of the spatio-temporal O2 dynamics at the baseof heterotrophic biofilms in combination with confocal imaging of bacterial biomass and biofilm structure.This allowed us to directly link biomass distribution to O2 distribution under both steady state and non-steadystate conditions. We demonstrate that the O2 dynamics in biofilms is governed by a complex interactionbetween biomass distribution, mass transfer and flow that cannot be directly inferred from structuralinformation on biomass distribution alone.

K-3000 Helsingør, Denmark.

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Molecular oxygen (O2) is a key molecule for important biogeo-chemical and metabolic processes (Fenchel and Finlay, 2008; Glud,2008). It is produced by oxygenic phototrophs (cyanobacteria, algaeand plants) and is the preferred terminal electron acceptor inbiological breakdown of carbohydrates since it generates the highestenergy yield compared to other electron acceptors. Higher O2 levelsare critical e.g. for anaerobic microorganisms and processes, and foraerobic organisms due to formation of reactive oxygen species thatcan damage cellular processes. The O2 distribution in biologicalsystems can thus have strong effects on biogeochemical conversionrates and growth yields affecting the morphology of tissues and cellclusters. Local variations in O2 respiration or production rates coupledwith mass transfer limitations, e.g. due to the presence of diffusiveboundary layers, can lead to steep spatial gradients of O2 that responddynamically to environmental parameters such as light or flow(Fenchel and Finlay, 2008). Oxygen concentration can vary strongly atμm to mm scale and precise quantification of the O2 distribution anddynamics is a prerequisite for understanding the performance andregulation of many metabolic conversion rates, biotechnological andbiomedical processes.

Oxygen dynamics can be monitored at high spatio-temporalresolution with electrochemical or fiber-optic O2 microsensors(Klimant et al., 1995, 1997; Kühl, 2005; Revsbech, 2005). However,the spatial coverage is limited due to the one-dimensional nature ofsuch measurements, and it is difficult to describe the spatialheterogeneity of systems with microsensor techniques, especiallyunder non steady state conditions. With the development of planaroptodes, a new tool for mapping the spatial distribution of O2 becameavailable (Glud et al., 1996; Kühl and Polerecky, 2008). Planar optodesuse luminescent O2 indicators immobilized in a polymeric matrix,which is permeable to O2 and can be fixed on foils or glass surfaces.The measuring principle is based on the dynamic collisional quench-ing of the indicator luminescence by O2 (DeGraff and Demas, 2005).Using sensitive gated CCD camera systems, O2 dependent levels ofluminescence and its exponential decay characteristics can be imaged,ultimately resulting in a description of the two dimensionaldistribution of O2 (Holst et al., 1998; Liebsch et al., 2000; Oguriet al., 2006).

Hitherto planar O2 optodes have mostly been based on the use ofeither ruthenium(II) polypyridyl complexes or metallo-porphyrinesas O2 indicators (Amao, 2003; Wolfbeiss, 2005). These indicatorsexhibit moderate luminescence brightness with luminescence life-times in the μs–ms range. Recently, novel optical sensor materialswere developed, based on the use of cyclometalated iridium(III)coumarin complexes as O2 indicators with an exceptionally brightluminescence (Borisov and Klimant, 2007; DeRosa et al., 2004). These

http://dx.doi.org/10.1016/j.mimet.2011.01.021



http://www.sciencedirect.com/science/journal/01677012

Fig. 1. Absorption and luminescence emission spectra of optical O2 indicatorsimmobilized in polystyrene: ruthenium(II)-tris-4,7-diphenyl-1,10 phenantroline(Ru-DPP), platinum(II)-meso-tetra(pentafluorophenyl)porphyrin (Pt-TFPP), and irid-ium(III) acetylacetonato-bis(3-(benzothiazol-2-yl)-7-(diethylamino)-coumarin) (IrC).

68 M. Staal et al. / Journal of Microbiological Methods 85 (2011) 67–74

new O2-sensitive dyes are suitable for application in ultrathintransparent sensor layers, enabling very short response times andspatial resolution at the single cell level. A thin layer and shortresponse time are required e.g. when fast changes in O2 concentra-tions under non steady state conditions are monitored.

Here we present new thin layer optodes based on the new iridium(III) coumarin complexes immobilized in a 1 μm thick layer on amicroscope cover slip. We compare the O2 measuring characteristicsof the new optodes with other thin planar optodes based onruthenium(II) polypyridyl and platinum(II) porphyrin complexes(Kühl et al., 2007) and apply the new optodes for combinedmicroscopic imaging of biomass, O2 dynamics and spatial gradientsin heterotrophic microbial biofilms under different flow regimes.

2. Materials and methods

2.1. Optode preparation

Three different types of luminescent O2 indicators were used in thisstudy: i) ruthenium(II)-tris-4,7-diphenyl-1,10 phenantroline (Ru-DPP),ii) platinum(II)-meso-tetra(pentafluorophenyl)porphyrin (Pt-TFPP),and iii) iridium(III) acetylacetonato-bis(3-(benzothiazol-2-yl)-7-(diethylamino)-coumarin) (IrC). Ru-DPP and IrC were synthesized asdescribed elsewhere (Klimant and Wolfbeiss, 1995; Borisov andKlimant, 2007). Pt-TFPP was obtained commercially (Frontier ScientificInc., USA). All luminescent dyes were immobilized in the samepolystyrene (PS) matrix (ST316310/1 LS223989 JV, Goodfellow Ltd.,Cambridge, UK) to guarantee similar gas diffusion characteristicswithinthe optode. Dye concentrations (mg indicator per g PS) were 18.75 mgRu-DPP/g, 25 mg Pt-TFPP/g, and 15 mg IrC/g. Thin-film optodes werefabricated by spin coating ~1 μm thick indicator layers onto 20×50 mmsilanized microscope coverslips (Kühl et al., 2007).

2.2. Optode calibration setup

Optodes were calibrated at 20 °C in freshwater with defined O2

concentrations. The optodes were placed in a holder on the window ofa small water filled tank. The temperature of the tank was controlledby a cryostat (Julabo F25 HD, Germany). Oxygen levels in thecalibration chamber were varied by flushing the water with definedgas mixtures of N2 and O2 at a flow rate of 0.6 lmin−1. Mixtures weregenerated with a PC-controlled programmable gas mixing systemusing electronic mass flow controllers (Sensorsense, Netherlands).The O2 concentration in the water was increased step wise, at timeintervals including at least 2 min of steady state at each O2

concentration. Additionally, O2 levels in the water were monitoredwith a fiber-optic O2 minisensor system (Fibox 3, Presense GmbH,Germany). The temperature dependence of the planar sensorluminescence was measured at two different O2 concentrations insteps of 5 °C over a range of 5–30 °C.

2.3. Life-time imaging system and image calibration

Luminescence life-time and intensity of the optodes was imagedwith a modular luminescence life-time imaging system (Holst et al.,1998) consisting of i) a fast gate-able 12 bit SVGA CCD (1280×1024 pixel) camera (SENSICAM-SENSIMOD, PCO AG, Germany)equipped with a macro lens (Xenoplan XNP 1.4/17, Schneider-Kreutznach, Germany) and a 590 nm (30 nm bandwidth) bandpassfilter, ii) a custom built trigger box driving two high power blue LED's(1 W Luxeon Star, 470 nm, Lumileds) for the excitation of the Ru-DPPand IrC optodes, or two UV power LED's (405 nm 1W, RoithnerLasertechnik GmbH, Austria) for excitation of the Pt-TFFP optode, andiii) a custom-built PC-controlled pulse-delay generator. Imageacquisition and hardware control were done with a custom madesoftware program (Holst and Grunwald, 2001).

Life-time imaging with the system was done by acquiringluminescence intensity images (using a binning of 2), within twodifferent time windows, w1 and w2, after the eclipse of an excitationlight pulse. The luminescence life-time (τ) was calculated as(Gerritsen et al., 1997):

τ =Δt

lnðIw1=Iw2Þð1Þ

where Δt is the time delay between the start time of recording of thetwo time windows, and Iw1 and Iw2 are the corresponding lumines-cence images.

Since the three O2 indicator dyes have different light absorptionand life-time characteristics, we optimized the measuring protocolsfor each dye. In this study, we used excitation light pulses of 5, 5 and7 μs for Ru-DPP, IrC and the Pt-TFPP optodes, respectively. Twoluminescence intensity images (Iw1 and Iw2 ) were acquired over 3 μs.Image one (Iw1) was acquired at 0.1 μs while the second image (Iw2)was acquired at 3.1 (Ru-DPP), 4.1 (IrC) or 7.1 μs (Pt-TFPP) after theexcitation light pulse. These measurements were repeated and imagepixel values integrated on the CCD chip over an exposure time periodup to 400 ms to improve the signal to noise ratio. The first intensitywindow image is acquired by accumulation of Iw1 pulses over theintegration time, and then the same is done for the second intensitywindow image. In this study we used exposure times of 100, 300 and400 ms for IrC, Ru-DPP and Pt-TFPP optodes, respectively. Differentexposure times were necessary to achieve comparable pixel intensityvalues on the camera, i.e., to obtain minimum grey values of N400after subtraction of a dark image.

Image analysis was done in the custom made software Look@-MOLLIdata (Polerecky, http://www.mpi-bremen.de/Binaries/Bina-ry4997/Polerecky-EADS-report-2005.pdf) and the freeware ImageJ(http://rsbweb.nih.gov/ij/).We analysed randomly selected regions ofinterest (ROI's) (n=6–10) in the acquired images, each containing~5000 pixels; regions at the very sensor edge or in the unevenlyilluminated periphery of the field of view were avoided. From theROI's we derived the average luminescence intensity (pixel value) ofimage 1 and image 2, as well as the average life-time and its standarddeviation. Stern–Volmer plots of the data were fitted with a modified

http://www.mpi-bremen.de/Binaries/Binary4997/Polerecky-EADS-report-2005.pdf

http://www.mpi-bremen.de/Binaries/Binary4997/Polerecky-EADS-report-2005.pdf

http://rsbweb.nih.gov/ij/

A

B

C

Fig. 2. Luminescence life-time (τ) vs. O2 concentration (A) and corresponding Stern–Volmer relationship (B) measured at 20 °C for the Pt-TFPP (black squares), IrC (opencircles) and Ru-DPP (open triangles) optodes. C. Inverse of Stern–Volmer relationshipwith O2, including curve fits estimating α and Ksv.

69M. Staal et al. / Journal of Microbiological Methods 85 (2011) 67–74

Stern–Volmer equation that accounts for a non quenchable fraction,α, within the optode (Carraway et al., 1991):

II0

=ττ0

=1−α

I + KsvC+ α ð2Þ

where I0 and I are the luminescence intensities at anoxia and in thepresence of O2 at concentration C, and τ0 and τ are the equivalentluminescence life-times, α is the non-quenchable fraction within thesensor matrix, and Ksv is the quenching coefficient. Both Ksv and αwere fitted from the τ/τ0 vs. O2 concentration curves using a non-linear fitting routine (Origin 7.5, OriginLab Corp, USA).

2.4. Microscopic O2 imaging set up

Microscopic luminescence life-time imaging was done on anOlympus BX50W1 fixed stage microscope equipped with 40× and60× water immersion objectives (Plan-Apo WLSM/40×, NA 0.9;Uplan-ApoW/60×; NA 1.20, Olympus, Japan). The imaging systemwas the same as described above, except that the CCD camera wasmounted on the microscope via a C-mount, and a blue LED (5 WLuxeon star, 470 nm, Lumileds) was mounted in the epifluorescenceport of the microscope (more details in Kühl et al., 2007).

Since light capturing was very efficient by the high NA microscopeobjectives, the integration time per intensity image was only 20 ms(Iw1, Iw2 and Idark). Maximal sampling rate was 1 life-time image per1.5 s. The highest sampling rate applied in this study was one O2

distribution image per 3 s. The same CCD camera was also used toacquire bright field images of biomass distribution, in order to allowdirect comparison of the biomass pictures with the O2 distributionimages.

After the O2 measurements, the biofilm was stained with a redfluorescent nucleic acid stain (Syto® 60, Molecular Probes, Inc.,Eugene, OR). A spinning disk confocal imaging system (Ultraview LCI,Perkin Elmer) mounted on the same microscope (Kühl et al., 2007)was used to acquire stacks of optical sections through the biofilm(using laser excitation at 647 nm and collecting emission from theSyto 60 dye at N700 nm). The CCD camera used for confocal imagedetection had 1344×1024 pixels (Hamamatsu ORCA ER, HamamatsuPhotonics Inc., Japan).

There was only little overlap between the emission spectra of theIrC optode and the Syto 60 dye (max excitation/emission 652/678 nm; http://probes.invitrogen.com/media/pis/td11341.pdf). How-ever, due to the high amount of IrC in the planar optode relative to theSyto 60 staining of the biofilm, there was significant luminescencedetected from the optode when we measured the biomass with theconfocal microscope close to the biofilm base. The first surface imagein the biomass image stack was therefore affected by the optode andwas excluded from the biomass reconstruction.

All image treatment was performed in Image J (freely available athttp://rsbweb.nih.gov/ij/). Image handling comprised backgroundcorrection, multiplication with a flat field correction image andsummation of the resulting image stack as a proxy for biomassdistribution. Since the confocal imaging and life-time imaging werenot made with the same camera, and the resolution of the life-timeimages and the biomass images were not equal, the pixel size of thelatter was reduced to the size of the life-time image. After resizing, thecontours of the structures were determined by simple thresholding,and these contours were pasted into the life-time images. In addition,profile plots of the biomass images were extracted along the samelines as O2 concentration profile plots extracted from O2 images toenable comparison of biomass distribution and O2 distribution.

2.5. Biofilm growth

Biofilms of heterotrophic bacteria were grown in small flow cells(Pamp et al., 2009; Stovall Life Science Inc, USA) with the glass coverreplaced by an IrC thin-film optode glued to the flow cell withUV-curing adhesive (Adhesive 426, LightWelder PC-3; Dymax EuropeGmbH, Germany). Each lane in the growth chamber was 4 cm long,4 mm wide and 1 mm deep. A biofilm inoculum was grown from asediment sample incubated in an aerated carbon rich mediumcontaining basic nutrients (0.01 gl−1 phosphate, 0.001 gl−1 magne-sium, 0.1 gl−1 glucose, 0.1 g l−1 Ca-acetate and0.1 gl−1 Na-succinate).The inoculum was flushed through the flow chamber at low flow(1 ml h−1) to allow initial attachment of bacteria to the optode surface.Thereafter, new medium was continuously pumped through the flowchamber (flow rate 4 ml h−1, resulting in a flow of 1 mh−1 and aretention time of 2.4 min)with an adjustable high precision peristalticpump (Minipulse 3, Gilson, France) to select for biofilm formingorganisms. At these rates, flow in this type of biofilm growth chamberis laminar (cf. Pamp et al., 2009). Visible biofilm structures formedwithin 5 days. All tubing's (made of O2 impermeable Tygon tubing)were replaced and fresh aerated mediumwas prepared prior to actualmeasurements to prevent substrate and O2 consumption before themedium reached the incubation chamber. During the flow experi-ments, the retention time of themedium varied from 2 s at the highestflow to 7 min at the lowest flow rate.

http://probes.invitrogen.com/media/pis/td11341.pdf

http://rsbweb.nih.gov/ij/


3. Results

3.1. Measuring characteristics of thin-film optodes

The absorption maxima of the IrC and Ru-DPP optodes are in theblue spectral region, while thePt-TFPP optode has its highestabsorption in the UV region (Fig. 1). The luminescence emissionmaxima are 565 nm, 605 nm and 650 nm for IrC, Ru-DPP and Pt-TFPPoptodes, respectively. The new IrC-based planar optodes showed athree times higher brightness than Ru-DPP-based optodes for thesame blue light excitation intensity, and in order to acquire images atsimilar pixel values, we used a three times longer integration time perimage for the Ru-DPP optode (100 vs. 300 ms). To reach similar pixelvalues in measurements with the Pt-TFPP optode we used anintegration time of 400 ms. However, the latter was partly due tothe use of different LED's (405 nm) with a less focused emissioncharacteristic.

There was a large difference in luminescence life-time between thethree tested optodes (Fig. 2). At 20 °C, the Pt-FTTP optode had life-timesranging from 11 to 39 μs, IrC had life-times between 4.5 and 10.6 μs,while Ru-DPP had life-times between 3.7 and 5.5 μs, respectively, at 0 to35% O2 in the gas phase. The IrC optodes exhibited an almost Stern–Volmer relationship. Fitting the inverse of the Stern–Volmer curveshowed that newly prepared IrC optodes exhibited a quasi ideal Stern–Volmer relationship following Eq. (2) with an α value close to zero, i.e.α=0.02±0.03 (mean±standard deviation). Ru-DPP and Pt-FTTP hadhigher α-values of 0.11±0.05 (mean±standard deviation).

All optodes showed decreasing luminescence life-times withincreasing temperature (SFig 1, 2). Under anoxic conditions, the τ0values of IrC, Pt-TFPP and Ru-DPP optodes showed a decreased inluminescence life-time of 3%, 7% and 7%, respectively, whentemperature increased from 10 °C to 30 °C. The temperature effectwas larger at 20% O2, where τ values of IrC, Pt-TFPP and Ru-DPPoptodes decreased 10%, 11% and 17%, respectively, over the sametemperature range. Similar results were found by linear fitting ofnormalized τ values (STable 1, SFig. 2).

Fig. 3. Light transmission microscope images (A–C) of the biomass distribution in an ~25 μ(A), central (B) and top (C) parts of the biofilm. Lines in the upper panel indicate where O2

fluorescence image stack of the Syto-60 stained biofilm. The image is false colored, wheO2 distribution images in the same biofilm at different flow rates. Flow rates (m h−1) areconspicuous biomass structures. The white arrow in the high flow image indicates an area

Old IrC optode kept over several months exhibited a reduction inτ0 and an increase in the non-quenchable fraction, α (SFig. 3). Similaraging effects were found for the Ru-DPP optode (data not shown).

3.2. Sensor homogeneity

The O2 distribution in the aquarium used for calibration measure-ments can be considered homogeneous, since the water was wellmixed, no organic substrate was present and the optodes werecarefully cleaned using a soft cotton stick and brief cleaning with 70%ethanol prior to the calibrations. Thus any spatio-temporal variationin luminescence life-time at a given constant O2 level was caused bynoise and optical artifacts in the camera system or inhomogeneities inthe optode. The standard deviation of pixel values in the analyzedROI's (~5000 pixels) was used as a proxy for homogeneity of theoptode. Standard deviations within the obtained life-time imageswere analyzed at different O2 concentrations. For all three optodetypes, the standard deviation of luminescence life-time valuesdecreased with increasing O2 concentration. Absolute standarddeviations were highest for the Pt-TFPP optodes and ranged from1.3 (high O2) to 4 μs (anoxia), while the IrC optode had the loweststandard deviations varying from 0.1 to 0.4 μs. Since the luminescencelife-time values depend on the O2 concentration as well as theindicator type, the standard deviation was also expressed aspercentage of the life-time. When expressed as percentage, therewas less effect of the O2 concentration on the standard deviation. Therelative standard deviation of the luminescent life-time was againlowest for the IrC optode, followed by the Ru-DPP optode and the Pt-TFPP optode (STable 1). Thus, IrC optodes exhibited the besthomogeneity in O2 sensing within ROI's used for the calibrationcurve. An example of the homogeneity of luminescence intensity andlife-time of all three types of optodes is shown in SFig. 4A.

Luminescence measurements with a scanning confocal micro-scope equipped with a 60× lens, showed a significantly bettermicroscopic homogeneity of IrC optodes in comparison to Ru-DPPoptodes (Fig. 4B).We did not have a laser line available tomeasure the

m thick heterotrophic biofilm grown on an IrC optode; images are focused on the baseprofiles were extracted (see Fig. 4). D. Maximal intensity projection image of a confocalre red indicates a high biomass. Scale bar indicates 10 μm. E. Montage of differentindicated in the lower right corner of panels. Black contours are inserted to indicatewith larger mass transfer resistance outside the biofilm structure.

Fig. 4. Oxygen concentrations extracted from biofilm O2 images obtained under highflow (red curves) and low flow (blue curves) conditions, respectively, along the pixellines shown in Fig. 3. Numbers in each panel denote the pixel lines (y) in Fig. 3. Theblack line indicates the relative biomass in the biofilm as determined from confocalimage stacks. Vertical dotted lines represent the edge of the biomass on the optode.


heterogeneity of the Pt-TFPP optode, but macroscopic measurementsclearly showed pixel heterogeneity in the luminescence image of thePt-TFPP, which were not caused by the camera system.

3.3. Microscopic O2 measurements in biofilms

Heterotrophic biofilms were grown on the surface of IrC thin-filmoptodes. Due to the high brightness of IrC and efficient light capturewith the high NA objectives, microscopic O2 imaging could beperformed at a much lower integration time than in macroscopicimaging. At 400× magnification, O2 was measured with a horizontaloptical resolution of 0.16 μm per pixel.

The biofilms had a maximum thickness of ~25–30 μm and showeddistinctive structures resulting in a heterogeneous distribution ofbiomass and O2 concentration. Fig. 3A–D shows a ~50 μm wide and~80 μm long cell aggregate together with two smaller ~25–30 μmwide cell clusters under the main structure with a thickness of ~15–20 μm. The same thickness was found for a part (lower half in theimage) of the large aggregate. The three structures were separated bya channel like structure.

The O2 conditions in the biofilm were strongly affected by themedium flow rate. The whole flow chamber became anoxic understagnant conditions, while a more heterogeneous O2 distribution wasinduced by higher flow rates (N5 mh−1) (Fig. 3E). In areas with no orvery thin biofilms (up to 5 μm thickness) O2 levels increased with anincrease in the flow rate. Pronounced O2 gradients were found at theedges of thicker biofilm structures, especially at high flow rates(N10 mh−1). The inner compartments of thicker biofilm partsremained anoxic, even at the highest flow rates, indicating that theO2 demand of these structures was higher than the maximumpotential O2 influx.

The flow direction of the medium was from the top in the shownimages, alleviating mass transfer limitation upstream of larger cellaggregates and increasing such limitations downstream. The gradientoutside the black lines in Fig. 3 shows that on the upper left side of themajor structure there is a very thin and steep O2 gradient (2–3 μm)where the flow hits the structure. In contrast, in less flow exposedcavities of the biofilm on the upper right side of the major structure(indicated by a white arrow in the 49.5 mh−1 frame, Fig. 3), the O2

concentration already decreased significantly outside the biofilmstructure. There was also a shading effect visible in the channel inbetween the three larger biofilm structures, though some biomasswas detected on the optode in that region, whichmay also explain thelower O2 concentration.

After correction for image size and shifts due to the use of twodifferent cameras it was possible to align biomass and O2 distribution.In total, 9 horizontal lines (5 pixels thick) were analyzed, in orderto visualize the relationship between biomass and O2 distribution(Fig. 4) at two different flow rates. At low flow rate (4 mh−1) the O2

levels at the biofilm base was much lower than at high flow rate(30 mh−1). We found no direct relationship between biomass ontop of the optode and the O2 concentration at the optode.Oxygen concentrations decreased already in regions surroundingthe biomass structures due to mass transfer limitation. Furthermore,O2 levels below larger cell aggregates remained anoxic despite a lowerbiomass in these regions (see Fig. 3 and the biomass indication linesin the graphs of line 5, 50, 78 and 100 in Fig. 4). At high flow, O2

was able to penetrate all the way to the biofilm base in the thinnerregions of the largest cell aggregate (Fig. 4, line 205, 243 and 298)alleviating anoxia. The width of the large structure was almostequivalent to the O2 penetration depth from the side, i.e. maximally15–20 μm at high flow rates and 8–14 μm at low flow rates (Fig. 4,line 5–140).

The measurements described above were performed under steadystate conditions with a low temporal resolution (one sample perminute). Additional measurements at higher time resolution (one

sample per 3 s) showed a more complex O2 dynamics (Fig. 5). At thestart of the experiment, the system was in steady state with no flow.At t=0, the flow was turned on to maximum flow rate (70 mh−1)and O2 concentrations reached their maximum values within 10 s. Forclarity we only show O2 dynamics and biomass data (Fig. 5A, B) fromline 50 (Fig. 3). The O2 concentrations remained at maximal levels forthe whole flow period outside large structures. Within the structuresthere was a decrease in O2 concentration over time, especially in thezones (~10 μm from the aggregate edge) proximate to the anoxic partof the biofilm. The central parts of aggregates also remained anoxicduring this experiment. Different O2 depletion rates are found in thebig aggregate structures (Fig. 5C). Fastest O2 depletion (indicatedwithwhite arrows) was found in the peripheral parts of larger cellaggregates reaching rates of up to ~0.5 μmol O2l−1 s−1. After 48 s theflow was turned off again, causing an immediate decrease in O2

concentration.

4. Discussion

Thin-film planar O2 optodes are highly transparent andwell suitedfor O2 imaging combined with structural analysis of biofilms usinglight or confocal microscopy (Kühl et al., 2007). The new IrC-basedoptodes were superior to earlier developed Ru-DPP and Pt-TFPP basedoptodes in terms of a significantly higher brightness, better homoge-neity of luminescence life-time images and thus less noisy O2 images.IrC-based optodes also exhibited a lower temperature dependency ofthe O2-dependent luminescence. The dynamic range of the

image of Fig.�4

Fig. 5. A. Oxygen dynamics under non-steady state conditions at particular positions inthe biofilm (distance between points is 10 pixels, i.e. 1.6 μm) on the horizontal line onpixel line 50 (see Fig. 3). Each point denotes an average of 5×5 pixels. The upper 3 lines(x-pixel 80, 110 and 140) represent positions in the area outside of the biofilm. Pixelposition 280 represents the permanent anoxic zone, while positions in between pixel140 and 290 represent a transient zone. At t=0 the flow was turned on at 70 m h−1,and flow was stopped again at t=48 s. B. Biomass distribution in the investigatedbiofilm region. C. Image of O2 depletion rate distribution in the biofilm calculated for theperiod between t=12 s and t=45 s. White arrows indicate zones with the strongestdecrease in O2.


luminescence life-time for the IrC optode falls in between Pt-TFPP andRu-DPP based optodes, which makes the IrC optode suitable for O2

imaging ranging from anoxic conditions up to about 2.5 timesatmospheric saturation. The new IrC-based optodes enabled micro-scopic imaging of complex patterns in the spatio-temporal distribu-tion of O2 in biofilms that could be related to flow conditions andbiomass distribution at a hitherto unreached resolution. In thefollowing we discuss the measuring characteristics and biofilmapplication in more detail.

4.1. Heterogeneity effects

Despite the fact that luminescence life-time imaging corrects forheterogeneity in dye distribution, both Ru-DPP and Pt-TFPP basedoptodes exhibited relative high standard deviations in their life-timevalues (STable 1) reducing the operational spatial resolution ofmicroscopic O2 imaging as compared to IrC based optodes. It wasreported that platinum porphyrin based optodes had a higher spatialresolution than ruthenium based optodes (Oguri et al., 2006).However, when expressing the standard deviations as percentage oftotal lifetime, as has been done in this study, the authors found a

similar standard deviation with ruthenium and porphyrin basedoptodes (K. Oguri, personal communication).

A comparison of the thin-film optodes at higher (600×) opticalmagnification showed a higher heterogeneity of luminescence in theRu-DPP optode than in the IrC-based optode. In a recent review(DeGraff and Demas, 2005), heterogeneity in O2 optodes wasattributed to i) macroheterogeneity in sensor materials due to non-optimal fabrication, ii) microheterogeneity due to phase separationand microcrystallization of the matrix polymer and/or the indicatordye, and e.g. microscopic cracks in the sensor material, andiii) nanoheterogeneity due to heterogeneity in molecular orientationand polymer nanostructure causing variations in O2 diffusion andsolubility. All these types of heterogeneity affect sensor response andcan lead to a non-ideal Stern–Volmer quenching behavior, i.e. non-linear Stern–Volmer plots.

The spin-coating technique used in our study enabled fabricationof b1–2 μm thick homogeneous layers on glass coverslips thatexhibited no significant macro heterogeneities (Kühl et al., 2007).However, our microscopic investigation (SFig. 4) showed pronouncedmicroheterogeneities supporting earlier observations on Ru-DPP andPt-porphyrin based optodes (Bedlek-Anslow et al., 2000; DeGraff andDemas, 2005; Eaton et al., 2004). A detailed discussion of mechanismscausing such heterogeneity and their effects on sensor performance isoutside the scope of this paper. While micro- and possibly nano-heterogeneity was present also in the new IrC optodes, it wassignificantly lower than in Ru-DPP based optodes.

4.2. Bleaching and aging of optodes

Over longer times, i.e. months, the new IrC optode exhibited adecrease in luminescence life-time at all O2 concentrations. Inaddition, it was found that the non quenchable fraction, α, increasedfrom virtually zero in freshly prepared IrC optodes to 0.18 in16 months old IrC optodes (SFig. 3). This increase could be causedby aging of the indicator/polymer matrix developing larger micro-and nano-heterogeneity (see above). However, an additional exper-iment showed that bleaching could reproduce the same effect on thelife-time and the non-quenchable fraction within the IrC optode asaging (SFig. 3). Borisov and Klimant (2007) found that the IrCcomplex was easily bleachable upon illumination with a Xe lightsource. In our study we did not find a strong bleaching induceddecrease in luminescence upon exposure to light from blue LED's or ahalogen lamp. However, prolonged exposure to 488 nm laser light onthe confocal microscope resulted in strong bleaching of the IrC as wellas the Ru-DPP optode, which could be an important problem for usingsuch indicators for more continuous luminescence intensity orratiometric imaging on laser confocal microscopes. Microscopicluminescence life-time imaging measurements with the optodesused a blue 5 W power LED (470 nm) and we did not find anydecrease in luminescence intensity measured over time under apermanent anoxic area, even after 250 consecutive O2 measurementsand several minutes exposure to the halogen lamp of the microscope.Thus, the new IrC optodes are well suited for luminescence life-timebased microscopic imaging of O2.

4.3. Response time and temperature effects

Luminescence signals from thin-film optodes are much lower thanfrom conventional optodes using the same dye concentration.Conventional planar optodes have a thickness of 10–40 μm (Kühland Polerecky, 2008), and the response time of such optodes isdependent on the equilibration time of the whole luminescent layerupon changes in O2. This limits the temporal resolution when imagingfast O2 dynamics, e.g. in studies of photosynthesis (Glud et al., 1999).Thin-film optodes have a much faster response time but may sufferfrom low luminescence. However, due to the high brightness of the


IrC-based optode it was possible to reduce the integration time by afactor of 3 in comparison to Ru-DPP and Pt-TFPP based thin-filmoptodes thus allowing a much higher temporal resolution. Thetemperature effect on the luminescence life-time was also smallestfor the IrC optodes allowing measurements over a broader temper-ature range without the necessity to recalibrate the optode. This isespecially convenient when O2 is measured in phototrophic systems,since supplied light can heat surfaces locally (Jimenez et al., 2008).

4.4. Overall comparison of the O2 optodes

All three types of O2 optodes are capable of sensing the spatio-temporal dynamics of O2, and excel under specific conditions due totheir specific ranges in decay times and absorption and emissionspectra. Generally, O2 sensitive dyes with a long luminescence life-time are more suitable to monitor spatial variation at low O2

concentrations, while short life-time dyes are more suitable tomonitor high O2 conditions (DeGraff and Demas, 2005; Oguri et al.,2006). Table 1 sums the overall performance and applicability of thethree types of O2 optodes tested in this study. The new IrC optodeperforms best as a general sensor for luminescence life-time basedmicroscopic O2 imaging at 0–40% O2.

4.5. Application in biofilms

In this study, we measured the O2 dynamics in biofilms at 400×magnification, but the magnification can easily be varied and adaptedto the scale relevant for the study object by changing microscopeobjectives. At lower magnifications (b200×), it was difficult to supplyenough excitation light to the optode via the objective, especially withobjectives having a large focal distance. In these cases excitation lightcould be supplied more efficiently externally via two LED's mountedclose to the objective.

Planar optodesmonitor the O2 conditions in a defined plane.Whenan undefined volume of water is between the optode and the object ofstudy, measurements done under steady state conditions will besubject to diffusive smearing. For a more accurate estimate of O2

production or consumption rates under non steady state conditionsan estimate of the water volume is required. In permeable systemsconvective transport will also alter the O2 dynamics (Polerecky et al.,2005; Precht et al., 2004). When used as a growth substratum, thebiofilm matrix is tightly connected to the optode surface. Therefore,thin-film optodes are ideal for studying the dynamics in spatial O2

distribution at the biofilm basis as a function of biomass cover andstructural heterogeneity.

Our study showed an O2 penetration depth of ~10–15 μmrevealing extremely steep O2 gradients inside the larger cellaggregates and persistent anoxic conditions inside cell aggregatesN25 μm even at elevated flow. Such steep gradients in freshly grownbiofilms have been measured before with microelectrodes (DeBeer

Table 1General performance of investigated thin-film optodes.

Property Optode material

IrC Pt-TFPP Ru-DPP

Brightness ++ + +Temperature dependence Low Low HighSensor homogeneity ++ + +Applicability for photosynthetic systems(high pO2)

+ + ++

Applicability for other systems (low pO2) + ++ −Interference from fluorescence inphototrophic systems

No Yes, Chl a Yes, cyanobacteria

Photo stability − + +Commercial availability of the indicator No Yes Yes

et al., 1994), while other studies report steep O2 gradients overN30 μm (e.g. Rasmussen and Lewandowski, 1998; Schramm et al.,1996). The measured gradients represent the outcome of masstransfer limitation and the volume specific consumption rate of O2.However, in our set-up it is presently difficult to use steady stateimages of spatial O2 distribution for simple flux calculations, since theoptode measurements integrate effects of heterogeneous biomassdistribution and complex mass transfer geometry inside as well asoutside of the biofilm.

Even in apparently clean parts of the optode images, it cannot becompletely excluded that a very thin biofilm is present, which alongwith the presence of a diffusive boundary layer can create diffusivesmearing. Nevertheless, the imaging of steady state O2 distribution inbiofilms clearly showed the presence of distinct hot spots of activity,which not always link directly to biomass distribution. This is animportant result as many studies of biofilms infer distribution offunction and microenvironments based on structural analysis withconfocal microscopy (Pamp et al., 2009). Our results clearly show thatsuch inference has its limitations as even small differences in biomassgeometry and a spatially heterogeneous mass transfer can haveprofound effects on O2 levels inside biofilms.

A better estimate of spatial O2 dynamics could be determined fromdynamic imaging of the O2 distribution over time when turning theflow on and off (Fig. 5). In regions with no bacterial biomass, the O2

concentration reached itsmaximumvalue (6 s) within twomeasuringpoints, i.e. over 6 s. The fast response of the optode thus allowed themonitoring of quick changes in O2 concentration under non steadystate conditions.

When maximal O2 levels were reached upon onset of flow after aperiod of stagnant conditions, several zones within larger biofilmstructures showed a subsequent lowering in O2 concentration(Fig. 5C), despite the fact that the flow was not changed. Wespeculate, whether such response reflects an acclimation to a shiftfrom anoxic to oxic conditions. Alternatively, this phenomenon mayreflect substrate limitation under stagnant and anoxic conditionsresulting in an initial limitation in the first couple of seconds after O2

became available. In principle, the O2 depletion rate after stopping theflow reflects the local consumption rates, which can be used to derivekinetic parameters e.g. from a fit of Michaelis–Menten kinetics in theregions where biomass is present. However, in the regions withoutbiomass the O2 depletion rate will be partly affected by the overlayingvolume which is an unknown factor. A study implementing suchanalyses by the use of O2 sensitive luminescent nano particles formapping the spatial distribution and O2 kinetics in biofilms will bepresented elsewhere.

5. Conclusion

The new IrC thin-film optodes allow spatial mapping of O2 inbiofilms at a hitherto unreached spatio-temporal resolution. Incombination with fluorescent staining of the biomass it is possibleto link O2 distribution and dynamics with biofilm biomass distributionand to detect local variations in O2 distribution due to biofilmstructure and its interaction with flow. It was possible to monitorchanges in O2 consumption around these structures under steadystate conditions as well as under non steady state conditions. Weshow here that O2 microenvironments inside biofilms are extremelydynamic over spatial scales of 10–20 μm and over time scales of a fewseconds. Oxygen conditions in biofilms exhibit dynamic changes inresponse to flow and substrate conditions and cannot be directlyinferred from structural microscopic investigations. Besides biofilmresearch, this new technique also has a large potential for otherresearch fields where biomass and physiological functions aremonitored e.g. in studies of cell and tissue cultures in biotechnologyand biomedical research.


Acknowledgements

This study was supported by the Danish Research Council forTechnology and Production Sciences (FTP), the Danish Natural ScienceResearch Council (FNU), and the Danish National Advanced Technol-ogy Foundation.

Appendix A. Supplementary data

Supplementary data to this article can be found online atdoi:10.1016/j.mimet.2011.01.021.

References

Amao, Y., 2003. Probes and polymers for optical sensing of oxygen. Microchim. Acta143, 1–12.

Bedlek-Anslow, J.M., Hubner, J.P., Carroll, B.F., Schanze, K.S., 2000. Micro-heterogeneousoxygen response in luminescence sensor films. Langmuir 16, 9137–9141.

Borisov, S.M., Klimant, I., 2007. Ultrabright oxygen optodes based on cyclometalatedlridium(III) coumarin complexes. Anal. Chem. 79, 7501–7509.

Carraway, E.R., Demas, J.N., DeGraff, B.A., 1991. Luminescence quenching mechanismfor microheterogeneous systems. Anal. Chem. 63, 332–336.

DeBeer, D., Stoodley, P., Roe, F., Lewandowski, Z., 1994. Effects of biofilom structures onoxygen distribution and mass-transport. Biotechnol. Bioeng. 43, 1131–1138.

DeGraff, B.A., Demas, J.N., 2005. Luminescence-based oxygen sensors. Rev. Fluoresc.125–151.

DeRosa, M.C., Hodgson, D.J., Enright, G.D., Dawson, B., Evans, C.E.B., Crutchley, R.J., 2004.Iridium luminophore complexes for unimolecular oxygen sensors. J. Am. Chem.Soc. 126, 7619–7626.

Eaton, K., Douglas, B., Douglas, P., 2004. Luminescent oxygen sensors: time-resolvedstudies and modelling of heterogeneous oxygen quenching of luminescenceemission from Pt and Pd octaethylporphyrins in thin polymer films. Sens. Act. BChem. 97, 2–12.

Fenchel, T., Finlay, B., 2008. Oxygen and the spatial structure of microbial communities.Biol. Rev. 83, 553–569.

Gerritsen, H., Sanders, R., Draaijer, A., Ince, C., Levine, Y., 1997. Fluorescence lifetimeimaging of oxygen in living cells. J. Fluoresc. 7, 11–15.

Glud, R.N., 2008. Oxygen dynamics of marine sediments. Mar. Biol. Res. 4, 243–289.Glud, R.N., Ramsing, N.B., Gundersen, J.K., Klimant, I., 1996. Planar optrodes: a new tool

for fine scale measurements of two-dimensional O2 distribution in benthiccommunities. Mar. Ecol. Prog. Ser. 140, 217–226.

Glud, R.N., Kühl, M., Kohls, O., Ramsing, N.B., 1999. Heterogeneity of oxygen productionand consumption in a photosynthetic microbial mat as studied by planar optodes.J. Phycol. 35, 270–279.

Holst, G., Grunwald, B., 2001. Luminescence lifetime imaging with transparent oxygenoptodes. Sens. Acta B Chem. 74, 78–90.

Holst, G., Kohls, O., Klimant, I., König, B., Kühl, M., Richter, T., 1998. A modularluminescence lifetime imaging system for mapping oxygen distribution inbiological samples. Sens. Acta B Chem. 51, 163–170.

Jimenez, I.M., Kühl, M., Larkum, A.W.D., Ralph, P.J., 2008. Heat budget and thermalmicroenvironment of shallow-water corals: do massive corals get warmer thanbranching corals? Limnol. Oceanogr. 53, 1548–1561.

Klimant, I., Wolfbeis, O.S., 1995. Oxygen-sensitive luminescent materials based onsilicone-soluble ruthenium diimine complexes. Anal. Chem. 67, 3160–3166.

Klimant, I., Meyer, V., Kühl, M., 1995. Fiberoptic oxygen microsensors, a new tool inaquatic biology. Limnol. Oceanogr. 40, 1159–1165.

Klimant, I., Kühl, M., Glud, R.N., Holst, G., 1997. Optical measurement of oxygen andtemperature in microscale: strategies and biological applications. Sens. Acta BChem. 38, 29–37.

Kühl, M., 2005. Optical microsensors for analysis of microbial communities. Meth.Enzymol. 397, 166–199.

Kühl, M., Polerecky, L., 2008. Functional and structural imaging of phototrophicmicrobial communities and symbioses. Aquat. Microb. Ecol. 53, 99–118.

Kühl, M., Rickelt, L.F., Thar, R., 2007. Combined imaging of bacteria and oxygen inbiofilms. Appl. Environ. Microbiol. 73, 6289–6295.

Liebsch, G., Klimant, I., Frank, B., Holst, G., Wolfbeis, O.S., 2000. Luminescence lifetimeimaging of oxygen, pH, and carbon dioxide distribution using optical sensors. Appl.Spectrosc. 54, 548–559.

Oguri, K., Kitazato, H., Glud, R.N., 2006. Platinum octaetylporphyrin based planaroptodes combined with an UV-LED excitation light source: an ideal tool for high-resolution O2 imaging in O2 depleted environments. Mar. Chem. 100, 95–107.

Pamp, S.J., Sternberg, C., Tolker-Nielsen, T., 2009. Insight into themicrobial multicellularlifestyle via flow-cell technology and confocal microscopy. Cytom. A 75A, 90–103.

Polerecky, L., Franke, U., Werner, U., Grunwald, B., de Beer, D., 2005. High spatialresolution measurement of oxygen consumption rates in permeable sediments.Limnol. Oceanogr. Meth. 3, 75–85.

Precht, E., Franke, U., Polerecky, L., Hüttel, M., 2004. Oxygen dynamics in permeablesediments with wave-driven pore water exchange. Limnol. Oceanogr. 49, 693–705.

Rasmussen, K., Lewandowski, Z., 1998. Microelectrode measurements of local masstransport rates in heterogeneous biofilms. Biotechnol. Bioeng. 59, 302–309.

Revsbech, N.P., 2005. Analysis of microbial communities with electrochemicalmicrosensors and microscale biosensors. Meth. Enzymol. 397, 147–166.

Schramm, A., Larsen, L.H., Revsbech, N.P., Ramsing, N.B., Amann, R., Schleifer, K.H., 1996.Structure and function of a nitrifying biofilm as determined by in situ hybridizationand the use of microelectrodes. Appl. Environ. Microbiol. 62, 4641–4647.

Wolfbeis, O.S., 2005. Materials for fluorescence-based optical chemical sensors. J. Mater.Chem. 15, 2657–2669.


10. A simple optode for tap water biofilms

M. Staal, E. Prest, H. Vrouwenvelder, L.F. Rickelt, and M. Kühl:

A simple optode based imaging technique to measure O2 distribution and dynamics in tap water

biofilms

Water Research 45: 5027-5037 (2011)

DOI: 10.1016/j.watres.2011.07.007

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 7

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

A simple optode based method for imaging O2 distributionand dynamics in tap water biofilms

M. Staal a,*, E.I. Prest a,b, J.S. Vrouwenvelder b,c, L.F. Rickelt a, M. Kuhl a,d

aMarine Biological Section, Department of Biology, University of Copenhagen, Strandpromenaden 5, DK-3000 Helsingør, DenmarkbWetsus, Centre of Excellence for Sustainable Water Technology, P.O. Box 1113, 8900 CC Leeuwarden, The NetherlandscDelft University of Technology, Department of Biotechnology, Julianalaan 67, 2628 BC Delft, The NetherlandsdPlant Functional Biology and Climate Change Cluster (C3), Department of Environmental Science, University of Technology Sydney,

Broadway NSW 2007, Australia

a r t i c l e i n f o

Article history:

Received 2 February 2011

Received in revised form

25 May 2011

Accepted 3 July 2011

Available online 13 July 2011

Keywords:

Oxygen sensing

Planar optodes

Lifetime

Imaging

Membrane fouling simulator

Biofilm

Abbreviations: ROI, region of interest; MFdiphenyl-1,10-phenanthroline; PS, polystyre* Corresponding author. Tel.: þ45 3532098; faE-mail address: [email protected] (M. Sta

0043-1354/$ e see front matter ª 2011 Elsevdoi:10.1016/j.watres.2011.07.007

a b s t r a c t

A ratiometric luminescence intensity imaging approach is presented, which enables spatial

O2measurements in biofilm reactorswith transparent planar O2 optodes. Optodes consist of

an O2 sensitive luminescent dye immobilized in a 1e10 mm thick polymeric layer on

a transparent carrier, e.g. a glass window. Themethod is based on sequential imaging of the

O2 dependent luminescence intensity, which are subsequently normalized with lumines-

cent intensity images recorded under anoxic conditions. We present 2-dimensional O2

distribution images at the base of a tap water biofilm measured with the new ratiometric

method and compare the results with O2 distribution images obtained in the same biofilm

reactor with luminescence lifetime imaging. Using conventional digital cameras, such

simple normalized luminescence intensity imaging can yield images of 2-dimensional O2

distributions with a high signal-to-noise ratio and spatial resolution comparable or even

surpassing those obtained with expensive and complex luminescence lifetime imaging

systems. The method can be applied to biofilm growth incubators allowing intermittent

experimental shifts to anoxic conditions or in systems, in which the O2 concentration is

depleted during incubation.

ª 2011 Elsevier Ltd. All rights reserved.

1. Introduction treatment and industrial processes (Nicolella et al., 2000),

Biofilms are surface-associated microbial communities

exhibiting spatio-temporal heterogeneity in their structure,

composition, physiology and chemical microenvironment

(Costerton et al., 1995; Stewart and Franklin, 2008). Such

communities represent the preferred lifestyle of many

microbes in natural ecosystems, and biofilms also play

important roles in more applied contexts such as waste water

S, membrane fouling simne.x: þ45 35321951.al).ier Ltd. All rights reserved

chronic infections (Costerton et al., 1999; Hall-Stoodley et al.,

2004), and e.g. corrosion and biofouling of materials (Ridgway

and Flemming, 1996; Vrouwenvelder et al., 2008).

The growth dynamics and complex structural heteroge-

neity of biofilms has been studied in great detail, especially

through application of various microscopic techniques (Neu

et al., 2010). However, a similar detailed mapping of the

chemical landscape and dynamics in biofilms is lacking in

ulator; mil, 1 mil equals 25.4 mm; Ru-dpp, Ruthenium(II)-tris-4,7-

.


http://www.sciencedirect.com

http://www.elsevier.com/locate/watres

http://dx.doi.org/10.1016/j.watres.2011.07.007



wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 75028

most studies. Electrochemical and fiber-optic microsensors

(Revsbech, 2005; Kuhl, 2005) can provide very detailed infor-

mation on local chemical dynamics, zonations of microbial

processes andmass transfer processes in biofilms (e.g. de Beer

et al., 1994; Stoodley et al., 1994; Kuhl and Jørgensen, 1992;

Kuhl et al., 1996), but such 1-dimensional characterization is

in most cases inadequate to assess the true spatial distribu-

tion and dynamics of the chemical microenvironment in

heterogeneous biofilms at similar resolution as biofilm struc-

ture can be resolved (Jørgensen and Des Marais, 1990).

Modeling of biofilm systems has shown that significant

differences can be found between 1- and 2-dimensional

descriptions of biofilms, since 1-dimensional models fail to

describe spatial heterogeneity other than over depth (z-plane).

It has also been argued from biofilm modeling approaches

that it is sufficient to do 2-dimensional measurements for

a proper description of spatial heterogeneity in microenvi-

ronments, since increasing model complexity from 2- to 3-

dimensions did not yield much extra insights (Eberl et al.,

2000). However, there is still a lack of real datasets linking

spatial biofilm heterogeneity to chemical heterogeneities and

microenvironments.

With the development of imaging techniques, several new

non-invasive methods became available, e.g. Confocal Laser

Scanning Microscopy (Neu et al., 2010), Raman Micro-

spectroscopy (Ivleva et al., 2010), lifetime imaging (Glud et al.,

1996; Hidalgo et al., 2009), Magnetic Resonance Microscopy

(Wagner et al., 2010). Thesemethods allowmeasurements of 2-

and 3-dimensional chemical and biological landscapes in bio-

films. Molecular oxygen (O2) is a key parameter in biogeo-

chemical andbiological studies (Glud, 2008; Fenchel andFinlay,

2008) and the introduction of planar optodes for mapping 2-

dimensional O2 concentration in natural systems (Glud et al.,

1996) was a big step forward for the study of the heterogeneity

of O2 distribution and dynamics in sediments and biofilms (e.g.

Glud et al., 1998, 1999; Kuhl et al., 2007). Such planar optode

measurements are based on luminescent indicator dyes,

immobilized in a polymericmatrix and cast onto a transparent

carrier. The measuring principle relies on the dynamic

quenching of indicator luminescence by O2. Both the lumines-

cence intensity (I) and luminescence lifetime (s) vary reversibly

with O2 concentration, and the process does not consume O2.

The ideal responseof suchopticalO2 sensors isdescribedby the

SterneVolmer relation (Bacon and Demas, 1987):

I

I0¼ s

s0¼ 1

1þ KSV½O2�5I0I¼ s0

s¼ 1þ KSV½O2� (1)

where s0 and s denote the luminescence lifetime in the

absence and in the presence of O2 respectively; I0 and I denote

the luminescence intensity in the absence or presence of O2;

KSV is the bimolecular quenching coefficient of the dye in its

specific polymeric matrix, and [O2] is the O2 concentration.

In practice, most planar optodes exhibit a non-ideal

SterneVolmer like response, which can be modeled with

a two-component model (Carraway et al., 1991), where only

a certain fraction of the O2 indicator dye remains quenchable

upon immobilization. This relationship can be described by

the equation:

II0

¼ ss0

¼ 1� a

1þ KsvCþ a (2)

where a is the non-quenchable fraction. Initial applications of

O2 planar optodes involved simple luminescence intensity

measurements in combination with planar optodes with

a black O2 permeable overcoat to avoid optical artifacts from

background light and sample backscatter (Glud et al., 1996).

Application of transparent O2 optodes on glass allows direct

alignment of O2 distribution to the structure of the sample

(e.g. Holst and Grunwald, 2001; Kuhl et al., 2007, 2008, Staal

et al., 2011) and nowadays oxygen imaging with planar

optods makes almost entirely use of luminescent lifetime

imaging systems (Holst et al., 1998; Oguri et al., 2006). Such

systems are mostly custom built and relatively expensive

(>30.000 V), which has been a bottleneck for the more wide-

spread distribution of planar O2 optode methodology.

Recently, however, a new ratiometric method has been pub-

lished using a normal digital single-lens reflex camera as

detector (Wang et al., 2010; Larsen et al., in press). In this

method, correction for an uneven light field was accom-

plished by inclusion of a luminescent O2 insensitive reference

dye in the planar optode matrix. Both dyes are excited by the

same excitation source, but the reference dye emits light in

the green region, and the O2 sensitive dye in the red region,

coinciding with the green and red channels of common color

camera CCD or CMOS detectors. A ratio of the red and green

channel indicates the O2 concentration.

Here we present a simple luminescence intensity imaging

approach enabling 2-dimensional mapping of biomass and O2

concentration in a biofilm growth incubator using simple

digital color cameras or monochromatic cameras combined

with suitable emission filters. Themethod uses the ratio of the

luminescence from a transparent planar optode under anoxia

to the luminescence under experimental O2 conditions to

correct forheterogeneity in excitation light.No referencedye is

required. As a proof of principle, a heterotrophic biofilm was

grown on top of a transparent planar optode, which created

heterogeneity in O2 concentrations. We show that such

a simple imaging approach yields information on the 2-

dimensional distribution of O2 in biofilms with a good signal-

to-noise ratio and with a spatial resolution of 36 mm/pixel

comparable or better than in more elaborate lifetime based O2

imaging.

2. Methods

The planar O2 optode used in this study was based on the

luminescent O2 sensitive dye Ruthenium(II)-tris-4,7-diphenyl-

1,10-phenanthroline (Ru-dpp) immobilized in a polystyrene

(PS) matrix (20 mg Ru-dpp/g PS). Such Ru-dpp based planar

optodes are suitable for O2 imaging over awide dynamic range

(up to full O2 saturation) (Kuhl and Polerecky, 2008). The PS/

Ru-dpp mixture was dissolved in chloroform (3.3 w/w%) and

cast onto a silanized glass plate (Kuhl et al., 2007). After slow

evaporation of the solvent in a semi-closed container, this

resulted in aw9 mmthick homogeneous O2 sensor layer on the

complete glass window of the incubator (6 cm � 28 cm).

The coated glass plate was mounted as a window in

a watertight flow-trough biofilm growth incubator, i.e.,

a membrane fouling simulator (MFS), where a polypropylene

spacer mesh was placed on top of the optode surface. The



wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 7 5029

biofilm growth chamber had external dimensions of

0.07 m � 0.30 m � 0.04m (Vrouwenvelder et al., 2006, 2007).

Coupons of feed spacer, membrane and product spacer were

placed in theMFS resulting in the same spatial dimensions and

orientation as in spiral wound membrane elements applied in

water treatment. For MFS experiments, membranes and

787 mm (31 mil) thick feed spacer sheets were taken from

a new, unused spiral wound reverse osmosis membrane

element. The development of fouling, i.e. biofilm formation,

was monitored by a rhodamine tracer dye imaging technique

(see below) and by measuring the pressure drop increase over

the feed spacer channel of the MFS. During operation, the MFS

windowwas covered with a light-tight lid to prevent growth of

phototrophic organisms.

The reactor system setup consisted of a pressure reducing

valve, manometer, dosage point (for biodegradable

compounds), MFS and flow controller (Fig. 1, Vrouwenvelder

et al., 2007). An extra T-connection was placed in the tubing

before the inlet of the biofilm monitor to allow injection of

saturated sodium dithionite solution to create anoxic condi-

tions, as well as injection of water colored by the dye rhoda-

mine WT (Chrompton & Knowles OT/US 04029NS). The dye

RhodamineWT is an inert, non-adsorbing and stable tracer for

flow visualization and is simple to quantify by light absorbance

(Huettel et al., 1996). TheMFSwas sterilized for aperiodof 5min

with 70% ethanol before the start of the experiment. Bacteria in

the biofilm originated from bacteria occurring in the drinking

water system. The MFS was operated at 16 �C. A pressure of

120 kPa was applied to avoid degassing. The feed water flow

was 16 L h�1 equal to a linear flow velocity of 0.16 m s�1,

representative for the linearflowvelocity in the leadmembrane

element in full-scale installations containing spiral wound

nanofiltration or reverse osmosis membrane elements

(Vrouwenvelder et al., 2009a). The MFSs were operated single

pass without (partial) recirculation. Pressure drop measure-

ments were performed with a pressure difference transducer

(Deltabar S: PMD70-AAA7FKYAAA, Endress & Hauser, The

Netherlands). The calibrated measuring range was 0e50 kPa

(Vrouwenvelder et al., 2009b).

Concentrated substrate was dosed into the feed water (tap

water), prior to the MFS by a peristaltic pump (Masterflex L/S

Fig. 1 e Schematic description of the membrane fouling simula

sensor, flow meter and the imaging setup for O2 detection. The

or a mEye color CMOS camera. The trigger-delay box and the con

only present in the lifetime setup with the fast gate-able digita

direction. There is no flow via the pressure-drop sensor.

brushless digital drive, HV 7523.70, 1.6e100 rpm, with an Easy

Load II pump rotor, Applicon Analytical, The Netherlands) at

a flow of 0.03 L h�1 from a 5L stock solution reservoir. The

dosage of substrate was checked periodically by measuring

the water volume pumped over a defined time interval. The

chemicals NaCH3COO, NaNO3 and NaH2PO4 were dosed in

a mass ratio C:N:P of 100:20:10 with a concentration of 1.0 mg

acetate-C L�1 in the feed water entering the MFS. The

substrates were dissolved in MilliQ water. To restrict bacte-

rial growth in the substrate dosage bottle, the pH was set at

11 by NaOH dosage. Stock solution bottles were replaced

every 5 days. The substrate flow rate (0.03 L h�1) was low

compared to the feed water flow rate (16 L h�1). Thus, the

effect of the substrate flow on the pH of the feed water was

insignificant.

2.1. Imaging systems and image calculations

Two different camera systems were used for O2 measure-

ments: (1) A monochrome 12 bit fast gate-able cooled

1280X1024 CCD chip camera (2/3" chip), denoted as the PCO

camerain this manuscript (SENSICAM-SENSIMOD, PCO AG,

Kehlheim, Germany) and controlled by custom made acqui-

sition software (Look@MOLLI, Holst and Grunwald, 2001) that

can perform intensity based as well as lifetime based O2

imaging (Holst et al., 1998); (2) A 8 bit color 1280X1024 CMOS

chip camera (1/2" chip), denoted as the mEye camerain this

manuscript (USB mEye SE,UI-1540-C, IDS Imaging Develop-

ment Systems GMBH, Obersulm, Germany) and controlled by

themanufacturers acquisition software. The mEye camera can

only perform intensity based O2 imaging. Both image acqui-

sition programs allowed manual programming of the expo-

sure time of the camera chips. The gain per color channel

could be set manually in the software for the mEye camera

allowing separate signal optimization of the three color

channels.

For measurements with the mEye camera, the camera was

coupled via a C-mount to a 2Xmagnification objective with an

additional focusing lens (Microbench basic set, Qioptiq,

Germany); the same optical setup was also mounted on the

PCO camera for direct comparisons. The focal distance was

tor (MFS) including a water mixing system, pressure-drop

camera was a cooled PCO monochrome cooled CCD camera

trol of the LED driver (both indicated with dashed lines) are

l camera (PCO camera). The black arrows indicate the flow



wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 75030

w1.5 cm. A long pass filter (>590 nm) was placed in between

the camera chip and the lens to exclude background and

reflected blue excitation light. With a binning of 2, the reso-

lution of this setup is w36 mm per pixel. Additional measure-

ments with the PCO camera were done with a Xenoplan XNP

1.4/17 objective (Schneider-Kreutznach, Germany) equipped

with a long pass filter (>590 nm, Schneider-Kreutznach,

Germany) and distance macro rings (2 mm) on the PCO

camera to shorten the minimum focal distance from 15 to

5e10 cm. The resolution with this setup is w72 mm per pixel.

Background light was further diminished by a black card box

mounted around the camera setup.

For the O2 measurements with the PCO camera, two

modulated power LEDs were used for excitation of the optode

(1W Luxeon Star, 470 nm, Lumileds, San Jose; USA), as

controlled by a custom built trigger-delay box. The two LEDs

were mounted next to the lens. For lifetime imaging, the PCO

camera was used in a modulated measuring mode, in which

luminescence intensity images were acquired sequentially by

integrating series of 3 ms imaging periods starting respectively

at 0.1 (IW1) ms and 4.1 (IW2) ms after the 4 ms long excitation

pulses, and followed by dark image acquisition (see also fig. X-

1 and table X-1, additional information). The total integration

time per measurement period was 500 ms (see Holst et al.,

1998 for a detailed description of the measuring method and

program). Lifetime (s) images were subsequently calculated

from the luminescence intensity images IW1 and IW2 images

according to

s ¼ Dt

ln

�IW1=IW2

� (3)

where Dt is the time difference between IW1 and IW2.

For the ratiometric imaging method, we calculated the O2

concentration from the ratio of luminescence images (I0/I )

taken under complete anoxic conditions (I0) and oxygenated

conditions (I ), respectively (see also fig. X-1 and table X-1,

additional information). In measurements with the PCO

camera system, we also used the IW1 images to calculate O2

images based on the ratio I0W1/IW1, thus allowing direct

comparison of image quality by both lifetime and intensity

based imaging. Additionally, intensity based measurements

were also done using the intensity image setting in Look@

MOlli, where the O2 dependent luminescence intensity is

measured during the blue LED excitation period, in contrast to

the Iw1 images taken 0.1 ms after the excitation flash. The

integration time for the intensity images was shorter (100 ms)

than for the lifetime measurements (to measure comparable

intensities as in the lifetime images), while the excitation

pulse length was equal (4 ms). The measuring time per pulse

was 3 ms. Dark images were automatically taken and sub-

tracted in the image acquisition program.

For the mEye color camera measurements, two power LEDs

(1W Luxeon Star, 470 nm, Lumileds, San Jose; USA) were used

for excitation of the optode. The LED light intensities were

controlled by a stable and adjustable DC voltage source (GPC-

3030DQ, GW Instek, Tucheng City, Taiwan) allowing precise

control of current and voltage. The exposure time for the mEye

camerawas 200ms (see also table X-1, additional information)

and the gain for the blue, green and red channel were set to 80,

85 and 12, respectively, to optimize the gray value distribution

over the three color channels.

Calibration of the optode in the MFS was done by circu-

lating water with different O2 concentrations at a stabilized

temperature. The O2 concentration in the water was changed

by flushing the water with a series of set gas mixtures, mixed

by a PC controlled automated gas mixing system based on

electronic mass flow controllers (SensorSense, Nijmegen,

Netherlands).

2.2. Rhodamine assay

Imaging of biofilm thickness in the MFS was performed to

compare O2 distribution with biomass distribution. Imaging

was performed with the mEye camera without the long pass

filter and samples were illuminated with 2 warm white LED

strips (Hide-a-lite, Electro Elco AB, Sweden). A volume of 20ml

of a dilute rhodamine WT solution was injected into the MFS

to measure the biofilm thickness. Rhodamine WT absorbs

green-orange light, and absorption was measured by the

green channel of the camera. If the measurement was carried

out immediately after injection of the solution, rhodamine

was only present in the flow channels, i.e. places were no

biofilm was present. Calibration of the relationship between

absorption and thickness was done with a triangle shaped

glass cuvette with a thickness range of 0e700 mm, filled with

the rhodamine solution (appendix, Fig X-2 and X-3,). A

membrane (equal to the one in the MFS) was placed behind

the cuvette to have the same light reflection characteristics as

in theMFS. Images of the calibration cuvette weremade at the

same camera distance and illumination geometry as used for

the biofilm thickness measurements in the MFS.

All image calculations (O2 sensing and rhodamine assay)

were done in the freeware ImageJ (version 1.45a; http://rsb.

info.nih.gov/ij). After import to ImageJ, the color RGB images

were split. All images were converted to a 32 bit floating point

format before initial thresholding. Thresholding was per-

formed to exclude low intensity pixels, i.e. pixels within areas

where no fluorescence was measured (for example areas with

marker ink), from further calculations (i.e ratio calculations,

lifetime calculations etc). Ratio images were calculated using

the “process/ image calculator” option in ImageJ. The color

camera images were split into their Red, Green and Blue

channels prior to ratio calculations, by using the

“image/color/split channels” option in ImageJ. False

coloring of the images was applied to visualize/emphasize

differences in O2 concentration. All data handling in ImageJ

was done manually.

3. Results

3.1. Calibration

For calibration, lifetime images and luminescent intensity

images of a clean optode (without biofilm) were recorded at 5

different O2 concentrations. From these images, we calculated

ratio images for the ratios s0=s, I0W1/IW1 of the PCO camera and

the I0/I of the red channel of the mEye camera at the different

O2 concentrations (Fig. 2), and we averaged the values of

mailto:Look@MOlli

mailto:Look@MOlli

http://rsb.info.nih.gov/ij

http://rsb.info.nih.gov/ij



wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 7 5031

a region of interest (1.6 � 105 pixels), comprising half of the

imaged area. Selection of the regions of interest was done to

exclude areas in the image without oxygen sensitive dye as

well as areas with pixel values below a threshold grey value

(<30). All three imagingmethods showed a linear relationship

with O2 concentration (r2 > 0.99), indicating that the non-

quenchable fraction (a) was small. While the calibration

curves of the different methods showed some variations, the

use of different linear correlations to convert I0/I into oxygen

concentrations (see caption in Fig. 2) allowed us to compare

the calculated O2 images based on measurements with the

different methods.

3.2. Comparison of different methods

To evaluate the performance of the ratiometric and the life-

time approach, we compared O2 images based on lumines-

cence images taken with the PCO camera (2X lens mounted)

using different calculation andmeasuringmethods (Fig. 3). All

pictures were taken from exactly the same area of the planar

optode in the biofilm growth incubator (MFS). The biofilm in

the MFS had been growing for a period of 10 days. The lumi-

nescence images under full anoxic conditions (I0 and I0w1)

were taken after injection of saturated sodium dithionite

solution into the MFS. Images (I and Iw1) and lifetime images

(Iw1 and Iw2) in the presence of O2 were taken, while oxygen-

ated medium was flowing through the MFS.

The O2 distribution on the optode surface was highly

heterogeneous due to variations in biofilm thickness. The

biofilm partially blocked the water flow causing formation of

Fig. 2 e Calibration curves of the O2 dependent ratios of s0=s

( y [ 1049.8x L 1048.5, circles, Ksv [ 9.53 3 10L4), the

intensity I0W1/IW1 ( y [ 1164.5x L 1160.2, squares,

Ksv [ 8.59 3 10L4) both measured with the PCO camera,

and the I0/I ratio of the luminescence intensity in the red

channel measured with the mEye camera ( y [ 974.8x L

973.6, triangles, Ksv [ 1.03 3 10L3). The linear correlation

(r2) for all curves was >0.999. The points represent

average values of 1.6 3 105 pixels in a square Regions Of

Interest (ROI) in the center of the image. Error bars denote

the standard deviations of the ratio values of the pixels.

flow channels in the exopolymer matrix. All three O2 images

showed the same spatial O2 distribution patterns, and the O2

concentrations were in the same range. However, the O2

images based on the ratio method I0W1/IW1 or I0/I (Fig. 3)

showed much more details, as compared to the lifetime

images.

The biofilm in the MFS formed up to 0.8 mm thick struc-

tures due to the support of the spacer mesh (Fig. 4). Formation

of such thick biofilms caused an increased pressure gradient

within the MFS (data not shown) forcing medium through

narrow channels. The flow channels were visualized by

injection of rhodamine WT solution into the MFS (Fig. 4C).

The mEye andmany other color cameras allowmanual gain

setting of the separate color channels to optimize the pixel

saturation per channel. For O2 sensing, we used the intensity

histograms of each color channel in the camera program at

anoxia (highest luminescence) to optimize the gain settings of

every channel. In this way, no pixels were oversaturated in

any of the channels, while ensuring optimal signal-to-noise

ratio. With such optimized gain settings, it was possible to

see differences in fluorescence intensity, even without split-

ting the three channels (Fig. 5A). However, proper calculation

of O2 images from the color camera .tiff format images still

required splitting of the three channels.

We used the red channel for O2 calculation since the

emission spectrum of the Ru-dpp indicator (max. emission

610 nm) coincided best with the spectral range of the red

channel. There was also a visible change in the green channel

upon changing O2 concentration, while the blue channel did

not show much variation. There was no biofilm growth in the

MFS during the calibration, and the O2 concentration was

considered homogeneous within the field of view. However,

there was a clear intensity difference visible (Fig. 5B), due to

spatial variation in excitation light and lens effects (Fig. 5C).

After calculation of the ratio image, such heterogeneity was

gone (Fig. 5D, E) and the image ratio showed a homogeneous

response of the planar optode area in the camera field of view

to the different O2 concentrations.

To compare the ratio images from the mEye setup with the

lifetime images of the PCO setup,we analyzed a position in the

MFS that was imaged by both camera systems, consecutively

(Fig. 6). Both lifetime and ratio imagesmeasuredwith the mEye

camera yielded relatively noisy images, as compared to the

ratio I0W1/IW1 image and the I0/I image made with the PCO

camera (Fig. 3B, D). However, a structured O2 distribution due

to flow channel formation within the biofilm was still clearly

visible. Closer inspection of the images showed that the O2

distribution did notmatch exactly in the two image types. This

difference is most likely caused by repetitive changes in flow

rate. Inour setup,wehad to stop theflowto inject the saturated

dithionite solution for theanoxicmeasurement before theflow

was started again. This can result in detachment of part of the

biofilm structure. In our example,most of the structure stayed

intact and the images are still comparable (Fig. 6).

3.3. Heterogeneity of O2 distribution and consumption

The highest O2 concentrations were generally found in areas

with no or little biofilm formation (Fig. 4D), while low O2 zones

developed away from the flow channels in areas with thick



Fig. 3 e Images of O2 distribution in the biofilm growth incubator (MFS) as measured with a 12 bit PCO camera system with

a 2X magnification microscope lens mounted. The O2 concentrations in the images are calculated from lifetime imaging (A),

the ratio of the IW1 images (I0W1/IW1) under anoxic conditions (I0W1) and under oxygenated conditions (IW1) (B), and the ratio

of the luminescence intensity (I0/I ) of images recorded under anoxic and oxygenated conditions, respectively (C). The color

bar indicates the O2 concentration in mmol lL1. (For interpretation of the references to colour in this figure legend, the reader

is referred to the web version of this article.)

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 75032

biofilm and little water movement. However, overlaying the

O2 image with the biofilm distribution image also showed

some zones with relative dense biofilm formation and relative

highO2 concentrations, aswell as some regions exhibiting low

O2 concentrations without thick biofilm formation. The lowO2

zones did not reach complete anoxia at the optode surface

(Fig. 4A, B). No differences were found between the two

imaging approaches. The average O2 concentration was

25 � 13 mmol l�1 in the largest O2 depleted zone (Fig. 7A left

side, Region Of Interest (ROI) 1), while the O2 concentration in

other depleted zones was 52 � 13 mmol l�1. The O2 concen-

tration reached zero within these regions w60e90 s after

the flow in the biofilm monitor was stopped. In comparison,

the O2 concentration in the biofilm flow channels was

152 � 23 mmol l�1 and became anoxic 100e150 s after the flow

was stopped.

We estimated the O2 depletion rate by measuring a series

of O2 images after stopping the flow with a frequency of 1

image per 10 s, and by subsequent subtraction of images in

such time series. Lifetime images were generally too noisy to

acquire accurate O2 depletion rate images based on two

sequential images, and averaging of 4e5 sequential images

was necessary. However, the ratio (I0w1/Iw1) based O2 images

were less noisy, and with these images it was possible to

calculate O2 depletion rates images without averaging.

The initial O2 depletion rate (Fig. 7) in the channels

measured by the ratio method was w3.57 � 1.97 mmol l�1 s�1

(average value of 6.3 � 104 pixels), which was 2e4 times the

depletion rate (1.30� 1.39mmol l�1 s�1 averagevalueof 2.9�104

pixels) found in the low O2 zones (ROI1 excluded). However,

this was partly caused by the difference in initial O2 concen-

tration at the moment the flow was stopped (Fig. 7B). Oxygen

depletion rates in thedifferent channel regionswereabout 30%

higher (1.85 � 1.98 mmol l�1 s�1) at w50 mmol l�1 (average

50 � 24 mmol l�1). The large zone with low O2 on the left site

(ROI1) had a lower O2 depletion rate 0.65 � 1.31 mmol l�1 s�1 at

25� 14 mmol l�1 (3X104 pixels,while the other lowO2 zones had

a depletion rate of 1.00 � 1.28 mmol l�1 s�1 at an average O2

concentration of 29� 12 mmol l�1. The O2 depletion rates in the

channels were 1.15 � 1.38 mmol l�1 s�1 at an average O2

concentration of 31 � 19 mmol l�1. The O2 depletion rates

seemed equal for all 10 ROI’s at O2 concentrations below

w20 mmol l�1 (Fig. 7B). The respiration rates found in this study

are within the ranges found using micro-electrodes

(0.3e5 mmol l�1 s�1) (Nielsen et al., 1990; Satoh et al., 2005).

4. Discussion

4.1. Comparison of the different imaging techniques

All three imaging methods showed a linear relationship

between O2 concentration and so=s, I0w1/Iw1 and I0/I, indicating

a low non-quenchable fraction of the indicator in the planar

optodes used in this study. In principle, this linear relation-

ship for the range 0e300 mmol l�1 O2 allows a simple two point

calibration.

The calibration images made with the color camera (Fig. 5)

showed that the ratio approach corrects for variation in the

luminescence intensity due to heterogeneity in the light field

and lens effects. This makes the method applicable for

imaging in systems even when the light distribution of the

excitation lights is not perfectly homogeneous. A prerequisite

is, however, that the measured area can be made anoxic

without physical change or movement of the setup. In our

case, we created anoxic conditions of the monitored area by

addition of dithionite via an injection port in the tubing or by

stopping the flow of the medium. But even though we took

care not to touch the camera or MFS, the image position

sometimes changed in between capturing images, due to

minutemovement of theMFS relative to the camera. The shift

was only 2 or 3 pixels, but is enough to create erroneous

results. The risk of movement is much bigger in our system

than in the lifetime setting, since nothing is physically

touched in between the recording of the lifetime images. The

time difference between the recording is much smaller than

for the I0/I images. However, within ImageJ it can be easily

checkedwhether the images are still aligned and, if necessary,

the images can be repositioned based on chosen reference

points within both images.



Fig. 4 e Images of O2 concentration in an MFS harboring mature biofilms (10 days old). The O2 was measured using

luminescence lifetime imaging (A), and by taking the ratio between luminescence intensity images measured under anoxia

and under experimental conditions (B). The heterogeneity in O2 concentration is mostly caused by channel formation

within the incubator. The color bar indicates the O2 concentration in mmol lL1. (C) Visualization of the water channels in the

same position. The color bar indicates the water volume of moving water based on light absorption in the green channel

after injection of rhodamine WT solution into the MFS. The color bar of C indicates the biofilm thickness in mm. After

tresholding the image in panel C a mask is created with two different ranges of biofilm thicknesses. The white areas of the

mask (D) indicate the open channel, the gray shaded zones indicate areas where the biofilm was 0.1e0.3 mm thick, while

black zones indicate >0.3 mm thick biofilms. Panel E shows the O2 image (B) combined with the a mask (D) made of the

biofilm distribution image. Panel F shows the oxygen image with the support mesh overlayed. (For interpretation of the

references to colour in this figure legend, the reader is referred to the web version of this article.)

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 7 5033

The ratio I0/I images from the color camera did not have the

same spatial accuracy as the ratio based oxygen images from

the PCO camera. This is not surprising, given that the PCO

camera had a more sensitive cooled CCD chip with a bit depth

of 12, while the color camera had a simple CMOS chip with

a depth of 8 bit. A bit depth of 12 results in a 16 times larger

dynamic range as compared to 8 bit cameras, and thus amuch

higher accuracy in the calculation of the ratio. However, the

resolution, even with our relatively simple 8 bit color camera

was almost as good as the resolution of the lifetime images.

The accuracy may easily be improved by using better cameras

e.g. with a bit depth of 12 per color channel. Cameraswith a bit



Fig. 5 e Luminescence intensity images at different O2 levels (expressed as % O2 in gas phase) (A) from an O2 sensitive planar

optode mounted in an MFS measured with the mEye camera (8 bit color). (B) the red channel of the color images, (C) the

average gray values of the red channel, (D) images of the ratioI0/I. (D) the average value of the ratio (E). (For interpretation of

the references to colour in this figure legend, the reader is referred to the web version of this article.)

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 75034

depth of 12 and cooled down to �30 �C below ambient

temperature can be found nowadays for prices starting under

1000 V (e.g. cameras from Tucsen inc., China or Optic Star,

United Kingdom), which is substantially below costs and

efforts involved in establishing a luminescent lifetime system.

The O2 images calculated from lifetimemeasurements had

a lower resolution than the images based on the ratio I0/I. The

higher resolution of the I0/I ratio approach thus allowed

visualization of O2 gradients, which were not clearly visible in

the lifetime images. The reason for the difference in spatial

Fig. 6 e Comparison of lifetime image obtained with the PCO ca

channel images of the 8 bit mEye color camera from approximate

correct for differences in pixel size. Positioning and resizing wa

glass window. The color bar indicates the O2 concentration in m

a black permanent marker on the window. (For interpretation o

referred to the web version of this article.)

resolution may be caused by the exponential character of

lifetime images. Taking the exponent of the ratio Iw1/Iw2, as is

done in the lifetime images, increases the noise level of the

oxygen image. There was no obvious difference in resolution

between the I0w1/Iw1 and I0/I.

Another imaging approach based on intensity images,

rather than on lifetimewas used in the very first planar optode

applications (Glud et al., 1996, 1999), where Ksv and a in

Equation (2) were determined by measuring luminescence

intensity images at 0%, 20%, and 100% O2. Subsequently, the

mera (A), and a I0/I image (B) as calculated from the red

ly the same area in the MFS. The I0/I was reduced in size to

s based on the black ink structure (upper left) drawn on the

mol lL1. The black areas are caused by marks made with

f the references to colour in this figure legend, the reader is



Fig. 7 e Relationship between O2 concentration and O2

depletion rate for 10 regions of interest (ROIs) in the MFS.

Panel A shows an O2 distribution image with a selection of

different ROIs from which the depletion rates were

measured. Panel Bshows the relationship in the areas

where water is flowing freely (ROI 9 and 10), as well as in

the low O2 zones, characterized by the absence of free

moving water (ROI 1-8). The points in the graphs indicate

the average O2 depletion rates at average O2 concentration

values of the ROI’s indicated in the legend.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 7 5035

Ksv and a images where used to convert experimental lumi-

nescence intensity images to O2 concentration images. While

yielding high quality O2 images, this approach required the

use of non-transparent planar optodes as pure intensity based

O2 imaging is prone to light field variations and e.g. scattering

artifacts from the sample structure (Holst et al., 1998). This

method also requires the camera not be moved relative to the

optode.

In the setup presented here, the O2-dependent image ratio

is calculated directly from the ratio of an image obtained

under anoxic conditions and an image under a given O2

concentration. The conversion from ratio to O2 image is

carried out by a linear correlation. This correlation can be

determined prior to or afterward actual experiments and is

not done at pixel level. Therefore it is not required that the

optode has exactly the same position relative to the camera

during calibration, yielding a relatively flexible and robust, yet

simple O2 imaging setup.

In our case, the image ratio had a linear relationship to O2

concentration over 0e300 mmol l�1, but also non-linear rela-

tionships can be used for the conversion within ImageJ. It has

been found that other types of fluorescent dyes have non-

linear relationships with O2 concentration e.g. due to

a substantial non-quenchable fraction of the immobilized

indicator (Carraway et al., 1991). For planar optodes exhibiting

a linear relationship between image ratios and O2 concentra-

tion, it is possible to calculate dynamic changes (e.g. during

experimental lightedark shift or stop-flow) without making

an image under anoxic conditions simply by calculating the

ratio of two consecutive luminescence intensity images

multiplied by the slope found in the O2 calibration curve.

4.1.1. O2 heterogeneity in the biofilm monitorThe O2 distribution was heterogeneous in the MFS after bio-

film development. There was a good agreement with the

distribution of free moving water and the higher O2 concen-

trations. It seems logic that O2 concentrations are higher in

channels since little biomass is present in the channels, while

water is flowing relatively fast. Advective transport is impor-

tant in these regions, and fast flowing water will result in thin

boundary layers, while little biomass is present. The O2

concentration at the optode surface depends on the thickness

of the diffusive boundary layer on top of the biofilm, the

thickness of the biofilm on top of the optode and the O2

consumption rate within the biofilm. However, some regions

seemed to deviate from the relationship between water

volume and higher O2 concentration. This may partly be

explained by the fact that the rhodamine method to estimate

the free moving water did not discriminate for the distance at

which the water flows from the optode. When half of the

space is filled with biofilm, the free moving water may flow

directly over the optode, but it may also be that the biofilm is

between the optode and the water. Both situations will give

the same absorbance, but will result in different O2 concen-

trations at the optode surface. This illustrates the current

limitations in our ability to experimentally resolve spatially

complex chemical landscapes and mass transfer phenomena

in biofilms.

It was surprising that O2 was not fully depleted in regions

without flowing medium. These regions did only become

anoxic after stopping the flow in the MFS, while the lowest O2

concentration under flow conditions in the MFS was

w25 mmol l�1. Advective flow through the biofilm as reported

in several studies (Costerton et al., 1999; Stewart and Franklin,

2008) could explain the observation. Since these channels are

extremely small (a few mm) advective flow will be relatively

slow due to a high resistance. This flow may have been

overlooked by the rhodamine method used in this study. We

only injected 20 ml rhodamine solution, prior to stopping the

flow. This was enough to penetrate the main channels, but

may not enough to show a flow in areas almost completely

filled with biofilm.

Another explanation may be that the diffusion of the

organic substrate into the biofilm is slow, and not in balance

with the diffusion of O2. This would result in the consumption

of all organicmaterial in the zones next to the channels, while



wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 75036

not all O2 is consumed. The stop-flow experiments showed

that the O2 depletion rate was indeed relatively low in the

biofilm zones (Fig. 4, left side). The O2 depletion rate was

almost 2 times lower compared to the depletion rate in the

channel regions, while the biomass was assumed to be much

higher in the biofilm zones. This could indicate that the bio-

film in the low O2 zones is either inactive or limited by organic

substrates rather than by O2. However, there was still some

consumption in these regions, as shown by the onset of

anoxic conditions when flow was stopped. Actually, biofilm

regions reached anoxia before the channel regions turned

anoxic.

4.1.2. O2 depletion and the stop-flow techniqueThe O2 concentrations in themedium at the inlet and outlet of

the MFS weremeasured daily with an O2 fiber optode showing

that the O2 concentration in the medium dropped by

25 mmol l�1 during its residence in the MFS, while O2 imaging

experiments were conducted. At the moment of the planar

optode measurement, the flow was reduced from 16 to

2.7 L h�1 due to resistance caused by the biofilm formation.

This would result in a total O2 consumption rate of

18.7 nmol s�1 in the MFS. The average O2 depletion rate

measured using the stop-flow technique was 2.2 mmol l�1 s�1

for the monitored region. The free volume of the MFS was

6.5 ml, resulting in a total O2 consumption rate of the whole

monitor of 14.3 nmol s�1. This value is rather close to the value

estimated bymeasuring the O2 decrease between the inlet and

the outlet, although it is a bit lower. One explanation for the

lower value may be that the O2 sensor used to measure O2

concentrations in the inlet and outlet measures in the tubing

a bit out of the monitor. Since biofilms will also form in the

tubing, this will add an extra consumption component to the

system, resulting in a higher consumption rate. The stop-flow

technique combined with the planar optode does not involve

the tubing. Another explanation is that the surface area that

was monitored does not reflect the behavior of the total

incubator perfectly. In general, these two methods yield the

same overall results, but the imaging method gives much

more information on the heterogeneity in process rates.

Many microelectrode studies are published, showing

oxygen gradients and oxygen fluxes in biofilms, but volu-

metric O2 respiration rates of biofilms were determined in

only few publications. Several volumetric respiration rates

found with microelectrode studies were within the same

range as the O2 depletion rates found in this study

(0.3e5 mmol l�1 s�1, e.g. see Nielsen et al., 1990; Satoh et al.,

2005). Therefore we feel that the O2 depletion rates found

with the ratiometric imaging approach are representative for

the respiration rates of the biofilm. More extreme values,

ranging from 0.03 (Polerecky et al., 2005) to 50 mmol l�1 s�1 (De

Beer and Costerton, 2006) have also been reported, but the

differences found may rather result from environmental

factors like temperature, availability of organic substrate, etc.

than from the method used. With the planar optode it can be

assumed that the oxygen depletion rate reflects the oxygen

consumption rate of the thin layer just on top of the optode,

especially when the oxygen decrease is measured during the

first 10 s after the flow is stopped. Stopping the flow will

disrupt the steady state O2 gradient and therefore reflect the

respiration rate (Staal et al., 2011). After 5e10 s the O2 deple-

tion rate will become increasingly affected by a change in flux

from overlaying layers.

5. Conclusions

It can be concluded that the I0/I ratio approach for O2 imaging

with transparent planar O2 optodes can be considered a good

and more simple alternative to more elaborate luminescence

lifetime imaging approaches. However, our approach can only

be used in systems were the O2 concentration can be reduced

to zero without physical movement of the MFS relative to the

camera. Such mechanical stability can easily be achieved in

most biofilm reactor setups. A very simple imaging setup for

imaging O2 in biofilm reactors can be established from inex-

pensive commercial CCD or CMOS cameras and high intensity

LEDs. The image analysis can be performed by powerful

freeware such as ImageJ. In line with other recent ratiometric

O2 imaging approaches (Wang et al., 2010; Larsen et al., in

press), this simplifies O2 imaging and makes it more acces-

sible for the research community. Here we presented O2

distribution imagesmeasured at the base of a biofilm, but with

some modifications in the experimental setup it is also

possible to measure the O2 distribution inside granules or

biofilms.

Appendix. Supplementary material

Supplementary data associated with this article can be found,

in the online version, at doi:10.1016/j.watres.2011.07.007.

r e f e r e n c e s

Bacon, J., Demas, J., 1987. Determination of oxygenconcentrations by luminescence quenching of a polymer-immobilized transition-metal complex. Analytical Chemistry59, 2780e2785.

Carraway, E.R., Demas, J.N., DeGraff, B.A., Bacon, J.R., 1991.Photophysics and photochemistry of oxygen sensors based onluminescent transition-metal complexes. AnalyticalChemistry 63, 337e342.

Costerton, J.W., Lewandowski, Z., Caldwell, D.E., Korber, D.R.,Lappin-Scott, H.M., 1995. Microbial biofilms. Annual Review ofMicrobiology 49, 711e745.

Costerton, J.W., Stewart, P.S., Greenberg, E.P., 1999. Bacterialbiofilms: a common cause of persistent infections. Science284, 1318e1322.

De Beer, D., Costerton, J.W., 2006. Microbial biofilms. In:Dworkin, M., Falkow, S., Rosenberg, E., Schleifer, K.H.,Stackebrandt, E. (Eds.), Prokaryoteseds. Springer, New York,pp. 904e937.

De Beer, D., Stoodley, P., Roe, F., Lewandowski, Z., 1994. Effects ofbiofilm structures on oxygen distribution and mass transport.Biotechnology and Bioenginering 43, 1131e1138.

Eberl, H.J., Picioreanu, C., Heijnen, J.J., van Loosdrecht, M.C.M.,2000. A three-dimensional numerical study on the correlationof spatialstructure, hydrodynamic conditions, and masstransfer and conversionin biofilms. Chemical EngineeringScience 55, 6209e6222.




wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 5 0 2 7e5 0 3 7 5037

Fenchel, T., Finlay, B., 2008. Oxygen and the spatial structure ofmicrobial communities. Biological Reviews 83, 553e569.

Glud, R.N., Ramsing, N.B., Gundersen, J.K., Klimant, I., 1996.Planar optrodes: a new tool for fine scale measurements oftwo-dimensional O2 distribution in benthic communities.Marine Ecology Progress Series 140, 217e226.

Glud, R.N., Santegoeds, C.M., de Beer, D., Kohls, O., Ramsing, N.B.,1998. Oxygen dynamics at the base of a biofilm studied withplanar optrodes. Aquatic Microbial Ecology 14, 223e233.

Glud, R.N., Kuhl, M., Kohls, O., Ramsing, N.B., 1999. Heterogenityof oxygen production and consumption in a photosyntheticmicrobial mat as studied by planar optodes. Journal ofPhycology 35, 270e279.

Glud, R.N., 2008. Oxygen dynamics of marine sediments. MarineBiology Research 4, 243e289.

Hall-Stoodley, L., Costerton, J.W., Stoodley, P., 2004. Bacterialbiofilms: from the natural environment to infectious diseases.Nature Reviews Microbiology 2, 95e108.

Hidalgo, G., Burns, A., Herz, E., Hay, A.G., Houston, P.L.,Wiesner, U., Lion, L.W., 2009. Functional tomographicfluorescence imaging of pH microenvironments in microbialbiofilms by use of silica nanoparticle sensors. Applied andEnvironmental Microbiology 75 (23), 7426e7435.

Holst, G., Grunwald, B., 2001. Luminescence lifetime imaging withtransparent oxygen optodes. Sensors & Actuators, B: Chemical74, 78e90.

Holst, G., Kohls, O., Klimant, I., Konig, B., Kuhl, M., Richter, T.,1998. A modular luminescence lifetime imaging system formapping oxygen distribution in biological samples. Sensorsand Actuators B: Chemical 51, 163e170.

Huettel, M., Ziebis, W., Forster, S., 1996. Flow-induced uptake ofparticulate matter in permeable sediments. Limnology andOceanography 41, 309e322.

Ivleva, N.P., Wagner, M., Horn, H., Niesner, R., Haisch, C., 2010.Raman microscopy and surface-enhanced Raman scattering(SERS) for in situ analsysis of biofilms. Journal of Biophotonics3 (8e9), 548e556.

Jørgensen, B.B., Des Marais, D.J., 1990. The diffusive boundarylayer of sediments: oxygen microgradients over a microbialmat. Limnology & Oceanography 35, 1343e1355.

Kuhl, M., Rickelt, L.F., Thar, R., 2007. Combined imaging ofbacteria and oxygen in biofilms. Applied and EnvironmentalMicrobiology 73 (19), 6289e6295.

Kuhl, M., Glud, R.N., Ploug, H., Ramsing, N.B., 1996.Microenvironmental control of photosynthesis andphotosynthesis-coupled respiration in an epilithiccyanobacterial biofilm. Journal of Phycology 32, 799e812.

Kuhl, M., Jørgensen, B.B., 1992. Microsensor measurements ofsulfate reduction and sulfide oxidation in compact microbialcommunities of aerobic biofilms. Applied and EnvironmentalMicrobiology 58, 1164e1174.

Kuhl, M., 2005. Optical microsensors for analysis of microbialcommunities. Methods in Enzymology 397, 166e199.

Kuhl, M., Polerecky, L., 2008. Functional and structural imaging ofphototrophic microbial commmunities and symbioses.Aqautic Microbial Ecology 53, 99e118.

Larsen, M., Borisov S.M., Grunwald B., Klimant I. and Glud R.N.A simple and inexpensive high resolution color ratiometricplanar optode imaging approach: application to oxygen andpH sensing. Limnology & Oceanography Methods, in press.

Neu, T.R., Manz, B., Volke, F., Dynes, J.J., Hitchcock, A.P.,Lawrence, J.R., 2010. Advanced imaging techniques forassessment of structure, composition and function in biofilmsystems. FEMS Microbiology Ecology 72, 1e21.

Nicolella, C., van Loosdrecht, M.C.M., Heijnen, S.J., 2000. Particle-based biofilm reactor technology. Trends in Biotechnology 18,312e320.

Nielsen, L.P., Christensen, P.B., Revsbech, N.P., Sørensen, N.P.,1990. Denitrification and oxygen respiration in biofilmsstudied with a microsensor for nitrous oxide and oxygen.Microbial Ecology 19, 63e72.

Oguri, K., Kitazato, H., Glud, R.N., 2006. Platinumoctaetylporphyrin based planar optodes combined with anUV-LED excitation light source: an ideal tool for high-resolution O2 imaging in O2 depleted environments.MarineChemistry 100, 95e107.

Polerecky, L., Franke, U., Werner, U., Grunwald, B., de Beer, D.,2005. High spatial resolution measurement of oxygenconsumption rates in permeable sediments. Limnology andOceanography: Methods 3, 75e85.

Revsbech, N.P., 2005. Analysis of microbial communities withelectrochemical microsensors and microscale biosensors.Methods in Enzymology 397, 147e166.

Ridgway, H.F., Flemming, H.F., 1996. Membrane biofouling. In:Mallevialle, J., Odendaal, P.E., Wiesner, M.R. (Eds.), WaterTreatment Membrane Processes. McGraw-Hill, New York, pp.6.1e6.62.

Satoh, H., Sasaki, Y., Nakamura, Y., Okabe, S., Suzuki, T., 2005.Use of microelectrodes to investigate the effects of 2-chlorophenol on microbial activities in biofilms.Biotechnology and Bioengineering 91, 133e138.

Staal, M., Borisov, S.M., Rickelt, L.F., Klimant, I., Kuhl, M., 2011.Ultrabright planar optodes for luminescence life-time basedmicroscopic imaging of O2 dynamics in biofilms. Journal ofMicrobial Methods 85, 67e74.

Stewart, P.S., Franklin, M.J., 2008. Physiological heterogeneity inbiofilms. Nature Reviews Microbiology 6, 199e210.

Stoodley, P., deBeer, D., Lewandowski, Z., 1994. Liquid flow inbiofilm systems. Applied and Environmental Microbiology 60,2711e2716.

Vrouwenvelder, J.S., van Paassen, J.A.M., Wessels, L.P., vanDam, A.F., Bakker, S.M., 2006. The Membrane foulingsimulator: a practical tool for fouling prediction and control.Journal of Membrane Science 281 (1e2), 316e324.

Vrouwenvelder, J.S., Bakker, S.M., Cauchard, M., Le Grand, R.,Apacandie, M., Idrissi, M., Lagrave, S., Wessels, L.P., vanPaassen, J.A.M., Kruithof, J.C., van Loosdrecht, M.C.M., 2007.Themembrane fouling simulator: a suitable tool for predictionand characterisation of membrane fouling. Water Science &Technology 55 (8e9), 197e205.

Vrouwenvelder, J.S., Manolarakis, S.A., van der Hoek, J.P., vanPaassen, J.A.M., van der Meer, W.G.J., van Agtmaal, J.M.C.,Prummel, H.D.M., Kruithof, J.C., van Loosdrecht, M.C.M., 2008.Quantitative biofouling diagnosis in full scale nanofiltrationand reverse osmosis installations. Water Research 42,4856e4868.

Vrouwenvelder, J.S., Hinrichs, C., Van der Meer, W.G.J., VanLoosdrecht, M.C.M., Kruithof, J.C., 2009a. Pressure dropincrease by biofilm accumulation in spiral wound RO and NFmembrane systems: role of substrate concentration, flowvelocity, substrate load and flow direction. Biofouling 25,543e555.

Vrouwenvelder, J.S., Van Paassen, J.A.M., Kruithof, J.C., VanLoosdrecht, M.C.M., 2009b. Sensitive pressure dropmeasurements of individual lead membrane elements foraccurate early biofouling detection. Journal of MembraneScience 338, 92e99.

Wagner, M., Manz, B., Volke, F., Neu, T.R., Horn, H., 2010. Online3assessment of biofilm development, sloughing and forceddetachment in tube reactor by means of magnetic resonancemicroscopy. Biotechnology and Bioengineering 107 (1),172e181.

Wang, X., Meier, R.J., Link, M., Wolfbeis, O.S., 2010. Photographingoxygen distribution. Angewandte Chemie 122, 5027e5029.



11. Appendix (Posters)

Cover glass optode measurementsA ratiometric approach was used for planar optodemeasurements in to different types of flowcells forbiofilm formation. Cover glasses was prepared withan inert reference dye with a green emission and a O2sensitive dye with a red emission. Both dyes wereexitateted with blue light. Thus, it was possible tocalculate the O2 concentration from the seperated redand green pixel of a simple colour camera.

FIGURE 5. The experimental setup used for O2measurements (here for a sediment sample). Exf.:470 nm shortpass exitation filter. Emf.: 530 nmlongpass emission filter (from ref. 2)

FIGURE 6. A 20 mm x 20 mm cover glass mounted ona square flowcell. The cell was at first flushed withanoxic and the picture is take short after the flow withair saturated water is initiated from the top.

FIGURE 7. The O2 distribution in a oblong flowcellmounted with 24 mm x 50 mm optode cover glass.The cell is inoculated with Pseudomonas aeruginosa.There are 3 seperated cells in the device and only themiddel is shown.

ref.

FIGURE 2. Modulation of reference lightref., is the anoxic response and

is the response at concentration c.

Minioptode measurementsThe O2 optode meters used for micro- and minioptodes usually measurethe luminescence lifetime with a phase-modulation technique, where thelifetime, τ, can be calculated from the measured phase angle shift, Φ,between the sinusoidal intensity modulated (at frequency, fmod) exitationand emission signals:Eq. (2) tan(Φ) = 2π.fmod .τΦ is then measured as the delay or difference between the reference light(similar to exitation light) and the emission light.

IntroductionMolekular oxygen (O2) is a key metabolic parameter for all plants and animals and for the majority of microorganisms. In most ecosystems,O2 concentration and dynamics not only outlines the activity and distribution of aerobic processes but also provides a proxy for overallgeochemical production and mineralization and in medicine it can be an indication of a patient’s condition.Otical O2 measurements is a reliable, effective and fast method to determinate O2 concentrations and their mechanical robusness providespossibilities that are beyohd what can be done with traditional electrochemical sensors. Futhermore, it is possible with relativly simplemeans to create 2-dimensional pictures of the O2 distribution in an object. This can be done with a macroscopic size up to severaldecimetres or under microscope with a resolution down to a few µm.Here we present 3 different applications of optical O2 sensors (O2 optodes):1.O2 optodes are well suited for environmental analysis due to their good long-term stability of measuring signals. Here we present a new,robust O2 optical sensor array enabling long-term in situ deployment and continuous monitoring of O2 penetration and temperature undervarying water level regimes over time periodes of several months.2. To determine the pO2 in the sinus of cystic fibrosis patients and compare this to values to healthy humans a minioptode was incorporatedin a catheter.3. For 2-dimensional measurements planar optodes are used. If the O2 concentration is calculated from life time determination special fastgate-able cameras are required, but with a ratiometric approach a simple colour camera is sufficient. The O2 distribution in 2 different flow-cell systems for biofilm formation is shown using a ratiometric system.

Datalogger

Multi-channeloxy-meter

thermocouple cablepolymer optical fibre

stainless steel bar

oxygen sensitive coating on Mylar foil

thermocouple

polyurethane rubber

silicone rubber

steel nut

silicone disc

steel disc

steel screw

FIGURE 3. Scematic drawing of a 1.3m long sensor arrayspear euipped with 10 sensorspots along the cylinder shaftfor in vivo O2 and temperature measurements.

A sensor array ready for use. Theblack spots along the blue tube arethe sensors covered with blacksilicone.

Optical O2 sensors (O2 optodes)and some applications

L.F. Rickelt1 and M. Kühl1,2,3

1Marine Biological Section, Department of Biology, University of Copenhagen, Strandpromenaden 5, DK-3000 Helsingør, Denmark2Plant Functional Biology and Climate Change Cluster, University of Technology Sydney, Ultimo Sydney NSW 2007, Australia

3Singapore Centre on Enviromental Life Sciences Engineering, School of Biological Sciences, Nanyang Technological University, Singapore

Dynamic Quenching of Luminescence1. Luminescence process in absence of O2

O2

absorption of lightin ground state excited state emmision of light

2. Quenching of luminophore by O2

O2

absorption of light

excited state

excited state(singlet)

O2ground state

(triplet)

energy transferby collision

FIGURE 1. All optical oxygen sensors are based on quenchingof a luminescent compound (fluorescent or phosphorescent) bymolecular oxygen. The luminescent response to the O2 concen-tration follows the Stern-Volmer relation:

Eq. (1)

where Φo, Io and τo are the phase angle, luminescence intensityand lifetime, respectively, of the indicator in the absence of O2,and Φc, Ic and τc are the phase angle, luminescence intensityand decay time of the indicator at a given O2 content, c. KSV is acharacteristic quenchingcoefficient of the immobilized indicatorand can be detemined as the slope of a plot of Io/Ic vs. the O2concentration.

= = =tan(Φ0) I0 τ0tan(Φc) Ic τc

1 + Ksv.c

The spear array in situ in peat soil at Maglemosen,20 km north of Copenhagen.

In situ application of a spear array with 10 minioptodesTo monitor the in sito variations of O2 content at different depths in a peat soil, a spearequiped with an array of 10 oxygen minioptodes and temperature sensors wasconstructed (FIGURE 3) and placed in the wetland area, Maglemosen, 20 km north ofCopenhagen in November 2008. The simultaneous registration of luminescent phaseangel shift and temperature allowed for a temperature correction of the signal. The waterlevel was also followed. As an exapel of the results, the data for a 3 week period in May,2009, are shown at depths: 2, 10, 20, 30, and 40 cm (FIGURE 4A+B). In general, the O2distribution in the peat soil was affected by the water level. When the water level reachedbelow a specific sensor depth, the local O2 content increased and vice versa. The surfacesensor exhibited diel O2 (FIGURE 4A) and temperature (FIGURE 4B) variations, butthe variations were not coincident. Thus the diel variations can best be ascribed torhizospheric oxidation from the vegetation, which mainly consisted of Phlarisarundinacea.

FIGURE 4. In situ measurements in a peat soil (Maglemosen)during a 3 week period in 2009. (A) O2 and water levelvariations. The dashed lines mark the depths of the respectiveO2 sensors in relation to the water level. (B) The measuredtemperature at the same depths.

[email protected]

1) Aanæs, K., L.F.Rickelt, H.K.Johansen, C.v.Buchwald, T.Pressler, N.Højby, and P.Ø.Jensen:Decreased mucosal oxygen tension in the maxillary sinuses in patients with cyctic fibrosis.Journal of Cystic Fibrosis (2011) 10: 114-1202) Larsen, M., S.M.Borisov, B.Grunwald, I.Klimant, and R.N.Glud:A simple and inexpensive high resolution color ratiometric planar optode imaging approach: application to oxygen and pH sensingLimnol. Oceanogr. Methods (2011) 9: 348-360

FIGURE 9. The catheter on its way tothe right maxillary sinus

FIGURE 8. The catheter with the incorporatedoptode.

Oxygen tension in the maxillary sinuses1

For pO2 in vivo measurments in the maxillary sinuses in cysticfibrosis patients a wing catheter was equipped with a 2 mmplastic optical fiber prepared with a O2 sensitive dye at the end(FIGURE 8). In FIGURE 9 is the catheter shown in action. Thepatient is under full anaesthesia. From a group of 20 CF patientscompared with a equal control group there was a significantdifference in the pO2 (FIGURE 10).

FIGURE 10. Results of the pO2 measurmentsin the mucosa of all maxillary sinuses (t-test)p < 0.0263

0 5 10 15 20 25 30 35 40 45 50 55 60 65

0

50

100

% a

irsatu

ration

time (seconds)

whitout recess

recess

Figure 5 – Response time curves for 16 flat cut sensors; 8 without recess (red)

and 8 with a ~25 µm deep recess (blue). Each fiber was moved from air

saturated water to anoxic water (1% Na2SO3) at time 0 s.

Figure 4 - Images of the light emission from different types of optical fiber tips.

The light source was a blue LED. The fiber tips were inserted in a dilute milk

suspension (left column) and in a solution of the water soluble O2 indicator

Ru(dpp(SO3Na)2)3 (right column). After pictures were taken of the flat-cut tips,

the tips were etched and an additional set of pictures were taken.

Figure 3 – Photographs of etched fiber tips. (A) A flat-cut optical fiber tip etched

in HF for 240 s. (B) A tapered optical fiber tip etched for 90 s

40 µm tip, 20 µm recess B

_____ 100 µm

133 µm tip, 75 µm recess A

100 µm _____

Responstime

100-0%

Amplitude

0% (anoxic)

Amplitude

100%

(airsaturated)

Untapered sensors

without recess 29.3±8.8 17,715±2,482 7,609±2,482

Untapered sensors

with ~25 µm recess 11.7±4.7 15,100±7,897 7,331±4,090

Table 1. Comparison of 95% response time and signal amplitude of flat cut fiber

O2 optodes with and without recess. Numbers indicate means±standard

deviation (n=8). The table shows faster response times for recessed sensors with

an amplitude equivalent to sensors without recess. This indicates a focusing of

light in the recess.

0 500 1000 1500 2000

0

10

20

30

40

50

60

70

80

0 200 400 600 800 1000 1200 1400

0

20

40

60

80

100

120

140

160

180

rad

ius (

µm

)

A

core

cladding

B

rem

ove

d f

iber

(µm

)

etching time (seconds)

recess depth 22oC

removed cladding 22oC

removed cladding 23oC

Figure 2 – Fiber radius of untapered optical fibers as a function of the etching time at 22°C

and 23°C. The position of the core and cladding is indicated (A). The depth of the recess

(22°C) and the amount of cladding material removed as a function of etching time (22°C

and 23°C) (B).

In the core of optical glass fibers GeO2 is used as dopant in the SiO2 for raising the refractive

index. The shape of the etching curve (A) is consistent with a decrease in refractive index from

the centre to the core-cladding interface. Because GeO2 is dissolved faster than SiO2 by

hydrofluoric acid the etching rate decreases with a gradual decreasing refractive index.

Overview:

To produce fiber-optic O2 with better mechanical stability and faster response time a recess

was etch at the sensing end of graded index multimode optical glass fibers. The new procedure

for making this recess is based on etching fiber tips in 40% hydrofluoric acid (HF) for defined

immersion times (Fig. 1). As the etching velocity decreases radially from the core center in

multimode graded index fibers (Fig. 2), a recess can be formed in the tip of flat-cut untapered

(Fig. 3A) or tapered fibers (Fig. 3B). Etched fiber tips showed improved focussing of

excitation light coupled into the fiber at the opposite end (Fig. 4), and very efficient excitation

of thin layers of optical O2 indicators immobilized into the recess resulting in shorter response

times (Fig. 5, Table 1). The sensor chemistry is well protected when immobilized in recessed

fiber tips and allows the construction of O2 microoptodes with improved mechanical stability

that can measure repeatedly even in very cohesive biofilms, tissue and dry soil (Fig. 1).

Figure 1 – Schematic diagram of the setup for etching optical fibers tips with hydrofluoric

acid (HF). The same setup was used for testing the mechanical stability of O2 microoptodes.

For this, the Eppendorf tube was replaced with a glass beaker containing the test media and a

microoptode was connected to a fiber-optic O2 meter. In a cohesive mat all sensors without

recess lost their entire signal and the sensor chemistry was torn off when the sensor tip was

pulled back, whereas recessed sensors still showed good signals.

Shaping the measuring tip of fiber-optic O2 microprobes

for enhanced performance Lars F. Rickelt and Michael Kühl

Marine Biological Section, Department of Biology, University of Copenhagen,

Strandpromenaden 5, DK-3000 Helsingør, Denmark.

[email protected]

Documents

Development and application of fiber-optic sensors in environmental