Ms

Ya

Ub

a

ARRAA

KPMSC

1

rtsgtmrttismfOeb

sr

h0

Sensors and Actuators B 221 (2015) 1094–1103

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

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

id-infrared holey and slotted photonic crystal waveguides inilicon-on-sapphire for chemical warfare simulant detection

i Zoua,∗,1, Swapnajit Chakravartyb,∗,1, Parker Wraya, Ray T. Chena,b,∗

Microelectronics Research Center, Department of Electrical and Computer Engineering, University of Texas at Austin, 10100 Burnet Rd. Austin, TX 78758,SAOmega Optics, Inc., 8500 Shoal Creek Blvd., Austin, TX 78757, USA

r t i c l e i n f o

rticle history:eceived 20 April 2015eceived in revised form 10 July 2015ccepted 14 July 2015vailable online 20 July 2015

a b s t r a c t

We provide the first experimental demonstration of the transmission characteristics of holey and slot-ted photonic crystal waveguides (PCWs) in silicon-on-sapphire at mid-infrared wavelength of 3.43 �m.Devices are studied as a function of lattice constant to tune the photonic stop band across the singlewavelength of the source laser. In the holey PCW, propagation losses less than 20 dB/cm are observedbelow the light line while more than 400 dB/cm propagation losses are observed above the light line,

eywords:hotonic crystal waveguidesid-Infrared waveguides

ilicon photonicshemical sensor

characteristic of the waveguiding behavior of photonic crystal line defect modes. Comparison betweenthe holey and slotted PCWs with PCWs without low-index inclusions shows the higher sensitivity torefractive index changes in chemical sensing for the holey and slotted PCWs compared to the regularPCWs. Triethylphosphate (TEP), a chemical warfare simulant, was detected down to a concentration of10 ppm (parts per million) with a 800 �m long holey PCW in gas phase via optical absorbance signature.

© 2015 Elsevier B.V. All rights reserved.

. Introduction

Slow light in PCWs [1] and slotted PCWs [2,3] has been used toeduce the optical absorption path length and achieve high detec-ion sensitivity in on-chip optical absorption spectroscopy for theelective detection of volatile organic compounds (VOC) [1,2] andreenhouse gases [3] based on unique analyte absorption signa-ures in the near-infrared (near-IR). In the mid-infrared (mid-IR),

ost compounds and gases of interest, such as xylene and methaneespectively, have at least two orders of magnitude larger absorp-ion cross-sections than in the near-IR. Since silicon is opticallyransparent from 1.1 �m to 8 �m [4], PCWs and slotted PCWsn silicon-on-insulator (SOI) (from 1.1 �m to 3.7 �m), silicon-on-apphire (SoS) (from 1.1 �m to 5.5 �m) and free-standing siliconembranes (from 1.1 �m to 8 �m) can serve as the ideal plat-

orm for highly sensitive optical absorption spectroscopy on chip.

f these, silicon membranes provides the largest bandwidth, how-ver, the fabrication yield and long waveguide structures tend toend and buckle due to the free-standing structure. The bending

∗ Corresponding authors. Tel.: +1 512 471 7035; fax: +1 512 471 8575.E-mail addresses: yzou@utexas.edu (Y. Zou),

wapnajit.chakravarty@omegaoptics.com (S. Chakravarty),aychen@uts.cc.utexas.edu (R.T. Chen).

1 Joint first authors.

ttp://dx.doi.org/10.1016/j.snb.2015.07.061925-4005/© 2015 Elsevier B.V. All rights reserved.

issue limits the maximum lengths of PCWs that can be effectivelyused for on-chip absorption spectroscopy. Recently Chiles et al.developed silicon rib optical waveguides [5] supported by a highindex silicon membrane on air cladding as a new platform for mid-infrared integrated photonics. However, the high index membranebelow the rib waveguide in [5] would reduce the photonic band gapin PC devices and thus make PC design challenging. Another plat-form in Silicon on nitride (SON), demonstrated by Khan et al. [6]has a transparency window that extends to 6.7 �m; however thehigher refractive index of cladding silicon nitride (n > 2, dependingon silicon fraction) compared with sapphire (n∼1.7) also reducesthe available photonic band gap in PC structures. Furthermore, at�∼3.4 �m, SON waveguides are also more lossy (5.2 dB/cm) [6]compared to SoS ridge waveguides (2.1 dB/cm) [7]. Thus SoS pro-vides an ideal platform for PC-based mid-IR photonic sensors till5.5 �m. PC research in the mid-IR has however been limited by theneed for bulky and expensive tunable laser sources [8,9] that madeit conceivably difficult to transition PC sensors from the lab to thefield.

We previously demonstrated the enhanced detection sensi-tivity of VOCs in PCWs versus strip and slot waveguides [10,11]in near-IR by infrared absorption spectroscopy on chip. We also

demonstrated slot waveguides in the mid-IR in SoS [7]. Recently,we provided the first demonstration of PCW characteristics in SoSat the mid-IR wavelength 3.43 �m with a fixed wavelength inter-band cascade laser (ICL) [12]. While slowdown of light increases the

tuato

imlwaglpclls

twhPdmatst

2

PhssooFbgbiaastmsbgnagoppmihpmTpadeiifi

Y. Zou et al. / Sensors and Ac

nteraction time between the analyte and the propagating opticalode, in PCWs the optical interaction in the slow light regime is

imited primarily to the in-plane evanescent optical mode overlapithin the holes of the PC lattice, in the first two or three rows

djacent to the PCW. It was shown previously [13] that slot wave-uides enhance the optical intensity of the propagating mode in aow index slot within a high index strip waveguides. Slotted PCWsrovide enhanced optical absorbance by the analyte due to theombined effects of enhanced optical mode overlap with the ana-yte in a slot at the center of the PCW, and the large slowdown ofight in the PCW. However, in the mid-IR, slotted PC waveguidetructures have not been experimentally demonstrated.

In this paper, we experimentally investigate, for the first time inhe mid-IR, two structures that enable enhanced intensity overlapith the analyte within a slow light PCW. The first structure is aoley PCW wherein holes with diameters smaller than the bulkC are etched as enclosed circular openings along the propagationirection within the PCW at the antinodes of the PCW propagatingode. The second structure is a conventional slotted PCW wherein

rectangular slot is etched uniformly in the center of the PCW fromhe input to the output. We finally detect the chemical warfareimulant triethylphosphate via optical absorbance signatures usinghe miniature chip-integrated slow light enhanced devices.

. Design and principle

A schematic of the holey PCW is shown in Fig. 1(a). The holeyCW device comprises a W1.2 PCW with a single missing row ofoles along the �–K direction in a hexagonal lattice of air holes inilicon, with lattice constant a. In the center of the PCW, a row ofmaller holes with radius rs = 0.625r, where r = 0.25a is the radiusf the holes in the bulk lattice, is etched at the lattice positionsf the PC. W1.2 indicates that the width of the PCW is 1.2 × √

3a.ig. 1(b) shows the dispersion diagram of the holey PCW obtainedy 3D plane-wave expansion (PWE) simulation. The dispersion dia-ram shows the guided mode separated from the dielectric bandy a stop gap. The light line for sapphire (refractive index n = 1.7)

s superimposed. The optimized structure was designed to ensure large guiding bandwidth for the propagating PCW guided modes well as a large electric field overlap with the analyte. Since aingle wavelength ICL is used in or research, it is also necessaryo ensure the existence of a sufficiently wide stop gap to enable a

ore accurate determination of the slow light guiding transmis-ion regime of the guided mode. Fig. 1(c) plots the guided modeandwidth below the sapphire light line and the width of the stopap between the guided mode and the dielectric band in terms oformalized frequency (a/�), as a function of the radius of the smallir holes, rs. A negative stop gap indicates that the frequency of theuided mode at the transmission band edge is lower than the zenithf the dielectric band. In order to ensure both good coupling to andropagation via the holey PCW guided mode, the stop gap must beositive. Hence rs must be greater than 0.6r. The holey PCW guidedode bandwidth however decreases monotonically with increas-

ng rs. Fig. 1(d) plots the percent electric field overlap with the airoles in the holey PCW as a function of rs. Fig. 1(d) also plots theercent electric field overlap and the peak electric field enhance-ent factor in the small air holes at the center of the holey PCW.

he percent electric field overlap increases with increasing rs. Theeak electric field enhancement factor is nearly constant at 3.5 as

function of rs. We therefore choose rs = 0.625r as our optimizedevice, taking into account the above design constraints. The 3D

lectric field intensity profile of the propagating slow light mode,ndicated by the red circle in Fig. 1(b) is shown in Fig. 1(e). Fig. 1(f)s a two-dimensional (2D) cross section of the field intensity pro-le in Fig. 1(e) through one of the small holes in the center of the

rs B 221 (2015) 1094–1103 1095

holey PCW. The cross-section is taken through the center of the sili-con slab. The cross-sectional profile shows approximately 3.5 timesenhancement of the peak electric field intensity in the small holes.

A schematic of a rectangular slotted PCW is shown in Fig. 2(a).The slotted PCW device comprises a W1.5 PCW with a single miss-ing row of holes along the �–K direction in a hexagonal lattice ofair holes in silicon, with lattice constant a. The radius of air holesis r = 0.25a. In the center of the PCW, a rectangular slot with widths = 0.155a, is etched from the input to the output. W1.5 indicatesthat the width of the PCW is 1.5 × √

3a. Fig. 2(b) shows the dis-persion diagram of the slotted PCW obtained by 3D plane-waveexpansion simulation. Similar to the holey PCW, the optimizeddevice geometry is determined by scanning the slot width. Fig. 2(c)plots the guided mode bandwidth and width of stop gap, in termsof normalized frequency as a function of slot width s. Since stopgap must be positive, the slot width must be bigger than 100 nm. Incontrast to the holey PCW, in the conventional rectangular slottedPCW, the guided mode bandwidth and peak enhancement factordecrease monotonically with the increasing slot width. Fig. 2(d)shows that the percent electric field overlap in the slot as well asall etched air regions in the rectangular slotted PCWs has its max-imum value around 160 nm. Taking the various design constraintsinto considerations, we choose the slot width 130 nm in this specificdesign, to provide an obvious stop gap, a relatively large bandwidthfor the guided mode, and large electric field overlap. It may be notedhere that in applications such as electro-optic polymer infiltratedslotted PCWs where only the polymer in the slot is poled and hencethe electric field overlap with the slot is of interest only. However,in sensing applications, the analyte infiltrates both the slot and theair holes of the bulk lattice. Hence, in both the holey PCW and therectangular slotted PCW demonstrated here, one would be inter-ested in the percent electric overlap with all air etched regions inthe PCW sensor device. The electric field intensity profile of thepropagating slow light mode, indicated by the red circle in Fig. 2(b)is shown in Fig. 2(e). Fig. 2(f) is a cross section of the field intensityprofile in Fig. 2(e), showing the factor of 13 enhancements in theelectric field intensity in the slot, larger than the holey PCW.

For the purpose of comparisons, the guided mode energy distri-bution in a conventional W1 PCW is shown in Fig. 3. Fig. 3(a) and (b)show the schematic of conventional W1 PCW and its dispersion dia-gram obtained by 3D plane-wave expansion. Fig. 3(c) and (d) showthe 3D electric field intensity profile and the cross-section inten-sity in a conventional W1 PCW. In contrast to the holey and slottedPCWs, the guided mode only interacts evanescently in-plane withthe holes adjacent to the PCW. Due to the transverse electric (TE)nature of the guided mode, similar to the holey and conventionalslotted PCW, the mode interacts with any analyte minimally viaevanescent interactions out-of-plane.

3. Results and discussions

3.1. Device characterization

The height of the silicon device layer according to wafer specifi-cations is 600 ± 60 nm. In this work, the height of the silicon devicelayer was determined to be h = 585 ± 1 nm from ellipsometry forthe holey PCW and h = 570 ± 1 nm from ellipsometry for the slottedPCW with mean square error (MSE) equal to 6 A. h/a of the siliconslab was calculated accordingly for each lattice constant. Deviceswere fabricated using a combination of electron beam lithogra-phy and inductively coupled plasma etching. Scanning electron

micrograph (SEM) images of the devices are shown in Figs. 4 and 5respectively.

Light at 3.43 �m from a source Thorlabs ICL is coupled into andout of the chip from a single mode ZrF4 fiber from Thorlabs using

1096 Y. Zou et al. / Sensors and Actuators B 221 (2015) 1094–1103

Fig. 1. (a) Schematic of the holey W1.2 PCW, which is formed by etching a row of smaller air holes with radius rs = 0.625r in the center of the PCW. (b) Dispersion diagramof the device in (a) by 3D plane-wave expansion. (c) Bandwidth and stop gap variations versus rs . (d) Peak electric field enhancement factor (blue) and percent electric fieldoverlap (filled black) in small air holes in the center of the PCW, as a function of rs . Open black triangles plot the percent electric field overlap with all air holes in the holeyPCW as a function of rs . The gray areas in (c) and (d) indicate forbidden design regions due to non-existence of a positive stop gap. The red star indicates the parameter wes light mr of the

siAmtl

elected in this design. (e) 3D electric field intensity profile of the propagating slow

ed dashed plane, at the center of the slab through the small air holes in the center

ubwavelength grating (SWG) couplers. The design and character-zation of SWGs used here has been described in detail before [7].

s in previous research [12,14,15], a PC group index taper is imple-ented at the input and output interfaces of the holey PCW with

he strip waveguide to reduce Fresnel reflection losses and enhanceight coupling efficiency into the slow light guided mode. The PC

ode in (b). (f) Cross-section of the electric field intensity profile in (e) taken on the holey PCW in (e).

group index taper is formed by gradually widening the PCW fromW1.2 to W1.27 at the interface between the strip waveguide and

holey PCW.

In the slotted PCW, light is first coupled into the chip fromSWGs into strip waveguide. Light from the strip waveguide is cou-pled next into a slot waveguide using a S-shape mode converter

Y. Zou et al. / Sensors and Actuators B 221 (2015) 1094–1103 1097

Fig. 2. (a) Schematic of the slotted W1.5 PCW with a rectangular slot s = 0.155a, in the center of the PCW. (b) Dispersion diagram of the device in (a) by 3D plane-waveexpansion. (c) Guided mode bandwidth and stop gap as a function of slot width s. (d) Peak electric field enhancement factor (blue) and percent electric field overlap (filledblack) within the slot in the center of the PCW, as a function of s. Open black triangles plot the percent electric field overlap with all air holes in the holey PCW as a function ofr ce of

( oss-sei

[todwp

s . The gray areas in (c) and (d) indicate forbidden design regions due to non-existene) 3D electric field intensity profile of the propagating slow light mode in (b). (f) Crn (e).

7]. Finally, light is coupled into the slotted PCW interface fromhe slot waveguide. The design, fabrication and characterization

f the slot waveguide and mode converter used here has beenescribed in detail previously [7]. Similar to the previous work,e chose a 130 nm slot with 600 nm rails as the input and out-ut slot waveguide dimensions, which had resulted in highest

a positive stop gap. The red star indicates the parameter we selected in this design.ction of the electric field intensity profile at the position as shown by red dash line

optical coupling efficiency into the slot waveguides in our simu-lation from input strip waveguides. A PC group index taper is also

implemented to match group velocities from the slot waveguideinto the slotted PCW. For the slotted PCW demonstrated here, theslotted PCW width is increased from W1.5 to W1.6 at the slot-ted PCW-slot waveguide interface. All device components for the

1098 Y. Zou et al. / Sensors and Actuators B 221 (2015) 1094–1103

F in (a)

s he ele

ra

lrtiddlaFcd

ig. 3. (a) Schematic of a conventional W1 PCW. (b) Dispersion diagram of the devicelow light mode in a W1 PCW at the Brillouin zone boundary. (d) Cross-section of t

espective holey and slotted PCWs are fabricated in the same stepnd post-processed as described before in Piranha/HF [7,16].

The light from the output SWG for short 50 �m (∼60 periods)ong holey PCW devices with several lattice constants but the same/a is plotted in Fig. 6(a). The plot is normalized with respect tohe highest and lowest power observed versus lattice constant. Its observed that for a less than 830 nm, the output power from theevices is practically zero. Propagation loss measurements wereone for several devices in Fig. 6(a) and plotted. Devices with

engths of 50 �m, 100 �m, 200 �m and 400 �m were fabricated

t different lattice constants. Transmission losses as observed inig. 6(b) were obtained. The lattice constants for propagation lossharacterization were chosen based on a comparison with previousata [12] which showed a similar bulge in the short 50 �m long

by 3D plane-wave expansion. (c) 3D electric field intensity profile of the propagatingctric field intensity profile at the position as shown by red dash line in (c).

PCWs at lattice constants for which the ICL emission propagates atwavelengths near the light line boundary. It was ensured from ellip-sometry that the selected chips for fabrication had h = 585 ± 1 nmwith mean square error (MSE) equal to 6 A. Fig. 6(a) also plots thecorresponding propagation losses versus lattice constant (open cir-cles). The measured propagation losses at a = 840 nm and 845 nmare 15 dB/cm and 13 dB/cm respectively. At a = 850 nm, a = 865 nm,a = 870 nm, and a = 890 nm, the propagation loss increases rapidlywith increasing length of the PCW, reaching as high as 730 dB/cmfor a = 870 nm. It can thus be concluded that the ICL emission prop-

agates above the light line for devices with a greater than 845 nm.The stop gap and light line boundaries are thus approximatelydemarcated based on the transmission measurements. In betweenthe two dashed line, shown in Fig. 6(a), light propagates in the holey

Y. Zou et al. / Sensors and Actuators B 221 (2015) 1094–1103 1099

Fig. 4. (a) Top view SEM image of the holey PCW. Top view magnified images showing the holey PCW (b) at the interface between the holey PCW and the input stripwaveguide, and (c) near the center of the device in (a).

F showi7

Pfgos(ridi

p

ig. 5. (a) Top view SEM image of the slotted PCW. (b) Top view magnified images

0o tilt view of the slot mode converter at the input (or output) of the slotted PCW.

CW mode below the light line. As the lattice constant is decreased,rom left to right in the inset of Fig. 6(a), the ICL emission propa-ates in the respective devices at higher group indices as a resultf which the propagation loss increases and the transmitted inten-ity decreases. When the device is immersed in tetrachloroethyleneC2Cl4), a liquid which is non-absorbing at 3.43 �m, the band cor-esponding to the holey PCW guided mode in Fig. 1(b) is loweredn frequency. Hence the fixed wavelength ICL can transmit light in

evices with smaller lattice constant, as shown by the dashed curve

n the inset of Fig. 6(a).Fig. 7(a) and (b) plot the transmission measurement results and

ropagation loss from the slotted PCW. Similar to the holey PCW,

ng the slotted PCW near the center of the device image in (a). (c) Top view and (d)

zero transmission is observed for lattice constants for which oursingle wavelength laser source at 3.43 �m propagates in the stopgap of the dispersion diagram. High propagation losses exceed-ing 600 dB/cm are observed for representative lattice constantsa = 850 nm and a = 860 nm for which the single wavelength at3.43 �m propagates above the sapphire light line. Significantlylower propagation losses are observed at representative latticeconstants a = 830 nm and a = 840 nm between the dashed lines in

Fig. 7(a) when the single wavelength source is propagated by theguided mode from Fig. 2(b) in the slotted PCW waveguide.

We observe that the guided mode propagation loss belowthe light line for the slotted PCW device, while still significantly

1100 Y. Zou et al. / Sensors and Actuators B 221 (2015) 1094–1103

Fig. 6. (a) Normalized transmitted intensity through an air-clad W1.2 holey PCWin SoS with r = 0.25a, rs = 0.625r as a function of a at � = 3.43 �m plotted in air(bold circles). Propagation losses were measured for devices with different latticeconstants a = 840 nm, 845 nm, 850 nm, 865 nm, 870 nm and 890 nm (open circles).Insets magnify the data for devices between a = 810 nm and a = 845 nm in air (•)and tetrachloroethylene (�). (b) Propagation loss for devices of different length ata = 840 nm (�), a = 845 nm ( ), a = 850 nm ( ), a = 865 nm ( ), a = 870 nm ( )aaa

ltdNia

3

haiisomitat

Fig. 7. (a) Normalized transmitted intensity through an air-clad W1.5 slotted PCWin SoS with r = 0.25a, s = 0.155a as a function of a at � = 3.43 �m plotted in air (boldcircles). Propagation losses were measured for few devices (open circles). Insetsmagnify the data for devices between a = 830 nm and a = 800 nm in air (•) and tetra-chloroethylene (�). (b) Propagation loss for devices of different length at a = 830 nm

nd a = 890 nm ( ). The five dash lines are the linear fit for a = 840 nm (black), = 845 nm (blue), a = 850 nm (red), a = 865 nm (green), a = 870 nm (magenta), and = 890 nm (cyan).

ower than above the light line, is of the order of 55 dB/cm. Inhe near-infrared, 115 nm wide slotted silicon PCWs in SOI haveemonstrated propagation losses of the order of 30 dB/cm [17].arrow slotted PCWs in SOI comparable to the scaled slot width

n the present case have demonstrated similar propagation lossess those demonstrated here.

.2. Sensing

From the electric field profiles, the slotted PCWs presentederein have the largest enhancement factor compared to the holeynd conventional PCW. The peak electric field enhancement factorsn the holey PCW and the conventional rectangular slotted PCWs 3.5 times and 13 times compared to a regular PCW without alot in the present design. Similarly, in contrast to a 5% percentverlap with air etched regions in a conventional PCW, in our opti-ized designs, the percent electric field overlap with the analyte

s approximately 8% and 15% in the holey and conventional slot-ed PCWs respectively. To quantify the relation between sensitivitynd electric field overlap, 3D FDTD simulation was further doneo determine the band shifts of the transmission band edge. C2Cl4

(�), a = 840 nm ( ), a = 850 nm ( ) and a = 860 nm ( ). The four dash lines are thelinear fit for a = 830 nm (black), a = 840 nm (red), a = 850 nm (green), and a = 860 nm(blue).

with refractive index around 1.5 was chosen as a top cladding forthe whole device. The simulation results are plotted in Fig. 8(a)–(c)for conventional PCWs, holey PCWs, and slotted PCWs respectively.

As shown in Fig. 8 the slotted PCW shows the largest shift intransmission band edge followed by holey PCW and then regularPCW due to the progressively larger mode overlap with analytes.The simulated shift of the transmission band edge for the slottedPCW is approximately two times the shift in the designed holeyPCW which agrees well with the two times larger percent electricfield overlap with the analyte in a conventional slotted PCWcompared with holey PCW. However, since the propagation lossesfor the holey PCW (about 15 dB/cm) were more than three timeslower than that for the slotted PCW (about 55 dB/cm), hence holeyPCW devices were selected for the detection of TEP (Triethylphos-phate, C6H15O4P) which has a strong absorption peak at 3.43 �m.The measurement setup is shown in Fig. 9. The transmitted lightthrough an 800 �m long holey PCW with a = 845 nm is measuredin the presence and absence of TEP. A measured quantity of TEP

was provided by controlling the gas flow-rate in a commercialKin-Tek vapor generator. And the vapor released via heated tubingat a height of approximately 1–2 mm from the surface of the chipto mimic a field version of the system with gas being sensed in

Y. Zou et al. / Sensors and Actuators B 221 (2015) 1094–1103 1101

Fig. 8. Simulated transmission with air-clad and C2Cl4-clad conditions for (a) conventional PCW, (b) holey PCW, and (c) slotted PCW.

used t

ftFtodTTirmnttPiawTst

Fig. 9. Schematic of the experimental setup

ree space. The chip surface was also heated to 70 ◦C to ensure thathe vapors of TEP did not condense onto the surface of the chip.ig. 10(a) shows the change in transmitted light intensity throughhe holey PCW with instantaneous switching on and switchingff of the TEP flow at 10 ppm. The transmitted signal intensityrops to 80% of its original intensity indicating the presence ofEP. In contrast in Fig. 10(b), the steady state in the presence ofEP is about 60% of its original intensity. The measurements arendependent of the flow rate of nitrogen at 10 ppm and 50 ppmespectively is indicated by the same starting intensity of trans-itted light through the holey PCW prior to switching on TEP. The

oise in the measurements is attributed to the electrical noise ofhe detector and the vibrations of the optical fiber that disturbshe alignment during the optical measurements. When the holeyCW section in Fig. 10 is replaced by a slot waveguide between thenput and output SWGs, no change in intensity of light transmittedt 3.43 �m is observed at 25 ppm TEP concentration in Fig. 11(a),

hile a small change in intensity is observed at a much larger

EP concentration of 28 pph in Fig. 11(b). When the holey PCWection in Fig. 10 is replaced by a strip waveguide, no change inransmitted intensity is observed at 28 pph TEP (Fig. 12).

o characterize our device and sense TEP gas.

According to Beer–Lambert infrared absorption spectroscopy,transmitted intensity I is given by:

I = I0 exp(−˛�L) (1)

where I0 is the incident intensity, is the absorption coefficient ofthe medium, L is interaction length and � is the medium-specificabsorption factor determined by dispersion enhanced light-matterinteraction. In conventional free-space systems, � = 1; thus L mustbe large to achieve a suitable sensitivity of measured I/I0. For lab-on-chip systems, L must be small, hence � must be large. Mortensenet al showed [18] using perturbation theory that

� = f × c/n

vg(2)

where c is the velocity of light in free space, vg is the group velocityin medium of effective index n and f is the filling factor denotingrelative fraction of optical field residing in the analyte medium.

From the propagation loss data in Fig. 6(a) that identifies the posi-tion of the light line, and following an analysis similar to ref. 10, ata = 845 nm, our single wavelength laser source propagates throughthe holey PCW at a group index ng = 20 compared to refractive index

1102 Y. Zou et al. / Sensors and Actuators B 221 (2015) 1094–1103

Fig. 10. Change in transmitted light intensity through an 800 �m long holey PCWwith a = 845 nm with introduction of (a) 10 ppm TEP and (b) 50 ppm TEP.

Fig. 11. Transmitted light intensity through a silicon slot waveguide in SoS in thepresence and absence of (a) 25 ppm and (b) 28 pph TEP.

Fig. 12. Transmitted light intensity through a silicon strip waveguide in SoS in thepresence and absence of 28 pph TEP.

n = 1 in air. Group index ng is inversely related to group veloc-ity in Eq. (2). The smaller holes in the center of the holey PCWalso provide enhanced electric field overlap as shown in Fig. 1(e).The slot waveguide enhances f, and therefore provides higher sen-sitivity than a strip waveguide. The holey PCW enhances both fand ng, and is therefore the most sensitive amongst the threedevices.

In the present devices, the analyte is already known to haveabsorbance at 3.43 �m. For an unknown analyte, it will be nec-essary to measure absorbance at multiple wavelengths. In orderto construct a complete lab-on-chip spectrometer, one can envi-sion several on-chip ICLs integrated in parallel via wafer bondingand coupled to separate holey PCWs on chip, the change in opticalintensity detected via on-chip detectors. We recently demon-strated such integration of lasers, detectors and waveguides onchip in an InP platform for a complete lab-on-chip absorptionspectrometer.

4. Conclusions

In summary, we demonstrated the propagation characteristicsof holey and slotted PCWs in SoS in the mid-IR. By engineeringthe lattice constant, we experimentally demonstrated the trans-mission characteristics with a single wavelength source. In theguided mode region below the light line, propagation losses below20 dB/cm were observed for group indices ng greater than 20 forthe holey PCW. Huge losses above the light line and zero trans-mission in the stop band were experimentally observed. Similarlosses above the light line and in the stop gap were observed in slot-ted PCWs. Below the light line, propagation losses observed weresimilar to losses observed in similar devices in SOI in the near-infrared. Finally, the chemical warfare simulant triethylphosphatewas identified by absorption signatures in an 800 �m long holeyPCW.

Acknowledgements

The authors thank NSF for partially supporting this work (SBIRGrant #IIP-1127251). S.C., P.W. and R.T.C. acknowledge the ArmySBIR Contract #W911SR-12-C-0046 for partially supporting thiswork.

References

[1] W.-C. Lai, S. Chakravarty, Y. Zou, R.T. Chen, Multiplexed detection of xyleneand trichloroethylene in water by photonic crystal absorption spectroscopy,Opt. Lett. 38 (2013) 3799–3802.

tuato

[

[

[

[

[

[

[

[

[

Y. Zou et al. / Sensors and Ac

[2] W.-C. Lai, S. Chakravarty, X. Wang, C. Lin, R.T. Chen, Photonic crystal slotwaveguide absorption spectrometer for on-chip near-infrared spectroscopyof xylene in water, Appl. Phys. Lett. 98 (2011) 023304.

[3] W.-C. Lai, S. Chakravarty, X. Wang, C. Lin, R.T. Chen, On-chip methane sensingby near-IR absorption signatures in a photonic crystal slot waveguide, Opt.Lett. 36 (2011) 984–986.

[4] R. Soref, Mid-infrared photonics in silicon and germanium, Nat. Photon. 4(2010) 495–497.

[5] J. Chiles, S. Khan, J. Ma, S. Fathpour, High-contrast, all-silicon waveguidingplatform for ultra-broadband mid-infrared photonics, Appl. Phys. Lett. 103(2013) 151106.

[6] S. Khan, J. Chiles, J. Ma, S. Fathpour, Silicon-on-nitride waveguides formid-and near-infrared integrated photonics, Appl. Phys. Lett. 102 (2013)121104–121113.

[7] Y. Zou, H. Subbaraman, S. Chakravarty, X. Xu, A. Hosseini, W.-C. Lai, et al.,Grating-coupled silicon-on-sapphire integrated slot waveguides operating atmid-infrared wavelengths, Opt. Lett. 39 (2014) 3070–3073.

[8] R. Shankar, R. Leijssen, I. Bulu, M. Loncar, Mid-infrared photonic crystalcavities in silicon, Opt. Express 19 (2011) 5579–5586.

[9] C. Reimer, M. Nedeljkovic, D.J. Stothard, M.O. Esnault, C. Reardon, L. O’Faolain,et al., Mid-infrared photonic crystal waveguides in silicon, Opt. Express 20(2012) 29361–29368.

10] W.-C. Lai, Y. Zou, S. Chakravarty, L. Zhu, R.T. Chen, Comparative sensitivityanalysis of integrated optical waveguides for near-infrared volatile organiccompounds with 1ppb detection, SPIE OPTO, in: International Society forOptics and Photonics, 2014, pp. 89900Z-Z-6.

11] Y. Zou, S. Chakravarty, W.-C. Lai, R.T. Chen, Silicon Chip Based Near-Infraredand Mid-Infrared Optical Spectroscopy for Volatile Organic CompoundSensing, CLEO: Science and Innovations, in: Optical Society of America, 2014,p. STh3 M. 6.

12] Y. Zou, S. Chakravarty, P. Wray, R.T. Chen, Experimental demonstration ofpropagation characteristics of mid-infrared photonic crystal waveguides insilicon-on-sapphire, Opt. Express 23 (2015) 6965–6975.

13] Q. Xu, V.R. Almeida, R.R. Panepucci, M. Lipson, Experimental demonstration ofguiding and confining light in nanometer-size low-refractive-index material,Opt. Lett. 29 (2004) 1626–1628.

14] Y. Zou, S. Chakravarty, D.N. Kwong, W.-C. Lai, X. Xu, X. Lin, et al.,Cavity-waveguide coupling engineered high sensitivity silicon photoniccrystal microcavity biosensors with high yield, selected topics in quantumelectronics, IEEE Journal of 20 (2014) 171–180.

15] Y. Zou, S. Chakravarty, L. Zhu, R.T. Chen, The role of group index engineeringin series-connected photonic crystal microcavities for high density sensormicroarrays, Appl. Phys. Lett. 104 (2014) 141103.

16] R. Shankar, I. Bulu, M. Loncar, Integrated high-quality factorsilicon-on-sapphire ring resonators for the mid-infrared, Appl. Phys. Lett. 102(2013), 051108–3.

17] A. Di Falco, M. Massari, M. Scullion, S. Schulz, F. Romanato, T. Krauss,Propagation losses of slotted photonic crystal waveguides, IEEE Photon. J. 4(2012) 1536–1541.

18] N.A. Mortensen, S. Xiao, Slow-light enhancement of Beer-Lambert-Bouguerabsorption, Appl. Phys.ics Lett. 90 (2007) 141108.

rs B 221 (2015) 1094–1103 1103

Biographies

Yi Zou received his Ph.D. degree in Electrical and Computer Engineering in 2014from the University of Texas at Austin, Austin. Currently he is a postdoc fellow inthe Department of Electrical and Computer Engineering in the University of Texasat Austin, Austin. His research interests include silicon photonics, photonic sensor,mid Infrared photonics and fiber optics.

Swapnajit Chakravarty obtained his B.E. degree in Electrical Engineering in 2001from Jadavpur University, Kolkata, India, M.S. in Electrical Engineering (ElectronicMaterials and Devices) from the University of Cincinnati in 2003 and Ph.D. inElectrical Engineering from the University of Michigan, Ann Arbor in 2007. Dr.Chakravarty has been an active researcher in photonic crystal microcavity devicesfor the past 10 years. His research area includes electrically and optically injectedphotonic crystal lasers and light emitting arrays, nanophotonic interconnects andphotonic crystal sensors. His current research focuses on utilizing III–V and siliconphotonic crystal waveguide devices and photonic crystal microcavity devices forchip-integrated optical sensing and spectroscopy with applications in water and airpollution monitoring, early cancer detection and the detection of various pathogens.Dr. Chakravarty has 7 issued patents and over 60 publications in refereed journalsand conferences. In 2013, Dr. Chakravarty was elected as Senior Member of IEEE andSenior Member of SPIE.

Parker Wray is an undergraduate student in Electrical Engineering Department inUniversity of Texas at Austin. His technical cores are “fields and waves” (the study ofEM phenomena) and signal processing. He is a member of Tau Beta Phi (the generalengineering honors society at UT). His research interests include metamaterials,photonics, non-linear optics, and their applications in communications and sensors.

Ray T. Chen received the B.S. degree in physics from National TsingHua University,Taiwan, in 1980, the M.S. degree in physics from the University of California, SanDiego, in 1983, and the Ph.D. degree in electrical engineering from the Universityof California, Irvine, in 1988. He worked as a Research Scientist, a Manager, andthe Director of the Department of Electrooptic Engineering in Physical Optics Cor-poration, Torrance, CA, from 1988 to 1992. He joined the University of Texas (UT),Austin, as a Faculty Member to start optical interconnect research program in theDepartment of Electrical and Computer Engineering in 1992. He is the Keys and JoanCurry/Cullen Trust Endowed Chair at The University of Texas Austin, and the Directorof Nanophotonics and Optical Interconnects Research Laboratory, MicroelectronicsResearch Center. He is also the Director of a newly formed AFOSR MultidisciplinaryResearch Initiative (MURI) Center for Silicon Nanomembrane involving faculty fromStanford, UIUC, Rutgers, and UT. He also serves as the founder and the Chairmanof the Board of Omega Optics, Inc., since its initiation in 2001. There are 46 stu-dents who have received the EE Ph.D. degree from his research group at UT Austin.His group has reported its research findings in more than 560 published papersincluding more than 70 invited papers. He holds 20 issued patents. His research

topics are focused on nanophotonic passive and active devices for optical phasedarray and interconnect applications, thin film guided-wave optical interconnec-tion and packaging for 2-D and 3-D laser beam routing and steering, and true timedelay wide band phased array antenna. He is a Fellow of the IEEE, the OSA andthe SPIE.