Embed Size (px)
Citation preview
UTILIZATION OF MULTIPLE HARMONICS OF
WAVELENGTH MODULATION ABSORPTION
SPECTROSCOPY FOR PRACTICAL GAS SENSING
A DISSERTATION
SUBMITTED TO THE DEPARTMENT OF MECHANICAL ENGINEERING
AND THE COMMITTEE ON GRADUATE STUDIES
OF STANFORD UNIVERSITY
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
Kai Sun
DEC 2013
This dissertation is online at: http://purl.stanford.edu/rb361sv6972
© 2013 by Kai Sun. All Rights Reserved.
Re-distributed by Stanford University under license with the author.
ii
I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.
Ronald Hanson, Primary Adviser
I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.
Mark Cappelli
I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.
Jay Jeffries
Approved for the Stanford University Committee on Graduate Studies.
Patricia J. Gumport, Vice Provost for Graduate Education
This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file inUniversity Archives.
iii
iv
v
Abstract
To meet more rigorous criteria for environmental-unfriendly emissions and to increase
energy efficiency, in-situ real-time sensors are needed to optimize the performance of
next-generation energy systems. The emergence of high-quality (narrow linewidth, fast
tuning capability) tunable diode lasers (TDLs) has enabled the use of wavelength
modulation spectroscopy (WMS) for harsh industrial applications. Compared to
conventional direct absorption measurements, WMS has the advantage of 10-100 times
better detection sensitivity, avoids the need to obtain a zero-absorption baseline, and
provides much better isolation from the beam steering, non-absorption transmission loss
(e.g., light scattering) or mechanical vibrations.
Many models have been developed to interpret the measured WMS signal into absolute
absorption. However, most of these models are limited to specific applications by a wide
variety of assumptions and approximation most of which deal with the simultaneous
intensity and wavelength modulation of injection-current-modulated diode lasers. In this
dissertation, two generalized approaches to analyze the WMS absorption signal were
developed that account for non-ideal simultaneous intensity modulation of laser output
when injection current variation is used for wavelength modulation. The first approach is
ideal for wavelength-fixed WMS (the laser mean wavelength is fixed) analysis and the
vi
second approach is ideal for wavelength-scanned (the laser mean wavelength is scanned)
WMS analysis, and both of them can be used for arbitrary modulation depth, or laser
architectures even when severe non-linear intensity modulation occurs simultaneously
with wavelength modulation. These new interpretations of WMS absorption signals
provide the potential for extended and improved use of WMS for practical gas sensing in
a much wider array of applications.
The first approach built on earlier work in our laboratory. The analysis of calibration-
free, 1f-normalized, WMS-2f absorption signals was extended to higher harmonics (for
example 3f, 4f…) using traditional Fourier analysis. The new approach and procedure
developed also accounts for non-ideal wavelength-tuning of the injection-current tuned
laser as well as etalon interference from the optical components in the laser line-of-sight
(LOS). This approach was validated using measurements of the CO transition of R (11)
in the 1st overtone band near 2.3µm in a laboratory cell at room temperature for a range
of CO mole fractions (0.21-2.8%) and pressures (5-20atm). For high-pressure gas
sensing, wavelength modulation spectroscopy with higher-order harmonic detection
(WMS-nf, n>2) was found to have less influence from the WMS background signals
when the selected modulation depth was near the optimal modulation depth for the
WMS-2f signal.
This WMS approach was then used for measurements in a pilot-scale entrained-flow coal
gasifier at the University of Utah. Even though the particulate scattering reduced the laser
transmission as much as 99.997%, and pressure broadening at the 18atm (~250psig)
operating pressure blended the absorption transitions, successful in-situ rapid-time-
vii
resolved 1f-normalized WMS-2f absorption measurements for gas temperature and H2O
mole fraction were made.
Based on lessons learned during the gasifier measurements at Utah and a desire to
eventually develop real-time sensors for long-term monitoring, a second approach for
WMS analysis was developed that differs from previous WMS analysis strategies in two
significant ways: (1) the measured laser intensity without absorption is used to simulate
the transmitted laser intensity with absorption and (2) digital lock-in and low-pass filter
software is used to expand both simulated and measured transmitted laser intensities into
harmonics of the modulation frequency, WMS-nf (n=1,2,3,…), avoiding the need for an
analytic model of intensity modulation or Fourier expansion of the simulated WMS
harmonics. The new method was demonstrated and validated with WMS of H2O dilute
in air (1atm, 296K, near 1392nm). WMS-nf harmonics for n=1 to 6 are extracted and the
simulations and measurements are found in good agreement for the entire WMS
lineshape.
This new analysis scheme was applied to monitor the synthesis gas output from an
engineering-scale transport reactor coal gasifier at the National Carbon Capture Center.
There the pressures ranged up to 15 atm (~220psig) and temperatures up to 650K.
Continuous monitoring of moisture level in the gasifier output with 2s time resolution
was performed by the TDL sensor for more than 500 hours, including the periods of
burner ignition, combustion heating with a propane flame, coal combustion, coal
gasification, and reactor shut-down via coal-feed termination.
viii
In addition, a novel and rapid approach to determine the collisional linewidth via the
WMS signals at different harmonics at the modulation frequency is presented. The peak
values of the WMS-nf absorption spectrum near the transition line center are used to infer
the absorption lineshape, which is exploited here to extract collision-broadening
halfwidth C from the ratio of WMS-4f/WMS-2f (or other even harmonics) signals
when the mean laser wavelength is tuned to line center. Measurement of the absorption
linewidth enables quantitative WMS measurements without the need for a collision-
broadening database. Alternatively, when collision-broadened spectral data are available,
a WMS-based pressure sensor can be realized, and a demonstration using the 4fpeak/2fpeak
ratio gives less than 0.7% difference for the pressure for cell measurements from 100 torr
to 753 torr.
These new WMS analysis schemes have been validated in near commercial environments
and illustrate the potential of their use to develop practical TDL sensors for a wide
variety of industrial applications.
ix
Acknowledgement
In my life time, I have been helped by numerous people. Some of them, I cannot even
remember their names or have never had a chance to know their names. However, it is
their help that gives me a positive attitude towards the world and humanity. And that’s
the main reason why I am motivated to help others, not matter if I know them or not.
Among all these people the help from some are particularly important to me. I sincerely
thank my advisor Prof. Ronald Hanson, who offers me the opportunity to study in such a
wonderful research group and university. In addition to learning the cutting-edge laser
absorption techniques from him, I gained the experience of managing a research group
and practicing high standards in my research. I also need to thank Dr. Jay Jeffries, for his
immense help for my research in all respects, some even beyond his responsibility. My
research work at Stanford would have been much harder without these help. I also would
like to thank Prof. Mark Cappelli for serving as member of my reading committee.
Wonderful suggestions have been given by him that improve the quality of this
dissertation. Last but not least, Prof. Chris Edwards and Prof. Adam Brandt’s attendance
in my PhD defense are greatly appreciated.
Also from the Stanford Hanson group, I would like to thank Dr. David Davidson for his
help in preparing and supporting the lab supply. And I should surely not forget my
x
labmates, almost all of whom I have obtained generous help from. But in particular I’d
like to thank Ritobrata Sur, Greg Rieker, Wei Ren, Shengkai Wang, Matt Campbell, Sijie
Li, Chris Goldenstein, Brian Lam, Chris Strand, Zekai Hong, Jason Porter, Mitchell
Spearrin, Ian Schultz, Yangye Zhu and Ivo Stranic.
Indeed all will not be possible without the support from my family. I’d like to thank my
parents for their endless love and encouragements that make the person who I am today.
Finally but most importantly, I’d like to thank my wife Xing Chao, who graduated from
the Hanson group as well. For years, we share our insights on all different matters and
solve the difficulties in life as well as research. From these unforgettable experiences, we
have set the fundamentals of our love. I also want to thank our baby girl, Eileana Yueran
Sun. She was born in the Stanford Children’s Hospital 8 months ago. Although she has
yet to speak the language of ours, her beautiful eyes tell it all and always render me
intoxicated in the joy and encouragements by even just a simple glimpse.
xi
Table of Contents
Chapter 1 Introduction ................................................................................. 1
1.1 Background and Motivation .......................................................................................1
1.2 Overview of dissertation ............................................................................................4
Chapter 2 Laser absorption fundamentals ................................................. 9
2.1 Beer-Lambert law .......................................................................................................9
2.2 Scanned-wavelength direct-absorption (scanned- DA) .........................................12
2.3 Wavelength modulation spectroscopy (WMS) ........................................................15
Chapter 3 Generalized 1f-normalized WMS-nf model using Fourier
analysis considering non-ideal diode laser performance ......................... 19
3.0 Motivation ................................................................................................................19
3.1 WMS-nf model ........................................................................................................21
3.2 Laser wavelength modulation characterization and optical system intensity
modulation characterization ...........................................................................................26
3.2.1 Laser wavelength modulation characterization ............................................................. 26
3.2.2 Optical system intensity modulation ............................................................................. 27
3.3 WMS mathematical and physical meanings and 1f-normalization strategy ............29
3.4 WMS-nf model validation by high pressure CO WMS spectra measurements .......33
3.4.1 WMS-nf model validation ............................................................................................. 33
3.4.2 High pressure CO sensor design based on WMS-nf detection..................................... 35
3.5 Advantages of using higher harmonics ....................................................................38
3.5.1 Advantage of WMS-nf (n>2) in reducing the background signal drift ........................ 38
3.5.2 Advantage of WMS-nf (n>2) in reducing the interference from neighbors ................. 44
Chapter 4 H2O absorption sensor using fixed-wavelength WMS in a
pilot-scale high pressure entrained-flow coal gasifier .............................. 47
4.0 Motivation ................................................................................................................47
4.1 Gasifier facility .........................................................................................................49
4.1.1 Entrained-Flow Gasifier and Sampling Locations ....................................................... 50
xii
4.1.2 System Operation ......................................................................................................... 52
4.2 H2O absorption sensor design ..................................................................................54
4.3 Laser absorption sensor setup and alignment ...........................................................57
4.4 Results and discussion ..............................................................................................59
4.4.1 Reactor measurements .................................................................................................. 59
4.4.2 Gasifier product syngas-stream measurements ............................................................. 64
4.4.3 Comparison between WMS-2f and 4f in high pressure and noisy environment........... 65
Chapter 5 Novel Strategy for calibration-free wavelength-scanned
WMS analysis .............................................................................................. 69
5.0 Motivation ................................................................................................................69
5.1 Overview of a WMS absorption experiment/simulation ........................................71
5.1.1 Transmitted intensity measurement: ............................................................................. 72
5.1.2 Intensity and wavelength modulation characterization ................................................. 73
5.1.3 Simulated transmitted laser intensity: ........................................................................... 74
5.1.4 Lock-in analysis ............................................................................................................ 75
5.1.5 Normalization to account for non-absorption losses ..................................................... 76
5.2 Example analysis of WMS absorption detection of H2O .........................................77
5.2.1 Transmitted intensity measurement for WMS detection of H2O .................................. 77
5.2.2 Laser characterization for WMS detection of H2O ....................................................... 78
5.2.3 Simulated transmitted laser intensity for WMS detection of H2O ............................... 79
5.2.4 Lock-in analysis for WMS detection of H2O ............................................................... 83
5.2.5 Issues for normalization by WMS-1fm .......................................................................... 85
6.2.6 Normalization to account for non-absorption transmission losses: .............................. 88
5.3 Comparison with Fourier analysis of WMS ............................................................89
Chapter 6 H2O absorption sensor using Calibration-free wavelength-
scanned WMS fitting strategy in an engineering-scale high pressure
fluidized coal gasifier .................................................................................. 91
6.0 Motivation ................................................................................................................91
6.1 Laboratory Validation Experiment...........................................................................92
6.2 Gasifier facility and measurement setup ..................................................................96
6.3 Measurement results .................................................................................................98
Chapter 7 Absorption lineshape from ratios of different WMS
harmonic signals ........................................................................................ 105
xiii
7.0 Motivation ..............................................................................................................105
7.1 WMS fundamentals and derivation .......................................................................108
7.2 An example case and laboratory demonstration ....................................................109
7.3 Pressure sensor developed using the 4fpeak/2fpeak ratio ...........................................114
Chapter 8 Summaries and future plans .................................................. 117
8.1 Summaries ..............................................................................................................117
8.1.1 A generalized 1f-normalized WMS-nf method using Fourier Analysis ...................... 117
8.1.2 Demonstration of the 1f-normalized WMS-2f strategy in a pilot-scale entrained-flow
high pressure coal gasifier ................................................................................................... 118
8.1.3 A novel strategy for WMS absorption analysis .......................................................... 119
8.1.4 Demonstration of the fitting strategy for wavelength-scanned WMS in an engineering-
scale fluidized-bed high pressure coal gasifier .................................................................... 120
8.1.5 Absorption lineshape from ratios of different WMS harmonic signals ...................... 121
8.2 Future plans ............................................................................................................122
8.2.1 Other species measurements in the fluidized-bed coal gasifier in NCCC .................. 122
8.2.2 Integrate the cavity enhanced techniques with WMS ................................................. 123
8.2.3 Species time-history measurements in shock tubes using CET/WMS ....................... 123
Appendix ..................................................................................................... 125
A.1 Laboratory measured spectroscopy parameters ....................................................125
A.2 Derivation for Eqn (7.4) ........................................................................................129
Reference .................................................................................................... 133
xiv
xv
List of Tables
Table 4.1 Gasifier specifications. ......................................................................................51
Table 4.2 Selected transitions for temperature and gas concentration sensors at different
locations of the coal-gasifier (* more than one transitions with similar E" in the selected
wavelength region, which form one apparent peak feature at high pressure, **details of
CO2 and CO work were present where else [70]) ..............................................................55
Table 5.1 Measured spectroscopic parameters for probed H2O transition near 7185.60
cm-1
and its neighbor near 7185.39cm-1
at 296K.. .............................................................81
Table 6.1 Laboratory measured spectroscopic parameters (linestrength, collisonal
broadening coefficients and their temperature dependence exponents) at 296K for the
target transition ..................................................................................................................93
Table 6.2 Typical conditions at the measurement location (the gas mixture is balanced by
N2) ......................................................................................................................................96
Table A.1 Comparison between the measured linestrength and those recorded in
HITEMP 2010 database for studied transitions (Tref = 296K) .........................................125
Table A.2 Measured H2O- H2O collision-broadening coefficients and those recorded in
HITEMP 2010 database for studied transitions (Tref = 296K) .........................................126
Table A.3 Measured H2O-CO2 collision-broadening coefficients for studied transitions
(Tref = 296K) ....................................................................................................................126
xvi
Table A.4 Measured H2O-CO collision-broadening coefficients for studied transitions
(Tref = 296K) ....................................................................................................................127
Table A.5 Measured H2O-H2 collision-broadening coefficients for studied transitions
(Tref = 296K) ....................................................................................................................127
Table A.6 Measured H2O-CO2 pressure shifting coefficients for studied transitions (Tref =
296K) ...............................................................................................................................128
Table A.7 Measured H2O-CO pressure shifting coefficients for studied transitions (Tref =
296K) ...............................................................................................................................128
Table A.8 Measured H2O-H2 pressure shifting coefficients for studied transitions (Tref =
296K) ...............................................................................................................................128
xvii
List of Illustrations
Figure 2.1 Cartoon schematic of the attenuation of laser transmission by absorption by
the gas along laser line-of-sight. ........................................................................................10
Figure 2.2 Left: measured transmitted laser intensity and the fitted zero-absorption
baseline. Right: the processed absorption(the grey area under the measured absorption
curve is the integrated absorbance area). ...........................................................................13
Figure 2.3 Left: Absorption spectrum of two transitions with distinct temperature
dependence. Right: The ratio of the linestrength of these two transitions with temperature.14
Figure 2.4 Schematic of a typical fixed-wavelength WMS measurement with two TDLs
for temperature measurements and the frequency spectrum of the received time-domain
signals after fast Fourier transform (FFT) .........................................................................16
Figure 3.1 Example of laser wavelength response to the injection-current modulation.
The blue circles are the relative frequency measured by a solid etalon of 0.02cm-1
FSR.
The measured modulation depth is 0.101cm-1
and initial phase of the wavelength
modulation is -2.1363 radian. ............................................................................................26
Figure 3.2 Transmission of a thin cavity built by two parallel surfaces (cavity length:
2mm, surface reflectivity: 2%) ..........................................................................................27
Figure 3.3 Upper: The simulated I0 and It in the WMS measurement and the sine fit to It
at 1f frequency. Bottom: the residual between I0 and It .....................................................30
Figure 3.4 The residual in figure 3-3 and its sine fit at 2f frequency ................................31
xviii
Figure 3.5 Experimental setup for high pressure CO gas sensing. ...................................34
Figure 3.6 Measured (square, red) and simulated (solid line, black) 1f-normalized WMS-
2f, 3f and 4f lineshape signals at different pressures. Gas mixture: 1.59% CO in N2; T =
296 K; optical pathlength L = 100.5 cm. ...........................................................................36
Figure 3.7 Measured CO mole fractions by 1f-normalized WMS-nf technique and
comparison with calibrated CO mole fractions (dashed line), bath gas: N2; T = 296K;
optical pathlength L = 100.5cm. ........................................................................................38
Figure 3.8 Measured 2f background magnitude versus time, normalized by the laser
intensity. (Measured with a 2.3µm Nanoplus TDL, traveling through an evacuated,
17.3cm cell with wedged windows. Modulation depth = 1.52cm-1
, modulation frequency
= 1kHz. No external intervention presented during measurement, conducted at Stanford
in 2011) ..............................................................................................................................39
Figure 3.9 Measured 1f-normalized WMS-2f and 4f signals as well as their background
signals. (0.21% CO in N2, P = 10atm, T = 296K, L = 100.5cm; a = 1.52cm-1
, f = 1kHz)40
Figure 3.10 Measured ratio of 1f-normalized WMS-nf signal to its 1f-normalized
background signal. ( a =1.52cm-1
, f =1kHz, T =296K, L =100.5cm) ..............................41
Figure 3.11 Measured ratios of 1f-normalized WMS-nf signal to the drift magnitude of
its 1f-normalized background signal. ( a =1.52cm-1
, f =1kHz, T =296K, L =100.5cm) 42
Figure 3.12 simulated WMS-nf background signals with the cavity length from 0.498mm
to 0.502mm for a modulation depth of 1.5cm-1
and a reflectivity of 4.3% .......................43
Figure 3.13 Background-drift induced WMS-nf detection limits at different pressures.
(CO in N2, T = 296K, L = 100.5cm; a = 1.52cm-1
, f = 1kHz) .........................................44
xix
Figure 3.14 Simulated kH for targeted transition near 4300.7cm-1 at 20atm ( a =1.52cm-
1, f =1 kHz, 0.21% CO in N2, T =296K, L =100.5cm) ...................................................45
Figure 4.1 University of Utah Gasification Research Facility. .........................................49
Figure 4.2 Schematic of entrained-flow gasification research facility. ............................50
Figure 4.3 Schematic to the pilot-scale, entrained-flow, coal-gasifier at University of
Utah ....................................................................................................................................52
Figure 4.4 Schematic to measurement locations at University of Utah: (1) the reactor-
core, (2) pre-quench, (3) post-quench, (4) after clean-up ..................................................53
Figure 4.5 Absorption linestrengths of H2O, CO2 and CO at 296 K from HITEMP 2010
database. .............................................................................................................................56
Figure 4.6 Simulations of the 1f-normalized WMS-2f peak magnitude ratio of the
7185.6cm-1
transition to the 6806.0cm-1
transition versus temperature at different
pressures and absorber concentrations (H2O). ...................................................................57
Figure 4.7 Schematic of the experiment setup in the control room and optical alignment
from the control room to the gasifier rig. ...........................................................................58
Figure 4.8 Photos of the connection of the long PM fiber and BNC cables from the
control room to the gasifier rig: (a) Fiber and BNC cables output from the control room,
(b) Aligned from the building of the control room to the gasifier, (c) connect the PM fiber
to a lens collimator mounted on the flange at location 2, (d) connect the BNC cable to the
detector mounted on the opposite flange ...........................................................................59
Figure 4.9 Measured optical emission from the combusting liquid or coal and light
transmission in the reactor core as the gasifier was pressurized (low-pressure fuel
xx
isopropyl alcohol switching to a pulverized coal slurry at 4.4atm). Measurements used
the optical filter and InGaAs detector described in the text...............................................60
Figure 4.10 Measured and simulated 1f-normalized WMS-2f absorption spectra of
7185cm-1
transition (left) and 6806.0cm-1
transition (right) in the reactor core (27% H2O
in syngas flow, P=11.2atm, T=1510K, L=12.5cm, transmission loss: 99.997%). ............62
Figure 4.11 Temperature in reactor core determined by laser absorption (~1s time
resolution) and by thermocouples in the walls. .................................................................63
Figure 4.12 Temperature determined by laser absorption in the pre-quench location with
a ~1s time resolution (data shown for four reactor pressures). Note unstable T observed
at 15atm caused by fluctuations in oxygen supply. ...........................................................64
Figure 4.13 Measured gas temperature and H2O mole fraction by the TDL sensors with a
~2.5s time resolution and thermocouple temperatures in the gasifier product-syngas
stream .................................................................................................................................65
Figure 4.14 Measured and simulated 1f-normalized WMS-2f and 4f spectra of 7185.6
cm-1
transition in the reactor core. (a) Measured WMS signal with the background signal
measured with N2 in the gasifier subtracted; (b) measured WMS signal with the
background signal measured in the laboratory prior to the gasifier experiments subtracted.
(25% H2O in syngas flow, P=18atm, T=1620K, L=12.5 cm, transmission loss: 99.99%,
a=0.96 cm-1
, f=10kHz) .......................................................................................................66
Figure 5.1 Measurement step: Determine measured transmitted intensity versus time
with absorber M
It(t) and without absorber M
I0(t). ..............................................................73
xxi
Figure 5.2 Wavelength characterization versus time ( )t and intensity versus time M
I0(t)
of a wavelength-scanned, wavelength-modulated laser including wavelength-dependent
transmission along the measurement path without absorption. .........................................74
Figure 5.3 Simulation of transmission intensity versus time of a wavelength-scanned
wavelength-modulated laser through a simulated absorption spectrum. ...........................75
Figure 5.4 Use of a digital lock-in and low-pass filter to expand the time series of
measured or simulated laser intensity into the WMS-nfm harmonic signals. ....................76
Figure 5.5 Schematic of the experimental setup for measuring the transmitted laser
intensity versus time for WMS detection of H2O in a gas cell. .........................................78
Figure 5.6 Measured laser intensity versus time in the absence of the absorber (scan rate
= 25Hz, scan amplitude = 2V, modulation frequency = 10 kHz, modulation amplitude =
0.1V). .................................................................................................................................79
Figure 5.7 Measured frequency response to the laser injection-current tuning and its best
fit (same modulation configuration as Figure 5.6).............................................................80
Figure 5.8 Simulated absorbance versus frequency for H2O transition near 7185.6cm-1
and its neighbor at 7185.39cm-1
(0.75% H2O in air ,P = 1 atm, T = 296 K, L = 100.5 cm).81
Figure 5.9 Simulated absorbance versus time ( ( ))t for the H2O transition near
7185.6cm-1
(for the absorbance versus wavelength shown in Figure 5.8). Note constant
peak values between 0.021 and 0.0225s are real (not detector saturation) as the
modulation is fast compared to the scan rate. ....................................................................82
Figure 5.10 Simulated transmitted laser intensity versus time ( )s
tI t for a single scan of
the modulated laser over the absorption feature (For the laser intensity in Figure 5.6 and
the absorbance versus time in Figure 5.9). ........................................................................83
xxii
Figure 5.11 Measured and simulated WMS-nfm spectra for H2O transition near 7185.6
cm-1
. (0.75% H2O in air, T = 296 K, P = 1atm, L = 100.5 cm, a = 0.081 cm-1
, fm = 10
kHz, optical depth = 0.101). Note the amplitude difference was produced by attenuating
the measurement laser intensity to mimic the influence of non-absorption losses. ...........85
Figure 5.12 Simulated 1fm spectra for H2O transition near 7185.6 cm-1
at optical depths
0.01, 0.1, and 1.0 at 1 atm with a modulation index of 1.8 (the laser characterization is the
same as Figure 5.6 and Figure 5.7; note the modulation index is 1.8, and the line center
includes pressure shift in 1 atm air). ..................................................................................86
Figure 5.13 Measured and simulated 1fm-normalized WMS-nfm spectra for H2O
transition near 7185.6 cm-1
. (same condition as Figure 5.11, optical depth = 0.101). .......89
Figure 5.14 Comparison of the 1fm-normalized WMS-nfm spectra using different
absorption analysis approaches for H2O transition near 7185.6 cm-1
. (same condition as
Figure 5.13) ........................................................................................................................90
Figure 6.1 Simulated absorption spectrum for H2O molecule at typical gasifier conditions93
Figure 6.2 Laboratory measurement setup for validation of the wavelength-scanned
WMS strategy for high pressure gas sensing .....................................................................94
Figure 6.3 Measured WMS-2f/1f spectra using different modulation depths and the best
fit results (best fit parameters: for a = 0.4cm-1
, xH2O = 0.0953% ,c =0.823 cm
-1, for a =
0.6cm-1
, xH2O = 0.0947% ,c =0.820 cm
-1, for a = 0.8cm
-1, xH2O = 0.0961% ,
c =0.802
cm-1
) ..................................................................................................................................95
Figure 6.4 Best-fit results for the C (left panel) and integrated absorbance and mole
fraction (right panel) at different pressures. (T = 296K, L = 100.5cm) .............................96
xxiii
Figure 6.5 Location of the TDL sensor in the syngas process piping 30 meters
downstream of the exit of the PCD. Note the lasers and supporting electronics were
remotely located in the instrumentation shelter. ................................................................97
Figure 6.6 Schematic (left) and photo (right upper) of the sensor installation showing the
mounting rail hanging on the syngas pipe with redundant actuated shut-off valves,
redundant window pairs, temperature and pressure alarm for window failure, and the
TDL transmitter and receiver housings. The TDL electronics in the shelter are shown in
the right bottom panel. .......................................................................................................97
Figure 6.7 In situ measurements of exhaust gas moisture during reactor start-up
including ignition of the propane burner, switch to coal combustion with pulsed coal
feed, transition to gasification, and reactor shutdown when the coal input was terminated.
The pressure trace was provided by NCCC. The region surrounded by the red dashed
rectangle is shown in Figure 6.9. .......................................................................................99
Figure 6.8 Measured (dashed line) WMS-2f/1f absorption spectra and the best-fit results
(solid lines) at different gasifier operation conditions (black: heating using the
propane/air flame, blue: coal combustion, red: coal gasification). fs = 25Hz, f = 10kHz, a
= 0.78cm-1
. Pressure and temperature data were provided by NCCC. ............................100
Figure 6.9 Left panel: TDL monitored moisture mole fraction from hour 42 to hour 43
and the reactor temperature measured by the thermocouple; Right panel: measured WMS
2f/1f absorption spectra at point 1(the lowest moisture mole fraction in a single pulse)
and at point 2 (the highest moisture mole fraction in a single pulse) ..............................101
Figure 6.10 TDL measured transition collisional width and its comparison to the
expected values. ...............................................................................................................102
xxiv
Figure 6.11 TDL recorded moisture content in the syngas without people attendance for
a duration of more than 400 hours. ..................................................................................103
Figure 6.12 Correlations of the TDL measured moisture content in the syngas product
flow to the reactor temperature measured by the thermocouple (left) and to the coal-
dispense vessel pressure (right) .......................................................................................104
Figure 7.1 Simulations of normalized WMS-nf spectra peak magnitudes versus C for
the H2O transition near 7185.6 cm-1
. The nf-harmonic signals were normalized by the DC
component of the incident laser power (0f) (T = 296 K, P = 1 atm, L = 100.5 cm, a =
0.081 cm-1
, f = 10 kHz) ....................................................................................................107
Figure 7.2 The simulated ratio WMS-4fpeak/WMS-2fpeak as a function of C for
different absorber mole fraction, pressure and pathlength conditions ( T = 296 K, a =
0.081 cm-1
, f = 10 kHz, =7185.59 cm-1
). ......................................................................110
Figure 7.3 Measured WMS-4fpeak/WMS-2fpeak with pressures (0.75% H2O in air, L =
100.5 cm, a = 0.081 cm-1
, f = 10 kHz) .............................................................................111
Figure 7.4 Measured C using WMS-4fpeak/WMS-2fpeak ratio with the comparison to
calculated C ad s function of pressure (0.75% H2O in air, L = 100.5 cm, a = 0.081 cm-
1, f = 10 kHz, =7185.59 cm
-1) .......................................................................................112
Figure 7.5 H2O mole fraction determined from WMS-2fpeak using C from the ratio of
WMS-4fpeak/WMS-2fpeak (0.75% H2O in air, L = 100.5 cm, a = 0.081 cm-1
, f = 10 kHz,
=7185.59 cm-1
) .................................................................................................................112
xxv
Figure 7.6 The WMS-4fpeak/WMS-2fpeak ratio as a function of C computed for peak
absorbance ranging from 0.2 to 1, corresponding to a pathlength increase from 50-250cm
(a = 0.081 cm-1
, f = 10 kHz, =7185.59 cm-1
) ................................................................113
Figure 7.7 measured gas total pressure using WMS-4fpeak/WMS-2fpeak ratio and its
comparison to the baratron result( 0.75% H2O in air, L = 100.5 cm,T = 296K, a = 0.081
cm-1
, f= 10 kHz, =7185.59 cm-1
) ..................................................................................115
xxvi
1
Chapter 1 Introduction
1.1 Background and Motivation
Tunable diode laser absorption spectroscopy (TDLAS) is an established method for in-
situ, non-intrusive, monitoring of gas composition, temperature, pressure and velocity [1-
15]. With the emergence of reliable, room temperature, narrow-linewidth, wavelength-
tunable diode lasers, such absorption sensors have transitioned in the past two decades
from laboratory sensors [16-20] into practical devices for industrial facilities [21-32].
Wavelength modulation spectroscopy (WMS) and direct absorption (DA) are the two
most common methods for TDLAS sensing.
In DA [33-38], the laser wavelength is typically scanned across an isolated transition and
the non-absorbing transmitted intensity (often called the baseline intensity) is determined
by extrapolating the laser intensity from the non-absorbing regions at the extremes of the
scan to the region with absorption to account for laser intensity variation with
wavelength. This baseline (incident) intensity and the transmitted intensity are combined
with the Beer-Lambert relation to determine the transition lineshape and the integrated
absorbance. For homogeneous gases the interpretation of DA data is quite
straightforward, as the integrated absorbance depends only on the line strength of the
transition, temperature, pressure, absorber mole fraction, and pathlength. DA is the
method of choice for systems with isolated transitions of sufficient strength (i.e., high
signal-to-noise (SNR) measurements of the absorption attenuation of the transmitted
2
intensity) and a transition linewidth small enough to allow the laser to be wavelength
scanned on/off the absorption transition.
By contrast, the WMS method is advantageous for applications with small absorbance,
high pressure or for absorbers with closely spaced transitions, which are blended in the
wings precluding measurement of a zero-absorption baseline. In WMS, the laser
wavelength is modulated at frequency f and signals are detected at the harmonics nf,
isolating the signal from low-frequency noise [39-42]. The majority of WMS applications
involve the detection of trace quantities of the target species with very small absorption
signal. Except for the first harmonic, the WMS signals are ideally detected against a zero
or near-zero background, while DA is detected as the difference between transmitted
intensity with and without absorption. Thus, the WMS SNR is improved compared to
DA by the ability to detect a small signal against a near-zero background at detection
frequencies well above low-frequency intensity noise. In addition, detection of the
transmitted intensity synchronously with modulation also provides WMS immunity from
optical emission from the measurement volume (although for highly luminous
applications care must be taken to insure the detector is not saturated).
The WMS signals at all of the harmonics are proportional to laser intensity, and the WMS
signal at 1f is dominated by the intensity modulation from injection-current modulated
TDLs for optically thin measurements. Thus normalizing the WMS-nf signals by the
WMS-1f signal can account for variations in laser intensity, including non-absorption
losses such as light scattering or beam steering [43-51]. This normalization enables
quantitative WMS absorption measurements without determining a zero-absorption
baseline [47, 48], making wavelength-scanned, 1f-normalized WMS-nf an attractive
3
strategy for absorption measurements in harsh (i.e. high-pressure, high-opacity, high-
emission, high-temperature) environments, especially where the laser cannot be tuned to
a zero-absorption wavelength [32, 47, 48, 52].
In this dissertation, two generalized approaches to analyze the WMS absorption signal
were developed that accounts for non-ideal simultaneous intensity modulation of TDL
output when injection current variation is used for wavelength modulation. The first
approach built on earlier work in our laboratory. The analysis of calibration-free, 1f-
normalized, WMS-2f absorption signals was extended to higher harmonics (for example
3f, 4f…) using traditional Fourier analysis. The second approach differs from previous
WMS analysis strategies in two significant ways: (1) the measured laser intensity
without absorption is used to simulate the transmitted laser intensity with absorption and
(2) digital lock-in and low-pass filter software is used to expand both simulated and
measured transmitted laser intensities into harmonics of the modulation frequency,
WMS-nf (n=1,2,3,…), avoiding the need for an analytic model of intensity modulation or
Fourier expansion of the simulated WMS harmonics. The first approach is ideal for
wavelength-fixed WMS and the second approach is ideal for wavelength-scanned WMS,
and both of them can be used for arbitrary modulation depth, or laser architectures even
when severe non-linear intensity modulation occurs simultaneously with wavelength
modulation. These new methods of determining the WMS background signals can be
applied to the entire line-of-sight optical path (previous methods accounted for
background signals resulting from intensity variation of the laser source only).
The second approach provides WMS background signals for scanned WMS over the
entire absorption lineshape (previous methods were valid only at line center) and for all
4
harmonics of the modulation frequency. Exploiting the simultaneous analysis of all
harmonics of the WMS signal, this research includes evaluation of the use of higher
harmonics (i.e. 3f, 4f) of the WMS signal. Sensors using these new analysis methods for
WMS were used to monitor H2O and temperature in the syngas products of coal
gasification. These sensors were evaluated for practical use in a pilot-scale entrained-
flow coal gasifier at University of Utah and in an engineering-scale fluidized-bed coal
gasifier at National Carbon Capture Center in Alabama. These applications illustrate the
use of the new WMS analysis scheme in extremely harsh environments where more than
99.99% of the laser intensity was scattered by the particulate [47,52].
Although wavelength modulation may also be applied to other laser-based diagnostics,
like laser-induced florescence (LIF), there are several complications involved in the
extension of LIF to WMS detection. Small LIF signals are shot-noise limited by the low
number of fluorescent photons collected, and thus, the use of other harmonics to
normalize the signal to account for laser power fluctuations can only be done when the
LIF signal is large. When the LIF signal is large, the detection of the signal in the time
domain provides quite satisfactory noise rejection and the complications of frequency
domain detection would be of limited value. The systems are also limited to modulation
frequencies that are slow compared to the fluorescent lifetime. Thus, it is not clear that
there is a significant advantage of using wavelength modulation for LIF and it will not be
further discussed in this dissertation.
1.2 Overview of dissertation
This dissertation is arranged into chapters to describe the technical work.
5
Chapter 2: An overview of fundamental spectroscopy concentrating on the direct
absorption technique and wavelength modulation spectroscopy is provided.
Chapter 3: A general method based on Fourier analysis for 1f-normalized
wavelength modulation absorption spectroscopy with nf detection (i.e., WMS-nf)
is presented that considers the performance of injection-current tunable diode
lasers (TDLs) including reflective interference induced by other optical
components in the line-of-sight (LOS). The optimization of sensitive detection of
optical absorption by species with structured spectra at elevated pressures is
explored using this analysis method. Advantages of using higher harmonics (i.e.
4f) are discussed and demonstrated.
Chapter 4: The analysis scheme in chapter 3 is demonstrated with a H2O
absorption sensor in a pilot-scale entrained-flow slagging high pressure coal
gasifier, located at Unviersity of Utah. The temperature-scanned WMS absorption
spectra of H2O transitions were measured at the gasifier reactor core, an extremly
harsh environment with more than 99.99% transmission loss, intense optical
emissions from hot particulate, high pressures up to 250psig (18atm) and high
tempratures up to 1800K. Quantitative monitoring of gas temperature and H2O
mole fraction in the gasifier product flow are presented. This work was the 1st
demonstration of laser absorption measurements in a pilot-scale high-pressure
coal gasifier.
Chapter 5: A novel strategy for quantitative analysis of scanned-wavelength
WMS using injection-current-tuned diode lasers is presented. The scheme is
applicable for arbitrary species, gas pressure, temperature, modulation depth and
6
harmonic order of the transmitted intensity. Instead of simulating the WMS
sbsorption signal via Fourier analysis, the method simulates the transmitted laser
intensity after absorption. The simulated laser transmission signal is then filtered
by the same digital lockin filter and lowpass finite-impulse-response (FIR) filter
to extract the nf components (WMS-nf) used to filter the experimental WMS
signals.
Chapter 6: The data analysis scheme of Chpater 5 was demonstrated with a real-
time, laser absorption sensor for continuous monitoring of the moisture content in
the product stream of an engineering-scale fluidized-bed high-pressure coal
gasifier. The scanned-wavelength wavelength modulation spectroscopy with 2nd
harmonic detection (WMS-2f) was used to determine the absorption magnitude. A
fitting strategy was developed to simultaneously determine the moisture mole
fraction and the collision-broadening halfwidth ( C ) of the transition from the
measured 2f/1f spectrum at pressures up to 15 atm. The scheme was analogous to
the fitting strategy for scanned-wavelength direct absorption measurement for an
isolated transition at low pressures. This strategy was validated in the lab
environment and then demonstrated in a campaign test in the US National Carbon
Capture Center. To demonstrate the sensor capability for long duration use, the
sensor was operated unattended from more than 500 hours and the moisture was
continuously monitored.
Chapter 7: A rapid approach to infer the transition collision-broadening
halfwidth from the ratio of WMS signals at different harmonics of the modulation
frequency was demonstrated. It is exploited here to extract collision-broadened
7
absorption lineshape from the WMS-4fpeak/WMS-2fpeak ratio (or other even
harmonics) signals when the mean laser wavelength is tuned to line center. Rapid
inference of the C enables quantitative calibration-free WMS measurements
without the need for a collision-broadening coefficient database. Alternatively,
when collision broadening spectral data are available, a WMS-based pressure
sensor can be realized and a demonstration using the ratio of WMS-4fpeak/WMS-
2fpeak for cell measurements from 100 torr to 753 torr was shown to yield the
pressure to better than 1%.
Chapter 8: The major contributions of the dissertation are summarized and
planned future work is discussed.
8
9
Chapter 2 Laser Absorption Fundamentals
Before developing new schemes to analyze WMS absorption data, the basic fundamentals
of laser absorption are discussed to define terms and units.
2.1 Beer-Lambert law
As a monochromatic laser beam travels through a gas media, laser photons can be
absorbed by gas molecules when the laser wavelength is tuned to resonance with
absorption transitions. This absorption excites the molecule from a low energy quantum
level to a higher energy level as illustrated in the cartoon of Figure 2.1. The transmitted
intensity of the laser light is related to the gas properties through the Beer-Lambert law
(for a uniform flow):
0
( ) exp ( )tIS T P x L
I
(2.1)
where is the fractional light transmission, the laser frequency (note that laser
wavelength and laser frequency ν are related λ1/ and the typical units for λ are nm
(or µm) while the units for are wavenumbers (cm-1
), 0I and tI are the incident and
transmitted laser intensities, S the linestrength, the lineshape function of the
absorption transition, T and P the gas temperature and pressure, x the mole fraction of
the gas species and L the optical pathlength.
The lineshape function at wavelength [cm-1
] is most often approximated by a Voigt
lineshape [53], which is a convolution of the Gaussian and Lorentzian lineshapes:
10
( ) ( ) ( )V D Cu u du
(2.2)
Figure 2.1 Cartoon schematic of the attenuation of laser transmission by absorption by
the gas along laser line-of-sight.
The Doppler lineshape function D has a Gaussian form:
2
02 ln 2( ) exp 4ln 2D
D D
(2.3)
where 0 [cm-1
] is the transition line center and D [cm-1
] is the Doppler full-width at
half maximum (FWHM) and is given by:
7
07.1623 10D
T
M (2.4)
where M [g/mol] is the molecular weight of the absorber.
11
The Lorentzian lineshape is expressed as:
2
2
0
1
2( )
2
CC
C
(2.5)
where C is the collision-broadening halfwidth of the transition (FWHM). In the limit
of binary collisions the width is proportional to pressure at constant temperature, and for
a multi-component gas mixture, C can be obtained by summing the contributions from
all components:
2 ( )C j j
j
P x T (2.6)
where jx is the mole faction of species j and j is the broadening coefficient due to
collisions with the jth
species. The relationship between j and T can be described as:
296 296( )
j
j
n
K
j TT
(2.7)
where jn is the temperature exponent of the collision-broadening coefficient.
Tunable diode lasers (TDLs) are a convenient source for absorption sensors, as these
devices are relatively inexpensive, rugged, electrically efficient, and in the near infrared
(NIR) well developed by the telecommunications industry. The TDL wavelength can be
tuned by adjusting the injection current and or the laser temperature. This facile
wavelength tuning has led to the development of two commonly employed TDL
absorption measurement techniques: (1) scanned-wavelength direct absorption (DA) and
12
(2) wavelength modulation spectroscopy with second harmonic detection (WMS-2f).
Both strategies have attractive properties for real-time gas sensing and are discussed in
the following two sections.
2.2 Scanned-wavelength direct-absorption (scanned- DA)
The Beer-Lambert law in Eqn (2.1) can be rewritten in terms of absorbance at the laser
frequency :
0
ln ( ) ( )tv v
v
Ik L S T P x L
I
(2.8)
where the absorption coefficient, vk ( ) ( )S T P x . Scanned wavelength direct
absorption can take advantage of the defined normalization of the lineshape ( ) 1v dv
, and the integrated absorbance can be expressed as:
( )i v iA dv S T P x L
(2.9)
The scanned- direct absorption measurement varies the laser wavelength over a range
that captures an entire transition feature. The zero-absorption baseline is fitted by the
zero-absorption portion of the scan to obtain 0I as illustrated in Figure 2.2 . The
integrated absorbance is then calculated by Eqn. (2.8) and (2.9), as shown in Figure 2.2.
The integrated absorbance from scanned-λ DA has a significant advantage: the integrated
absorption is free from detailed knowledge of the lineshape including collision
broadening (also called pressure broadening). However, the laser wavelength must be
13
scanned across the entire feature to extract 0I . Increasing pressure eventually overlaps or
blends absorption from neighboring features. Thus the pressure range of this DA
advantage is limited by the pressure broadening.
Figure 2.2 Left: measured transmitted laser intensity and the fitted zero-absorption
baseline. Right: the processed absorption (the grey area under the measured absorption
curve is the integrated absorbance).
Scanned-λ DA can be used to monitor temperature by taking the ratio R of the integrated
absorbance from two absorption transitions with linestrengths with different temperature
dependence as given (also see in Figure 2.3):
2 2 2
1 1 1
( ) ( )( )
( ) ( )
A S T P x L S TR T
A S T P x L S T
(2.10)
where the temperature dependence of the linestrength is explicitly given as:
1
0
00
0
"
000 exp1exp1
11exp
)(
)()()(
kT
hcv
kT
hcv
TTk
hcE
T
T
TQ
TQTSTS
(2.11)
14
where ( )Q T is the partition function of the absorbing molecule; h [J·S], c [cm/s] and k
[J/K]are Planck’s constant, speed of light in vacuum and the Boltzmann constant,
respectively; "
1E and "
2E are the lower state energies [cm-1
] of the two selected absorption
transitions; 0T [K] is the reference temperature of the spectral database, normally 296 K.
The temperature sensitivity of a two-line absorption sensor is proportional to the
difference in the E" of the selected pair of transitions. In later discussions of temperature
sensor design and sensitivity, the two lines will be denoted as high E" line and low E"
line.
Figure 2.3 Left: Absorption spectrum of two transitions with distinct temperature
dependence. Right: The ratio of the linestrength of these two transitions with temperature.
The temperature is given by spectral data and the ratio R by:
2 0
01 0
" "
2 1
" "
2 1ln ln
hck
S T hckTS T
E ET
R E E
(2.12)
15
Once the temperature is measured, the mole fraction of the absorption species can be
determined by the integrated absorbance of either line. One drawback of scanned- DA is
the need to determine the zero-absorption baseline (Io), which can be a challenge to
determine in high-pressure environments such as the coal gasifier. Efficient gasifiers
operate at elevated pressures [54], where the collision broadening of the absorption
transition can blend the target transition with its neighbors, and there is no region of zero-
absorption baseline within the wavelength scan range of a typical diode laser. In
addition, there can be significant attenuation of the laser intensity due to scattering by
particulate in the synthesis gas.
2.3 Wavelength modulation spectroscopy (WMS)
There is a long history of using wavelength modulation spectroscopy (WMS) for low-
noise measurements of small values of absorbance. However, here other advantages of
WMS using injection-current tuned diode lasers are exploited, which leads to
normalization strategies that offer solutions to the challenges of high pressure absorption
measurements in particulate laden gases. For WMS, the laser wavelength is modulated at
a frequency f. When TDLs are driven with a modulated injection current, there is a
simultaneous modulation of both the laser wavelength and the laser intensity. After the
modulated laser beam is directed through the gas sample (e.g., Figure 2.4), the frequency
content of the transmitted laser intensity is analyzed. As shown in Figure 2.4, the
wavelengths of two lasers were modulated at different frequencies, and the fast Fourier
transform of the transmitted laser intensity shows signals at the harmonics of the
modulation frequency. The signal at twice the modulation frequency ( 2 fS ) is
proportional to the absorption signal and the laser intensity (Io), while the signal at the
16
modulation frequency ( 1 fS ) is dominated by the laser intensity (Io). Thus, the ratio of
2 1/f fS S provides a signal independent of laser intensity [43-51].
Similar to the scanned direct absorption strategy, two-line thermometry can be applied in
WMS-2f for temperature measurements. By optimizing the modulation parameters, the
developed WMS T-sensor can be sensitive to temperature only:
2 1 1
2
2 1 2
( / )( )
( / )
f f transition
f
f f transition
S SR T
S S
(2.13)
Figure 2.4 Schematic of a typical fixed-wavelength WMS measurement with two TDLs
for temperature measurements and the frequency spectrum of the received time-domain
signals after fast Fourier transform (FFT)
17
Once the temperature is measured, the gas mole fraction can be determined from either
1f-normalized WMS-2f signal.
When this work began, 1f-normalized WMS-2f was routinely used in our laboratory
using the methods reviewed by Reiker et al [48]. This scheme required the generation of
a database of collision-broadening coefficients for all major components of the gas and
was limited to systems where a reasonable estimate of the gas composition was available.
In this dissertation, three major improvements were made: (1) the 1f-normalized WMS-2f
method is extended to the 1f-normalized WMS-nf; (2) a better strategy was developed for
charactering the intensity modulation that includes the intensity non-linearity of the entire
optical system and (3) a new WMS method is presented that avoids the need for a
collision-broadening database. These improvements illustrate the potential of the use of
WMS to develop practical TDL sensors for a wide variety of industrial applications.
18
19
Chapter 3 Generalized 1f-normalized WMS-
nf model using Fourier analysis considering
non-ideal diode laser performance
3.0 Motivation
Wavelength modulation absorption spectroscopy with 2nd
harmonic detection (WMS-2f)
is an established technique for gaseous molecular absorption measurements [55-63]. In
WMS-2f, the laser wavelength is modulated at frequency f and signals are detected at
frequency 2f, isolating the signal from low-frequency noise. Its unique advantages of
noise-rejection and AC coupled capability enable sensitive detection of small absorption
signals. The method becomes even more powerful when a current-injection tunable diode
laser is used as the light source. WMS signals at all of the harmonics are proportional to
the laser intensity, and for an injection-current-tuned diode laser the 1f signal is
dominated by the intensity modulation. Thus, normalizing the WMS-2f signals by the
WMS-1f signal can account for variations in laser intensity, including non-absorption
losses such as light scattering or beam steering. This normalization enables quantitative
WMS absorption measurements without determining a zero-absorption baseline, making
1f-normalized WMS-2f an attractive strategy for absorption measurements in various (i.e.
high-pressure, high-opacity, high-emission, high-temperature) environments.
20
Due to the use of WMS for many applications, models have emerged to interpret the
measured WMS signal into absolute absorption. However, most models are limited to
specific applications and may not be applicable to others. For example, some models are
valid only when the intensity modulation can be neglected [39], the modulation depth is
small [59], or the modulation frequency is low [45]. Others are only accurate when the
intensity modulation is linear, and may not be suitable for external-cavity lasers where
the non-linearity in intensity modulation can be large [63]. Previous related work in our
laboratory accounted for non-linear modulation of the laser and the finite phase shift
between intensity and wavelength modulation in its analysis of optically-thin WMS at
transition line center. Later this work was expanded to calibration-free measurements
with larger optical depth [48]. These models become even more complex when the
optical system has additional wavelength-dependent intensity variations (e.g., wavelength
dependent transmission interference (etalons) or the use of a semiconductor optical
amplifier with wavelength-dependent gain to increase the laser power).
In this chapter, a general model for 1f-normalized wavelength modulation absorption
spectroscopy with 2f detection is presented that considers the performance of injection-
current-tuned diode lasers and the reflective interference induced by other optical
components in the line-of-sight (LOS). This model can be used for arbitrary modulation
depths and optical depths and is extended to higher harmonics, for example WMS-3f, 4f.
Advantages of using these higher harmonics (nf, n>2) are discussed as well. More
specifically, the contents of this chapter are organized into five sections:
21
1) A generalized WMS-nf model for arbitrary modulation depths, applicable for both
low-pressure and high-pressure gas sensing is presented. The result is not limited to
optically-thin conditions and accounts for the performance characteristics of the
complete optical system including the injection-current tuned TDL and the LOS
optics.
2) A method to evaluate the laser wavelength modulation characteristics and intensity
modulation characteristics of the entire optical systems is then presented.
3) The WMS methematical and physical meanings are discussed, to privide prior
knowledge useful to understand the 1f-normalization strategy.
4) The TDL-WMS-nf model was validated by measurements of the 1f-normalized
WMS-nf lineshapes of the R(11) transition in the 1st overtone band of CO molecule
near 2.3µm at 4 different pressures ranging from 5atm to 20atm with a range of CO
mole fractions from 0.21% to 2.8%, all at room temperature.
5) The potential for WMS-nf (n>2) to reduce the background signal drift and
interference absorption from neighboring transitions is then discussed.
3.1 WMS-nf model
The Li et al. [44] model for WMS-2f including the performance of injection current-tuned
diode lasers is extended to WMS-nf including higher harmonic terms that were ignored in
most past WMS-2f studies.
Typically, for a WMS absorption measurement, the injection current of a TDL is
modulated by a sinusoidal function at frequency f , resulting in simultaneous modulation
of the laser wavelength (or frequency) and intensity,
22
( ) cos(2 )t a ft , (3.1)
0 0
1
( ) 1 cos( 2 )m m
laser laser laser laser
m
I t I i m ft
(3.2)
where [cm-1
] is the center (mean) wavelength (or frequency) of the laser under
modulation, a [cm-1
] the modulation depth, the initial phase of the wavelength
modulation, , 0
laserI the laser intensity emitted from the laser cavity, and 0
laserI the laser
intensity without modulation, m
laseri the mth
Fourier coefficient of the laser intensity, and
m
laser the initial phase of the mth
order intensity modulation.
For practical measurements, the measured laser intensity in the data acquisition system
can be modeled as:
00( ) ( ( )) ( ( ))T ( ( )) ( )optics laserI t G t t t I t (3.3)
where G is the detector responsibility [V/W], the non-absorption transmission loss due
to beam scattering or steering, Toptics is the transmission of the entire optical system. For
typical WMS measurements, the tuning range of the laser wavelength is small (< 1nm),
thus G and can be taken as constants with wavelength. The measured laser intensity is
a periodic function with period of 1/f, and can be expanded into Fourier series as:
00 0
1
( ) T ( ( )) ( ) 1 cos( 2 )optics laser
m m
m
I t G t I t I i m ft
(3.4)
where 0I is the detector-measured laser intensity without modulation; mi is the mth
Fourier coefficient of the measured detector signal with intensity modulation, and m the
initial phase of the mth
order intensity modulation.
23
Most previous WMS-2f studies utilize Eqn. (3.2) instead of Eqn. (3.4) and include only
the first-order, linear intensity-modulation term 1i , and the second-order, non-linear
intensity modulation amplitude 2i ; all other higher-order Fourier components are ignored.
This assumption is accurate when the laser intensity response to the injection-current
modulation has small nonlinearities and transmissions of the optical components in the
system are not significantly wavelength dependent over the wavelength region of
concern. These assumptions are not always valid; for example an external-cavity TDL
produces a significant interference (etalon) pattern in the output intensity, resulting in
non-negligible 3i and 4i terms even with small modulation depths (< 0.1cm-1
) [63].
Therefore the model developed here includes all the Fourier components to represent a
more general case. The highest order needed will depend on the specific laser
architecture, modulation depth and non-linearity of other optics on the laser line-of-sight
(LOS).
In the model, the choice of time-zero is arbitrary. To be consistent with the theory
derived by Kluczynski et al. [56] and Li et al. [44], the time-zero is defined as / 2 f ,
which sets the initial phase of the wavelength modulation as 0; Eqn. (3.1) and (3.4) are
then rewritten:
( ) cos(2 )t a ft , (3.5)
0 0
1
( ) 1 cos( 2 )m m
m
I t I i m ft m
. (3.6)
When the laser beam travels through the gas absorbing medium, the wavelength-
dependent transmission is described by the Beer-Lambert law:
24
0
exptj j i
j
IS P x L
I
, (3.7)
where jS and j are the linestrength and line shape function of transition j , P is the total
pressure of the gas, ix is the mole fraction of absorber i and L is the path length.
Because the laser wavelength is modulated at frequency f, the transmission also becomes
a periodic function and can be expanded into Fourier series:
0
( ) cos( 2 )k
k
t H k ft
, (3.8)
where kH is the kth
order Fourier coefficient defined as:
0
1( cos )cos
(1 )k
k
H a k d
. (3.9)
Note that no assumption has been made about the optical depth and this model is valid for
arbitrary absorbance.
The transmitted intensity with the presence of absorbers becomes:
0 0
1 0
( ) ( ) ( cos(2 )) 1 cos( 2 ) cos( 2 )t m m k
m k
I t I t a ft I i m ft m H k ft
. (3.10)
The harmonics at the modulation frequency of the transmitted laser intensity are extracted
by lock-in amplifiers with a bandwidth determined by the low-pass filter. The X-
component [detector signal × cos(n·2πft)] and Y-component [detector signal ×
sin(n·2πft)] of the lowpass filtered nf signal are written as:
25
0
1
1 1( (1 ) ) cos( )
2 2nf n n k nk k kn k
k
X I H H H i k
, (3.11)
0
1
1 1( (1 ) ) sin( )
2 2nf n k nk k kn k
k
Y I H H i k
, (3.12)
For cases where non-linear intensity response is insignificant (laser without large etalons
in the intensity tuning) or the modulation depth is small, the following approximations
can be used:
0 1 1 1 11
1 1( (1 ) ) cos( )
2 2nf n n n n
X I H H H i
(3.13)
0 1 1 1 11
1 1( (1 ) ) sin( )
2 2nf n n n
Y I H H i
. (3.14)
The WMS-nf signal magnitude can be written:
2 2
nf nf nfR X Y . (3.15)
When no absorbers are present, 0k kH , the absorption-free WMS-nf background
signals in the system are given:
0
0
1cos( )
2nf n nX I i n , (3.16)
0
0
1sin( )
2nf n nY I i n , (3.17)
0 0 2 0 2
0
1( ) ( )
2nf nf nf nR X Y I i , (3.18)
0
0 0fR I . (3.19)
26
After vector-subtracting the background, the WMS-nf signal becomes:
0 2 0 2[( ) ( ) ]nf nf nf nf nfS X X Y Y . (3.20)
3.2 Laser wavelength modulation characterization and optical
system intensity modulation characterization
3.2.1 Laser wavelength modulation characterization
To obtain an accurate WMS simulation, the absorption is described by the spectroscopic
parameters including linestrength and collisional broadening coefficients as described in
chapter 2, and the parameters that describe the injection current-tuning behavior of the
specific TDL. These parameters include a [cm-1
], the modulation depth, and ψ, the initial
phase of the wavelength modulation. Figure 3.1 shows an example laser characterization
result of a DFB laser near 1352nm. The method to evaluate the parameters that quantify
the injection-current tuning characteristics of the TDLs has been discussed in detail
previously, and the reader is referred to the literature [42,44].
Figure 3.1 Example of laser wavelength response to the injection-current modulation.
The blue circles are the relative frequency measured by a solid etalon with 0.02cm-1
FSR.
The measured modulation depth is 0.101cm-1
and initial phase of the wavelength
modulation is -2.1363 radian.
0 0.2 0.4 0.6 0.8 1
x 10-3
-0.1
-0.05
0
0.05
0.1a= 0.101cm
-1; phi= -2.1363rad
Time [s]
Fre
quency [cm
-1]
characterized wavelength response result
27
3.2.2 Optical system intensity modulation
The light transmission can vary as a function of laser wavelength due to the optical
components along the LOS. Most often such variation is due to interference from
reflection at parallel optical surfaces, commonly called "etalon" effects. At some
wavelengths, the interference is constructive, resulting in a transmission close to the
unity, and at other wavelengths, this interference can be destructive, resulting in a lower
transmission (see Figure 3.2). To minimize this interference, some surfaces of the optical
components are slightly wedged, but still it is very hard, or nearly impossible to avoid all
such interference in the entire optical system (especially for components with thin
parallel surfaces). For example, many IR-detectors are protected with flat windows in
front of the active area that can result in reflective interference when the laser wavelength
is tuned. As this interference is wavelength dependent, it can produce background signals
at the harmonics of the laser modulation frequency. Even larger interference can be
produced by other optical components such as measurement volume windows (non-
wedged), optical filters, or optical amplifiers.
Figure 3.2 Transmission of a thin cavity built by two parallel surfaces (cavity length:
2mm, surface reflectivity: 2%)
6385 6390 63950
0.2
0.4
0.6
0.8
1
frequency [cm-1]
cavity t
ransm
issio
n
28
As a result, it will be more accurate to evaluate the intensity modulation characteristics (
mi and m ) of the entire optical system rather than simply the laser. mi and m are
extracted from the WMS background signals for a measurement with no absorbers in the
LOS using Eqn. (3.16) and (3.17). During the measurements, time zero is determined by
the data acquisition and laser modulation triggering, and it is practical to use Eqn. (3.1)
and (3.4) (instead of Eqn. (3.5) and (3.6)) to define the laser intensity modulation; Eqn.
(3.16) and (3.17) then become:
0
0
1cos
2nf n nX I i , (3.21)
0
0
1sin
2nf n nY I i . (3.22)
Then m can be obtained as:
1 0 0 0
1 0 0 0
tan ( / ) 0
tan ( / ) 0
mf mf mf
m
mf mf mf
Y X for Y
Y X for Y
(3.23)
and mi can be obtained with Eqn. (3.18) and (3.19) as:
0 0
02 /m mf fi R R . (3.24)
This illustrates that the parameters ( mi and m ) of the intensity modulation used for
WMS lineshape simulations can be obtained from the WMS zero-absorption background
signals that combine the performance of the LOS optics and the TDL. For practical
measurements, when some external interference (e.g., etalons) cannot be easily
eliminated, or if the system includes optical components that are spectrally sensitive,
characterizing the overall system will be more relevant than characterizing only the laser.
A more detailed investigation of WMS background signals will be given in section 3.5.1.
29
3.3 WMS mathematical and physical meanings and 1f-
normalization strategy
One advantage of WMS is that the 1f signal can be used to normalize other harmonic
signals. The normalization can account for the laser intensity variations, allowing robust
absorption measurement in harsh environments where the non-absorption transmission
loss or optics vibrations are varying with time. It is helpful for understanding the 1f-
normalization strategy if we take a look at the mathematical and physical meanings of the
WMS harmonics first. Due to the modulation of the laser wavelength, the transmitted
laser intensity is a periodic function with the base frequency at f, and can be expanded
into Fourier series as:
2 2
( ) Y sin(2 ) cos(2 )
sin(2 )
sin(2 )
t n n
n n
n n n
n
nf n
n
I t DC AC DC nft X nft
DC X Y nft
DC S nft
. (3.25)
From the equation, the mathematical meaning of the WMS-nf signal is clear, i.e. Snf is the
Fourier coefficients in the Fourier series expanded from the transmitted laser intensity. Its
physical meaning can be observed in the following example. The blue dashed curve
showed in Figure 3.3 is the simulated transmitted laser intensity without absorption, I0;
the green curve is the simulated transmitted laser intensity with absorption, It; the red-
dashed curve is the sine wave fit at the 1f frequency for It. In the simulation, the laser
center wavelength is set at the linecenter of the transition.
It can be observed that the fitted laser intensity at 1f frequency after absorption is very
close to the laser intensity modulation profile without absorption. This illustrates that the
1f signal at transition linecenter is dominated by the laser intensity modulation
30
magnitude. And this magnitude is proportional to the transmitted laser power on the
detector. Thus the physical meaning of the 1f signal at transition linecenter is the laser
power in this simulated case.The lower part of the figure shows the residual between It
and its fit by the sine wave at the 1f frequency. For each modulation period, the laser
wavelength passes the same transition twice, and the residual is close to a periodic
function of frequency at 2f. This function is then fitted by a sine function at the 2f
frequency and is shown in Figure 3.4.
Figure 3.3 Upper: The simulated I0 and It in the WMS measurement and the sine wave fit
to It at the 1f frequency. Bottom: the residual between I0 and It
0 1 2
x 10-4
0
1
2
3
4
5
6
7
Time [s]
Dete
cto
r sig
nal [A
.U.]
0 1 2
x 10-4
-2
-1
0
Time [s]
Resid
ual [A
. U
.]
I0
It
It sine fit
It - I
t sine fit
31
The amplitude of the sine wave fit at the 2f frequency is the 2f signal. If there is no
absorption in the measurement, the residual is zero assuming a linear intensity
modulation, resulting in a zero amplitude of the sine function fit. Due to the absorption,
the residual has strong components at the 2f frequency (Figure 3.4), and this component
is proportional to absorption. Thus, the essence of the 2f signal has a physical meaning of
the magnitude of the absorption. The same logic can be applied to the other higher
harmonics, for example, 3f and 4f signals.
Figure 3.4 The residual in figure 3-3 and its sine fit at the 2f frequency
According to Eqn. (3.25), both the 1f signal and the 2f signal are proportional to the
magnitude of the transmitted laser intensity, as:
( ) sin(2 )t nf n
n
C I t C DC C S nft (3.26).
A C times increase in the transmitted laser intensity will result in a C times increase in
the WMS-nf signal. Thus the WMS-1f signal (dominated by the laser intensity
0 1 2
x 10-4
-2
-1.5
-1
-0.5
0
0.5
Time [s]
Resid
ual [A
.U.]
It - I
t sine fit
sine fit at 2f
32
modulation amplitude) can be used to normalize all harmonics of the WMS signals
(dominated by the product of laser intensity modulation amplitude and absorption),
similar to the 1f-normalized WMS-2f scheme developed previously [48]:
1/nf normalized nf fS S S (3.27) for cases where the background signals can be
negligible.
And
2 20 0
0 0
1 1 1 1
nf nf nf nf
nf normalized
f f f f
X X Y YS
R R R R
(3.28) for cases where a
background subtraction is necessary.
The normalization accounts for the laser intensity and responsibility (and gain) of the
detector, making the normalized measurement independent of the laser intensity
impinging on the detector and thus suitable for environments involving large non-
absorption transmission losses or variations.
In some cases, for example, when the absorption optical depth is large (~1) and the
modulation depth (the wavelength modulation range) is comparable or less than the
transition broadening width, the 1f signal can be dominated by both the absorption and
the intensity modulation. However the 1f normalization strategy still works in such cases
due to two factors (1) according to Eqn (3.11), (3.12) and (3.26), all harmonic signals are
proportional to the transmitted laser intensity 0I , so the normalization cancels the
dependence on the zero-absorption baseline during an absorption measurement. (2) The
normalization does not cancel the absorption signal. Unlike 2f signal, the increase of the
absorption will decrease the 1f signal (due to the interaction between the intensity
modulation and absorption). And the resulted 2f/1f signal increases with the absorption. A
more detailed discussion about this is presented in section 5.2.5.
33
3.4 WMS-nf model validation by high pressure CO WMS
spectra measurements
3.4.1 WMS-nf model validation
The WMS-nf model presented above was validated by measurements of CO WMS
lineshapes at different pressures and absorber mole fractions. The probed CO transition
was R(11) of the 1st overtone band near 4300.7cm
-1. The measurements included gas
pressures ranging from 5atm to 20atm and CO mole fractions ranging from 0.21% to
2.8% diluted in N2, all at room temperature.
The average pressure investigated was 12.5atm. At this pressure, the optimal modulation
depth (maximum signal magnitude) for WMS-2f is 1.56cm-1
(modulation index, m =
a/Δν ~2.2, where the transition half-width at half-maximum (HWHM) Δν is 0.71cm-1
at
12.5atm ). The optimal modulation depths for WMS-nf signals (n>2) are larger than
WMS-2f], for this case, 2.93cm
-1 for 4f and 4.33cm
-1 for 6f. However, the maximum
modulation depths achievable by the available TDLs are limited by their injection current
limits. Although the modulation frequency can be decreased to obtain a larger
modulation depth, this will reduce the time resolution of the measurement, and make
WMS less effective in rejecting low-frequency noise. As a result, a modulation depth of
1.52cm-1
and a modulation frequency of 1kHz were selected. The selected modulation
depth is very close to the optimal modulation depth for WMS-2f, but smaller than the
optimal for WMS-nf signals (n>2), especially when the pressures are larger than 10atm.
The WMS lineshapes were measured at room temperature (296K) in a 100.5cm-long cell
(Figure 3.5), equipped with wedged sapphire windows. A DFB laser (center wavelength
4300cm-1
from Nanoplus) was used with output coupled into a single-mode fiber
34
(corning, SMF-28). The collimated-laser beam was directed through the cell and focused
(mirror of 5cm focal length) onto a New Focus extended- InGaAs photo-receiver of
700kHz bandwidth with 1mm2 active area. The detector voltage signal was then sampled
using a 12-bit National Instrument PCI-6110 DAQ card, at a sampling rate of 2.5 MHz.
Gas mixtures (CO in N2) were prepared in a mixing tank and allowed to diffusively mix
overnight before the measurements.
The injection current of the laser was modulated with a 1kHz sine wave. The center
frequency of the modulated laser was scanned by stepping the laser temperature in 0.1 C
steps resulting in 0.04cm-1
laser frequency resolution. After each step, a 15 second delay
was given for the laser temperature to stabilize. The WMS lineshape was collected for
wavelength tuning over the range from 4298cm-1
to 4304cm-1
.
Figure 3.5 Experimental setup for high pressure CO gas sensing.
The background signals were measured before introducing the CO absorber into the cell
to obtain mi and m as described in section 3.2.2. The WMS-nf signal simulations used
linestrength and broadening coefficients from Chao et al. [42], and the pressure-induced
linecenter-shift coefficients from HITEMP 2010 [64].
35
Figure 3.6 compares the measured background-subtracted WMS-nf spectra with
simulations for the R(11) transition with 1.59% CO in N2 at pressures from 5atm to
20tam. Although at these pressures, a closed-form Lorentzian lineshape could have been
used in the simulations without introducing significant errors, the more general Voigt
lineshape function was used to facilitate validation of the measurements over a wider
range of pressure (both high and low pressure applications). At all pressures, the
measured WMS-2f, 3f and 4f lineshapes agreed very well with the simulations based on
the model described in section 3.1 and 3.2, with an average deviation of less than 2%.
An overall measurement uncertainty was estimated to be about 2.5%, including 0.5% in
the gas mixture mole fraction, 1% in the spectroscopic parameters, 0.5% in laser
characterization parameters and 0.5% in pressure. Thus, the observed deviation between
the measurements and simulations for WMS-2f, 3f and 4f lineshapes were all less than the
system uncertainty.
3.4.2 High pressure CO sensor design based on WMS-nf detection
After the WMS-nf model was validated by measuring wavelength-scanned WMS-nf
lineshapes, a sensor adopting fixed-wavelength WMS was evaluated for practical sensing
of gas mole fraction at elevated pressures. This fixed-wavelength scheme can have much
faster time resolution than with the slow wavelength scanning using temperature tuning.
A modulation frequency of 1kHz and a lowpass filter with bandwidth of 100Hz was used
to give 20ms temporal resolution for CO mole fraction measurements. To validate the
sensor, mixtures with CO mole fractions ranging from 0.21% to 2.8% were used and the
experiment conditions were same as the previous section. The measured CO mole
fractions were determined from the peak values of the 1f-normalized WMS-nf spectra as
shown in Figure 3.7 for n = 2-6.
36
Figure 3.6 Measured (square, red) and simulated (solid line, black) 1f-normalized WMS-
2f, 3f and 4f lineshape signals at different pressures. Gas mixture: 1.59% CO in N2; T =
296 K; optical pathlength L = 100.5 cm.
The results were compared with the calibrated CO mixture mole fractions. The averaged
deviations between measurements of WMS-2f, 3f, 4f and signals simulated for known gas
mixtures were all less than the measurement uncertainty analyzed in section A. However,
the measured WMS-5f and 6f signals had larger deviations due to the limited modulation
depth.
37
The fixed-frequency WMS results show that the lower harmonics (for n = 2-4) have the
advantage of larger signal magnitude in gas mole fraction measurements at high
pressures. This explains why the 2f signal is favored for most applications, especially
trace species measurements in a simple optical system without parallel surfaces
(windows, mirrors, etc.), where etalon interferences are small compared with those often
present in practical high-pressure applications (e.g., coal gasifiers, gas turbine or internal
combustion engines). Here the measurement time was only a few minutes and the WMS-
background signals were measured just before the absorption measurements. The
accuracy of the CO mole fraction measurements indicates that the background drifts were
sufficiently small for this short time period and that no significant errors were incurred.
However, for applications where the measurement period is long (e.g., more than one
day), or when large variations in measurement conditions are present, or when
background measurements are not feasible, significant measurement error due to the
WMS background can occur, which could potentially increase the measurement
uncertainty and affect the accuracy of the absorption measurements. For these cases when
the drift of the WMS background cannot be regularly measured during an absorption
measurement, WMS higher harmonic signals can have the advantage of smaller
background drifts, which will be discussed in details in the next section.
38
Figure 3.7 Measured CO mole fractions by 1f-normalized WMS-nf technique and comparison
with calibrated CO mole fractions (dashed line), bath gas: N2; T = 296K; optical pathlength L =
100.5cm.
3.5 Advantages of using higher harmonics
3.5.1 Advantage of WMS-nf (n>2) in reducing the background signal
drift
Among all harmonics of WMS, the second harmonic of the transmitted laser signals,
called WMS-2f , is the one most commonly employed for gas sensing, owing to its
relatively large magnitude. However, at high pressures >10atm, when an optimal
modulation depth for WMS-2f is used with an injection current-tuned diode laser, the
2f background signal can be significant. An inaccurate measurement of the
background signal can generate dramatic error in the gas mole fraction measurements.
In a practical absorption measurement, with interference patterns generated from
reflections between parallel surfaces (etalons), the background signal can drift over
time due to thermal expansion or other movements of the optical windows [65,66],
temperature change of the laser components [67], and slight variation in the optical
39
alignment [68]. Even under quiet laboratory conditions in an evacuated gas cell with
the temperature stabilized at 296±2K, when a relatively large modulation depth of
1.52cm-1
(Nanoplus TDL, near 2325nm) was used, a 33% change in the 2f background
was observed over ~ 15 hours as shown in Figure 3.8. As simultaneous monitoring of
the WMS background is not possible for practical TDL absorption sensor applications,
schemes are desired to reduce the impact of WMS background variations.
In this section, a detailed investigation of the WMS-nf background signals and their
drifts was examined for a large modulation depth (a>= 0.5cm-1) required by high
pressure measurements. As in such cases, the background signal is much more
important due to the small ratio of the absorption signal to the background signal.
0 2 4 6 8 10 12 14 160.024
0.026
0.028
0.030
0.032
0.034
0.036
33% change
2f b
ackgro
und s
ignal, n
orm
aliz
ed
by the laser
inte
nsity (
unitle
ss)
Time [hour]
Normalized 2f background signal
Figure 3.8 Measured 2f background magnitude versus time, normalized by the laser
intensity. (Measured with a 2.3µm Nanoplus TDL, traveling through an evacuated,
17.3cm cell with wedged windows. Modulation depth = 1.52cm-1
, modulation
frequency = 1kHz. No external intervention presented during measurement, conducted
at Stanford in 2011.)
At high pressures, the magnitude of the absorption-induced 2f signal (background
subtracted) can be comparable to its absorption-free background signal. When the
pressure is higher than 10atm and a relatively large modulation depth is used, such as
40
one close to the optimal modulation depth, the 2f signal can even be much smaller
than its background signal in small absorbance (<0.1) cases. Figure 3.9 shows the
measured 1f-normalized WMS-2f and 4f signals (background subtracted) as well as
their background signals for 0.21% CO in N2 at 10atm in a 100.5cm long cell (peak
absorbance ~0.085). Note that at these experiment conditions, the 2f signal peak is
smaller than its background signal at the same wavelength, while the 4f signal peak is
much larger than its corresponding background signal. To quantify the relative
magnitudes of the background signal, we here define a signal-to-background ratio
(SBR) as the absorption-induced signal to the absorption-free background signal at the
same wavelength. Therefore, in the above measurement, the 2f and 4f signal peaks
have SBRs of 0.7 and 35, respectively.
4299.5 4300.0 4300.5 4301.0 4301.50.00
0.01
0.02
0.03
0.04
0.05
2f peak
4f peak
0.21% CO in N2
P = 10 atm, T = 296 K
1f-
no
rma
lize
d W
MS
-2f (4
f) s
ign
al [u
nitle
ss]
Frequency [cm-1]
1f-normalized 2f
1f-normalized 2f background
1f-normalized 4f
1f-normalized 4f background
Figure 3.9 Measured 1f-normalized WMS-2f and 4f signals as well as their
background signals. (0.21% CO in N2, P = 10atm, T = 296K, L = 100.5cm; a =
1.52cm-1
, f = 1kHz)
Such measurements were carried out for various pressures, and the 1f-normalized
WMS-nf peak-to-background ratios are plotted versus pressure in Figure 3.10.
Although the absolute WMS-2f signal was larger than all other harmonic signals, the
ratio of the 2f signal to its background signal was at least one order-of-magnitude
41
smaller than for the other harmonics. The result also indicates that for P ≥ 10atm, the
absorption-induced 2f signal peak was smaller than its background signal for the
condition studied here. The small SBRs will not result in significant errors in mole
fraction measurements as long as the background signal is well determined and does
not change significantly during the measurement period as for measurements shown in
Figure 3.7. However, if the background signal drifts over time due to changes in the
beam path, e.g., signals associated with optical interferences and reflections, this may
lead to significant error in the mole fraction determination if the SBR is small. Higher
WMS harmonic signals have much larger SBRs than WMS-2f and potentially can thus
have less relative error in absorption measurements from the background drift. In
addition, since the peak signals of the higher harmonics (n>2) is much larger than their
background signal, knowledge or accurate evaluation of the background signals may
not be required.
5 10 15 20
1
10
100
1000
0.21 % CO in N2
ratio o
f 1
f-norm
aliz
ed W
MS
-nf
peak m
agnitude t
o
its 1
f-norm
aliz
ed b
ackgro
und m
agnitud
e
Pressure [atm]
2f
3f
4f
5f
6f
Figure 3.10 Measured ratio of 1f-normalized WMS-nf signal to its 1f-normalized
background signal. ( a =1.52cm-1
, f =1kHz, T =296K, L =100.5cm)
To quantify the effect of the background drift, the background signal of the optical
system used in the previous measurement was monitored overnight, and Figure 3.11
42
shows the ratio of the WMS-nf signal peak to the change in background, the latter
defined as the standard deviation of the background signal over the measurement
period (~10 hours). The result is consistent with the SBR results in Figure 3.10. It
shows that the background drift has less effect on higher harmonics measurements.
For the optical systems described in Figure 3.5, the windows on the cell were wedged,
and the parallel surfaces in the system include the protective windows before the
detector photodiode, the connection between the fiber end and the collimator. All
these parallel surfaces result in thin etalons with a cavity length less than 1mm. Slight
temperature or alignment changes can result in the changes of cavity lengths. Figure
3.12 shows the simulated WMS-nf background signals with the cavity length from
0.498mm to 0.502mm for a modulation depth of 1.5cm-1
and a reflectivity of 4.3% (a
typical value for the uncoated glass material). The result shows that a 0.27μm change
in the cavity length can potentially cause a drift of the 2f background signal from
0.058 to 0.01. Such drifts are at least one order-of-magnitude smaller in 4f, 5f and 6f
background signals.
5 10 15 2010
-1
100
101
102
103
104
105
0 2 4 6 80.0395
0.0400
0.0405
0.0410
0.04151f-norm alized W M S-2f bg.
S tandard dev. = 4.5 X 10-4
2f
ba
ck
gro
un
d
T im e [hour]
ratio
of
1f-
no
rma
lize
d W
MS
-nf
pe
ak m
ag
nitu
de
to
its 1
f-n
orm
aliz
ed
ba
ckg
rou
nd
drift
ma
gn
itu
de
Pressure [atm]
2f
3f
4f
5f
6f
Figure 3.11 Measured ratios of 1f-normalized WMS-nf signal to the drift magnitude
of its 1f-normalized background signal. ( a =1.52cm-1
, f =1kHz, T =296K, L
=100.5cm)
43
0.498 0.499 0.500 0.501 0.5020.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
WM
S-n
f backgro
und s
ignal [A
.U.]
L [mm]
2f
3f
4f
5f
6f
Figure 3.12 simulated WMS-nf background signals with the cavity length from
0.498mm to 0.502mm for a modulation depth of 1.5cm-1
and a reflectivity of 4.3%
For mixtures where the change of the absorber concentration does not significantly
change the transition broadening width and the peak absorbance is small (<0.1), the
1f-normalized WMS-nf signal peak is linearly proportional to the absorber
concentration. This proportionality allows the detection limit to be estimated from the
measured concentration and the percentage drift of the background signal with respect
to the 1f-normalized WMS-nf peak magnitudes. Figure 3.13 shows the detection limits
for 2100ppm CO concentration measurements (R(11) transition in the 1st overtone
band) estimated from the drift in the TDL-WMS background. The detection limits are
much smaller for WMS-nf signals with n>2 than for WMS-2f. For 20atm pressure, the
estimated detection limit using 2f was 62ppm, corresponding to about 3% for the
measurement of 0.21% (2100ppm) CO, which was larger than the measurement
uncertainty analyzed in section 3 for short term use of the sensor. This detection limit
will be increased when the pressure is higher. If a 50ppm detection limit in mole
fraction measurements is required, the highest pressure condition measurable by
WMS-2f will be less than 16atm, whereas for WMS-nf with n>2, pressures up to
30atm will be feasible. These results are representative for laboratory conditions,
44
whereas larger background drifts and higher detection limits are anticipated for
practical field conditions. Of course, if a CO transition in the fundamental band is
used, due to the much higher absorption than the transitions in the 1st overtone band,
the background drift will cause much less uncertainty.
5 10 15 20
0
10
20
30
40
50
60
70
B
ackgro
ud-d
rift
induced
WM
S-n
f d
ete
ction lim
it [
ppm
]
Pressure [atm]
2f
3f
4f
5f
6f
Figure 3.13 Background-drift induced WMS-nf detection limits at different pressures.
(CO in N2, T = 296K, L = 100.5cm; a = 1.52cm-1
, f = 1kHz, absorption transition: R(11)
in the first overtone of CO)
3.5.2 Advantage of WMS-nf (n>2) in reducing the interference from
neighbors
One advantage of WMS-2f over DA is the higher sensitivity to the absorption
lineshape curvature, giving larger signals near the transition line center than in the
wings. This is attractive for elevated pressure measurements where the transitions are
more blended by collisional broadening. For WMS-2f measurements, the signal peak
usually appears right close to the line center, where sensitivity is highest; therefore
interference from the wings of neighbor transitions will be minimized. It follows that
this advantage can be even more pronounced for higher harmonics measurements, e.g.,
WMS-4f. The WMS-2f signal is dominated by the Fourier components from 0H to
4H ,
45
whereas WMS-nf (n>2) is dominated by higher Fourier components kH as shown in
Eqn. 3.14 and 3.15. For example, WMS-4f signal can be written as
4 0 4 5 3 1 1 6 2 2 2
1 1 1( ) cos( ) ( ) cos( 2 )
2 2 2fX GI H H H i H H i
,
(3.29)
4 0 5 3 1 1 6 2 2 2
1 1 1( ) sin( ) ( ) sin( 2 )
2 2 2fY GI H H i H H i
. (3.30)
Figure 3.14 shows the simulatedkH 's for an individual, isolated CO transition near
4300.7cm-1
. The higher the order of the Fourier component, the faster its magnitude
decays to zero as the wavelength deviates from the line center. As WMS higher
harmonics include contributions mostly from higher orderkH 's, transitions detected
with WMS higher harmonics will have less interference from neighboring transitions.
4298 4300 4302 4304
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04 H
0-1
H1
H2
H3
Frequency [cm-1]
4298 4300 4302 4304-0.015
-0.010
-0.005
0.000
0.005
0.010
0.015
Frequency [cm-1]
H4
H5
H6
Figure 3.14 Simulated kH for targeted transition near 4300.7cm-1 at 20atm ( a
=1.52cm-1, f =1 kHz, 0.21% CO in N2, T =296K, L =100.5cm)
When the 1f-normalized WMS-nf detection method is used to make the absorption
measurement, this advantage of WMS-higher harmonics will be degraded to a certain
extent, since the 1f signal used for normalization includes contributions from mainly
46
low orderkH . This is especially the case for large absorbance cases (>0.5) where the 1f
signal will be more dependent on the absorption. However, in the optically thin case
(absorbance < 0.1), where the 1f signal is mainly dominated by the variation of laser
power from the injection current-tuning, this advantage will remain significant.
47
Chapter 4 H2O absorption sensor using fixed-
wavelength WMS in a pilot-scale high
pressure entrained-flow coal gasifier
4.0 Motivation
Integrated gasification combined cycle (IGCC) is a promising power generation
technology that offers potential for high-efficiency CO2 sequestration. As the key
component in such a system, the gasifier converts the raw carbonaceous materials into
synthesis gas (syngas), comprised primarily of H2, CO and CO2, which is used as fuel in
the combustor of a gas turbine. The gasifier performance and refractory longevity are
directly associated with the operation temperature. However, due to the harsh working
conditions, the thermocouples normally used for sensing temperature have short lifetimes
in the gasifier reactor and are usually recessed in the large-heat-capacity refractory,
resulting in a slow response to variations in gas temperature. In addition, high-
performance gas turbines require precise fuel/air stoichiometry to optimize the operation
[69]. Traditional gas analysis such as gas chromatography (GC) requires extractive gas
sampling, and is usually associated with low time resolution and long-time delays. For
real-time gasifier control and optimization, rapid and robust in-situ sensors are needed in
the gasifier for temperature and species concentrations.
Tunable diode laser (TDL) sensors offer a promising solution to these needs in that the
optical-absorption measurements can be non-intrusive, in-situ, and species-selective with
48
fast-time resolution and fast response. TDL techniques can provide quantitative
measurements for both gas temperature and concentrations, and such sensors have been
demonstrated previously in various combustion systems including coal-fired power plants
and waste incinerators. However, conditions in a coal gasifier have made this a more
challenging environment for laser-based absorption sensors. In particular, the gasifier is a
high-pressure and particulate-laden reactor, in which pressure-broadened absorption
features lead to reduced detection sensitivity and to problems with interference from
other species and from neighboring transitions of the targeted species. In addition, the
laser intensity can be significantly attenuated due to light scattering by particulate. This is
exacerbated by the fact that the thermal emission from the particles in the reactor core
can produce large emission intensities. Due to these challenges, only a few laser
absorption measurements have been reported in coal-gasifier reactors [23], and most of
these measurements were at atmospheric pressure, whereas modern gasifiers are designed
for high pressures.
In this chapter, the H2O absorption measurement in a pilot-scale entrained-flow slagging
high pressure coal gasifier is presented. The H2O sensor employed a 1f-normalized
WMS-2f strategy introduced in Chapter 3 and multiple TDLs near 1.4µm are used to
monitor H2O absorption transitions, from which the gas temperature and H2O mole
fraction were inferred. This temperature sensor offers the potential to capture transient
temperature changes in real-time to be used as feedback signals in a closed-loop control
system. And as one of the major components in the syngas, the H2O mole fraction in the
syngas output flow was monitored.
49
4.1 Gasifier facility
The measurements were performed in a pilot-scale, pressurized, oxygen-blown,
entrained-flow “Texaco-style” or “GE-style” gasifier located at the University of Utah’s
Industrial Combustion and Gasification Research Facility (Figure 4.1). The gasifier is
located indoors in a dedicated laboratory building, which offers excellent access for
research and monitoring. The control room has adequate space for the sensor electronics
and control, and the reactors can be reached with modest length (~30 m) optical fibers
and signal cables. The research nature of this facility was ideal for the proof-of-concept
testing and investigation of the optimum engineering of optical view ports for gasifier
TDL sensors. The Utah gasifier provided a unique test environment of high pressure,
high temperature and particle loading for the test of TDL sensors, which was not
available at Stanford. Fundamental TDL absorption strategies could be tested in a
realistic gasifier environment.
Figure 4.1 University of Utah Gasification Research Facility.
50
4.1.1 Entrained-flow gasifier and sampling locations
A schematic diagram of the entrained-flow gasification system is shown in Figure 4.2.
Technical details of the gasifier are presented in Table 4.1. The heart of the system is a
20-cm diameter, 1.5-m long down-fired refractory-lined reactor (Figure 4.3). An injector
positioned at the top of the reactor uses oxygen to atomize a water-based slurry of
pulverized (~70 micron) coal. Five B-type thermocouples flush with the inner wall of the
refractory along the length of the reactor monitor the reactor temperature.
Figure 4.2 Schematic of entrained-flow gasification research facility.
51
Table 4.1 Gasifier specifications.
Specification Units Typical Max Units Typical Max
Pressure psig 250 425 atm 18 30
Temperature °F 2600 3100 °C 1425 1700
Coal feed rate lb/hr (dry) 75 135 mton/day (dry) 0.8 1.5
Thermal input MMBtu/hr 1.0 1.7 kWth 300 500
Slurry flow rate gal/hr 15 30 liter/hr 55 115
Slurry solids content wt% 59 65 wt% 59 65
Six pairs of opposing sample ports along the length of the gasifier allow optical access
across the flow. For the tests reported here, the fourth set of ports, approximately 0.7m
downstream from the injector, was used. This position is referred to as Location 1 in
Figure 4.4. Below the reactor core, several flat spray nozzles inject water into the flow to
cool the products. This rapid cooling quenches the gasifier reactions and causes liquid
slag to solidify. During normal operation, four spray nozzles are used, but for these tests
two opposing spray nozzles were removed and the remaining two nozzles were pointed
downwards at angle of roughly 30 degrees. The two empty ports were used to provide
optical access into this region of transition between the hot reactor and the water quench.
This position is indicated as Location 2 in Figure 4.4. After the quench, the slag was
separated from the syngas flow and the product syngas was transported through a 6.5 cm
diameter pipe from the gasifier through a pressure control valve. This transport pipe was
modified to mount two window ports for laser absorption measurements and is indicated
as Location 3 in Figure 4.4. The gasification facility has a high-pressure candle-style
filter for removing particulate from the syngas stream. The post-filter syngas line is
52
shown as Location 4 in Figure 4.4. For the tests reported here the filter was bypassed, so
no results from Location 4 are presented.
4.1.2 System operation
The gasifier was operated on pulverized coal or liquid ethanol during the day and idled
on natural gas at atmospheric pressure during the night. Before the first fed into the
system, the reactor was heated with natural gas for approximately three days to ensure
that the refractory was thoroughly heated and to allow the system to come to thermal
steady state.
Figure 4.3 Schematic to the pilot-scale, entrained-flow, coal-gasifier at University of
Utah
53
Figure 4.4 Schematic to measurement locations at University of Utah: (1) the reactor-
core, (2) pre-quench, (3) post-quench, (4) after clean-up
To prepare for introduction of coal, the natural gas burner was removed and the slurry
injection lance was installed and tightened down. After a final safety check of all
systems, the feed pump was turned on to begin feeding fuel to the reactor. For startup,
either ethanol or isopropyl alcohol was used. Alcohol is much more combustible than the
coal slurry and is used to establish a flame and heat the reactor to the target temperature.
Once it was confirmed that fuel was flowing through the injector, oxygen flow was
initiated at a flow rate corresponding to a stoichiometric ratio of roughly λ = 0.5.
Presence of a flame was confirmed both by UV flame detectors and by a rise in reactor
temperature. Shortly after the temperature began to rise, the pressure control valve was
closed and the system was allowed to pressurize to the target pressure. When the
pressure reaches approximately 60 psi the fuel was switched from alcohol to coal slurry.
54
Significant production of soot when feeding alcohol (essentially alcohol cracking) was
observed. The gasifier was then pressurized to a target pressure and stabilized for
gasification.
4.2 H2O absorption sensor design
The temperature was inferred by the ratio of 1f-normalized WMS-2f absorption signals
from two H2O transitions with different temperature dependence, known as two-line
thermometry. Pairs of H2O transitions near 1.4 µm were selected to measure the gas
temperature at three different gasifier locations as listed in Figure 4.4. Each measurement
location has a different range of temperature and thus a different line pair was needed to
optimize temperature sensitivity of the absorption ratio as well as the absorbance.
These transitions were selected based on four criteria:
(1) The transition should provide enough absorption. Previous WMS studies suggest
an optically-thin absorption of less than 10% to simplify the data reduction.
However, due to the harshness of the gasifier environment, we found that a larger
than 10% absorption was desired for a good signal to noise ratio (SNR).
(2) The transision, or set of close transitions of similar E” should be the dominant
feature in its wavelength region. Due to the high pressure environment in the
gasifier, some promising transitions for atomspheric or low pressure measurement
can not be used.
(3) The difference in the E” of the selected pair of transitions used for T
measurements should be as large as possible. The difference should at least be
larger than the measured temperature [in kelvin] divided by 1.44, as:
55
" [ ] /1.44E T K . To make the sensor more sensitive to the temperature, in this
study, we set the criterion as : " 2 [ ] /1.44E T K .
(4) Least interference from other species. As seen in Figure 4.5, in the telecom
wavelength region, H2O transitions are dominant and have negligible interference
from other major components in the syngas.
Table 4.2 Selected transitions for temperature and gas concentration sensors at different
locations of the coal-gasifier (* more than one transitions with similar E" in the selected
wavelength region, which form one apparent peak feature at high pressure, **details of
CO2 and CO work were present where else [70])
λ
[nm]
ν
[cm-1
] Lower state
energy
E" [cm-1
]
Location in the gasifier
1469 6806.0 3291 T sensor , reactor core, location 1
1392 7185.6 1045
1339* 7466.0* 2600 T sensor , pre-quench, location 2
1347* 7426.1* 1300
1405* 7117.3* 420
T sensor (paired with 7426.1cm-1
)
H2O % sensor, post-quench, location
3
2017** 4957.1 234
CO2 % sensor, post-quench, location
3
2325** 4300.7 253
CO % sensor, post-quench, location
3
A large range of pressures and non-uniformities in absorber concentration were expected
in the gasifier reactor core, and transitions were selected to maximize the sensitivity of
the absorption ratio to temperature (as shown in Figure 4.6 for the pair of 7185.6cm-1
and
6806.0cm-1
transitions) by optimizing the modulation depths. Our analysis (see Figure
56
4.6) indicates that a ~20% variation of absorber mole fraction and/or the total gas
pressure will result in less than 2% uncertainty in gas temperature.
1.2 1.5 1.8 2.1 2.4
10-23
10-22
10-21
10-20
10-19
21
1
2
3
1
2
222
1
T = 296KCO
Commercial TDL with fiber output
at extended NIR wavelengthTelecom region
CO2
H2O
Lin
estr
en
gth
[cm
-1/(
mo
lecu
le.c
m-2)]
Wavelength [m]
Figure 4.5 Absorption linestrengths of H2O, CO2 and CO at 296 K from HITEMP 2010
database.
The transitions selected for CO, CO2 and H2O mole fraction measurements in the post-
quench location (gasifier product syngas stream) are listed in Table 4.2. For all
measurements at location 1 (reactor core) and 3 (post-quench), WMS with a modulation
frequency of 10 kHz was used, and a time-demultiplexing technique was used when
multiple lasers were used. For measurements at location 2 (pre-quench), the 1347 nm
laser was modulated at 10 kHz and the 1339 nm laser was modulated at 13 kHz, allowing
separation of the laser signals with frequency-demultiplexing. To improve SNR, the
measured data were averaged for 1 s unless otherwise specified. The absorption-free
WMS background signals were measured when the gasifier was filled with pure N2 gas.
57
1000 1200 1400 1600 1800 20001
2
3
4
5
6
7
8
Ratio o
f (2
f/1f)
peak o
f 7185.6
cm
-1 t
ransitio
n
to (
2f/
1f)
peak o
f 6806.0
cm
-1 t
ransitio
n
Temperature [K]
9 atm, 25% H2O
11 atm, 25% H2O
13 atm, 25% H2O
11 atm, 20% H2O
11 atm, 30% H2O
Figure 4.6 Simulations of the 1f-normalized WMS-2f peak magnitude ratio of the
7185.6cm-1
transition to the 6806.0cm-1
transition versus temperature at different
pressures and absorber concentrations (H2O).
4.3 Laser absorption sensor setup and alignment
The syngas is poisonous and explosive. For safety concerns, the lasers were remotely
operated from a control room 20 m away from the gasifier rig (see Figure 4.7). The H2O
sensor was controlled by a PC using a LabVIEW program to drive two NEL fiber-
coupled DFB lasers (output power ~ 15 mW) via an ILX LDC-3900 power source. The
laser beams were multiplexed by a fused-fiber combiner, transported to the measurement
location via a 30 m-long polarization-maintaining (PM) single-mode fiber, lens-
collimated into free space and directed through the syngas flow. The detected signals
were subsequently transported back to the PC in the control room by a 30 m-long BNC
cable. The transmission signals were remotely recorded by a National-Instrument PCI-
6110 DAC card installed in the same PC. Photos of the connection of the long PM fiber
and BNC cables from the control room to the gasifier rig are shown in Figure 4.8.
58
Figure 4.7 Schematic of the experiment setup in the control room and optical alignment
from the control room to the gasifier rig.
59
Figure 4.8 Photos of the connection of the long PM fiber and BNC cables from the
control room to the gasifier rig: (a) Fiber and BNC cables output from the control room,
(b) Aligned from the building of the control room to the gasifier, (c) connect the PM fiber
to a lens collimator mounted on the flange at location 2, (d) connect the BNC cable to the
detector mounted on the opposite flange
4.4 Results and discussion
4.4.1 Reactor measurements
Significant laser intensity attenuation due to light scattering in the particulate-laden
environment and strong emission were the two major challenges experienced by the TDL
absorption measurements in the reactor core. One of the goals of these first measurements
was to evaluate the level of emission and light scattering versus gasifier pressure. Figure
4.9 shows the measured emission and light transmission at the point of optical access in
the reactor core when the gasifier pressure increased from 1atm to 11.2atm (at 1atm
pressure the gasifier was fueled on isopropyl alcohol, which was switched to coal slurry
at 4.4 atm). The measurement was performed by scanning the 1469nm laser with
60
injection current below the threshold for a portion of each scan. This allowed a monitor
of the emission (from the period when laser was off) and total transmission (by
comparing the transmitted laser intensity scan-amplitude when the laser was tuned off the
absorption line before and after the gasifier operation began) in every single scan. Note
that the transmitted laser intensity was also attenuated by absorption, but this only
contributed a small fraction of the beam attenuation compared to the non-absorption
losses. The non-absorption transmission loss was found to increase exponentially with
pressure, mainly because of the larger particulate loading at higher pressures. The optical
emission also increased by a factor of four as the reactor pressure increased (see Figure
4.9). Thus, the challenges of measuring the absorption in the gasifier reactor core
significantly increase with pressure, as the ratio of transmitted laser intensity to
background light emission decreased by a factor of 20,000 as the reactor pressure was
increased from 1 to 11atm (see Figure 4.9).
0 3 6 9 120.00
0.04
0.08
0.12
0.16
0.20
0.24
0.28
Switch isopropyl alcohol
to coal slurry
De
tecte
d e
mis
sio
n s
ign
al [V
]
Pressure [atm]
Lig
ht tra
nsm
issio
n [%
]
0.01
0.1
1
10
100
Emission
Transmission
Figure 4.9 Measured optical emission from the combusting liquid or coal and light
transmission in the reactor core as the gasifier was pressurized (low-pressure fuel
isopropyl alcohol switching to a pulverized coal slurry at 4.4atm (65 psig)).
Measurements used the optical filter and InGaAs detector described in the text.
61
To evaluate the performance of the H2O absorption sensor in the reactor core where the
environment was most hostile, wide wavelength scans of the 1f-normalized WMS-2f
spectra for the 1392nm and 1469nm lasers were performed by tuning the wavelength
with laser temperature in increments of 0.1 oC. At each laser temperature, fixed-
wavelength WMS was performed with f = 10kHz, and the two lasers were time-
demultiplexed at 100Hz. Sample averaging (100 times, resulting in 1s time resolution)
was used to decrease the noise. Figure 4.10 shows the measured spectra when the gasifier
was fueled on liquid ethanol at 11.2atm with temperature near 1510K as indicated by
thermocouple readings. The non-absorption transmission loss was as large as 99.997%
and the ratio of the emission to the transmitted laser intensity was as large as 200, even
using the spatial and spectral filters. However, the lineshape of the pressure-broadened
features agreed well with simulations, and SNRs of 11.2 for the 7185.6cm-1
transition and
7.4 for the 6806.0cm-1
transition were achieved for the complete 1f-normalized WMS-2f
spectra. The largest discrepancies were observed in the wings of the lineshape where
interference from the neighbor transitions was most significant. The temperature
measurements are based on the peak values of the 1f-normalized WMS-2f spectra at line
centers, and here the SNR increased to 42 for the 7185.60cm-1
transition and 14 for the
6806.0cm-1
transition. To summarize, even with the large emission background and
extreme attenuation of the transmission by scattering, the 1f-normalized WMS-2f strategy
provided robust absorption measurements in this extremely hostile environment.
62
Figure 4.10 Measured and simulated 1f-normalized WMS-2f absorption spectra of
7185cm-1
transition (left) and 6806.0cm-1
transition (right) in the reactor core (27% H2O
in syngas flow, P=11.2atm, T=1510K, L=12.5cm, transmission loss: 99.997%).
The gas temperatures in the reactor core were measured for the gasifier fueled on liquid
ethanol and fueled on pulverized-coal slurry, at 4.4atm and 11.2atm, respectively. The
TDL measured temperature results (Figure 4.11) showed a higher (by ~ 300K)
temperature for the coal gasification, in good agreement with the upstream and
downstream thermocouples installed at the wall of the reactor. However, due to the large
heat capacity of the wall and refractory, the thermocouple readings were very stable and
did not capture transient changes of the gas phase temperature. The temperature
variations in Figure 4.11 suggest that the location of the coal reaction zone may be
oscillating in the reactor.
63
0 60 120 180 240600
900
1200
1500
1800
2100
Tem
pera
ture
[K
]
Time [s]
4.4 atm, ethanol (Transmission: 0.3%)
Upstream TC readings
Downstream TC readings
11.2 atm, coal (Transmission: 0.02%)
Upstream TC readings
Downstream TC readings
Figure 4.11 Temperature in reactor core determined by laser absorption (~1s time
resolution) and by thermocouples in the walls.
Time-resolved temperature measurements in the pre-quench location (location 2) during
the coal gasification process are shown in Figure 4.12. For reactor pressures up to 12atm,
the temperature was relatively constant, indicating stable gasifier operation. However,
for some experiments, unstable behavior in the laser measured temperature was observed
when the reactor was operated at 15atm. It was later discovered that for these operating
conditions the oxygen supply system became unstable at the flow rate required to support
the 15atm gasifier operation. The oxygen supply problem was noticed by the facility
operators several minutes after the unstable temperature was identified by the laser
absorption sensor. The fast time-response of the absorption sensor thus allowed
successful capture of the unstable reactor operation due to these flow instabilities.
64
0 30 60 90 120 150 1800
200
400
600
800
1000
Unstable T
Time[s]
Tem
pera
ture
[K
]
7 atm
9 atm
12 atm
15 atm
Figure 4.12 Temperature determined by laser absorption in the pre-quench location with
a ~1s time resolution (data shown for four reactor pressures). Note unstable T observed
at 15atm caused by fluctuations in oxygen supply.
4.4.2 Gasifier product syngas-stream measurements
An important goal for these gasifier measurements was to monitor the heating value of
the product syngas. Although the beam scattering from the coal particles at the post-
quench location (location 3) was less severe, more than 99.9% transmission loss was still
observed due to window fouling by unreacted coal particles. Much of this window
fouling occurred during the transition of the gasifier from alcohol to coal fuel. Even with
these dirty windows, successful measurements were made for gas temperature and H2O
concentration at this location as shown in Figure 4.13. The measured temperature was
consistent with the upstream and downstream thermocouple readings, and the measured
H2O mole fraction remained close to the saturated H2O mole fraction calculated by the
average temperature and pressure during the measurement period. The agreement of laser
absorption measurements for gas concentrations and temperature confirm the potential of
65
TDL sensors for accurate monitoring of the syngas LHV. A more detailed discussion
about measuring the syngas LHV is described in Ref. [70].
0 200 400 6000
100
200
300
400
P = 8 atm
H2 O
mo
le fra
ctio
n [%
]
Te
mp
era
ture
[K
]
Time [s]
Measured temperature
Upstream TC readings
Downstream TC readings
0
4
8
12
16
20
Measured H2O mole fraction
Saturated H2O mole fraction
Figure 4.13 Measured gas temperature and H2O mole fraction by the TDL sensors with a
~2.5s time resolution and thermocouple temperatures in the gasifier product-syngas
stream
4.4.3 Comparison between WMS-2f and 4f in high-pressure and noisy
environment
Advantages of using WMS-higher harmonics (e.g., WMS-4f ) to decrease the
measurement uncertainty from WMS background signals and interference from
neighbors have been previously reported for benchtop laboratory measurements [56].
These advantages are expected to be especially attractive for high-pressure WMS
absorption measurements, where the background signals and interference from neighbors
can be large. The performance of 2f and 4f signals, together with a large modulation
depth (a=0.96cm-1
, f=10kHz) was evaluated in the reactor core measurements when the
gasifier was fueled on coal-slurry at 18 atm. Figure 4.14 shows the measured 1f-
normalized WMS-2f and 4f signals along with simulations.
66
0.00
0.05
0.10
0.15
0.20
0.25
0.300.00
0.05
0.10
0.15
0.20
0.25
0.30
(b) Laboratory measured background subtracted
(a) Gasifier measured background subtracted
P = 18 atm
718871877186718571847183
Wavelength [cm-1]
1f-
no
rmaliz
ed
WM
S-2
f (4
f )
Measured 2f
Measured 4f
Simulated 2f
Simulated 4f
0.64cm-1
1cm-1
Figure 4.14 Measured and simulated 1f-normalized WMS-2f and 4f spectra of 7185.6
cm-1
transition in the reactor core. (a) Measured WMS signal with the background signal
measured with N2 in the gasifier subtracted; (b) measured WMS signal with the
background signal measured in the laboratory prior to the gasifier experiments subtracted.
(25% H2O in syngas flow, P=18atm, T=1620K, L=12.5 cm, transmission loss: 99.99%,
a=0.96 cm-1
, f=10kHz)
With the use of the same modulation depth, the full-width at half-maximum of the 4f
absorption lineshape was about 36% narrower than that of the 2f lineshape,
demonstrating less interference to (from) neighboring features. A 10% change was
observed in the WMS-2f peak signals when background signals subtracted were based on
a laboratory measurement versus a background measurement made in the gasifier when
filled with nitrogen. However, for the 4f signal the change was less than 2.5%. A
practical gasifier operates continuously without opportunity for a WMS background
measurement, and thus the 4f approach may reduce the uncertainty in the background
signals compared to 2f. Of course this advantage is partially mitigated by the smaller
peak magnitude of 4f signal relative to the 2f signal. Considering that the 4f peak
wavelength is very close to the 2f peak wavelength, a strategy using scanned-wavelength
67
WMS to acquire both the 2f and the 4f peak signals may provide the optimum approach,
and of course, both the 2f and 4f signals could be evaluated in near real-time to check for
consistency and optimum data reduction strategy.
68
69
Chapter 5 Novel Strategy for calibration-free
wavelength-scanned WMS analysis
5.0 Motivation
The WMS absorption signal is the product of terms proportional to absorbance and the
lineshape function of the absorbing transition convoluted with the modulation. Thus, the
analysis of WMS absorption is more complex than DA as discussed in previous chapters.
The simulation of the WMS absorption signal introduced in Chapter 3 uses the Beer-
Lambert relation to combine an analytic model of the modulation of wavelength (and
intensity) with a simulated absorption spectrum to calculate WMS-nf harmonics via
Fourier expansion. Unfortunately, injection-current-modulated TDLs have simultaneous
intensity modulation requiring an additional analytic model of the TDL intensity versus
time. Thus, the Fourier expansion becomes quite complicated when the simultaneous
modulation of laser intensity and wavelength are combined with a realistic absorption
lineshape.
Such difficulties are even more pronounced for wavelength-scanned WMS
where the laser-dynamics of injection-current-tuned TDLs cannot be accurately described
by a Fourier series of a single modulation frequency [71].
However, wavelength-scanned WMS is crucial for practical TDL sensors. A
wavelength-scanned approach is needed to avoid problems of a drift of the mean laser
wavelength with time, where the mean wavelength is defined as the center wavelength of
the modulating laser. Such drifts move fixed-wavelength WMS measurements off the
line center of the transition. Without independent calibration or wavelength monitoring,
such drifts produce unacceptable uncertainty for a practical fixed-wavelength WMS
70
sensor. In addition, wavelength-scanned WMS can be used to measure velocity and/or
pressure (via transition lineshape).
In this chapter, a new comprehensive and accurate approach to analyze wavelength-
scanned WMS absorption signals at all the harmonics of the modulation frequency is
presented. To distinguish the modulation frequency from the scanned frequency fs, the
modulation frequency will be denoted as fm in this chapter. The new method differs from
previous WMS analysis strategies in two significant ways: (1) the measured intensity
versus time of the wavelength-scanned (at frequency fs), wavelength-modulated (at fm)
laser light is used to simulate the transmitted laser intensity versus time, and (2) digital
lock-in and low-pass filter software is used to expand the time series of simulated and
measured transmitted laser intensity into harmonics of the modulation frequency, WMS-
nfm (n=1,2,3,…). Using the measured laser intensity of the scanned and modulated laser
avoids the need to develop an analytic model to describe the variation of laser intensity
versus time. However, the wavelength variation of the scanned and modulated laser
versus time is characterized prior to measurements, similar to the traditional analysis.
The use of the lock-in and filter software to expand the WMS signal into modulation
frequency harmonics avoids the difficulties in the Fourier expansion of the time-varying
laser intensity. The new analysis scheme is valid at any optical depth, and modulation
index, and at all values of the mean laser wavelength. This enables the WMS-nfm
lineshape to be fit to determine gas properties such as absorber concentration,
temperature, and pressure via the integrated absorbance (demonstrated in chapter 6).
The details of the analysis scheme are described in this chapter in the context of a
demonstration experiment to measure WMS absorption of H2O dilute in air, in a cell at
room temperature and atmospheric pressure. First, in section 5.1, an overview of the
measurement and simulation of WMS-nfm is provided. Then, in section 5.2, the details of
71
the experiment and data analysis are discussed. The laser characterization and
measurements conducted prior to the WMS experiment are described and examples of the
data for the demonstration experiment are used for illustration. The absorption spectrum
of the target species is then simulated and combined with the measured transmitted laser
intensity without absorption and laser wavelength characterization data to simulate the
transmitted laser intensity with absorption using the Beer-Lambert relation. Then a digital
lock-in software is used to expand both measured and simulated transmitted intensity into
the WMS-nfm harmonics for n=1 to 6. Finally, the use of 1fm-normalization is shown to
account for non-absorption losses. The measured and simulated lineshapes agree for 1fm-
normalized WMS-nfm signals for n=2-6 without any calibration or adjustable parameters,
providing the ability to use this new WMS analysis scheme for calibration-free extraction
of gas parameters from best-fit analysis of WMS lineshapes.
Compared to past WMS analysis strategies, this new method is much easier to
implement. The use of the digital lock-in and low-pass filter software to extract the
WMS-nfm harmonics from the simulated transmitted intensity avoids the complex Fourier
expansion of the simulated absorption of the scanned and modulated laser intensity and
wavelength. This new scheme is valid for all WMS-nfm harmonics, at any optical depth,
and at all values of the mean laser wavelength even in the wings of the absorption away
from line center.
5.1 Overview of a WMS absorption experiment/simulation
The WMS experiment and data analysis consists of five steps: (1) Measurement of the
transmitted intensity of the scanned and modulated laser through the target gas sample,
(2) characterization of the laser wavelength and measurements of the transmitted laser
intensity without absorption versus time in response to the time-varying laser-injection
72
current, (3) simulation of the absorption spectrum using the characterization data and the
Beer-Lambert relation to calculate simulated transmitted intensity versus time, (4)
expansion of the simulated and measured transmitted laser intensity versus time into
harmonics of the WMS signal using the same lock-in and low-pass filter software, and
(5) normalization of the harmonics of the WMS signal by the 1fm-harmonic exploiting the
intensity variation of injection current modulated TDLs to account for non-absorption
losses. The measurement and the characterization can be performed in either order,
afterwards the other three steps must occur in the order listed. The absorption signal can
then be determined by two ways: (1) the peak signals of the measured WMS absorption
spectrum as discussed in previous chapters, (2) fitting the measured WMS absorption
lineshape by varying the integrated absorbance and collisional width, analogues to the
scanned DA measurement. The second way will be demonstrated in detail in the next
chapter.
5.1.1 Transmitted intensity measurement:
The measurement of the transmitted intensity of a scanned and modulated laser is
illustrated in the diagram of Figure 5.1. The TDL injection current is rapidly modulated at
frequency fm superimposed upon a slow scan of the mean injection-current of the
modulated laser at frequency fs. In this paper superscript M and S will distinguish the
time-dependent measured transmitted intensity M
It(t) from the simulated transmitted
intensity SIt(t) by superscripts, where the subscript “t” denotes transmission through the
absorbing gas, and the subscript “0” will denote the intensity measured without absorber
present M
I0(t). Ideally the intensity versus time would be measured with absorber M
It(t)
and without absorber M
I0(t); however some applications do not lend themselves to an in
situ measurement without absorber, and the M
I0(t) must be inferred from the optical
system (absorption free) as close as to the applications. The simplest optical system of
negligible absorption is to directly connect the laser to the detector.
73
Figure 5.1 Measurement step: Determine measured transmitted intensity versus time
with absorber M
It(t) and without absorber M
I0(t).
5.1.2 Intensity and wavelength modulation characterization
The laser intensity versus time without absorbers M
I0(t) is measured at the same digitizer
rate as the WMS measurements avoiding the need to model the laser intensity response
with injection current. The best intensity versus time characterization is the in situ
measurement using the application test volume evacuated or purged of the absorbing gas
as this accounts for any wavelength-dependent optical components. However, for some
practical implementations, an in situ absorption-free background measurement of laser
intensity is not possible; successful measurements have been performed in such
applications by laboratory characterization of the intensity including as many of the field
measurement optics and windows as possible. When the actual field measurement is
performed, care in alignment and set up is taken to minimize any wavelength-dependent
transmission. The dominant time variation of the laser intensity is usually produced by
the time-varying injection current, however additional time-varying intensity
contributions can arise from the wavelength tuning (and modulation) if any optics or
windows have wavelength-dependent transmission (e.g., a material near the edge of its
transmission range, or more likely interference from components with parallel surfaces
(etalons)).
74
The performance of the laser tuning is characterized to determine an analytic expression
for the laser wavelength versus time ( )t as the injection current of the laser is driven by
a combination of the scan and modulation frequencies, as shown in Figure 5.2. ( )t is
determined from a combination of measurements using an etalon and absorption
transitions of known wavelength as a function of injection current, and these data are fit
to a model of the wavelength tuning.
Figure 5.2 Wavelength characterization versus time ( )t and intensity versus time M
I0(t)
of a wavelength-scanned, wavelength-modulated laser including wavelength-dependent
transmission along the measurement path without absorption.
5.1.3 Simulated transmitted laser intensity:
The simulation of the transmitted laser intensity is illustrated in the flow chart in Figure
5.3. First a spectral database such as HITRAN or HITEMP is used to determine the
absorption spectrum ( ) near the target transition for an approximate gas composition
illustrated in the figure as initial guesses, a prelude to iterative fitting of the WMS
lineshapes; the collision-broadening coefficients in the database are used to estimate the
C of the transitions scanned. Note that the WMS analysis scheme developed here can
be used for isolated transitions or multiple transitions even if they are unresolved and/or
blended by collision broadening. The characterization of the laser tuning versus time
( )t is used to convert the absorbance spectrum ( ) to an absorbance time series
75
( ( ))t . The non-absorption laser intensity versus time M
I0(t) is combined with the
absorbance time series ( ( ))t using the Beer-Lambert relation to calculate the simulated
transmitted laser intensity versus time SIt(t) for the same time steps as the measured
transmitted intensity, M
It(t).
Figure 5.3 Simulation of transmission intensity versus time of a wavelength-scanned
wavelength-modulated laser through a simulated absorption spectrum.
5.1.4 Lock-in analysis
Both the measured transmitted intensity versus time M
It(t) and the simulated transmitted
intensity versus time SIt(t) are processed with a digital lock-in and a low-pass filter to
isolate the WMS signals at the harmonics of fm (shown in Figure 5.4). The lock-in
analysis of the simulated transmitted intensity extracts the WMS-nfm harmonics while
avoiding the complex mathematics of a Fourier expansion of simultaneous wavelength
and intensity modulation. Using the same lock-in analysis of the simulated and measured
transmitted intensity avoids explicitly evaluating these background signals as their
76
contribution is equally included in the simulated and measured WMS-nfm harmonics
signals. Because these background signals are proportional to the laser intensity, the
normalization of the WMS harmonics by the WMS-1fm signal can be performed without
explicit background corrections.
Figure 5.4 Use of a digital lock-in and low-pass filter to expand the time series of
measured or simulated laser intensity into the WMS-nfm harmonic signals.
5.1.5 Normalization to account for non-absorption losses
All of the harmonics of the WMS signal are proportional to the laser intensity. For
optically thin conditions, the WMS-1fm is dominated by the injection-current modulation
and it has long been recognized that other WMS harmonics could be 1fm-normalized at
line center to account for non-absorption losses in transmitted laser intensity. The
wavelength-scanned WMS-1fm signal has a large contribution from the laser intensity
modulation and contributions from gas absorption with lineshapes asymmetric
(“dispersion-like”) and symmetric (absorption) with respect to the transition line center.
At line center the “dispersion-like” contribution vanishes and 1fm-normalization of
WMS-nfm harmonics is easily understood. However, the magnitude of the asymmetric
77
contribution is also proportional to laser intensity, and although normalization by the
wavelength-scanned WMS-1fm distorts the WMS-nfm lineshapes this normalization can
still be used to account for non-absorption losses in transmitted laser intensity as the
distortion is identical for measurement and simulation of the WMS signals.
5.2 Example analysis of WMS absorption detection of H2O
A step-by-step discussion of the measurement and analysis of WMS absorption for an
example problem, H2O dilute in air, provides the context for a detailed description of this
new WMS analysis scheme.
5.2.1 Transmitted intensity measurement for WMS detection of H2O
The example experiment was conducted with known amounts of H2O dilute in air at
atmospheric pressure as illustrated in Figure 5.5. A DFB laser (NEL) near 1392 nm with
single-mode fiber output was used to probe the H2O transition near 7185.6cm-1
.
Computer-driven outputs (National Instruments PCI-6110) controlled the diode laser
injection current (ILX Lightware LD-3900). The injection current was modulated with a
sine function at fm=10 kHz superposed on a linear scan fs=25 Hz. The light exiting the
fiber was collimated into a beam, directed through a gas cell with wedged windows to
avoid etalon interference in the transmission of the wavelength-scanned (and modulated)
light. The transmitted light was then focused onto a near-infrared (NIR) photo-diode
detector (Thorlab PDA-10CS, bandwidth: 775 kHz at 30db gain). The laser intensity
signal was sampled (same PCI-6110 card, 12 bits) at a rate of 2.5 MHz. The optical path
external to the cell was purged with pure N2 to eliminate the absorbance in the ambient
environment. The measured transmitted intensity versus time M
It(t) was acquired for H2O
dilute in air in the cell.
78
Figure 5.5 Schematic of the experimental setup for measuring the transmitted laser
intensity versus time for WMS detection of H2O in a gas cell.
5.2.2 Laser characterization for WMS detection of H2O
The laser intensity versus time M
I0(t) including any losses or wavelength variation in the
optical components was then measured in the evacuated cell as illustrated in Figure 5.6.
Characterizing the laser wavelength tuning ( )t is more complex. First the scan-current
of the modulating laser was recorded for the peak WMS-2fm signal from a selected
transition. The known position of this transition was used to calibrate the absolute
wavelength for the scan of the modulating laser. The wavelength tuning of the
modulated laser around this calibration point was measured using a fiber input/output
etalon with 0.02 cm-1
FSR (Micron Optics).
( )t is modeled as: ( ) cos(2 ) ( )mt a f t F t (5.1)
where is the laser wavelength (or laser frequency) without injection-current tuning,
the phase of the frequency modulation (here -2.048 radian), ( )F t the function
describing the wide near-linear scan of the mean laser wavelength, expressed here as a
79
fourth-order polynomial. The measured frequency-tuning response is shown in Figure
5.7 and the best-fit result for the specific laser used in the demonstration was:
1 2 3 2 5 3 6 4
4
( )[ ] 7182.159 1.3775 10 2.4977 10 1.1702 10 1.3699 10
0.081 cos(2 10 2.0483)
t cm t t t t
t
(5.2)
Figure 5.6 Measured laser intensity versus time in the absence of the absorber (scan rate
= 25Hz, scan amplitude = 2V, modulation frequency = 10 kHz, modulation amplitude =
0.1V).
5.2.3 Simulated transmitted laser intensity for WMS detection of H2O
Single mode DFB TDLs have very narrow (~5MHz) linewidth and thus the Beer-
Lambert relation describes the simulated transmitted intensity versus time in terms of the
incident intensity and absorbance:
0( ) ( ) exp[ ( ( ))]S M
tI t I t t , (5.3)
80
where M
I0(t) is the intensity versus time of the modulated and scanned laser, measured
when the gas cell was empty (vacuum), ( )t , the laser wavelength-tuning characteristics
including scan and modulation, and ( ( ))t , the absorption spectrum of the target gas in
the region of the laser wavelength scan.
0.018 0.020 0.022 0.024
7185.0
7185.5
7186.0
Fre
quency [cm
-1]
0.0206 0.0208 0.0210
7185.51
7185.60
7185.69
Time [s]
Etalon peaks
Fit (4th order poly. + cos)
Figure 5.7 Measured frequency response to the laser injection-current tuning and its best
fit (same modulation configuration as Figure 5.6)
Water vapor in this example is measured by scanning the laser over a pair of transitions,
one near 7185.60 cm-1
(lower state energy, E" = 1045cm-1
) and its neighbor near 7185.39
cm-1
(E" = 447cm-1
). These transitions are overlapped by collisional broadening at
atmospheric pressure and provide a good test of the new WMS analysis method to
recover the complex WMS lineshape for a pair of unresolved transitions. To accurately
simulate the absorption spectrum for this example, the spectroscopic parameters
81
including the line strength, H2O-H2O broadening (2γself), and H2O-Air broadening (2γair)
coefficients at 296K were measured; the data are listed in Table 5.1.
Table 5.1 Measured spectroscopic parameters for probed H2O transition near 7185.60
cm-1
and its neighbor near 7185.39cm-1
at 296K..
Parameter 7185.60cm-1
7185.39cm-1
S (atm-1
cm-2
) 0.0195 0.00121
2γself (cm-1
/atm) 0.410 0.792
2γair (cm-1
/atm) 0.088 0.162
The absorption spectrum over the scan range shown in Figure 5.8 was simulated with a
Voigt lineshape at the measurement conditions (P = 1 atm, T = 296 K, L = 100.5 cm,
0.75% H2O in air).
7185.0 7185.5 7186.0 7186.50.00
0.02
0.04
0.06
0.08
0.10
0.12
0.75% H2O in air
T = 296 K
P = 1 atm
L = 100.5 cm
Absorb
ance
Frequency [cm-1]
Simulation (Voigt)
Figure 5.8 Simulated absorbance versus frequency for H2O transition near 7185.6cm-1
and its neighbor at 7185.39cm-1
(0.75% H2O in air ,P = 1 atm, T = 296 K, L = 100.5 cm).
82
Using ( )t from Eqn. 5.2 and the simulated absorption spectrum in Figure 5.8, the
absorbance can be written as a function of time ( ( ))t for the scanning and modulating
laser. For this example ( ( ))t for a single laser scan is illustrated in Figure 5.9; note that
the flat top of the absorbance is not saturation but is the modulation of the laser as its
wavelength traverses the peak of the absorption seen in Figure 5.9. The 25 Hz
wavelength scan is slow enough compared to the 10 kHz modulation that the modulated
wavelength includes the line center of the transition more than twenty times during the
scan. Using Eqn. 5.3 to combine the data for laser intensity versus time M
I0(t) from Figure
5.6 with the absorbance versus time in Figure 5.9, ( ( ))t , the simulated transmitted
intensity versus time SIt(t) can be calculated as shown in Figure 5.10.
0.0150 0.0175 0.0200 0.0225 0.02500.00
0.02
0.04
0.06
0.08
0.10
0.12
Absorb
ance
Time [s]
Simulation
Figure 5.9 Simulated absorbance versus time ( ( ))t for the H2O transition near
7185.6cm-1
(for the absorbance versus wavelength shown in Figure 5.8). Note constant
peak values between 0.021 and 0.0225s are real (not detector saturation) as the
modulation is fast compared to the scan rate.
83
0.018 0.021 0.0241.0
1.5
2.0
Tra
nsm
itte
d laser
inte
nsity [V
]
Time [s]
Simulation
Figure 5.10 Simulated transmitted laser intensity versus time ( )s
tI t for a single scan of
the modulated laser over the absorption feature (For the laser intensity in Figure 5.6 and
the absorbance versus time in Figure 5.9).
5.2.4 Lock-in analysis for WMS detection of H2O
The measured and simulated transmitted laser intensities versus time were both
numerically post-processed by using lock-in and finite-impulse-response (FIR) low-pass
filter (bandwidth of 2 kHz) software. The transmitted intensities M
It(t) and SIt(t) are each
multiplied by cos( 2 )n ft (and sin( 2 )n ft ) to expand the X-component (and the Y-
component ) of the measured and simulated signals at each of the nfm harmonics. A low-
pass (LP) filter was used to extract these components by taking convolution, as:
84
: ( ) cos( 2 )
: ( ) cos( 2 )
S S
nf t m
M M
nf t m
X I t n f t LP filter
X I t n f t LP filter
(5.4)
: ( ) sin( 2 )
: ( ) sin( 2 )
S S
nf t m
M M
nf t m
Y I t n f t LP filter
Y I t n f t LP filter
(5.5)
Note the FIR low-pass filter bandwidth must be less than fm /2 to avoid distorting the
modulation and large enough to avoid distorting the wavelength-scanned WMS lineshape
features (depends on scan range and rate, transition width and digitization rate). Because
the simulated and measured signals use the same filter, any distortion will be common to
both quantities. The absolute magnitude of the simulated and measured WMS-nfm
signals becomes:
2 2
2 2
S S S
nf nf nf
M M M
nf nf nf
S X Y
S X Y
. (5.6)
Figure 5.11 shows the comparison between the simulated and measured WMS-nfm
signals (the time-cost for simulating each harmonic is ~ 0.1s with a desktop computer
(Dell XPS8500, CPU: i7-3770, 8GB ram); the shapes of each harmonic signal WMS-nfm
agrees well between simulation and measurement but there is a significant difference in
magnitude. The laser intensity for this demonstration measurement was reduced by a
fiber attenuator added to simulate non-absorption laser intensity loss. The difference in
magnitude between simulation and measurement in Figure 5.11 with the fiber attenuator
illustrate how each of the harmonic signals is proportional to laser intensity. The use of
the attenuator to mimic non-absorption intensity loss provides an illustration of the use of
WMS-1fm-normalization of the other WMS-nfm harmonic signals to account for time-
varying laser intensity or non-absorption losses. Without the fiber attenuator there was
85
no such difference between simulation and measurement illustrating the fidelity of this
new scheme to simulate WMS-nfm harmonics.
Figure 5.11 Measured and simulated WMS-nfm spectra for H2O transition near 7185.6
cm-1
. (0.75% H2O in air, T = 296 K, P = 1atm, L = 100.5 cm, a = 0.081 cm-1
, fm = 10 kHz,
optical depth = 0.101). Note the amplitude difference was produced by attenuating the
measurement laser intensity to mimic the influence of non-absorption losses.
5.2.5 Issues for normalization by WMS-1fm
The use of WMS-1fm to normalize the higher harmonics WMS-nfm (n≥2) is well
developed in the literature at line center [44]. However, the normalization of
86
wavelength-scanned WMS has not previously been discussed. As seen in Figure 5.11,
the wavelength-scanned WMS-1fm signal is asymmetric with-respect-to the line center of
the transition. Thus normalization will distort the WMS lineshapes; however, if this
distortion is common to both simulation and measurement, quantitative fitting of the
WMS lineshapes to determine gas parameters will still be possible [71]. As first noted by
Cassidy and Reid [45] and described in detail in ref [73,74], the WMS-1fm signal has
contributions from the laser intensity modulation and from the absorption by the target
species. Here the highlights of the wavelength-scanned WMS-1fm lineshape are
described to understand its use for WMS-nfm normalization.
Figure 5.12 shows the wavelength-scanned WMS-1fm signal for the H2O absorption
demonstration at three optical depths (0.01, 0.1, and 1). For the optically thin case (0.01)
the 1fm-signal shown by the solid black line in the left panel of Figure 5.12 is dominated
by the amplitude of the laser intensity modulation. The value of the laser intensity
modulation contribution to the WMS-1fm signal depends on the modulation depth and the
laser intensity variation versus injection current; larger modulation depth produces a
larger WMS-1fm value at scan wavelengths away from the absorption transition.
Figure 5.12 Simulated 1fm spectra for H2O transition near 7185.6 cm-1
at optical depths
0.01, 0.1, and 1.0 at 1 atm with a modulation index of 1.8 (the laser characterization is the
87
same as Figure 5.6 and Figure 5.7; note the modulation index is 1.8, and the line center
includes pressure shift in 1 atm air).
As the optical depth increases the gas absorption contribution to the WMS-1fm signal
increases and the blue dashed line in the left panel of Figure 5.12 illustrates an optical
depth of 0.1 (same as the experiment described in detail above and shown in Figure 5.11).
The asymmetric shape of the absorption contribution to the wavelength-scanned WMS-
1fm signal becomes apparent. This asymmetric shape traces the first derivative of the
absorption lineshape from Figure 5.8.
For large optical depths attenuation of the WMS-1fm near line center must also be
considered as illustrated by the inset in Figure 5.12. The differences in the 1fm signal at
line center arises from the absorption of the laser intensity modulation contribution to the
1fm signal, which increases with optical depth and decreases with wavelength modulation
depth.
When the optical depth is increased even further, for example the red dotted line in the
left panel of Figure 5.12, the absorption contribution to the 1fm signal becomes larger
than the contribution from the laser-modulation amplitude. Because WMS-nfm harmonic
signals are defined in Eqn. 5.6 to always be greater than zero the 1fm signal does not go
negative, but it does approach near zero values at two wavelengths in the scan.
Normalization by dividing by a signal with near-zero values, such as seen in the red dot-
dashed curve in Figure 5.12, can artificially weight only a few points in a fit of the 1fm-
normalized WMS-nfm lineshapes. For wavelength-scanned WMS at moderate optical
depths, especially at large wavelength modulation depth, this issue can be avoided by
normalization using the value for the mean WMS-1fm for each scan.
88
5.2.6 Normalization to account for non-absorption transmission losses
Even though the odd harmonics are asymmetric about the transition line center, all of the
WMS-nfm harmonics are proportional to laser intensity; thus the WMS-1fm signal can be
used to normalize signals at other harmonics WMS-nfm to account for non-absorption
losses:
1/nf
normalized
nf fS S S (5.7)
Figure 5.13 shows the WMS-1fm-normalized values of WMS-nfm (n=2-6) for the
measured and simulated data from Figure 5.11. The absorption component of the
wavelength-scanned WMS-1fm signal in Figure 5.11 is asymmetric with-respect-to the
line center; thus, the 1fm-normalized wavelength-scanned WMS-nfm lineshapes in Figure
5.13 are distorted. The peak-signal for even harmonics of 1fm-normalized WMS-nfm
(n=2, 4, & 6) are no longer at the line center of the transition (note the WMS-2fm peak
signal is used to calibrate the laser wavelength scale and not the peak normalized signal).
The WMS simulation scheme has the same normalization distortion as the measurement
and thus, the simulations can be best fit to determine gas parameters. The differences
between simulations and measurements are also plotted as residuals in Figure 5.12, and
these values show agreement (defined as the ratio of the root mean square of the residual
to the peak 1f-normailized WMS-nf signals) within 3.2% for all harmonics from 2-6. The
differences between simulation and measurement are dominated not by the differences in
peak values, which differ by <1%, but are dominated by quite small differences in the
wavelength scale between simulation and measurement. The fidelity of the
demonstration measurement was limited by the wavelength scale and the digitizer time
resolution. This demonstrates that the 1fm-normalized WMS analysis provides an
accurate WMS-nfm lineshape for n=2-6. Note the relative magnitude decreases as the
harmonic increases. Additional digital resolution and perhaps an absorption lineshape
89
more accurate than Voigt are needed to recover the lineshapes for WMS-nfm harmonics
for n>6; for this demonstration experiment a 12 bit digitizer was used with a Voigt
lineshape.
Figure 5.13 Measured and simulated 1fm-normalized WMS-nfm spectra for H2O
transition near 7185.6 cm-1
. (same condition as Figure 5.11, optical depth = 0.101).
5.3 Comparison with Fourier analysis of WMS
The new WMS analysis method compares well with the established Fourier analysis
approach as illustrated in the Figure 5.14, which compares the H2O demonstration
measurement of the lineshape for 1fm-normalized WMS-2fm to that simulated using the
new approach and that simulated using traditional Fourier analysis where the intensity
modulation is characterized only at transition line center [44,48]. For this experiment
with a relatively small modulation depth, a DFB single-mode laser with highly linear
0
0.2
0.4
0.6
2f m
/1f m
0
0.2
0.4
3f m
/1f m
0
0.1
0.2
0.3
4f m
/1f m
0
0.05
0.1
0.15
5f m
/1f m
0
0.02
0.04
0.06
0.08
6f m
/1f m
7185.4 7185.6 7185.8-0.05
0
0.05
Frequency [cm-1
]
Resid
ual
7185.4 7185.6 7185.8-0.05
00.05
Frequency [cm-1
]
Resid
ual
7185.4 7185.6 7185.8-0.02
00.02
Frequency [cm-1
]
Resudual
7185.4 7185.6 7185.8-0.02
00.02
Frequency [cm-1
]
Resid
ual
7185.4 7185.6 7185.8-0.02
00.02
Frequency [cm-1
]
Resid
ual
Measurements
Simulations
Residual
4fm
/1fm
6fm
/1fm5f
m/1f
m
3fm
/1fm
2fm
/1fm
90
response to injection current was used, and the two simulations are in good agreement
with the measurements for the entire center lobe of the lineshape, and only disagree in the
wings by ~10%. However, in the wings on either side of line center, the new analysis
approach is in better than 1% agreement with the measurement for the entire lineshape.
The match in simulation and measurement over the entire lineshape suggests the potential
of fitting the 1fm-normalized, wavelength-scanned WMS-nfm lineshapes analogous to
wavelength-scanned direct absorption, and this fitting strategy is demonstrated in the next
chapter for H2O absorption sensor in an engineering-scale high pressure coal gasifier.
Figure 5.14 Comparison of the 1fm-normalized WMS-nfm spectra using different
absorption analysis approaches for H2O transition near 7185.6 cm-1
. (same condition as
Figure 5.13)
7185.3 7185.6 7185.90
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Frequency [cm-1]
1f m
-norm
aliz
ed W
MS
-2f m
[unitle
ss]
Measurement
Simulation by approach in this study
Simulation by Fourier analysis
91
Chapter 6 H2O absorption sensor using
calibration-free wavelength-scanned WMS
fitting strategy in an engineering-scale high
pressure fluidized coal gasifier
6.0 Motivation
The use of fixed-wavelength WMS for monitoring synthesis gas (called here syngas)
composition and temperature in a pilot-scale entrained-flow high-pressure coal gasifier
has been demonstrated in chapter 4, at locations from the gasification reactor to the
syngas output stream. Here we improve the WMS through the use of wavelength-
scanning, enabling monitoring of the moisture content in the syngas output of an
engineering-scale transport reactor high-pressure coal gasifier. Compared to the fixed-
wavelength WMS technique used for previous applications, the wavelength-scanned
WMS technique is less affected by drifts in the laser wavelength or pressure shifts in the
absorption transition center wavelength. In addition, wavelength-scanned WMS
simultaneously determines the absorption lineshape and thus minimizes the errors due to
the variations of the transition collisional width caused by the change of gas composition.
Thus, the wavelength-scanned WMS approach demonstrated in this chapter is more
suitable for long-term (days) calibration-free monitoring in complex gas flow, where the
92
laser wavelength may shift and variations in the collisional width of the absorption
transition due to changes in the gas composition are hard to estimate.
In this chapter, we first demonstrated in a laboratory static cell that fitting the measured
wavelength-scanned WMS H2O absorption spectrum using a DFB laser near 1352nm can
yield simultaneously the integrated absorbance and C of the probed transition, over a
pressure range from 3 to 16atm. The H2O sensor was then tested in an engineering-scale
transport reactor high pressure coal gasifier at National Carbon Capture Center in
Wilsonville, AL, where continuous monitoring of the moisture content in the output
syngas was demonstrated for more than 27 days.
6.1 Laboratory validation experiment
A transition near 7394.84cm-1
(lower state energy, E" ~ 744cm-1
) was selected to probe
H2O mole fraction in the gasifier output (see Figure 6.1). The absorption feature is in fact
two closely spaced transitions with the same E" positioned at 7393.79cm-1
and
7393.84cm-1
. The laboratory measured spectroscopic data [32] and HITRAN database
suggest this line pair can be considered as a single transition with one set of collision-
broadening parameters at high pressure. The spectroscopic data for this transition are
listed in table 1. This transition pair was selected for four reasons: (1) the absorption is
strong at the measurement temperature (near 600K), providing good signal-to-noise ratio
in gasifier environment; (2) the transition is well isolated from neighbors at elevated
pressures; (3) the 2f/1f absorption signal is not sensitive to temperature changes in the
region near 600K; (4) and the collisional width is small, providing narrow lineshape to
enhance WMS signals.
93
Table 6.1 Laboratory measured spectroscopic parameters (linestrength, collisonal
broadening coefficients and their temperature dependence exponents) at 296K for the
target transition
Linecenter
[cm-1
]
Linestrength
[cm-2
atm-1
]
Lower state
energy [cm-1
]
2γH2O-H2O
[cm-1
atm-1
]
(nH2O-H2O)
2γH2O-CO2
[cm-1
atm-1
]
(nH2O-CO2)
2γH2O-CO
[cm-1
atm-1
]
(nH2O-CO)
2γH2O-N2
[cm-1
atm-1
]
(nH2O-N2)
2γH2O-H2
[cm-1
atm-1
]
(nH2O-H2)
7393.84 0.0690 744 0.530
(0.650)
0.139
(0.957)
0.119
(0.620)
0.088
(0.575)
0.082
(0.463)
7392 7394 73960.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Target transitionXH2O
= 7.5%
P = 10 atm
T = 600 K
L = 20 cm
A
bsorb
ance
Frequency [cm-1]
Simulated absorbance
Figure 6.1 Simulated absorption spectrum for H2O molecule at typical gasifier conditions
To validate the H2O sensor in the laboratory, absorption near 7394cm-1
was measured for
a known H2O mole fraction as a function of pressures in a 100.5cm-long cell at room
temperature (Figure 6.2). The cell was first evacuated to measure a background signal,
then filled with H2O/Air mixture at constant pressure. The mixture was premixed and
stored overnight to ensure homogeneity. A DFB laser (NEL) near 1352 nm with single-
mode fiber output was used to probe the H2O transition near 7393.8cm-1
. The laser
injection current was modulated with a sine function at 10 kHz, superposed on a sawtooth
function at 25 Hz. Computer-driven outputs (National Instruments PCI-6115) controlled
the diode laser injection current (ILX Lightware LD-3900). After collimation, the laser
beam travelled through the cell and was focused onto a NIR photo-diode detector
94
(Thorlab PDA-10CS). The voltage signal on the detector was then sampled and
numerically post-processed by using a lock-in program and a digital FIR low-pass filter
(bandwidth of 2 kHz). The line-of-sight (LOS) pathlength external to the cell was purged
by pure N2 gas to minimize absorption from ambient atmospheric moisture.
Figure 6.2 Laboratory measurement setup for validation of the wavelength-scanned
WMS strategy for high pressure gas sensing
Figure 6.3 shows the measured 1f-normalized WMS-2f spectrum at three different
modulation depths as well as the best-fit results at known conditions of P = 10atm, T =
296K. The calibrated H2O moisture mole fraction in the mixture is 0.096% and the
expected FWHM collisional width is 0.81cm-1
(2
2 H O Air = 0.081cm-1
/atm, determined by
DA measurements at sub-atmosphere pressures). The comparison reveals better than 1%
agreement, demonstrating the feasibility of using the wavelength-scanned WMS 2f/1f
strategy for simultaneous determination of absorber mole fraction and transition C at
high pressures.
Figure 6.4 presents the C , the integrated absorbances and the resulted H2O mole
fractions that best fit the measured WMS absorption spectra at pressures ranging from 3.2
atm to 15.8 atm, measured with a modulation depth of 0.8 cm-1
. The small deviation (<
1.3%) between the measurements and the expected values indicates that: (1) the fitting
95
strategy for wavelength-scanned WMS simultaneously yields the transition integrated
absorbance and collisional width, even at high pressures up to 15.8 atm; (2) the non-ideal
Lorentzian behavior, such as line-mixing and the finite-collision breakdown do not have
significant influence on the WMS-2f signals for this target transition pair and this range
of conditions.
Figure 6.3 Measured WMS-2f/1f spectra using different modulation depths and the best-
fit results (best fit parameters: for a = 0.4cm-1
, xH2O = 0.0953% ,c =0.823 cm
-1, for a =
0.6cm-1
, xH2O = 0.0947% ,c =0.820 cm
-1, for a = 0.8cm
-1, xH2O = 0.0961% ,
c =0.802
cm-1
)
7393 7393.5 7394 7394.5 73950
0.01
0.02
0.03
0.04
0.05
0.06
frequency [cm-1
]
2f/1f [unitl
ess
]
a = 0.4cm-1
a = 0.4cm-1
a = 0.6cm-1
a = 0.6cm-1
a = 0.8cm-1
a = 0.8cm-1
96
2 4 6 8 10 12 14 16
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Expected C
Measured C
C [
cm
-1]
Pressures [atm]
2 4 6 8 10 12 14 16
0.06
0.08
0.10
0.12
Inte
gra
ted
ab
so
rba
nce
[cm
-1]
0.094
0.098
H2O mole fraction
H2O
mo
le f
ractio
n [
%]
Pressure [atm]
0.02
0.04
0.06
0.08
0.10
0.12
Integrated absorbance
Figure 6.4 Best-fit results for the C (left panel) and integrated absorbance and mole
fraction (right panel) at different pressures. (T = 296K, L = 100.5cm)
6.2 Gasifier facility and measurement setup
The laser sensor performance was evaluated in an engineering-scale transport reactor
gasifier at the National Carbon Capture Center operated for US Department of Energy by
Southern Company Services in Wilsonville, Alabama, United States. A location for
optical access was selected 30 meters downstream from the exit of the particulate control
device (PCD) of the gasifier, with the laser transmitter and receiver optics located on
either side of the syngas output pipe, as illustrated in Figure 6.5, and typical conditions
are listed in Table 6.2 Typical conditions at the measurement location (the gas mixture is
balanced by N2).
Table 6.2 Typical conditions at the measurement location (the gas mixture is balanced by
N2)
Property Values
Temperature [K] 600
Pressure [atm] 15
Pathlength [cm] 20
H2O [%] 6-10
CO2 [%] 6-10
CO [%] 8-12
H2 [%] 6-10
Trace species
(H2S, NH3, etc) [%]
< 1
97
Flow rate [kg/h] 12,500
Figure 6.5 Location of the TDL sensor in the syngas process piping 30 meters
downstream of the exit of the PCD. Note the lasers and supporting electronics were
remotely located in the instrumentation shelter.
Figure 6.6 Schematic (left) and photo (right upper) of the sensor installation showing the
mounting rail hanging on the syngas pipe with redundant actuated shut-off valves,
redundant window pairs, temperature and pressure alarm for window failure, and the
TDL transmitter and receiver housings. The TDL electronics in the shelter are shown in
the right bottom panel.
98
6.3 Measurement results
The wavelength-scanned WMS strategy as validated by the laboratory measurements was
used to continuously monitor the moisture content in the syngas product flow at different
gasifier operation conditions. The facility operation began by igniting a propane/air
burner in the reactor. When the reactor was hot enough, pulverized coal was added into
the reactor to first initiate coal combustion and then begin the gasification process. The
coal-fuel feeding continued for more than 10 hours for this 1st start up attempt, but was
terminated at hour 52 to correct a problem elsewhere in the gasifier.
The moisture level recorded for hours 0-52 with 2s time resolution is shown in Figure 6.7
(although the scan rate was 25Hz, the measurements were collected only every 2s to
reduce computational and data storage requirements). The moisture mole fraction was
inferred by fitting the measured scanned-WMS 2f/1f spectrum using two variables: the
absorber mole fraction and the transition collision-broadening width (shown in Figure
6.8). Due to the severe transmission losses due to beam steering and particulate
scattering, the light transmission was typically less than 0.01%. However, the 1f-
normalized, 2f absorption spectrum remained stable, providing a signal-to-noise ratio
larger than 20 (25Hz bandwidth) in the measured WMS-2f /1f spectra of 2.3% H2O,
corresponding to a detection limit of ~0.02% H2O mole fraction for 1Hz bandwidth.
99
048
1216
Pressure (atm)
Figure 6.7 In situ measurements of exhaust gas moisture during reactor start-up
including ignition of the propane burner, switch to coal combustion with pulsed coal feed,
transition to gasification, and reactor shutdown when the coal input was terminated. The
pressure trace was provided by NCCC. The region surrounded by the red dashed
rectangle is shown in Figure 6.9.
From hour 0 to 4, several attempts to ignite the reactor burner were made, but the
propane/air flame was unstable. This abnormal process was captured by TDL
measurements as the H2O mole fraction spiked and dropped rapidly several times during
this period. At hour 8 a stable propane flame was established and the warm-up period
began. The moisture content of the syngas steadily increased at the measurement location
during warm-up; this variation was expected as the fuel/air ratio of the propane burner
was varied and as the output gas piping warmed up to eliminate condensation.
100
7393.0 7393.5 7394.0 7394.5 7395.00.0
0.2
0.4
0.6
0.8
1.0
1.22f/
1f
[unitle
ss]
Frequency [cm-1]
2.3% H2O, P = 4.5atm, T=390K
8.0% H2O, P = 8.8atm, T = 541K
13.3% H2O, P = 13.5 atm, T = 584K
Figure 6.8 Measured (dashed line) WMS-2f/1f absorption spectra and the best-fit results
(solid lines) at different gasifier operation conditions (black: heating using the
propane/air flame, blue: coal combustion, red: coal gasification). fs = 25Hz, f = 10kHz, a
= 0.78cm-1
. Pressure and temperature data were provided by NCCC.
At hour 38, propane heating ceased and the coal feeding began. The large fluctuations in
H2O mole fraction during the gasification were not caused by signal noise but by
variations in the coal feed rate. Two factors support this conclusion: (1) the independent
measurement by the reactor thermocouple were correlated with TDL moisture
measurements, as seen in the left panel of Figure 6.9; (2) as shown in the right panel, the
measured WMS 2f/1f lineshapes were free of noise and the increase of absorption at
measurement time point 2 was obviously larger than the one at point 1, indicating more
H2O molecules in the laser line-of-sight. This moisture variation corresponded to the
pulsing of the coal feed rate to slow the warm-up rate of the facility and prevent damage
to the ceramic lining of the gasifier and the output plumbing. With the 2s time resolution,
101
the TDL sensor captures the time-varying moisture in the exhaust gas due to the changing
fuel content in the reactor. Such results were not found in the GC measurement results (5
minute time resolution) made downstream for sampled CO2. This demonstrates the
advantage of in-situ TDL measurement over sampled GC measurements for time-
resolved monitoring of the transient gasification process.
42.0 42.5 43.00.04
0.06
0.08
0.10
0.12
0.14
2
1
Reacto
r tem
pera
ture
[K]
H2O
mole
fra
ction
Time [hours]
1060
1080
1100
1120
7393.0 7393.5 7394.0 7394.5 7395.00.0
0.5
1.0
1.5
2f/
1f
[un
itle
ss]
Frequency [cm-1]
Measurement
Best fit
Measurement
Best fit
Figure 6.9 Left panel: TDL monitored moisture mole fraction from hour 42 to hour 43
and the reactor temperature measured by the thermocouple; Right panel: measured WMS
2f/1f absorption spectra at point 1(the lowest moisture mole fraction in a single pulse) and
at point 2 (the highest moisture mole fraction in a single pulse)
Figure 6.10 shows the measured C after the burner was ignited and the comparison to
the expected results. The expected C collisional width was based on the GC gas
composition results, the TDL measured moisture mole fraction, and the collision-
broadening coefficients pre-measured in the lab using Eqn. (2.6) and (2.7). The
comparison reveals better than 3% agreement, which is very reasonable as the GC
measurement has 2-3% uncertainty and the collision-broadening coefficient database has
1-2% uncertainty. Note the variations in the C measurements from hour 40-52
correlate with the moisture variation due to coal feed rate, as H2O is a much more
effective collision partner for collisional broadening than other components of the syngas
102
(see Table 6.1). The precision and accuracy in the in-situ determination of the transition
collisional width demonstrates that the scanned-WMS can be analyzed without prior
knowledge C of the transition, even at the high pressures of ~ 15atm. Thus a practical
sensor can be quantitatively analyzed without calculating or estimating C , which
requires assembly of a collision-broadening coefficient database and a reasonably
accurate estimate of the gas composition.
10 20 30 40 50
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
C [
cm
-1]
Time [hours]
Measured C
Expected C
Figure 6.10 TDL measured transition C and its comparison to the expected values.
The TDL moisture sensor operated unattended from hour 140 until hour 575, and
continuous records of H2O and temperature were collected except for three short (~10
min) periods when the valves on the optical access were closed due to independent
gasifier upsets. During the entire period, the transmission of the TDL sensor remained
stable (transmission ~ 10-5
), and the windows did not exhibit any indications of fouling.
103
The syngas moisture record is illustrated in Figure 6.11 and compared with samples of
liquid water extracted from the syngas stream. These two H2O measurement methods
have similar mean values over the 435 hour record, with larger variations seen in the
liquid water sampling.
The H2O mole fraction data of Figure 6.11 varies with time by approximately ±0.002,
which might suggest that the TDL sensor has a ±3% statistical uncertainty. However,
when the H2O data are examined on a more expanded time scale, the measured H2O
content has a distinct oscillation (period of ~ 10 minutes) as illustrated in Figure 6.12.
These oscillations in H2O are strongly correlated with the reactor temperature (left panel)
and the coal-dispense vessel pressure (right panel). As this pressure increases so does the
coal feed rate assuming constant reactor pressure, NCCC engineers speculated that the
variations of the moisture content are due to coal feed rate oscillations, but it is notable
that this was first observed by in-situ TDL measurements.
Figure 6.11 TDL recorded moisture content in the syngas without people attendance for
a duration of more than 400 hours.
104
Figure 6.12 Correlations of the TDL measured moisture content in the syngas product
flow to the reactor temperature measured by the thermocouple (left) and to the coal-
dispense vessel pressure (right)
105
Chapter 7 Absorption lineshape from ratios
of different WMS harmonic signals
7.0 Motivation
Previous calibration-free WMS approaches [47,48] require the knowledge of the
collision-broadening halfwidth ( C , full-width half-maximum, FWHM, cm-1
/atm) as
the normalized WMS signal is dependent on both the absorption and transition lineshape.
For most applications, the lineshape function can be described by a Voigt function, which
is given by the convolution of a contribution from Doppler broadening (inhomogenous,
Gaussian) and a contribution from collision broadening (homogenous, Lorentzian). The
Voigt lineshape is characterized by the two FWHM line widths: D for Doppler and
C for collision broadening. The Doppler-broadened width is a function of the gas
temperature and can be calculated by known temperature or the result from a two-line
thermometry measurement. For many applications, this temperature needs only be
approximately known as the collision-broadening contribution dominates the absorption
lineshape: for example at 300K and atmospheric pressure, / 3C D for H2O in the
overtone band near 1400nm. C is a function of gas pressure, temperature, and gas
composition often written as:
2 ( )C j ij
j
P x T , (7.1)
106
where jx is the mole fraction of species j in the gas mixture and ( )ij T is the collision-
broadening coefficient at temperature T of absorber i under the collisional perturbation
by species j .
An accurate estimation of C relies on two factors: (1) an accurate evaluation of the
collision-broadening coefficients between the targeted species and its collision partners;
and (2) the gas composition during the absorption measurements. The broadening
coefficients for air- and self-collisions are contained in the database HITRAN (reported
as γ, HWHM, cm-1
/atm), but laboratory cell measurements of collision-broadening
coefficients are needed for other collision partners. For example in combustion gas
exhaust, CO2 and H2O are typically present at significant mole fraction and broadening
coefficients for these collision partners are needed. Even when a well-validated
collision-broadening-coefficient database is available for the absorption transition,
uncertainty in the gas composition can also contribute to the uncertainty in the WMS
analysis. Knowledge of C is critical for accurate analysis of WMS absorption; thus,
alternative approaches to determine C are needed.
The harmonics (nf) of the WMS have different dependences on the absorption lineshape.
Here we consider a 296K atmospheric pressure example where collision broadening
dominates the absorption lineshape. Figure 7.1 illustrates the laser intensity (0f)
normalized WMS-nf signals for n=2-6 for a specific NIR absorption transition near
7185.6 cm-1
of water vapor as a function of C . (We assume the bath gas contains an
unknown collider species that dominates C .) Note as C increases by a factor of 3,
107
the WMS-2f signal decreases by a factor of 3 and the WMS-6f signal decreases by a
factor of 15.
It has been demonstrated recently that by fitting the entire 2f/1f absorption lineshape in a
wavelength-scanned WMS measurement, the integrated absorbance and C of the target
transition can be measured simultaneously, in a way analogous to the scanned-
wavelength direct absorption measurement. However, this fitting approach may cause
significant time-cost in data analysis (~10s), which is not acceptable for some
applications where second or sub-second real-time measurements are important.
0.04 0.06 0.08 0.10 0.120.00
0.01
0.02
0.03
0.040.75% H
2O in air
T = 296K, P = 753torr
L = 100.5cm
a= 0.081cm-1
0f-
norm
aliz
ed
WM
S-n
f sp
ectr
a
pe
ak m
agn
itu
de
[un
itle
ss]
Collisional broadening width [cm-1]
2f
3f
4f
5f
6f
Figure 7.1 Simulations of normalized WMS-nf spectra peak magnitudes versus C for
the H2O transition near 7185.6 cm-1
. The nf-harmonic signals were normalized by the DC
component of the incident laser power (0f) (T = 296 K, P = 1 atm, L = 100.5 cm, a =
0.081 cm-1
, f = 10 kHz)
Here we introduce a novel and rapid approach to determine C via the ratio of WMS
signals from different harmonics of the modulation frequency, demonstrated here using
the ratio WMS-4f/WMS-2f. Instead of fitting the entire WMS lineshape, only the peak
values of the WMS-nf absorption spectrum near the transition line center are needed,
108
denoted as WMS-nfpeak below. First we show that for optically thin conditions (peak
absorbance < 0.1) for a relatively isolated transition, the signal ratio from different
harmonics is a function only of the absorption lineshape. Second, a laboratory
demonstration of the determination of C using the ratio of WMS-4fpeak/WMS-2fpeak is
described from the absorption of H2O using a DFB laser near 1392nm. Third, for a set of
cell experiments, we demonstrate that gas pressure can be determined by the ratio of
WMS-4fpeak/WMS-2fpeak signals with a precision of <1%.
7.1 WMS fundamentals and derivation
The analytical expression for the X- and Y-component of the WMS-nf signals for a small
modulation depth can be expressed:
0 1 1 1 1
1 1( ) cos( )
2 2nf n n nX GI H H H i
(7.2)
0 1 1 1 1
1 1( ) sin( )
2 2nf n nY GI H H i
(7.3)
where G is the detector gain. For optically-thin absorption of an isolated transition, the
WMS signals at the nth
harmonic (for n≥2) can be expressed as (see appendix A.2 for a
full derivation):
2 2
0
1( ) ( , , )
2nf nf nf iS X Y SPx L GI F n
, (7.4)
where F is a function of the order of the harmonic used for absorption detection, the laser
wavelength, and the transition lineshape function.
109
The ratio of the WMS signals for different harmonics (n and m) can be expressed as:
( , , )
( ) / ( )( , , )
nf
nf nf mf mf
mf
F nS S
F m
. (7.5)
This reduces to a function of the selected harmonic order (n and m), center wavelength
for the harmonic signal wavelength (nf and
mf ), and the transition lineshape function.
For even harmonics of the WMS, the peak signals occur at the transition linecenter. Thus,
for example the ratio WMS-4fpeak/WMS-2fpeak is a function of the lineshape function
only:
4 4 2 2
(4, , )( ) / ( )
(2, , )peak peak
linecenterf f f f
linecenter
FS S
F
. (7.6)
7.2 An example case and laboratory demonstration
Although the ratio of WMS-4fpeak/WMS-2fpeak is a sensitive function of C , this ratio is
insensitive to total absorbance as illustrated in Figure 7.2 by the single overlapped curve
even though the absorber mole fraction was varied from 0.25 to 0.75%, the pressure from
250-750 torr and the pathlength from 5 to 15cm (for an absorbance range from 0.22% to
0.67%). For the entire range of conditions, the ratio WMS-4fpeak/WMS-2fpeak varies only
with C . Thus C determined from the measured ratio WMS-4fpeak/WMS-2fpeak can be
used for WMS absorption analysis.
110
0.00 0.02 0.04 0.06 0.08 0.10 0.12
0.2
0.3
0.4
0.5
0.6
0.7
0.8
C[cm
-1]
4f p
ea
k/2
f pe
ak[u
nitle
ss]
0.25% H2O, P = 500torr, L = 10cm
0.50% H2O, P = 500torr, L = 10cm
0.75% H2O, P = 500torr, L = 10cm
0.50% H2O, P = 250torr, L = 10cm
0.50% H2O, P = 750torr, L = 10cm
0.50% H2O, P = 500torr, L = 5cm
0.50% H2O, P = 500torr, L = 15cm
Figure 7.2 The simulated ratio WMS-4fpeak/WMS-2fpeak as a function of C for different
absorber mole fraction, pressure and pathlength conditions ( T = 296 K, a = 0.081 cm-1
, f
= 10 kHz, =7185.59 cm-1
).
The measured wavelength-scanned WMS-2f and WMS-4f absorption lineshapes are
shown in the left panel of Figure 7.3 at 296K and 753 torr. The simulated WMS-
4fpeak/WMS-2fpeak ratio as a function of pressure is shown in the right panel. The peak
absorbance is less than 0.10 for all these measurements and hence the absorption can be
considered optically thin. Figure 7.4 shows C inferred from the measured ratio WMS-
4fpeak/WMS-2fpeak agrees within 0.7% with the known values calculated from Eqn. (7.1)
using measured spectroscopic parameters.
111
100 200 300 400 500 600 700
0.3
0.4
0.5
0.6
0.7
0.8
0.75% H2O in air
T = 296K
L = 100.5cm
Modulation depth = 0.081cm-1M
easu
red
4f p
ea
k/2
f pe
ak[u
nitle
ss]
Pressure [torr]
Measurements
Figure 7.3 Measured WMS-4fpeak/WMS-2fpeak with pressures (0.75% H2O in air, L =
100.5 cm, a = 0.081 cm-1
, f = 10 kHz)
The values of C inferred from the WMS-4fpeak/WMS-2fpeak ratios were used to simulate
the WMS-2fpeak signal and a best fit was determined by varying the H2O mole fraction. If
a 2-D map of the simulated WMS-2fpeak signals as a function of C and absorber mole
fraction is prepared prior to the measurements, the computational cost of this iteration on
a standard PC is less than 1ms. The H2O mole fraction determined in this fashion is
shown in Figure 7.5 and agrees well with the known H2O mole fraction (0.75%). These
results demonstrate the feasibility of rapid (kHz) fixed-wavelength, calibration-free WMS
measurements of species mole fraction without the need for a database for collision-
7185.2 7185.4 7185.6 7185.8 71860
0.005
0.01
0.015
0.02
Frequency [cm-1
]
WM
S-n
f sig
na
l [A
.U]
2f
4f
4fpeak
2fpeak
112
broadening coefficients and known composition of the major species in the gas; note this
conclusion is valid for a relatively isolated transition with optically-thin absorbance
(<0.1).
100 200 300 400 500 600 700
0.02
0.04
0.06
0.08
0.10
C [cm
-1]
Pressure [torr]
Measured by 4fpeak
/2fpeak
Expected
Figure 7.4 Measured C using WMS-4fpeak/WMS-2fpeak ratio with the comparison to
calculated C as a function of pressure (0.75% H2O in air, L = 100.5 cm, a = 0.081 cm-1
,
f = 10 kHz, =7185.59 cm-1
)
100 200 300 400 500 600 700
0.4
0.6
0.8
1.0
Measured H2O%
0.73
0.77
H2O
concentr
ation [%
]
Pressure [torr]
Figure 7.5 H2O mole fraction determined from WMS-2fpeak using C from the ratio of
WMS-4fpeak/WMS-2fpeak (0.75% H2O in air, L = 100.5 cm, a = 0.081 cm-1
, f = 10 kHz,
=7185.59 cm-1
)
113
Even though the above derivation and demonstration assumed optically-thin conditions,
the errors are modest for larger absorptions. For the non-optically-thin case, the ratio of
WMS-4fpeak/WMS-2fpeak has a weak dependence on the peak absorbance as shown in
Figure 7.6 for values ranging from 0.2 to 1. Thus, C determined from the ratio WMS-
4fpeak/WMS-2fpeak might have an error as large as +/-0.002 cm-1
for =1, which might be
more than 10% of the total value, producing a large error (>10%) in the mole fraction
inferred from the WMS-2fpeak value. However, if we iterate on mole fraction and C
using the procedure described in Ref. [48], the values rapidly converge to acceptable
uncertainty (<2% of the mole fraction). Thus, the use of the WMS-4fpeak/WMS-2fpeak
ratio to determine C can be readily used for peak absorbance less than one.
0.00 0.02 0.04 0.06 0.08 0.10 0.12
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.5%H2O, T = 300K, P = 750 torr
4f p
ea
k/2
f pe
ak[u
nitle
ss]
C [cm
-1]
50cm
100cm
150cm
200cm
250cm
Figure 7.6 The WMS-4fpeak/WMS-2fpeak ratio as a function of C computed for peak
absorbance ranging from 0.2 to 1, corresponding to a pathlength increase from 50-250cm
(a = 0.081 cm-1
, f = 10 kHz, =7185.59 cm-1
)
One drawback of this approach is that the optimal modulation depth to maximize the
WMS signal cannot be determined in advance as it requires an estimate of the transition
114
linewidth before the absorption measurement. Empirically, this could be done with a
manual trial-and-error approach during the actual field measurements. Or alternatively,
the modulation depth can be varied during the measurement to ensure an optimal
modulation depth is adopted.
7.3 Pressure sensor developed using the 4fpeak/2fpeak ratio
In the previous section, C was determined from the ratio of WMS-4fpeak/WMS-2fpeak.
However, for the case when a database of collision-broadening coefficients is available
and when the major species of the gas mixture are known, this WMS-4fpeak/WMS-2fpeak
ratio can be used to determine pressure.
1
2 ( )C
j ij
j
Px T
(7.7)
Figure 7.7 shows the pressure determined from the measured collision-broadening
coefficients in Table 5.1. Because this pressure is determined from path-integrated
absorption over the line-of-sight, this method could be used to determine local (off-the-
wall) pressure in a compressible flow with non-uniform temperature or gas composition.
For example, if the gas composition is not uniform and the absorber is confined to the
core flow, such measurements would infer core flow pressure. Alternatively, even if the
gas composition is uniform but the gas temperature is not, line selection could bias the
absorption measurement to a “hot” or a “cold” region in the flow, and the local pressure
could be determined from the ratio of WMS-4fpeak/WMS-2fpeak. This strategy also
provides an alternative to using facility pressure data for WMS absorption analysis.
115
0 100 200 300 400 500 600 700 8000
100
200
300
400
500
600
700
800
averaged deviation = 0.64%
Measurements
Measure
d p
ressure
[to
rr]
Calibrated pressure [torr]
Figure 7.7 measured gas total pressure using WMS-4fpeak/WMS-2fpeak ratio and its
comparison to the baratron result (0.75% H2O in air, L = 100.5cm, T = 296K, a = 0.081
cm-1
, f= 10 kHz, =7185.59 cm-1
)
116
117
Chapter 8 Summaries and future plans
8.1 Summaries
Two different strategies for WMS analysis were developed that offer simplified and more
reliable WMS use for practical laser absorption sensors.
8.1.1 A generalized 1f-normalized WMS-nf method using Fourier
Analysis
A generalized model for 1f-normalized WMS-nf detection with an injection current-tuned
diode laser was presented that accounts for performance by the laser and etalon
interference by the optical components in the LOS. This model was validated using
measurements of the CO transition of R (11) in the 1st overtone band near 2.3µm at room
temperature for a range of CO mole fractions (0.21-2.8%) and pressures (5-20atm). For
high-pressure gas sensing, wavelength modulation spectroscopy with higher-order
harmonic detection (WMS-nf, n>2) was found to have less influence from the WMS
background signals when the selected modulation depth was close to the optimal
modulation depth for the WMS-2f signal. Similar levels of accuracy in CO mole fraction
measurements were observed by using 1f-normalized WMS-2f, 3f and 4f techniques, but
WMS-3f and WMS-4f detection showed better accuracy than WMS-2f when uncertainty
in the WMS background signal was significant (i.e. from the long term drift of the
background signals due to thin etalons). The large signal-to-background ratio of WMS
higher harmonics (WMS-nf, n>2) potentially offers the advantage that knowledge or
accurate evaluation of the background signals may not be required. This advantage is
118
useful for practical sensing applications such as that in a continuously running coal-
gasifier where opportunities for zero-absorption background signal are difficult to
achieve. In optically thin cases, absorption measurements of the targeted transition had
less interference from neighboring transitions by using detection of 1f-normalized WMS
higher harmonics. However, WMS-nf (n>2) detection required a larger modulation depth
than WMS-2f to achieve the optimal signal-to-noise ratio, which may be achieved at a
lower modulation frequency, but can in turn result in decreasing the measurement time
resolution. In addition, as the absolute amplitudes of the higher harmonic signals are
smaller than those for 2f, using higher harmonics may lower the signal-to-noise ratio.
8.1.2 Demonstration of the 1f-normalized WMS-2f strategy in a pilot-
scale entrained-flow high pressure coal gasifier
Successful in-situ rapid-time absorption measurements for gas temperature and
concentrations by 1f-normalized WMS-2f in a pilot-scale entrained-flow coal gasifier
with non-absorption transmission loss up to 99.997%, pressure up to 250psig and
temperature up to 1800K were demonstrated. The temperature sensors by pairs of H2O
transitions are accurate, robust and rapid enough to monitor the temperature and transient
instabilities in the gasification process, which can be used to control the gasifier to
optimize the gasification performance. The syngas compositions measured in the output
of the gasifier by simultaneous measurement of CO, CO2 concentrations (reported in
Ritobrata Sur’s work [70]) and simultaneous measurement of H2O concentration and gas
temperature prove the capability of laser absorption measurements in monitoring the
syngas heating value.
119
8.1.3 A novel strategy for WMS absorption analysis
The development of a new method for the analysis of WMS absorption measurements
using injection-current modulated TDLs was described in detail. The method was
validated with water vapor detection near 7185.6cm-1
in a static cell at 1 atm and 296 K
using a distributed feedback (DFB) laser. WMS-nf harmonics for n=1 to 6 were extracted
and the 1f-normalized WMS-nf were in good agreement with the measurement over the
entire WMS-nf lineshape.
This new analysis schemes differs from previous WMS analysis strategies in two
significant ways: (1) the measured intensity versus time of the wavelength-scanned (at
frequency fs), wavelength-modulated (at fm) laser light is used to simulate the transmitted
laser intensity versus time, and (2) digital lock-in and low-pass filter software is used to
expand the time series of simulated and measured transmitted laser intensity into
harmonics of the modulation frequency, WMS-nfm (n=1,2,3,…). This new approach has
nine distinct advantages versus traditional analysis of WMS. (1) The use of measured
laser intensity to simulate the Beer’s law absorption signals avoids the need for an
analytic model of laser intensity in its response to scanning and modulating the injection
current. (2) The use of measured intensity for the simulation also accounts for any
wavelength dependent transmission of other optical components in the apparatus. (3)
The use of the digital lock-in and low-pass filter software to extract the WMS-nfm
harmonics from the simulated transmitted intensity avoids the complex Fourier expansion
of the simulated absorption of the simultaneously scanned and modulated laser intensity
and wavelength. (4) The scheme is valid for all WMS-nfm harmonics, (5) at any optical
depth, (6) at any modulation index, and (7) at all values of the mean laser wavelength
120
even in the wings of the absorption away from line center. (8) This scheme is valid for
WMS using unresolved blended transitions. (9) Using the same software for both
simulation and measurement provides equal contributions of any non-ideal performance
of the lock-in and low-pass filter software.
8.1.4 Demonstration of the fitting strategy for wavelength-scanned
WMS in an engineering-scale fluidized-bed high pressure coal gasifier
A 1352nm DFB laser-based H2O absorption sensor employing the new wavelength-
scanned WMS 2f/1f technique described above was first validated in a high-pressure cell
in the laboratory and then successfully applied to monitor the synthesis gas output from
an engineering-scale transport reactor coal gasifier at the National Carbon Capture
Center. There the pressures ranged up to 15 atm (~220psig) and temperatures up to 650K.
Continuous monitoring of moisture level in the gasifier output with 2s time resolution
was performed by the TDL sensor for more than 500 hours, including the periods of
burner ignition, combustion heating with a propane flame, coal combustion, coal
gasification, and reactor shut-down via coal-feed termination. As expected for coal
synthesis gas applications, beam steering and particulate scattering were severe during
the measurements, resulting in less than 0.01% light transmission. However, the robust
2f/1f normalization strategy provided a SNR of better than 20 for measurements of 2.3%
mole fraction of H2O at 25Hz bandwidth, corresponding to a 0.02% detection limit for
moisture mole fraction for a 1Hz bandwidth. With 2-second time resolution, the TDL
sensor captured the time-varying moisture level in the gas exhaust due to the changing
fuel content in the reactor. This observation had been anticipated, but was observed for
the first time by the TDL moisture sensor. The sampled GC gas analysis on this facility
did not have the time resolution needed to observe this behavior. These results
121
demonstrate the feasibility of using the TDL sensors to provide realtime control signals to
optimize the gasification process.
The wavelength-scanned WMS measurements determined the C of the target transition
via lineshape fitting. Comparisons between the measured results and the expected values
provided better than 3% agreement (within the combined uncertainty of the synthesis gas
composition and the collision- broadening coefficient database). These results suggest
that wavelength-scanned WMS can be used, similar to scanned-wavelength direct
absorption, to determine the absorber mole fraction without knowledge of the transition
collisional width C prior to the measurement. This is important for ensuring accuracy
of the measurements, as for many applications, the gas composition can be difficult to
estimate, and the uncertainty in estimating the transition collisional width can result in
large errors in determining the absorber mole fractions via traditional analysis of WMS.
This successful demonstration of water vapor measurement in the synthesis gas output of
a large-scale industrial facility shows advantages of using wavelength-scanned WMS
techniques. Based on these successful H2O measurements, future work will include
extension to wavelength-scanned WMS TDL sensors for other major species in the
synthesis gas output such as CO.
8.1.5 Absorption lineshape from ratios of different WMS harmonic
signals
The analytical expression for the ratio of WMS harmonics (WMS-nf/WMS-mf) is derived
for the optically-thin case with an isolated (or nearly isolated) transition. This ratio is a
function of the absorption lineshape, and can be used to extract the lineshape data needed
for calibration-free interpretation of fixed-wavelength WMS signals without the need for
122
a database of collision-broadening coefficients and knowledge of the major species
composition of the gas. Thus, a practical sensor using calibration-free WMS can be
realized with in-situ determination of C . The method was demonstrated to accurately
determine collision-broadening coefficients in a laboratory cell experiment. Although the
derivation and demonstration of this approach using the WMS-4fpeak/WMS-2fpeak ratio
was based on optically-thin conditions for a relatively isolated transition, the use of this
approach is not limited to such cases. For the non-optically-thin case with the peak
absorbance ranging from 0.2-1, we demonstrate that the absorber mole fraction from the
simulated WMS-2fpeak and C determined using the ratio of WMS-4fpeak/WMS-2fpeak
rapidly converge in a few iterations. Alternatively, if collision broadening and gas
composition are known, the WMS-4fpeak/WMS-2fpeak ratio can be used to determine
pressure. This WMS-nf ratio strategy offers an alternative approach to measure the gas
pressure and potentially allows the development of a TDL sensor of species mole fraction
without independent pressure measurements. For compressible flows with non-uniform
temperature and/or gas composition, the method could be used to determine “off-the-
wall” pressure.
8.2 Future plans
8.2.1 Other species measurements in the fluidized-bed coal gasifier in
NCCC
The success of H2O measurements described in chapter 6 sprovides fundamentals of
absorption measurements for other species like CO, CO2 and CH4 in practical systems
such as the fluidized-bed coal gasifier in NCCC. A measurement campaign at NCCC has
123
been scheduled early next year (2014) to test the syngas heating-value sensor by
simultaneously measurements of four different species using the strategy described in
chapter 6 and 7.
8.2.2 Integrate the cavity enhanced techniques with WMS
WMS is an effective technique to decrease the noise in the absorption measurement.
Cavity enhanced techniques (CET) can enhance the absorption signal via increasing the
pathlength. It mostly uses two mirrors of high reflectivity and the laser light reflects back
and forth between these mirrors for thousands of times, yielding a large effective
pathlength. The combination of WMS with cavity-enhanced techniques provide a
potential strategy for extremely sensitive detection of trace species or species of weak
transitions. This will allow the detection of trace radicals or molecules in the combustion
environment to facilitate the combustion kinetics studies.
8.2.3 Species time-history measurements in shock tubes using
CET/WMS
The detection sensitivity of the some species in shock tubes can be limited by the
dimension of the shock tube. For example, sub-ppm detection sensitivity of NO at 1MHz
measurement bandwidth is very hard to achieve in a shocktube with an inner diameter of
~10cm. The combination of CET and WMS can potentially be used to increase the
detection sensitivity of absorption measurements in a shock tube by several orders of
magnitudes, allowing measurements of trace radicals formed in chemical reactions that
have not been measured previously.
124
125
Appendix
A.1 Laboratory measured spectroscopy parameters
Table A.1 to A.8 list the measured spectroscopy parameters measured in the laboratory
for the development of the H2O absorption sensor. The measurement procedure is
described in Ref [75].
Table A.1 Comparison between the measured linestrength and those recorded in HITEMP 2010
database for studied transitions (Tref = 296K)
Frequency
υ0
[cm-1
]
Lower state energy
E”
[cm-1
]
Measured
Linestrength
[cm-1
/atm-1
]
HITEMP 2010
Linestrength
[cm-1
/atm-1
]
6806.03 3291 6.14E-7 6.54E-7
7117.24 447 5.63E-2 5.46E-2
7117.42 447 1.61E-1 1.60E-2
7117.75 399 1.76E-1 1.75E-2
7185.60 1045 1.95E-2 1.98E-2
7426.11 1280 3.01E-3 3.25E-3
7426.14 1327 3.71E-3 4.20E-3
7426.45 1200 3.16E-3 3.23E-3
7426.60 1293 3.86E-3 4.05E-3
7435.62 1558 1.93E-3 1.95E-3
7435.73 1719 4.10E-3 4.21E-4
7435.94 1525 1.40E-3 1.45E-3
7436.00 1525 4.81E-4 4.93E-4
7436.91 1446 2.82E-3
(counted as one line)
2.18E-3
7436.92 1283 8.50E-4
7437.19 1202 5.31E-3 5.20E-3
7465.61 1999 9.84E-5 1.05E-4
7465.90 2631 1.17E-5 1.19E-5
7466.30 2661 1.17E-5 1.17E-5
126
Table A.2 Measured H2O- H2O collision-broadening coefficients and those recorded in HITEMP
2010 database for studied transitions (Tref = 296K)
Frequency
υ0
[cm-1
]
Measured
2γH2O-H2O
[cm-1
/atm]
HITEMP 2010
2γ H2O-H2O
[cm-1
/atm]
Measured
n H2O-H2O
6806.03 0.258 0.240 0.41
7117.24 0.921 0.940 0.92
7117.42 1.061 0.890 1.06
7117.75 1.116 0.920 1.16
7185.60 0.410 0.390 0.61
7426.11 0.572 0.562 1.32
7426.14 0.348 0.562 0.47
7426.45 0.486 0.666 0.72
7426.60 0.360 0.548 0.45
7435.62 0.340 0.440 0.38
7435.73 0.570 0.600 0.61
7435.94 0.396 0.380 0.42
7436.00 0.415 0.482 0.57
7436.91 0.541
(counted as one line)
0.650 0.49
7436.92 0.630
7437.19 0.762 0.800 0.65
7465.61 0.630 0.806 0.63
7465.90 0.224 0.610 0.38
7466.30 0.270 0.260 0.60
Table A.3 Measured H2O-CO2 collision-broadening coefficients for studied transitions (Tref =
296K)
Frequency
υ0
[cm-1
]
Measured
2γH2O-CO2
[cm-1
/atm]
Measured
n H2O-CO2
6806.03 0.102 0.96
7117.24 0.175 0.87
7117.42 0.178 0.73
7117.75 0.267 0.83
7426.11 0.0804 0.75
7426.14 0.0846 0.72
7426.45 0.1090 0.50
7426.60 0.1090 0.69
7185.60 0.156 0.74
7435.62 0.112 0.91
7435.73 0.148 0.47
7435.94 0.132 0.72
7436.00 0.098 0.60
7436.91 0.154
(counted as one line)
0.59
(counted as one line) 7436.92
7437.19 0.166 0.43
7465.61 0.202 0.59
7465.90 0.089 0.72
7466.30 0.041 0.21
127
Table A.4 Measured H2O-CO collision-broadening coefficients for studied transitions (Tref =
296K)
Frequency
υ0
[cm-1
]
Measured
2γH2O-CO
[cm-1
/atm]
Measured
n H2O-CO
7117.24 0.203 0.77
7117.42 0.196 0.61
7117.75 0.236 0.72
7426.11 0.0684 0.70
7426.14 0.0644 0.49
7426.45 0.1032 0.69
7426.60 0.0908 0.45
7185.60 0.117 0.64
7435.62 0.069 1.01
7435.73 0.128 0.27
7435.94 0.076 0.62
7436.00 0.069 0.66
7436.91 0.122
(counted as one line)
0.76
(counted as one line) 7436.92
7437.19 0.147 0.52
7465.61 0.258 1.00
7465.90 0.0322 0.14
7466.30 0.0284 0.20
Table A.5 Measured H2O-H2 collision-broadening coefficients for studied transitions (Tref =
296K)
Frequency
υ0
[cm-1
]
Measured
2γH2O-H2
[cm-1
/atm]
Measured
n H2O-H2
6806.03 0.0744 0.45
7117.24 0.096 0.39
7117.42 0.113 0.49
7117.75 0.154 0.67
7185.60 0.085 0.52
7426.11 0.0368 0.40
7426.14 0.0526 0.54
7426.45 0.0646 0.50
7426.60 0.0662 0.45
7435.62 0.064 0.45
7435.73 0.094 0.52
7435.94 0.076 0.48
7436.00 0.077 0.54
7436.91 0.088
(counted as one line)
0.57
(counted as one line) 7436.92
7437.19 0.101 0.54
7465.61 0.050 0.01
7465.90 0.024 -0.23
7466.30 0.0388 0.16
128
Table A.6 Measured H2O-CO2 pressure shifting coefficients for studied transitions (Tref = 296K)
Frequency
υ0
[cm-1
]
Measured
δ H2O-CO2
[cm-1
/atm]
Measured
m H2O-CO2
7117.24 -0.0211 0.71
7117.42 -0.0126 0.78
7117.75 -0.0102 0.79
7435.62 -0.0330 0.97
7435.73 -0.0312 0.85
7435.94 -0.0185 0.84
7436.00 -0.0202 1.18
7436.91 -0.0280
(counted as one line)
0.88
(counted as one line) 7436.92
7437.19 -0.0124 0.99
Table A.7 Measured H2O-CO pressure shifting coefficients for studied transitions (Tref = 296K)
Frequency
υ0
[cm-1
]
Measured
δ H2O-CO
[cm-1
/atm]
Measured
m H2O-CO
7117.24 -0.0139 0.90
7117.42 -0.0078 0.91
7117.75 -0.0090 0.78
7435.62 -0.0300 1.39
7435.73 -0.0167 0.72
7435.94 -0.0200 1.24
7436.00 -0.0263 1.24
7436.91 -0.0216
(counted as one line)
1.24
(counted as one line) 7436.92
7437.19 -0.0097 0.84
Table A.8 Measured H2O-H2 pressure shifting coefficients for studied transitions (Tref = 296K)
Frequency
υ0
[cm-1
]
Measured
δ H2O-H2
[cm-1
/atm]
Measured
m H2O-H2
7117.24 -0.0158 1.12
7117.42 -0.0106 1.25
7117.75 -0.0119 1.01
7435.62 -0.0141 1.34
7435.73 -0.0118 1.03
7435.94 -0.0117 1.13
7436.00 -0.0152 1.34
7436.91 -0.0133
(counted as one line)
1.31
(counted as one line) 7436.92
7437.19 -0.0122 0.82
129
A.2 Derivation for Eqn (7.4)
The analytical expression for the X- and Y-component of the WMS-nf signals can be
expressed as[48]
:
0 1 1 1 1
1 1( ) cos( )
2 2nf n n nX GI H H H i
(A-1)
0 1 1 1 1
1 1( ) sin( )
2 2nf n nY GI H H i
, (A-2)
where
0 0
1 1( cos )cos exp ( cos ) cos
(1 ) (1 )k j j i
jk k
H a k d S a P x L k d
. (A-3)
For optically-thin conditions:
0
0
( cos )cos(1 )
ik k j j
jk
Px LH S a k d
. (A-4)
And for conditions of nearly isolated transition, i.e., the absorption spectra of the
wavelength region in concern is dominated by the targeted transition:
0
0
( cos )cos(1 )
ik k
k
SPx LH a k d
, (A-5)
Substituting (19) into (15) and (16), we can obtain:
130
1 10
1 1
( cos )cos
cos1( cos )cos( 1)
2 2
cos( cos )cos( 1)
2
nf i
a n d
iX SPx L GI a n d
ia n d
, (A-6)
and
1 1
0
1 1
sin( cos )cos( 1)
21
2 sin( cos )cos( 1)
2
nf i
ia n d
Y SPx L GIi
a n d
. (A-7)
It is convenient to write the WMS-nf signal (for n≥2) as:
2 2
0 1 1
1( , , , , , )
2nf nf nf iS X Y SPx L GI F n a i
, (A-8)
where F is a function of the parameters listed in the parenthesis. Note that the first
parameter in the parenthesis is only related to the order of the harmonic and the four
parameters that follow are the laser tuning parameters which are pre-measured and assumed to
be unchanged during the absorption measurements. Thus, when the tuning behavior of a specified
and characterized laser is known, the expression for the WMS signal can be further simplified to:
0
1( ) ( , , )
2nf iS SPx L GI F n
. (A-9)
Then the ratio of the WMS signals at different harmonics can be expressed as:
131
( , , )( ) / ( )
( , , )
nf
nf nf mf mf
mf
F nS S
F m
. (A-10)
132
133
Reference
[1] R.K. Hanson, “Applications of quantitative laser sensors to kinetics, propulsion and
practical energy systems”, Proc. Combust. Inst. 33, (2011) 1–40
[2] M. G. Allen, “Diode laser absorption sensors for gas dynamic and combustion flows”
Measure. Sci. and Tech. 9, (1998) 545-562
[3] P. Werle, “A review of recent advances in semiconductor laser based gas Monitors”,
Spectrochimica Acta Part A 54, (1998)197-236
[4] J. Wolfrum, “Lasers in combustion: from basic theory to practical devices”, Proc.
Combust. Inst. 27, (1998) 1–41
[5] J.A. Silver, “Frequency-modulation spectroscopy for trace species detection: theory
and comparison among experimental methods” Appl. Opt. 31, (1992) 707-717
[6] G. Hancock, J.H. van Helden, R. Peverall, G.A.D. Ritchie and R.J. Walker “Direct
and wavelength modulation spectroscopy using a cw external cavity quantum cascade
laser”, Appl. Phys. Lett. 94, (2009) 201110
[7] T. R. Meyer, S. Roy, T. N. Anderson, R. P. Lucht, R. Jimenez and J. R. Gord, “10
kHz detection of CO2 at 4.5 µm by using tunable diode-laser-based difference-frequency
generation”, Optics Lett. 30, (2005) 3087-3089
[8] M. Seiter and M. W. Sigrist, “On-line multicomponent trace-gas analysis with a
broadly tunable pulsed difference-frequency laser spectrometer”, Appl. Opt. 38, (1999)
4691-4698
134
[9] K. Kohse-Höinghaus and J.B. Jeffries, “Applied Combustion Diagnostics” Taylor and
Francis, London (2002)
[10] K. Kohse-Höinghaus, R.S. Barlow, M. Aldén, and J. Wolfrum, “Combustion at the
focus: laser diagnostics and control,” Proc. of the Comb. Institute. 30, (2005) 89
[11] L. S. Chang, C. L. Strand, J. B. Jeffries, R. K. Hanson, G.S. Diskin, R. L. Gaffney
and D. P. Capriotti, “Supersonic mass-flux measurements via tunable diode laser
absorption and nonuniform flow modeling”, AIAA Journal 49, (2011) 2783-2791
[12] G. Wysocki, A.A. Kosterev and F.K. Tittel, “Spectroscopic trace-gas sensor with
rapidly scanned wavelengths of a pulsed quantum cascade laser for in situ NO monitoring
of industrial exhaust systems,” Appl. Phys. B. 80, (2005) 617
[13] R. Engeln, G. Berden, R. Peeters, and G. Meijer, “Cavity en- hanced absorption and
cavity enhanced magnetic rotation spectroscopy,” Rev. Sci. Instrum. 69, (1998) 3763–
3769
[14] A. O’Keefe, J. J. Scherer, and J. B. Paul, “cw Integrated cavity output
spectroscopy,” Chem. Phys. Lett. 307, (1999) 343–349
[15] B.W.M. Moeskops, S.M. Cristescu and F.J.M Harren, “Sub-part-per-billion
monitoring of nitric oxide by use of wavelength modulation spectroscopy in combination
with a thermoelectrically cooled, continuous-wave quantum cascade laser,” Opt. Lett. 31,
(2006) 823
[16] H. Li, A. Farooq, J.B. Jeffries, R.K. Hanson, “Near-infrared diode laser absorption
sensor for rapid measurements of temperature and water vapour in a shock tube,” Appl.
Phys. B 89, (2007) 407
135
[17] V.L. Kasyutich, P.A. Martin and R.J. Holdsworth, “An off-axis cavity-enhanced
absorption spectrometer at 1605nm for the 12CO2/13CO2 measurement” Appl. Phys. B 85,
(2006) 413-420
[18] G.S. Engel. W.S. Drisdell, F.N. Keitsch, E.J. Moyer and J.G. Anderson,
“Ultrasensitive near-infrared integrated cavity output spectroscopy technique for
detection of CO at 1.57um: new sensitivity limites for absorption measurements in
passive optical cavities” Appl. Opt. 45, (2006) 9221-9229
[19] D.S. Baer, J.B. Paul, M. Gupta and A. O’Keefe, “Sensitive absorption measurements
in the near-infrared region using off-axis integrated-cavity output spectroscopy”, Appl.
Phys. B 75, (2002) 261-265
[20] J. B. Paul, L. Lapson, and J. G. Anderson, “Ultrasensitive absorption spectroscopy
with a high-finesse optical cavity and off-axis alignment,” Appl. Opt. 40, (2001) 4904–
4910
[21] V. Ebert, H. Teichert, P. Strauch, T. Kolb, H. Seifert and J. Wolfrum, “Sensitive in
situ detection of CO and O2 in a rotary kiln-based hazardous waste incinerator using 760
nm and 2.3 µm diode lasers,” Proc. of the Comb. Institute. 30, (2005) 1611
[22]H. Pitz, T. Fernholz, C. Giesemann and V. Ebert, “Diode-laser-based in-situ-CH4-
detection for the surveillance of ignition processes in gas-fired power-plants,” Trends
Opt. Photon. 36, (2000) 111
136
[23] P. Ortwein, W. Woiwode, S. Fleck, M. Eberhard, T. Kolb, S. Wagner, M. Gisi and
V. Ebert, “Absolute diode laser-based in situ detection of HCl in gasification processes,”
Exp Fluids 49, (2010) 961
[24]R.P. Lucht, T.N. Anderson, S. Priyadarsan, S. Arumugam, R. Barron-Jimenez, J.A.
Caton and K. Annamalai, “Diode-laser-based sensor measurements of nitric oxide in
particulate-laden combustion exhaust streams,” Proc. Of the Twentieth Annual
International Pittsburgh Coal Conference, 20, (2003) 476
[25]T.N. Anderson, R.P. Lucht, S. Priyadarsan, K. Annamalai and J.A. Caton, “In situ
measurements of nitric oxide in coal-combustion exhaust using a sensor based on a
widely tunable external-cavity GaN diode laser,” Appl. Opt. 46, (2007) 3946
[26]X. Chao, J.B. Jeffries, R.K. Hanson, “Real-time, in situ, continuous monitoring of
CO in a pulverized-coal-fired power plant with a 2.3 μm laser absorption sensor,” Appl.
Phys. B, 110, (2013) 359-365
[27] G.B. Rieker, J.B. Jeffries, R.K. Hanson, T. Mathur, M.R. Gruber and C.D. Carter
“Diode laser-based detection of combustor instabilities with application to a scramjet
engine”, Proc. Combust. Inst. 32, (2009) 831–838
[28] S. T. Sanders, J. A. Baldwin, T. P. Jenkins, D. S. Baer and R. K. Hanson, “Diode-
laser sensor for monitoring multiple combustion parameters in pulse detonation engines”,
Proc. Combust. Inst. 28, (2000) 587–594
[29] Y. Deguchi, M. Noda and M. Abe, “Improvement of combustion control through
real-time measurement of O2 and CO concentrations in incinerators using diode laser
absorption spectroscopy”, Proc. Combust. Inst. 29, (2002) 147-153
137
[30] V. Drasek, W. Wehe and M.G. Allen, “Laser-based multiple gas species sensor for
harsh combustion process control applications” Trends in Optics and Photonics,
Conference on Lasers and Electro-Optics (2004) 96
[31] A.A. Kosterev, C. Roller, F.K. Tittel and W. Flory, “ Development of a QC-laser
based system for industrial gas monitoring” (2003) OSA/CLEO
[32] R. Sur, K. Sun, J. B. Jeffries, R. K. Hanson, “Multi-species laser absorption sensors
for in-situ monitoring of syngas composition”, Appl. Phys. B (2013) DOI:
10.1007/s00340-013-5567-2
[33] W. Ren, A. Farooq, D.F. Davidson, R.K. Hanson, “CO concentration and
temperature sensor for combustion gases using quantum-cascade laser absorption near
4.7μm” , Appl. Phys. B 107, (2012) 849-860
[34] S. Li, A. Farooq and R.K. Hanson, “H2O temperature sensor for low-pressure flames
using tunable diode laser absorption near 2.9µm”, Measure. Sci. and Tech. 22, (2011)
125301
[35] X. Chao, J.B. Jeffries and R.K. Hanson, “Absorption sensor for NO in combustion
gases with a 5.2µm quantum-cascade laser”, Proc. Combust. Inst. 33, (2011)725–733
[36] X. Liu, J.B. Jeffries and R.K. Hanson, “Measurement of non-uniform temperature
distributions using line-of-sight absorption spectroscopy”, AIAA journal 45 , (2007) 390-
419
[37] X. Liu, J.B. Jeffries and R.K. Hanson, “Development of a tunable diode laser sensor
for measurements of gas turbine exhaust temperature”, Appl. Phys. B 82, (2006) 469-478
138
[38] X. Zhou J.B. Jeffries and R.K. Hanson, “Development of a fast temperature sensor
for combustion gases using a single tunable diode laser”, Appl. Phys. B 81, (2006) 711-
722
[39] J. Reid and D. Labrie, “Second-harmonic Detection with tunable diode lasers-
comparison of experiment and theory”, Appl. Phys. B 26, (1981) 203-210
[40]P. Werle, “Spectroscopic trace gas analysis using semiconductor diode lasers”,
Spectrochimica Acta Part A, 52, (1996) 805-822
[41] R. Sur, T. J. Boucher, M. W. Renfro and B. M. Cetegen, “In-situ measurements of
water vapor partial pressure and temperature dynamics in a PEM fuel cell”, J.
Electrochem. Soc. 157 (2010) B45-B53
[42] X. Chao, J.B. Jeffries and R.K. Hanson, “Absorption sensor for CO in combustion
gases using 2.3 μm tunable diode lasers”, Measure. Sci. and Tech. 20, (2009) 115201
(9pp)
[43] D.T. Cassidy and L.J. Bonnell, “Trace gas detection with short-external-
cavity InGaAsP diode laser transmitter modules operating at 1.58 µm”, Appl. Opt. 27,
(1988) 2688–2693
[44] H. Li, G.B. Rieker, X. Liu, J.B. Jeffries and R.K. Hanson, “Extension of
wavelength-modulation spectroscopy to large modulation depth for diode laser
absorption measurements in high-pressure gases”, Appl. Opt. 45, (2006)1052-1061
[45] D.T. Cassidy, J. Reid, “Atmospheric pressure monitoring of trace gases using
tunable diode lasers”, Appl. Opt 21,(1982) 1185-1190
139
[46] T. Fernholz, H. Teichert and V. Ebert, “ Digital, Phase-Sensitive Detection for In
Situ Diode Laser Spectroscopy under Rapidly Changing Transmission Conditions”,
Appl. Phys. B 75, (2002) 229–236
[47] K. Sun, X. Chao, R. Sur, J.B. Jeffries and R.K. Hanson, “Wavelength modulation
diode laser absorption spectroscopy for high pressure gas sensing”, App. Phys. B 110,
(2013)497-508
[48] G.B. Rieker, J.B. Jeffries and R.K. Hanson, “Calibration-free wavelength-
modulation spectroscopy for measurements of gas temperature and concentration in harsh
environments”, Appl. Opt. 48, (2009) 5546-5560
[49] K. Uehara and H. Tai, “Remote detection of methane with a 1.66-µm diode laser”,
Appl. Opt. 31, (1992) 809–814
[50] K. Sun, R. Sur, X.Chao, J.B. Jeffries and R.K. Hanson, R. J. Pummill, K.J. Whitty,
“TDL absorption sensors for gas temperature and concentrations in a high-pressure
entrained-flow coal gasifier”, Proc. Combust. Inst. 34, (2013) 3593-3601
[51] X. Chao, J.B. Jeffries and R.K. Hanson, “Development of laser absorption
techniques for real-time, in-situ dual-species monitoring (NO/NH3, CO/O2) in
combustion exhaust”, Proc. Combust. Inst. 34, (2013) 3583-3592
[52] K.Sun, R.Sur, J.B. Jeffries and R.K. Hanson, “Calibration-free wavelength-scanned
WMS H2O absorption measurements in an engineering scale high pressure coal gasifier”,
Submitted to App. Phys. B in 08/2013
[53] A.B. Mclean, C.E.J. Mitchell and D.M. Swanston, “Implementation of an efficient
analytical approximation to the Voigt function for photoemission lineshape analysis”, J.
Electron Spectrosc. 69,(1994) 125-132
140
[54] T.F. Wall, “Combustion processes for carbon capture”, Proc. Combust. Inst.
31,(2007) 31-47
[55] A.N. Dharamsi, “ A theory of modulation spectroscopy with applications of higher
harmonic detection”, J. Phys. D:Appl. Phys. 29, (1996) 540-549
[56] P. Kluczynski, J. Gustafsson, A.M. Lindberg and O. Axner, “Wavelength
modulation absorption spectrometry - an extensive scrutiny of the generation of signals”,
Spectrochim Acta B 56,(2001) 1277-1354
[57] S. Schilt, L. Thevenaz and P. Robert, “Wavelength modulation spectroscopy:
combined frequency and intensity laser modulation”, Appl. Opt. 42, (2003) 6728-6738
[58] G.V.H. Wilson, “Modulation broadening of NMR and ESR line shapes”, J. Appl.
Phys. 34, (1963) 3276-3285
[59] L. C. Philippe and R. K. Hanson, “Laser diode wavelength-modulation spectroscopy
for simultaneous measurement of temperature, pressure, and velocity in shock-heated
oxygen flows”, Appl. Opt. 32, (1993) 6090–6103
[60] P. Kluczynski, A.M. Lindberg, O. Axner, “Wavelength modulation diode laser
absorption signals from Doppler broadened absorption profiles”, J. Quant. Spectrosc. and
Radiat. Transfer 83, (2004) 345-360
[61] R. Arndt, “Analytical line shapes for Lorentzian signals broadened by modulation”,
J. Appl. Phys. 36, (1965) 2522-2524
[62] O. Axner, P. Kluczynski and A.M. Lindberg, “A general noncomplex analytical
expression for the n:th Fourier component of a wavelength-modulated Lorentzian line-
shape function”, J. Quant. Spectrosc. and Radiat. Transfer 68, (2001) 299-317
141
[63] X. Chao, J.B. Jeffries and R.K. Hanson, “Wavelength-modulation-spectroscopy for
real-time, in situ NO detection in combustion gases with a 5.2 µm quantum-cascade
laser”, Appl. Phys B, 106, (2012) 987-997
[64] L.S. Rothman, I.E. Gordon, R.J. Barber, H. Dothe, R.R. Gamache, A. Goldman, V.
Perevalov, S.A. Tashkun and J. Tennyson “HITEMP, the high-temperature molecular
spectroscopic database” J. Quant. Spectrosc. and Rad. Transfer 111, (2010) 2139-2150
[65] J. Gustafsson and O. Axner, “ 'Intelligent' triggering methodology for improved
detectability of wavelength modulation diode laser absorption spectrometry applied to
window-equipped graphite furnaces” Spectrochimica Acta Part B 58, (2003) 143-152
[66] J. Gustafsson, N. Chekalin and O. Axner, “Characterization of 2f-, 4f-, and 6f-
background signals in wavelength modulation diode laser absorption spectrometry in
graphite furnaces”, Spectrochimica Acta Part B 58, (2003) 123-141
[67] P. Kluczynski, A. Lindberg, and O. Axner, Background signals in wavelength-
modulation spectrometry with frequency doubled diode-laser light. II. experiment,” Appl.
Opt. 40, (2001) 794– 804
[68] C. R. Webster, "Brewster-Plate Spoiler: a Novel Method for Reducing the
Amplitude of Interference Fringes that Limit Tunable-Laser Absorption Sensitivities," J.
Opt. Soc. Am. B 2, (1985).1464–1470
[69]S.J. Clayton, G.J. Stiegel, J.G. Wimer, US DoE report DOE/EF-0447 (2002)
[70] R. Sur, K. Sun, J.B. Jeffries, R.K. Hanson, R.J. Pummill, T. Waind, R.R. Wagner
and K.J. Whitty, “TDLAS-based sensors for in-situ measurement of syngas composition
in a pressurized, oxygen-blown, entrained flow coal gasifier”, App. Phys. B (2013) DOI:
10.1007/s00340-013-5644-6
142
[71] C. L. Strand and R. K. Hanson, “Thermometry and velocimetry in supersonic flows
via scanned wavelength-Modulation absorption spectroscopy”, 47th AIAA /ASME /SAE
/ASEE Joint Propulsion Conference & Exhibit 5600 (2011)
[72] C.S. Goldenstein, C. L. Strand, I.A. Schultz, K. Sun, J. B. Jeffries and R. K. Hanson,
“Fitting of calibration-free scanned-wavelength-modulation spectroscopy spectra for
determination of gas properties and absorption lineshapes” submitted to Appl. Opt. in
08/2013
[73] G. Stewart, W. Johnstone, J.R. Bain, K. Ruxton, K. Duffin, “Recovery of Absolute
Gas Absorption Line Shapes Using Tuneable Diode Laser Spectroscopy with Wavelength
Modulation – Part I: Theoretical Analysis”, Journal of Lightwave Technology 29, (2011)
811-821
[74] J. R. P. Bain, W. Johnstone, K. Ruxton, G. Stewart, M. Lengden, K. Duffin,
“Recovery of Absolute Gas Absorption Line Shapes Using Tuneable Diode Laser
Spectroscopy with Wavelength Modulation – Part 2: Experimental Investigation”,
Journal of Lightwave Technology 29, (2011) 987–996
[75] X. Liu, J.B. Jeffries and R.K. Hanson, “Measurements of spectral parameters of
water-vapor transitions near 1388 and 1345 nm for accurate simulation of high-pressure
absorption spectra”, Measure. Sci. and Tech. 18, (2007) 1185-1194
