Content uploaded by Scott Saleska
Author content
All content in this area was uploaded by Scott Saleska on Mar 18, 2016
Content may be subject to copyright.
A high precision pulsed quantum cascade laser spectrometer for
measurements of stable isotopes of carbon dioxide
J. B. MCMANUS*y, D. D. NELSONy, J. H. SHORTERy,
R. JIMENEZz, S. HERNDONy, S. SALESKA§ and M. ZAHNISERy
yAerodyne Research, Inc., 45 Manning Road, Billerica, MA US 01821
zHarvard University, Department of Earth and Planetary Sciences,
20 Oxford Street, Cambridge, MA, US 02138
§University of Arizona, Department of Ecology and
Evolutionary Biology, Tucson, AZ 85721, USA
(Received 15 February 2005; in final form 2 July 2005)
We describe a prototype instrument using a Peltier cooled quantum cascade
laser for precise measurement of stable carbon (13C/12 C) isotopologue ratios in
atmospheric CO
2
. Using novel optics and signal processing techniques in a
compact instrument, we are able to detect the difference between sample and
reference with a precision of 0.1ø(2standard error of mean of 11 samples)
in 10 min of analysis time. The standard deviation of 0.18øfor individual 30 s
measurements shows that this prototype instrument already approaches the best
reported literature values using continuous wave lead alloy tunable diode lasers.
The application of pulsed near room-temperature quantum cascade lasers to this
demanding problem opens the possibility of field worthy rapid response isotopic
instrumentation and attests to the maturity of these lasers as spectroscopic
sources.
1. Introduction
The measurement of isotopic ratios in the natural environment is a challenging
problem that can be addressed with high resolution laser spectroscopic techniques
[1–9]. Most of the prior work in laser spectroscopic measurement of isotopic
ratios has been with continuous wave (CW) cryogenically cooled lasers (especially
mid-infrared lead alloy semiconductor lasers) [1–7]. The advent of pulsed near room-
temperature quantum cascade lasers (QCLs) [10, 11] opens the possibility of field
worthy rapid response instrumentation for measurement of stable isotopologues.
New isotopic measurement modes may be implemented, such as eddy covariance
flux determination. The successful application of QCLs to isotopic measurements
also attests to their maturity as spectroscopic sources.
*Corresponding author. Email: mcmanus@aerodyne.com
Journal of Modern Optics
Vol. 52, No. 16, 10 November 2005, 2309–2321
Journal of Modern Optics
ISSN 0950–0340 print/ISSN 1362–3044 online #2005 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/09500340500303710
Measurement of the ratios of atomic isotopes in the molecules that participate
in environmental exchange processes is an important tool for understanding those
processes and the natural environment [12–15]. This is because there are charac-
teristic isotopic enrichments or depletions generated by the chemical reactions or
biochemical processes leading to sources and sinks of atmospheric trace gases. Stable
isotope abundances are typically expressed in terms of delta units (,ø), which are a
comparison of the ratio of two isotopes in a sample (e.g. Rx¼13 C=12 C), to the same
ratio in a standard, R
std
. The delta unit is defined as: x¼½ðRxRstdÞ=Rstd1000.
Isotopic deltas of interest in the natural environment may be 1øor less, and the
desired measurement precision is 0.1ø.
Isotopic ratio measurements are difficult because the ratio variations are small,
the minor isotopes are relatively rare, and the molecules of interest may be quite
dilute in the natural environment. As an example, consider one of the stronger
isotopic enhancements due to biological processes, the fixing of carbon in photo-
synthesis [12, 13]. Different classes of plants discriminate differently against uptake
of 13CO2relative to 12 CO2, so that woody plants (‘C3’ photosynthesis) have
13C16øand grasses (‘C4’ photosynthesis) have 13 C4ø, relative to
atmospheric CO
2
. One can measure the ratio 13 C/12C to identify the class of plants
involved in atmospheric sources and sinks of CO
2
[14, 15]. In atmospheric sampling,
the average concentration of CO
2
is 380 ppm and the concentration of 13CO2
is 3.8 ppm. The difference in concentration of 13CO2produced by burning C3
versus C4 plants (with the resulting CO
2
diluted to near background levels) would be
0.05 ppm. The standard method for determining trace gas stable isotope ratios
is isotope ratio mass spectrometry (IRMS) [12, 13], which provides 13 C with
precision of 0.01–0.05ø. However, IRMS has a number of experimental draw-
backs, especially in the great care needed in sample preparation (often involving
chemical processing or purification). Commercial IRMS units require permanent
installation for reliable operation, so that field samples are normally returned to
the laboratory and analysed long after they are taken.
Real time continuous isotopic measurements would greatly facilitate the
investigation of environmental exchange processes, allowing guidance of sampling
by real time analysis and enabling new techniques. If isotopic measurements can be
made with sufficient accuracy at a rate of 10 Hz, then the correlation of upward
wind velocity with concentration yields area flux, a technique called eddy covariance
flux measurement. Recently, lead alloy tunable diode lasers (TDLs) have been
applied to continuous in situ atmospheric isotope ratio measurements [1–7].
Several of these systems have been reported with precisions on the order of 0.1ø
[1, 4–6]. Lead alloy TDLs represent an advance in isotope measurement methods
but they are far from ideal, due to the need for liquid nitrogen (LN2) cooling, and
the lasers often change characteristics after cooling cycles.
Quantum cascade lasers (QCLs) provide alternative sources of high resolution
mid-infrared radiation that have overcome some of the difficulties associated with
lead alloy TDLs. QCLs can operate in the mid-infrared, where many molecules
display their strongest fundamental vibrational bands, for the highest sensitivity.
QCLs can operate near room temperature, and offer single-mode continuous tuning
and good power output. Since QC lasers were first demonstrated [16], they have
2310 J. B. McManus et al.
undergone rapid development [10, 11] and been used for atmospheric trace gas
measurements, in CW operation with LN2 cooling [17] and pulsed with thermo-
electric cooling [8, 9, 18–22].
In this paper we describe an instrument based on a pulsed quantum cascade laser
for precise measurement of carbon isotope ratios (13C/12 C) and present laboratory
results. The new QC laser spectrometer employs a dual path length optical absorp-
tion cell [1] and improved signal processing techniques. The precision and minimal
detectable absorbance we obtain with pulsed QC lasers is comparable to that
achieved with LN2 cooled CW TDLs, despite the broader linewidths inherent to
pulsed operation.
2. Instrument design
In the design of this instrument we have addressed problems that are specific to
isotopes and general problems for precision measurements with pulsed QC lasers.
At the needed level of precision (of 104, or 0.1ø) a number of subtle effects can
limit the ultimate performance, such as the method of operation of the laser, or
choices of spectral features and the method of deriving concentrations. The isotopic
specific measurement problems include the large difference in abundances of
the minor and major isotopes, and the temperature dependence of absorption
strengths. Our group at Aerodyne Research, Inc. has worked over the past several
years improving the precision of pulsed QCL spectrometers for several applications,
resulting in many of the general approaches described here [20–22]. The QCL
spectrometer combines commercially available QC lasers, an optical system, and
a computer-controlled electronic system for driving the laser and data processing.
The spectrometer is designed for simultaneous measurement of sample, pulse
normalization and frequency-lock spectra. We have used commercially available
distributed feedback InGaAs–AlInAs/InP QC lasers designed for pulsed operation
(Alpes Lasers, Neuchaˆ tel, Switzerland). An essential aspect of this work is that
it has been possible to select a QC laser that covers the frequency range from 2310
to 2314 cm
1
, which meets the line selection criteria described below.
Pulsed laser operation presents two main problems for high precision mea-
surements, noise due to pulse-to-pulse energy variations and the increase in laser
linewidth. We have greatly reduced the problem of pulse-to-pulse variations by
normalizing [20–22]. By taking advantage of the time delay of light propagation
through the multipass cell, we can use a single infrared detector for both the
absorption signal and pulse normalization. The problem of laser linewidth and
the associated problem of lineshape stability remain precision limiting issues. Even
though we minimize the linewidth by using short pulses (10 ns) and operate the
laser close to threshold, the laser width is greater than the molecular (Doppler)
absorption linewidth.
Stability against temperature changes is a key factor in precision measurements
and the basic thermal stability may be improved by careful selection of the two
absorption lines. Infrared linestrengths vary with temperature, so drift in the instru-
ment temperature can mask small isotopic deltas [5–7]. However, if the transitions
A high precision pulsed quantum cascade laser spectrometer 2311
for both isotopes have similar lower state energy levels, the temperature variation
of the two line strengths are closely matched. If the instrument temperature is to
within 0.1 K, the lower state energy level separated can be only 100 to 300 cm
1
to
maintain a precision of 0.2 to 0.5ø[6]. We have previously reported surveys of
thermal sensitivity for lines for the isotopes of CO
2
and CH
4
[1]. The line pairs
around 2314 cm
1
require a temperature stability of only 0.2 K to obtain a precision
of 0.1ø[1]. The temperature stability requirement is negligible using the line pairs
at 2311 cm
1
suggested by Weidmann et al. [8], which we have used extensively in
this study. The choice of spectral features to reduce thermal sensitivity may be less
optimal for gases other than CO
2
, since the following factors also need considera-
tion.
1. When using a single laser, lines from the major and minor isotopes must be
within the frequency scanning range of the laser.
2. The lines should have large transition moments, for maximum sensitivity.
3. The lines should be free of interference from other atmospheric gases.
A key design feature of our laser spectrometer is a dual-path multipass cell used
to compensate for the large difference in concentration between major and minor
isotopologues of CO
2
, an effect limiting precision. When long path lengths are
used to increase the absorption depth due to the minor constituent, then the major
constituent absorption depth may be excessive, giving essentially no transmission at
line centre. Our approach uses two different paths through the same measurement
volume for the major and minor isotope, where the ratio of path lengths is similar to
the abundance ratio, yielding a similar absorption depth for both isotopologues
[1]. The dual path-length cell is based on an astigmatic Herriott cell [23], with
the two paths corresponding to either 174 or 2 traverses of the 0.32 m base length [23].
A diagram of the cell coupling arrangement is shown in figure 1. The beam for the
174 pass pattern (for 13CO2) is injected at an angle to the centre axis and it exits
on the opposite side of the axis. The 2 pass pattern is generated by placing a
beamsplitter (BaF
2
) on the centre axis, reflecting 8% of the light straight into
the cell. The 2-pass beam then returns on the same axis after reflecting from the
back mirror. The resulting optical path lengths of 56.08 and 0.753 m have a ratio
of 74:1, nearly balancing the 12CO2=13CO2isotopomer ratio of 91:1. The dual
pathlength concept also greatly lessens the effect of temperature and pressure
variations since both isotopomers are measured in the same volume.
The instrument optical system, shown in figure 2, collects light from the QC laser
and directs it along four different paths: two paths through the absorption cell,
plus the pulse normalization and frequency-locking reference paths. The highly
divergent QC laser beam is collected by a reflecting microscope objective and
imaged at a pinhole (used only for initial alignment). After the first focus,
the beam passes through a wedged BaF
2
beamsplitter. The front reflection beam is
used for the pulse normalization spectrum, which defines the wavelength dependence
of the laser power, along with its temporal variability. The optical path length
of the pulse normalization beam is matched to the sample path length outside the
absorption cell, insuring that residual atmospheric absorption in that path is
cancelled. The second reflection from the beamsplitter is directed through a short
2312 J. B. McManus et al.
absorption cell with low pressure CO
2
, and its spectrum is used to actively control
the average frequency of the laser by changing the Peltier temperature. The beam
transmitted through the beam splitter is re-imaged into the astigmatic Herriott
multipass cell [23]. In this study, we sample at a relatively low pressure (7 torr),
which serves to reduce consumption of sample gas, minimize overlap of adjacent
lines and reduce optical depth at ambient concentrations. The optical board is
thermally controlled to better than 1 K and the instrument enclosure is purged with
CO
2
free dry air.
The spectral measurement is based on rapidly sweeping the laser across the
selected absorption lines, digitizing the direct-absorption signals and quantitatively
174 Pass Outpu
t
2-Pass Output
Laser Input
BaF2 Beamsplitter
Astigmatic
Herriott Cell
56m @ 174 passes
0.75m @ 2 passes
Figure 1. Optical arrangement to produce dual pathlength optical absorption cell.
A high precision pulsed quantum cascade laser spectrometer 2313
fitting the spectra to derive absolute concentrations. A unified computer system,
using ‘TDL Wintel’ software developed at Aerodyne Research, Inc., drives the
laser and synchronously analyses data in real time. A diagram of the signal
processing system is shown in figure 3. The laser is excited just above threshold
with 10–20 ns electrical pulses at 1 MHz. Spectral scans (over 0.5 cm
1
) are
Dual
Detector
QCL in housing
Trace Laser
Beamsplitter
Absorption Cell
15x
Main Path
Norm. Path
174 2
0.33m
0.64m
Figure 2. Optical layout of single quantum cascade laser instrument. The optical base
measures 0:33 0:64 m.
~100
Pulse
s
~0.5 cm−1
Ramp
Peltier
100 µs
1µs
Current
QCL
i(t)
Absorption
Detector
Frequency
t
time
V(t)
S
PN
PN S PN S
…..
1µs
100 µs
100 µs
VN(t)
VS(t)
Average
Divide
Fit spectra
Calc. Conc.
Adj. Peltier
~10 ns
Figure 3. Laser control and signal processing schematic.
2314 J. B. McManus et al.
obtained by simultaneously applying a sub-threshold current ramp, which modulates
the laser temperature and thus its spectral frequency. Fast (5 MHz) DAC/ADC
boards synchronously trigger the pulse electronics and integrate the resulting signals
from LN2-cooled InSb photodiodes. The DAC board also generates a TTL gate that
defines the ramp duration, including a short laser-off period at the end of each sweep
to measure the detector offset. Since this instrument employs separate beam paths
and detectors to monitor each isotope simultaneously, we employ dual data
acquisition processes operating in parallel, to collect data through separate
acquisition boards. We run two instances of TDL Wintel with processes that are
precisely synchronized by operating the instances as ‘master and slave’. The master
instance controls the laser and sends timing signals to the slave, which then
constructs appropriate timing gates for its detector. A typical spectral sweep consists
of 100–400 pulses, giving a spectral sweep rate of 2.5–10 kHz. Concentrations and
laser linewidths are determined in real time from the spectra through a nonlinear
least-squares fitting algorithm (Levenberg-Marquardt) that uses fitting functions
composed from Voigt [24] molecular profiles (based on the HITRAN [25] database
plus measured pressure and temperature), convolved with a (Gaussian) laser line-
shape and added to a polynomial baseline. Derived mixing ratios are typically
accurate to 5% without calibration.
3. Laboratory data
The basic data produced by the instrument are the two absorption spectra, for 13 CO2
using the long path (56 m) and 12CO2using the short path (0.74 m). The spectral pair
near 2311 cm
1
is shown in figure 4, for a sample with CO
2
concentration 350 ppm
at a pressure of 7.5 torr. The data here are reported at 1 Hz, where the concentration
noise is 150 ppb. Each panel in figure 4 displays three curves, (1) the experimental
absorption spectrum, (2) a spectral fit to the data, a convolution of the Voigt
molecular lineshape with the laser lineshape, and (3) a spectral simulation of only
the molecular lineshape. We employ a simple Gaussian instrument profile in the
convolution, which does a good job of capturing the measured lineshape and area,
even though some details of the observed shape are missed by using a symmetrical
profile. Comparing the molecular line to the data shows the degradation in
resolution due to the significant laser linewidth. For the 13 CO2absorption line
the apparent peak absorbance is 0.1, but the actual peak absorbance is three times
greater. The 12CO2line at 2311.105 cm
1
(figure 4(a)) is optically black at centre
even though that is not apparent from the experimental spectrum. Nevertheless,
the analysis does a good job of reproducing the experimental 12 CO2spectrum. These
lines are strong enough that the absorbance area is not conserved and accurate
results can only be obtained by doing a convolution of the molecular spectrum with
the instrumental width.
In figure 5 we show in more detail the contributions to the spectrum in the
long absorption path for 13CO2. The underlying 12CO2lines contribute significantly
to the absorption area in the wings of the 13 CO2absorption region, despite the low
pressure and good separation. To compensate for this effect, the fitting procedure is
A high precision pulsed quantum cascade laser spectrometer 2315
done for the two isotopomers simultaneously. Although it would be possible to vary
both species in the fit, there is insufficient 12 CO2absorption in the 13 CO2region to
accurately retrieve its concentration. Instead, the 12 CO2concentration is set to the
more accurate value determined from the short path absorption spectrum. This
greatly reduces the dependence of the 13 CO2=12CO
2
ratio determination on
variations in the 12CO2concentration.
The long term stability of the instrument is evaluated using the Allan variance
technique, which distinguishes high frequency random noise from drifts at
longer time scales [26]. The contributions of random noise to the variance (2)
decrease as t
1
, so a plot of variance with increasing integration time is expected to
have a (log-log) slope of 1 at short times, while at longer times the variance curve
levels off and rises due to drifts. In figure 6 we show time series data and an Allan
1.0
(a)
(b)
0.8
0.6
0.4
0.2
0.0
TRANSMISSION
2311.42311.32311.22311.12311.0
FREQUENCY (cm−1)
1.0
0.8
0.6
0.4
0.2
0.0
TRANSMISSION
174 PASS - 56m
2 PASS - 0.74 m
SPECTRA
CONVOLVED FIT
VOIGT SHAPE
7.5 Torr
12CO2 350 ppm
Laser Line Width 0.0105 cm−1
13CO2
2311.399
12CO2
2311.106
Figure 4. CO
2
isotopologue spectra with pulsed QC laser. The cell contains 350 ppm
CO
2
in air at 7.5 torr. The laser linewidth is 0.0105 cm
1
hwhm. The graphs show recorded
spectra (large dots), fit functions (solid lines) and simulated Voigt absorption lines (dashed
lines), for the long path (upper panel, (a)) and short path (lower panel, (b)).
2316 J. B. McManus et al.
plot of both isotopologues and their ratio for a 10 h period. The 13 CO2=12CO
2
ratio
has a 1 s standard deviation of 0.5ø. However, the variance does not decrease with
t1 as would be expected for random (white) noise, but rather behaves as f0:5
‘pink’ noise, indicating that that there are correlations in noise sources at time scales
of 1–100 s. The variance minimum at 200 s corresponds to a minimum
Allan
of 0.1ø.
The precision of an isotopic ratio measurement can be improved by alternating
between a sample and reference, with the time for each measurement less than
the averaging at which the Allan variance begins to increase. We present an
alternating measurement in figure 7, sampling gases from two tanks: the ‘sample’
containing 350 ppm total 12 CO2and unknown 13 C, was compared with a ‘reference’
mixture of CO
2
with 13CPDB ¼49:4ø(determined by mass spectrometry
courtesy of Professor Dan Schrag at Harvard University). For the data in figure 7,
the averaging time for each sample was 25 s, with 5 s between samples for gas
flushing. For this experiment we observe that the sample air has 13CCO2¼14:9ø,
relative to PDB. The difference between sample and reference is obtained with a
precision of 0.1ø(2standard error of mean of 11 samples) in 10 min of analysis
time. The standard deviation of 0.18øin individual 30 s intervals shows that this
prototype instrument already approaches the best reported literature values using
1.00
0.98
0.96
0.94
0.92
TRANSMISSION
2311.452311.402311.352311.302311.252311.20
FREQUENCY
(
cm
−1
)
SPECTRUM
12
CO
2
350 ppm (FIXED)
13
CO
2
336 ppm (FIT)
Pressure 7.5 Torr
Laser Line Width 0.0105 cm
−1
Figure 5. Detail of the contributions to the fit spectrum in the long absorption path.
There are small contributions from a weak 12CO2line that is nearly coincident with the
13CO2line.
A high precision pulsed quantum cascade laser spectrometer 2317
continuous lead-alloy TDLs [4]. While the averaged data yields a best precision
in delta of 0.1ø, the large-signal accuracy of the spectroscopically derived delta is
substantially less (10%). The spectroscopic accuracy is limited by uncertainty
in the tabulated HITRAN line strengths, and by uncertainty in the spectral baseline,
which relates to uncertainty in the laser lineshape. To achieve higher accuracy it is
necessary to calibrate with standards.
4. Discussion
In the prototype instrument described above, we achieve a precision in isotopic ratio
measurements of 0.18øin 30 s (1) and 0.1øin 600 s (2). This result is achieved
by carefully controlling numerous system parameters, especially laser base tempera-
ture and average centre wavelength, as well as the temperature of the optical
10−8
10−7
Allan Var (σ2)
Allan Var (σ2)
100101102103104
INTEGRATION TIME (s)
0.988
0.987
0.986
0.985
0.984
0.983
RATIO
1/20/2005
9:00AM 12:00 PM 3:00 PM
364×103
362
360
358
CO2 (ppb)
103
2
4
104
2
4
σAllan=45 ppb
τAllan=200 s
σAllan=0.1 ‰
τAllan=200 s
12CO2
13CO2 / 0.01124
1 s rms 0.5 ‰
Figure 6. Noise in isotope ratio and Allan plot showing long term drift behaviour.
2318 J. B. McManus et al.
table and the pulse electronics. We believe the largest source of noise and drift in the
1 to 100 s time scale is residual laser frequency fluctuations, affecting both the peak
position and the laser linewidth. Another significant drift effect is variation in the
residual CO
2
in the optical enclosure.
The success to date of our QC laser spectrometer for isotopic measurements
provides the opportunity to comment on the utility of these lasers as spectroscopic
sources. The measurement method presented is possible because a high-quality QCL
was available at the desired wavelength, a significant factor given the current scarcity
of QC laser manufacturers. The laser is very well behaved in tuning with temperature
and current. The reproducibility of the output as we ramp and pulse the current
contributes to the final precision performance of the instrument. Pulse energy noise
can be divided out with our ‘pulse normalization’ approach, greatly reducing
multiplicative noise. With careful system control, we can achieve noise performance
approaching the best reported literature values using CW TDLs [4]. The laser used
in this study has operated continuously for more than three months with negligible
change in characteristics.
There are several directions available to improve the performance of this
instrument. We continue to improve stabilization and control of the laser and
345×103
340
335
330
325
CO2 (ppb)
8:52 AM
1/21/2005
8:54 AM 8:56 AM 8:58 AM 9:00 AM 9:02 AM
0.97
0.96
0.95
0.94
RATIO / 0.01124
35.5×10−3
35.0
34.5
34.0
33.5
DIFFERENCE
STD DEV = 0.18‰
SAMPLE-REFERENCE
mean = 34.5 ± 0.1 ‰ (2σ,n=11)
12CO213CO2 / 0.01124
RMS = 0.46 ‰ (1 Hz)
SAMPLE
REFERENCE
Figure 7. Alternating measurements of 13CO2=12CO2in sample and reference air from
tanks. Reference was made by diluting pure CO
2
(13CPDB ¼49:4ø) with dry N
2
to match
sample gas ½12CO2of 350 ppm. Reference gas ½12CO2of 345 ppm (bottom panel) and
isotope ratio of 0:940 ¼60ø(middle panel) are directly retrieved (uncalibrated) values.
The flow rate of 0.5 SLPM in a cell volume of 0.5 l at pressure of 7 torr corresponds to a cell
flushing (1/e) time of 0.6 s. The first 5 s of each 30 s interval were discarded.
A high precision pulsed quantum cascade laser spectrometer 2319
instrument overall, which has been a key element of the performance achieved to
date. While we have focused to date on 13 CO2and 12CO2, we could also apply these
techniques to, e.g. 13CH4=12 CH4or 12 C18O16 O=12 C16O2. If we can achieve sufficient
precision with fast response time then we could measure isotopic fluxes by eddy
covariance. We have previously presented a dual QCL instrument [22] and we plan
to modify that design to include the isotope dual-path optics. The instrument
presented here has a LN2 cooled detector, but with continued improvement
in thermoelectrically cooled infrared detectors a fully non-cryogenic instrument
is possible. This would allow completely unattended long term operation and greatly
advance the application of infrared spectroscopy to isotope analysis.
5. Summary and conclusions
A high precision pulsed-quantum cascade laser spectrometer operating in the
mid-infrared (2311 cm
1
) has been demonstrated to determine isotopic abundances
by measuring 13CO2=12 CO
2
in atmospheric air. The instrument uses a dual path-
length absorption cell with a longer path (56 m) for the minor isotope and a
shorter path for the major isotope (0.74 m), with absorption lines selected for
similar lower state energies to negate the effects of sample temperature variation.
By stabilizing the pulse electronics and normalizing pulse-to-pulse intensity varia-
tions, measurement precision for each isotopologue of 5 104(1 s rms) relative to
their ambient mixing ratios may be achieved. The difference in the isotopologue
ratio between two samples of ambient air can be determined by alternating between
sample and reference with a replicate precision of 0.18ø(1) standard deviation
for 30 s averaging intervals. This corresponds to a standard error of mean over
11 cycles of 0.1ø(2) in 10 min of measurement time alternating between sample
and reference in a flowing system.
Acknowledgments
We gratefully acknowledge the contributions of Patrick Kirwin and Jeff Mulholland
(Aerodyne Research, Inc.) to the engineering and construction of the QCL spectro-
meter. We also thank Steve Wofsy and Bruce Daube (Harvard University) and
Frank Tittel (Rice University) for valuable advice and discussions. We thank
Dan Schrag at Harvard University for isotopic analysis. We thank Antoine Muller,
Yargo Bonneti, Guillaume Vanderputte and Stephan Blaser of AlpesLasers for
providing the QC laser and other high quality components used in this system.
Funding for instrument development has been provided by the US Department of
Energy STTR and SBIR programs.
References
[1] J.B. McManus, M.S. Zahniser, D.D. Nelson, et al., Spectrochim. Acta A 58 2465 (2002).
[2] K. Uehara, K. Yamamoto, K. Kikugawa, et al., Sens. Actuators B 74 173 (2001).
2320 J. B. McManus et al.
[3] T.J. Griffis, J.M. Baker, S.D. Sargent, et al., Agric. Forest Meteorol. 124 15 (2004).
[4] D.R. Bowling, S.D. Sargent, B.D. Tanner, et al., Agric. Forest Meteorol. 118 1 (2003).
[5] P. Bergamaschi, A.M. Brenninkmeijer, M. Hahn, et al., J. Geophys. Res. 103 8227 (1998).
[6] P. Bergamaschi, M. Schupp and G.W. Harris, Appl. Opt. 33 7704 (1994).
[7] J.F. Becker, T.B. Sauke and M. Loewenstein, Appl. Opt. 31 1921 (1992).
[8] D. Weidmann, G. Wysocki, C. Oppenheimer, et al., Appl. Phys. B Lasers Opt. 80
255 (2005).
[9] A.A. Kosterev, R.F. Curl, F.K. Tittel, et al., Opt. Lett. 24 1764 (1999).
[10] J. Faist, F. Capasso, C. Sirtori, et al., Electron. Lett. 32 560 (1996)
[11] F. Capasso, R. Paiella, R. Martini, et al., IEEE J. Quantum Electron. 38 511 (2002).
[12] P. Ghosh and W.A. Brand, Int. J. Mass Spectrom. 228 1 (2003).
[13] S.E. Trumbore, in Biogenic Trace Gases: Measuring Emissions from Soil and Water,
edited by P.A. Matson and R. C. Harriss (Blackwell Science Ltd, Oxford, 1995).
[14] C.-T. Lai, A.J. Schauer, C. Owensby, et al., J. Geophys. Res. 108(D18) 4566 (2003).
[15] S.R. Saleska, S.D. Miller, D.M. Matross, et al., Science 302 1554 (2003).
[16] J. Faist, F. Capasso, D.L. Sivco, et al., Science 264 553 (1994).
[17] C.R. Webster, G.J. Flesch, D.C. Scott, et al., Appl. Opt. 40 321 (2001).
[18] A.A. Kosterev and F.K. Tittel, IEEE J. Quantum Electron. 38 582 (2002).
[19] A.A. Kosterev, F.K. Tittel, R. Kohler, et al., Appl. Opt. 41 1169 (2002).
[20] D.D. Nelson, J.H. Shorter, J.B. McManus, et al., Appl. Phys. B Lasers Opt. 75 343 (2002).
[21] D.D. Nelson, J.B. McManus, S. Urbanski, et al., Spectrochim. Acta A 60 3325 (2004).
[22] R. Jimenez, S. Herndon, J.H. Shorter, et al., SPIE Proc. 5738 37 (2005).
[23] J.B. McManus, P.L. Kebabian and M.S. Zahniser, Appl. Opt. 34 3336 (1995).
[24] J. Humlı
´cek, JQSRT 27 437 (1982).
[25] L.S. Rothman, A. Barbe, D.C. Benner, et al., JQSRT 82 5 (2003).
[26] D.W. Allan, Proc. IEEE 54 221 (1966).
A high precision pulsed quantum cascade laser spectrometer 2321
