Content uploaded by Jérôme Faist
Author content
All content in this area was uploaded by Jérôme Faist
Content may be subject to copyright.
Content uploaded by Edward Anthony Whittaker
Author content
All content in this area was uploaded by Edward Anthony Whittaker
Content may be subject to copyright.
February 1, 1998 / Vol. 23, No. 3 / OPTICS LETTERS 219
Sensitive absorption spectroscopy with a room-temperature
distributed-feedback quantum-cascade laser
K. Namjou, S. Cai, and E. A. Whittaker
Department of Physics and Engineering Physics, Stevens Institute of Technology, Hoboken, New Jersey 07030
J. Faist, C. Gmachl, F. Capasso, D. L. Sivco, and A. Y. Cho
Bell Laboratories, Lucent Technologies, 700 Mountain Avenue, Murray Hill, New Jersey 07974
Received September 18, 1997
We report what we believe are the first spectroscopic measurements to be made with a room-temperature
quantum-cascade distributed-feedback laser. Using wavelength modulation spectroscopy, we detected N2O
and CH4in the chemical f ingerprint wavelength range near 8 mm. The noise equivalent absorbance for our
measurement was 5 parts in 105, limited by excess amplitude modulation on the laser output, which corresponds
to a 1-Hz bandwidth detection limit of 250 parts N2Oin 109parts N2in a 1-m path length. 1998 Optical
Society of America
OCIS codes: 140.5960, 300.1030, 300.6340, 280.3420.
There is a wide range of measurement and control ap-
plications that require, in situ, chemically selective de-
tection of molecular species in the gas phase. Laser
absorption spectroscopy is ideally suited for this pur-
pose, and low detection limits have been reported.1–3
Because the strongest line strengths for molecules oc-
cur in the middle-infrared portion of the spectrum,
the lead salt family of tunable semiconductor diode
lasers has been the standard choice for use in this
spectral region. Effective methods for detecting weak
absorbances have been developed.3,4 However, as a
source in a practical sensing instrument these devices
have two limitations: Lead salt lasers must always be
operated cryogenically, and they are available commer-
cially only as Fabry – Perot cavity oscillators and hence
are subject to cavity mode hopping and tuning gaps.
The quantum-cascade (QC) laser5–8 overcomes these
two drawbacks, and we report here what is to our
knowledge the first application of the QC laser to the
detection of dilute samples of molecules through their
absorption spectrum. The QC laser derives its gain
from the transition of electrons between two excited
states in the conduction band of a coupled quantum-
well structure. Inversion is obtained by careful
engineering of the carrier lifetimes, determined by
longitudinal phonon scattering. The active region
and a matching electron injector are cascaded many
times to increase output power. The lasing wave-
length is determined by the thickness of the active
region’s quantum-well layer, essentially independently
of the semiconductor bandgap. Devices with wave-
lengths ranging from 4 to 11 mm have been fabricated
with an AlInAsyInGaAs lattice matched to InP.7This
spectral range overlaps the most important part of
the molecular fingerprint region and includes a
substantial fraction of the lead salt lasing range.
However, unlike the lead salt laser, the QC laser is
capable of operating at room temperature and above.9
Furthermore, room-temperature QC lasers with a
distributed-feedback (DFB) architecture have also
been demonstrated to have continuous, single-mode
tuning of .20 cm21.10
Figure 1 diagrams a sensitive detection measure-
ment that we implemented by using such a QC DFB
laser. The laser, designed to emit near 8 mm,10 is
mounted in a temperature controlled, evacuated cham-
ber. The laser is driven by a 2.6-A amplitude, 1-MHz
train of 11-ns-wide current pulses (,1% duty fac-
tor), resulting in 10-mW collected average laser out-
put power. The thermoelectric cooler temperature is
set to 4 ±C. This setup tunes the laser resonance close
to the N2O absorption line of interest. We fine tune
the laser by superimposing upon the pulsed waveform
a4-Hz current ramp, which alone would be insuffi-
cient in amplitude to bring the laser above thresh-
old but which causes a temperature modulation on
the laser suff icient to sweep the lasing frequency by
approximately one wave number. Finally, we com-
bine the ramp with a 1.8-kHz sinusoidal dither, which
adds a second, higher-frequency temperature modu-
lation, resulting in a concomitant wavelength modu-
lation (WM).11
Fig. 1. Experimental arrangement for wavelength modu-
lation spectroscopy with the quantum-cascade DFB laser.
0146-9592/98/030219-03$10.00/0 1998 Optical Society of America
220 OPTICS LETTERS / Vol. 23, No. 3 / February 1, 1998
The modulated laser beam is collimated and then
directed through a 10-cm-long gas cell equipped with
wedged CaF2windows. The transmitted light is fo-
cused onto a HgCdTe photoconductive detector, cooled
to 77 K. The detector is coupled to a matched pream-
plifier whose output is detected by a lock-in amplifier.
The wavelength-modulation dither serves as the lock-
in reference, and the lock-in output is recorded with
a14-bit digital signal-processor-equipped computer.
The computer records individual ramp scans and can
be set to average these scans. The data reported here
were recorded mostly by use of 16-scan averaging with
a lock-in bandwidth of 250 Hz, resulting in a net
detection bandwidth of 16 Hz. Because the dither fre-
quency is much lower than the underlying pulse rep-
etition frequency, this arrangement results in WM
absorption spectroscopy.
The gas cell was filled with commercially premixed
volumes of diluted gas, which were further diluted with
high-purity dry N2. For most of the measurements
reported here we used samples of 10%, 1%, and 0.1%
N2OinN
2diluted in steps of five as many as three
times with total pressures from 1 to 20 Torr.
By fixing the temperature of the thermoelectrically
cooled laser head, we obtainedsets of scans by using the
4-Hz current ramp, whose center frequency could be es-
timated from the wavelength calibration of the QC DFB
laser obtained independently with a Fourier-transform
infrared spectrometer.10 Figure 2 shows several such
scans overlaid to provide a continuous temperature
scan from ,217 ±Cto6±
C, corresponding to a 2.5-cm21
laser frequency scan starting at 1280 cm21. The spec-
tral data show the characteristic derivative line shape
of WM spectroscopy obtained from a 10%N2Oy90%N2
premixed sample with a total pressure of 1 Torr. We
checked the frequency calibration by comparing the
data with the known frequencies and line strengths of
the N2O10
0
0–0000 band obtained from the HITRAN
database.12 These are shown on the bottom of the fig-
ure. As an additional check on frequency calibration
we include a scan obtained from a 10%CH4y90%N2
sample at 10-Torr total pressure.
By switching the laser modulation from frequency
dither to mechanical chopping we were able to mea-
sure direct absorption and thus calibrate the WM
signal to absolute absorbance. Figure 3(a) shows the
resultant spectral scans obtained with 40 Torr of the
10,000-parts-in-106(ppm; 1%) N2O-in-N2sample.
The solid curve shows the absorption dip associated
with one of the strong N2O lines from Fig. 2. The
dotted curve shows a background trace with the gas
cell evacuated, and the dashed–dotted curve shows
the trace with the beam blocked. We obtained the
dashed trace by passing the beam through a 1.44-GHz
free-spectral-range solid germanium etalon.
Figure 3(b) shows the fractional attenuation of the
laser beam on line center as a function of N2O con-
centration inferred from the magnitude of the ab-
sorption dip for the 10,000-ppm sample and two 53
dilution steps (circles) and the undiluted 1000-ppm
sample (square). The straight line is a least-squares
fit to the data including the point (0 absorbance, 0
ppm). The slight deviation at low concentration is due
to accumulated error in dilution and the low signal-to-
noise ratio at low absorbance.
WM affords a substantial improvement over the di-
rect measurement in sensitivity to absorbance. Using
the two premixed samples of N2O and the dilution pro-
cedure outlined above, we measured the WM signal
amplitude for one of the strongest N2O lines as a func-
tion of N2O concentration. We calibrated the WM sig-
nal for absorbance by matching the WM signal to the
directly measured absorbance at the highest concentra-
tions. The result is plotted in Fig. 4. The solid line
is a linear least-squares fit from the data from dilu-
tions of both premixed samples, where we forced the
Fig. 2. WM signal versus laser frequency for 10% diluted
samples of N2Oand CH4with corresponding HITRAN data
plotted below.
Fig. 3. (a) Transmitted laser power versus frequency with
the N2Osample (solid curve), a calibration etalon (dashed
curve), or an opaque block (dotted– dashed line) in the
beam. (b) Absorbance versus concentration, obtained by
dilution of the samples as discussed in the text.
February 1, 1998 / Vol. 23, No. 3 / OPTICS LETTERS 221
Fig. 4. Plot of WM signal amplitude versus N2O concen-
tration for 20-Torr total pressure samples. The error bars
represent the amplitude of the background signal f luctua-
tions depicted in Fig. 5.
Fig. 5. (a) WM signal that is due to a 1000-ppm N2O
sample at 10-Torr total pressure. (b) Background trace for
an evacuated sample cell. (c) Background trace with the
laser beam blocked.
fit to pass through point s0, 0das before. Below a dilu-
tion of 20 ppm we were unable to distinguish the WM
signal from background fluctuations, which are indi-
cated by the error bars on the figure. To estimate
the noise equivalent absorbance we show in Fig. 5(a)
the WM signal from the 1000-ppm sample at 10 Torr
and in Fig. 5( b) the trace obtained with the gas cell
evacuated to less than 10 mTorr. The rms amplitude
of the random component of the signal was found to
be approximately 10 mVypHz. Comparing this sig-
nal level with the absorbance sensitivity deduced from
Fig. 4, we estimate the noise equivalent absorbance to
be 5 31025ypHz. With the strongest N2O line in this
spectral region, this absorbance corresponds to a detec-
tion limit of ,0.25 ppm-mypHz.
We show in Fig. 5(c) the background obtained with
the laser beam blocked. The 280-nVypHz amplitude
is consistent with that expected from 300-K back-
ground radiation. We conclude that the limiting back-
ground of our measurement is due to excess amplitude
modulation that appears on the output of the laser.
In conclusion, we have demonstrated sensitive detec-
tion of dilute samples of N2O, using a quantum-cascade
DFB laser operated near 4 ±C. The noise equivalent
absorbance of 5 31025ypHz is limited by two nonfun-
damental factors. Sensitivity to absorbance is reduced
because of excess laser line broadening.11 The appar-
ent laser linewidth of 720 MHz is due to the frequency
chirp that results from the pulsed operation and the
heating during the current pulse. This is in strong
contrast to the linewidth observed for continuous wave
operation.13 Furthermore, amplitude noise from the
drive electronics imparts an additional background
at the dither frequency. A factor-of-3–5 improvement
in both factors should be possible with improved de-
vice characteristics and optimized drive electronics.
Combined with the use of high-frequency-modulation
techniques, these improvements should bring the sen-
sitivity of the QC DFB laser to within an order of
magnitude of the 6 31028ypHz minimum detectable
absorbance reported for frequency-modulated lead salt
lasers.14 We conclude that, combined with the advan-
tage of room-temperature laser operation, the detection
sensitivity demonstrated here will make the QC DFB
laser an excellent choice for sensitive absorption spec-
troscopy detection of species with strong vibrational
features in the chemical fingerprint spectral region.
We acknowledge partial support from the Defense
Advanced Research Projects Agency and the U.S.
Army Research Office under contract DAAH04-96-C-
0026 (Bell Labs) and the National Science Foundation
(Stevens Institute) under grant DMI-9313320.
References
1. J. A. Silver, Appl. Opt. 6, 707 (1992).
2. C. B. Carlisle and D. E. Cooper, Appl. Phys. Lett. 56,
805 (1990).
3. P. Werle, Appl. Phys. B 60, 499 (1995).
4. H. C. Sun and E. A. Whittaker, Appl. Phys. Lett. 63,
1035 (1993).
5. J. Faist, F. Capasso, D. L. Sivco, C. Sirtori, A. L.
Hutchinson, and A. Y. Cho, Science 264, 553 (1994).
6. J. Faist, F. Capasso, D. L. Sivco, A. L. Hutchinson,
C. Sirtori, and A. Y. Cho, Appl. Phys. Lett. 65, 2901
(1994).
7. F. Capasso, J. Faist, C. Sirtori, and A. Y. Cho, Solid
State Commun. 102, 231 (1997).
8. J. Faist, F. Capasso, C. Sirtori, D. L. Sivco, J. N.
Baillargeon, A. L. Hutchinson, S. N. G. Chu, and
A. Y. Cho, Appl. Phys. Lett. 68, 3680 (1996).
9. C. Sirtori, J. Faist, F. Capasso, D. L. Sivco, A. L.
Hutchinson, and A. Y. Cho, IEEE Photon. Technol.
Lett. 9, 294 (1997).
10. J. Faist, C. Gmachl, F. Capasso, C. Sirtori, D. L. Sivco,
J. N. Baillargeon, and A. Y. Cho, Appl. Phys. Lett. 70,
2670 (1997).
11. J. M. Supplee, E. A. Whittaker, and W. Lenth, Appl.
Opt. 33, 6294 (1994).
12. L. S. Rothman, R. R. Gamache, A. Goldman, L. R.
Brown, R. A. Toth, H. M. Pickett, R. L. Poynter, J.
M. Flaud, C. Camy-Peyret, A. Barbe, N. Husson, C.
P. Rinsland, and M. A. H. Smith, Appl. Opt. 26, 4058
(1997).
13. F. Capasso, D. L. Sivco, A. L. Hutchinson, and A. Y.
Cho, Appl. Phys. Lett. 67, 3057 (1995).
14. P. Werle, F. Slemr, M. Gehrtz, and C. Br¨
auchle, Appl.
Phys. B 49, 99 (1989).