Development of An Automated Diode-Laser-Based Multicomponent Gas Sensor
Development of An Automated Diode-Laser-Based Multicomponent Gas Sensor
The implementation and application of a portable fiber-coupled trace-gas sensor for the detection of
several trace gases, including CO2, CH4, and H2CO, are reported. This particular sensor is based on a
cw fiber-amplified near-infrared distributed Bragg reflector diode laser and an external cavity diode
laser that are frequency converted in a periodically poled lithium niobate crystal to the mid-IR spectroscopic fingerprint region 3.3 4.4 m. A continuous absorption spectrum of CH4 and H2CO from 3.37
to 3.70 m with a spectral resolution of 40 MHz 0.0013 cm1 demonstrated the spectral performance
that can be achieved by means of automated wavelength tuning and phase matching with stepper motor
control. Autonomous long-term detection of ambient CO2 and CH4 over a 3- and 7-day period was also
demonstrated. 2000 Optical Society of America
OCIS codes: 190.2620, 300.6340.
1. Introduction
The spectroscopic detection and quantification of numerous trace-gas species has become an important
aspect in many industrial, urban, and environmental
applications. More importantly, there is a need to
selectively detect several different molecular gas species simultaneously by means of a single, tunable,
and portable sensing device capable of high sensitivity and rapid response time.
Characteristic absorption bands of most molecules
can be found in the near-infrared overtone and midinfrared fundamental rovibrational spectral region.
The latter range has the advantage that the spectrum
is less congested and exhibits 20 200 times stronger absorption cross sections. Established techniques for multicomponent gas analysis include
nonlaser-based Fourier-transform infrared spectrometers that can rapidly cover a broad spectrum 1 s
with a typical resolution of 15 GHz.1 For highly
selective 50-MHz absorption spectroscopy of trace
gases, various narrow-linewidth laser sources are
used in the mid-IR, including cw cyrogenically cooled
leadsalt diode lasers2 4 and, more recently, cw and
The authors are with the Rice Quantum Institute, Rice University, Houston, Texas 77251-1892. The e-mail address for F. K.
Tittel is [email protected].
Received 4 January 2000; revised manuscript received 24 May
2000.
0003-693500244444-07$15.000
2000 Optical Society of America
4444
Fig. 1. Schematic of an automated DFG multicomponent gas sensor. M1, M2, and M3 are stepper motors for the ECDL, PPLN crystal
stage, and reference gas assembly, respectively. WDM, wavelength division multiplexer; DL, diode laser; CTRL, control.
with serial ports. Furthermore, the fast measurement response time of typically 510 s, including averaging and automated data processing, allows the
real-time monitoring of in situ and remote chemical
and atmospheric trace-gas species.
2. Gas Sensor Configuration
4445
Fig. 2. Diagram of the electronic control and data-acquisition system for the DFG-based multicomponent gas sensor. DAQ, data
acquisition; TEC, thermoelectric cooler.
recorded phase-matching function. Both the singlemode fiber-launched pump and the signal beams possess Gaussian beam profiles, and when mixed in a
19-mm-long PPLN crystal they produce a diffractionlimited mid-IR DFG beam of 2.9-W power at
3.5-m wavelength.8 The DFG beam is then collimated by use of an f 50-mm CaF2 lens, passing
through a Ge filter to block the unconverted pump
beams and directed to a multipass cell physical
length, 30 cm; actual path length, 18 or 36 m.10
The sensor operated with a typical sensitivity of 2
104 limited by the occurrence of etalon effects introduced by the optical elements. For convenient identification and tracking specific absorption line center
positions within the scan width, up to five individual
reference gas cells L 5 cm can be rotated into the
DFG beam path by a stepper-motor-driven assembly.
The sensor is also operated and controlled with a
LabView software code running on the laptop PC.
Figure 2 shows the electronic diagram of the gas
sensor. Both stepper motor controllers are interfaced to the laptop PC by a universal serial bus port.
A 16-bit PCMCIA data-acquisition card is used to
acquire the absorption signal synchronized to a trigger signal, the gas sampling pressure, and temperature. The card also controls a shutter through a
digital transistortransistor logic line for background
light subtraction that is necessary for an absolute
4446
Fig. 5. Measured absorption spectra of formaldehyde and methane over 270 cm1 centered at 2850 cm1. Estimated total pressure was 50 Torr 8% CH4 and 0.65% H2CO; absorption path
length was 5 cm.
This value can be compared with a theoretically computed phase-matching bandwidth of 12.5 cm1 at 3.6
m for normal-incidence nondiverging pump beams.
Similar to the ECDL stepper motor interface, a miniature joint coupling was used to connect to the PPLN
crystal linear translation stage Vernier set screw,
0.5-mm linear translation per revolution. In addition, a 48-pitch, 1:1 Miter gear was implemented to
realize a compact stepper motor arrangement with no
noticeable loss of precision.
The simultaneous computer control of wavelength
tuning and quasi-phase matching allows multispecies gas measurements within the instrument tuning
range from 3.3 to 4.4 m. For example, Fig. 5 shows
a continuous absorption spectra of methane CH4
and formaldehyde H2CO acquired over a 270-cm1-
4447
Fig. 6. Comparison of the relative frequency drift for a steppermotor-controlled and free-running ECDL operation as a function of
time.
Fig. 7. Ambient CO2 laboratory sampling over a 42-h time period.
the drift of the absorption by appropriately wavelength tuning the ECDL if the peak location was
outside the programmed maximum and minimum
boundary conditions. For comparison, the total
wavelength drift that can occur if the gas sensor system is not actively controlled is also shown in Fig. 6.
4. Spectroscopic Evaluation of the
Difference-Frequency Generation-Based Gas Sensor
The DFG-based gas sensor was used in several longterm tests. In a first test, the sensor was tuned to a
CO2 line at 2385.77 cm1 4.2 m and monitored the
CO2 concentration in ambient laboratory air continuously at reduced pressure of 80 Torr for a 3-day
period as depicted in Fig. 7. The rovibrational CO2
absorption lines at 4.2 m have a frequency spacing
of 0.7 cm1 21 GHz. In this case, it is advantageous to perform spectroscopic absorption measurements of CO2 at a reduced sampling pressure to avoid
a zero baseline offset. The CO2 concentration increases and decreases rapidly, respectively, at times
with and without human activity in the laboratory as
shown in Fig. 7. The dynamic detection range at
this CO2 absorption line is 0.52200 parts per million
ppm 100% absorption at a pressure of 80 Torr for
an 18-m optical absorption path-length-configured
multipass cell and a residual optical noise of 2
104. After 2 h into the CO2 sampling test, human
breath was briefly sampled and caused a sharp increase to CO2 levels of more than 2200 ppm, saturating the absorption signal. However, it is possible to
tune the DFG wavelength to a weaker CO2 transition
to measure higher concentrations or to lower frequencies to access a stronger CO2 transition at 4.2 m
when low CO2 concentrations are measured. The
DFG-based gas sensor measured higher ambient
background levels than the atmospheric concentrations measured by the National Oceanic and Atmospheric Administration.17 This may be due to the
fact that the CO2 and other gas species concentration
levels tend to be higher in urban areas such as Houston, Texas.
Fig. 9. Intercomparison of long-term CH4 concentration measurements by two independent DFG-based gas sensors indicating both CH4
concentrations and deviation in parts per billion.
20 August 2000 Vol. 39, No. 24 APPLIED OPTICS
4449