PNNL-13501 - Pacific Northwest National Laboratory

containing nitrous oxide, an optical cavity (with piezo
adjustable end mirror), and a digital oscilloscope used for
data averaging and export to a computer. Liquid nitrogen
cooled HgCdTe photovoltaic detectors are used with
custom-built low-noise preamplifiers. The optical cavity
used in this work incorporated plano-concave ZnSe
mirrors (radius = 100 cm) with a reflectivity of 99.85%.
The cavity itself consists of a 40-cm stainless steel tube
with vacuum-sealable mirror mounts (incorporating a
piezo element used to modify the cavity length).
Approximately 4 MW of infrared emission is collimated
from the quantum cascade laser, 10% of which is diverted
to a reference cell for wavelength calibration prior to the
optical cavity.
Figure 1. Experimental infrared integrated cavity
absorption spectrometer layout
Since the cavity length is 40 cm and the radius of
curvature of the resonator mirrors is 100 cm, a myriad of
longitudinal modes of the cavity are seen (in
transmission) as the wavelength of the quantum cascade
laser is scanned. Typical scans (accomplished by
applying a sawtooth modulation on the injection current)
are up to 0.5 cm -1 at a repetition rate of 1 to 10 kHz. At
first, this quasi-continuum of transmission modes may
seem to be detrimental to the absorption measurement;
however, if the optical cavity is being used to increase the
effective path length, then the higher number of modes is
advantageous. If fact, our apparatus is intentionally
misaligned (to reduce back-reflections to the laser) and
the spatial mode profile of the quantum cascade laser is
non-optimal for coupling to the primary mode of the
cavity. These two effects act to increase the number of
accessible modes of the cavity. During an experiment,
the cavity length is dithered (typically 10 to 100 Hz) using
various waveforms that are not synchronized to the
wavelength scan of the laser. Therefore, when many
cavity transmission scans are averaged on the digital
oscilloscope, the resulting spectrum appears as a nearly
flat baseline with a small direct current offset, rather than
the multitude of sharp cavity resonance peaks. The lower
panel in Figure 2 compares a signal-sweep trace and an
averaged trace (2500 sweeps) of the cavity transmission
with 100 ppmv of nitrous oxide present in cavity. The
upper panel in Figure 2 shows an expanded version of the
averaged data.
Cavity Transmission
Single Sweep
2500 averages
1172.1 1172.4
Laser Frequency ν (cm -1 )
Figure 2. Lower panel depicts a comparison of single (gray)
and averaged (black) transmission spectra with
approximately 100 ppmv of nitrous oxide present in the cell
at 25 torr. Upper panel is an expanded version of the
averaged trace.
As can be seen in the upper panel in Figure 2, the noise
floor is very regular and is similar to etaloning commonly
seen in laser absorption measurements. In this case, the
periodicity of the baseline fluctuations results from an
incomplete averaging of clusters of cavity modes.
Ideally, the cavity modes would be uniformly spaced in
the transmission spectra. Conventional cavity modes
calculations support this observation and suggest that this
effect can be dramatically reduced using a slightly shorter
cavity.
The effective path length (i.e., cavity-enhancement factor)
of this arrangement was found by multiplying the cavity
length by the ratio of cavity-enhanced absorbance to the
single-pass absorbance (found by replacing the mirrors
with plain windows). The comparison of the cavityenhanced
absorption spectrum with the single-pass
absorption is shown in Figures 3a and 3b, respectively.
The effective path length in the cavity-enhanced case was
found to be 38 meters (cavity-enhancement factor of ~90).
This initial effective path length can be improved upon
with the use of high reflectivity mirrors; a factor of ten
improvement would yield path lengths not easily obtained
with multi-pass cells.
Sensors and Electronics 385