Use of diffuse reflections in tunable diode laser absorption
spectroscopy
Dackson Masiyano, Dan Francis, Jane Hodgkinson and Ralph Tatam
Engineering Photonics Group, School of Engineering, Cranfield University, Cranfield,
Bedfordshire, MK43 0AL
j.hodgkinson@cranfield.ac.uk
Abstract.
Tunable diode laser absorption spectroscopy (TDLAS) has great potential in trace gas
detection, with applications in health, safety and environmental monitoring. The
attractiveness of TDLAS includes high sensitivity, specificity and high detection
speed. However the sensitivity of many TDLAS practical systems is limited by the
formation of unintentional Fabry-Perot interference fringes generated in the optical
path between the source and detector. Although it might be possible to design out the
problem, such measures are complicated and can be difficult to maintain in field
instruments. In recent years there has been interest in using diffuse reflections within
the optical path, in the following diverse areas; use of remote “laser pointer” style gas
detectors where the light is backscattered from a rough surface at ground level, use of
integrating spheres as multi-path gas cells and insertion of diffuse reflecting or
transmitting materials in the optical path. We present the preliminary results of
investigating the deliberate use of optical scattering in TDLAS by using diffusely
reflecting materials.
1. Introduction
Tunable diode laser absorption spectroscopy is an attractive technique for measurements of trace gases
because of its sensitivity, specificity, high detection speed and the possibility of simultaneous in situ
measurements[1]. Most TDLAS systems are limited in sensitivity by optical fringes superimposed on
the measured spectrum[1]. These result from unwanted etalons formed by reflections and scattering in
the optical system. These etalons often exhibit a free spectral range similar to the linewidth of the
absorbing species and appear as periodic spectral features with sufficient amplitude to obscure weak
absorption signals[1].
Design techniques to reduce the etalon formation include use of reflective optics to avoid on-axis
scattering from lenses, wedging and angling of all windows, antireflection coating of windows and
lens surfaces, and angle polishing optical fibre ends. Techniques have also been proposed to eliminate
or reduce the amplitude of the fringe signal. These include: asynchronous longitudinal dithering of
optical elements, dithered Brewster angle plates, introduction of an asynchronous current in addition
to the usual modulation current through the laser diode and electronic low pass filtering[2,3]. The
majority of the techniques effectively use the periodic nature of the etalon fringe with low pass
filtering to reject the fringe signal. A critical limitation of this low pass filtering approach is that the
line width of the absorption feature being measured must be much greater than the FSR of the etalon
fringe. The need for signal integration also limits the overall detection bandwidth[2]. Both the design
measures and the fringe reduction schemes outlined above generally add to instrument complexity or
can be difficult to maintain in field instruments.
Here, we present an alternative approach based on use of diffuse reflections, known in some
circumstances to reduce interference fringes[4]. This could lead to several benefits: (a) improved
detection sensitivity, (b) reduced complexity and costs in instrument manufacture, and (c) making
systems less susceptible to misalignment, thereby increasing field robustness. However, their use
introduces laser speckle that can contribute a random, rather than periodic, uncertainty to gas detection
measurements. We present preliminary results of a systematic study of these effects.
2. Theory
A speckle pattern is a random intensity distribution that is formed when coherent light is reflected
from a rough surface by random scatterers[5]. At any particular point, the resultant phasor, or speckle,
is the sum of many contributions from elementary phasors, whose amplitudes and phases are
statistically independent and are also independent of those of all other elementary phasors [5]. The
phases of these waves are uniformly distributed between π and –π[6].
Laser
illumination
Focal length, L
Diffusely
scattering
surface
Detector
Figure 1 Formation of a subjective speckle pattern
The statistical properties of speckle have been covered by a number of authors [5,7]. It has been
shown that; the irradiance at a point for linearly polarized Gaussian speckle follows negative
exponential statistics, the most probable brightness of the speckle is zero, and the contrast of a
polarized speckle field is always unity[5]. Sirohi[6] has summarised the expected speckle size for
subjective speckle (Figure 1), as follows:
lateral speckle size
 s  2.44  1  M  F #
(1)
Where M is the magnification of the detector objective and F# is the ratio of its focal length to its
aperture.
First, consider the level of speckle noise on a detector of size d x d. On the detector will be a
number N of uncorrelated speckles given by
N

d2
(2)
 s2
Goodman[8] has studied the statistics of speckle. We expect the level of speckle-related intensity noise
for a single speckle field to be given by
s
I
N
1
I

N


N
d
(3)
This level of noise is relevant to the use of direct spectroscopy, in which a zero measurement might be
made at a given wavelength and compared at some future time with an intensity signal.
Second, consider the level of uncertainty that relates to wavelength modulation. In this case we are
considering an intensity measurement made at one wavelength with a second measurement made
within a short modulation period, at a second wavelength. Sirohi has summarised the theoretical basis
for speckle interferometry[6], in which a phase difference δ is created between successive speckle
fields, such that

 2
L
2

(4)
where L is the optical path difference between the reflected and reference beams.
The value of δ determines whether the observed speckle fields are correlated or uncorrelated, the
correlation coefficient being given by
1  r 2  2r cos 
(5)
   
2
1  r 
where r is the ratio of the mean intensities of the two speckle fields. Since we expect r=1 here,
equation (5) reduces to
1  cos 
   
(6)
2
Thus, for increasing δ, the speckle fields alternate between (δ) = 1 (correlated) and (δ) = 0
(uncorrelated). If the speckle fields are correlated, we expect the speckle-related noise to reduce to
zero for differential measurements. If the induced phase change takes the value δ = π the speckle
fields are uncorrelated, or for larger phase changes we cannot predict whether they are correlated or
uncorrelated, so in the worst case we expect a speckle-related intensity noise given by equation (3).
3. Tunable diode laser absorption spectroscopy
The basic principle of TDLAS, its applications and comparisons of the different TDLAS techniques
have been described elsewhere in the literature[1]. For our investigations, we have chosen to employ
wavelength modulation spectroscopy (WMS) because of its relative ease of implementation.
3.1. Wavelength modulation spectroscopy
The schematic diagram of our experimental setup is shown in Figure 2. In WMS the laser wavelength
is modulated by varying the laser drive current at a frequency of several kilohertz.
Temperature
controller
Gas cell
Detector/amp
Laser
driver
Saw tooth
generator

Lock-in amplifier
Labview
software
Sine generator
Figure 2 Schematic diagram of a typical experimental setup for diode laser based WMS
The laser wavelength can be simultaneously slowly scanned across the a single gas absorption line by
a large amplitude, low frequency ramp signal as illustrated in Figure 3. The resulting signal is then
demodulated by a phase sensitive detector at the modulating frequency or at harmonics of this. One
advantage of this technique is to shift the detection to higher frequencies, at which laser excess noise is
reduced. The use of phase sensitive detection leads to significant improvement in signal to noise ratio.
(a)
To phase sensitive
detector
Modulation signal
(a)
% transmission
% transmission
1651nm
Slow line scan
Wavelength /µm
(b)
wavelength /µm
Figure 3 (a) Methane gas absorption lines around 1650nm from Hitran database[9] (b)Schematic diagram
of the principle of WMS at the 1650nm line; the laser wavelength is slowly scanned across an individual
gas absorption line whilst being simultaneously modulated at several kilohertz.
Figure 4 illustrates problem of interference fringes compared to the measured gas signal. One
common source of fringes is the gas cell, where an etalon can be formed between the windows (figure
4(a)) or within one window (figure 4(b)). For this example, the gas cell was purposely aligned to
maximise the fringe amplitude.
broad fringe
from empty gas
cell
narrow fringes from
empty gas cell
Direct absorption
gas signal
Incident and
reflected beams
(a)
Direct absorption gas
signal
Gas cell
Incident and reflected
beams
Gas cell
(b)
Figure 4 Oscilloscope traces showing direct spectroscopy signal obtained with a filled and an evacuated
gas cell. (a) Interference fringes from multiple reflections between windows of an empty gas cell.
(b) Interference fringes from multiple internal reflections within a window of an evacuated gas cell.
4. Experimental investigation
We conducted a series of experiments to study the effects of speckle noise in a system that simulated a
TDLAS based gas detector. By working at 800µm we were able to simulate the situation at 1651nm
and use a low cost camera to view the speckle directly.
The experimental configuration shown in Figure 5 includes: (i) a tunable diode laser driven so as to
simulate a TDLAS based gas detection, (ii) a simulated gas cell containing a diffusely scattering
surface and a number of different types of window (a wedged window is illustrated), (iii) an
interrogation system employing a silicon CCD camera. The setup is a model of the longer wavelength
(1651nm) gas detection and used a standard geometry for speckle interferometry experiments. Light
from a Sharp near infra red laser diode (LT022MDO) was collimated and passed to a polarisation
insensitive beam splitter. Light reflected from the interferometer formed by the window front surface
and the PTFE block was diverted to a camera (Pearpoint 800 pixel CCD camera, P1760). The wedged
window provided a reference wave front. Reflections from the two surfaces combined coherently to
form a speckle pattern. The speckle images were captured and stored using the CCD camera and a
frame grabber housed in a personal computer running Labview software.
The CCD chip (10.8×10.4mm) was larger than the aperture of our detector (1×1mm). By obtaining
the measured intensity for individual pixels, we simulated the estimated uncertainty in the measured
intensity for a series of ninety-four 1×1mm segments over the entire CCD area. The mean intensity
within each segment and the standard deviation of these mean intensities were used to measure the
speckle related measurement uncertainty in a series of experiments.
Optical isolator
Optically rough
material
Simulated gas cell
Lens L1
Laser
diode
Lens L2
Wedged
window
Adjustable
aperture
CCD
camera
PC running
Labview
software
Recorded speckle
image
Figure 5 Experimental setup for investigating the reduction of interference fringes in TDLAS using
speckle interferometry.
This experimental setup gives us the ability to obtain speckle patters while tuning the laser diode to
simulate wavelength modulation spectroscopy. The laser diode was temperature stabilised to reduce
changes in wavelength due to ambient temperature drift.
We confirmed that the level of the background noise was constant over the duration of the
experiment. Fixed pattern noise under dark conditions was measured and subtracted from all
subsequent images. We deliberately exaggerated the speckle noise by using smaller apertures to
facilitate the study.
5. Results and discussion
The results presented here are from initial experiments with subjective speckle. The subjective speckle
is obtained by misaligning the reference beam so that only the speckle pattern form the PTFE reaches
the camera. Future experiments will be done using interferometric speckle.
The results of the dependence of the expected noise level on the aperture size are presented in
figure 6 below. We divided the speckle pattern into segments representing the size of our detection
system. From the individual pixel data we calculated the mean intensity of each segment. We then
calculated the speckle related measurement uncertainty from the average intensity and the standard
deviation of the mean intensities of all the segments. The results confirm our expectation that the
speckle related intensity noise is inversely related to the aperture size. It appears that we have reached
the noise floor of our current configuration and we are investigating ways of reducing it.
0.14
error in the mean
0.12
0.1
0.08
0.06
0.04
0.02
0
0
5
10
15
20
aperture/mm
Figure 6: The uncertainty in intensity measurement for a 1mm ×1mm segment, as a function of the
diameter of the aperture of the detection system
6. Conclusion
Methods that have been proposed to reduce the effects of interference fringes include careful
alignment of optical components or mechanically jittering the offending components. Evidence also
suggests that in the right circumstances use of diffusely scattering materials may reduce fringes. We
have investigated their use and the consequent introduction of random uncertainty associated with
generation of laser speckle. We have developed a methodology for investigating the associated noise
and presented preliminary data for a simulated gas cell. Future work will involve investigation of
different diffusely reflecting materials and different optical path geometries including integrating
spheres.
7. Acknowledgements
This work is supported by the Engineering and Physical Sciences Research Council (EPSRC), UK
under grant No.GR/T04601/01. Jane Hodgkinson is supported by an EPSRC Advanced Fellowship,
GR/T04595/01.
8. References
[1] Werle, P. (1998), 'A review of recent advances in semiconductor based gas monitors',
Spectrochimica Acta Part A, 54, pp. 197-236.
[2] Sun, H. C. and Whittaker, E. A. (1992), 'Novel etalon fringe rejection technique for laser
absorption spectroscopy', Applied Optics, 31, pp. 4998-5002.
[3] Silver, J. A. and Stanton, A. C. (1988), 'Optical interference fringe reduction in laser absorption
experiments', Applied Optics, 27, pp. 1914-16.
[4] Tranchart, S., Bachir, I. H. and Destombes, J.-L. (1996), 'Sensitive trace gas detection with nearinfrared laser diodes and an integrating sphere', Applied Optics, 35, pp. 7070-7074.
[5] Goodman, J. W. (1975), 'Statistical properties of laser speckle patterns', in Laser speckle and
related phenomena, Springer-Verlag, pp. 9-75.
[6] Sirohi, R. S. (2002), 'Speckle interferometry', Contemporary Physics, 43, pp. 161-180.
[7] Jones, R. and Wykes, C. (1989), Holographic and Speckle Interferometry (2nd edition), CUP.
[8] Goodman, J. W. (1976), 'Some fundamental properties of speckle', Journal of the Optical Society
of America, 66, pp. 1145-1150.
[9] Rothman et al, L. S. (1992), 'The HITRAN molecular database: editions of 1991 and 1992',
Journal of Quantitative Spectroscopy and Radiative Transfer, 48, pp. 469–507.