A Compact, Pulsed Infrared Laser-Excited ... - OSA Publishing

A Compact, Pulsed Infrared Laser-Excited Photothermal Deflection Spectrometer OLUWATOSIN O. DADA* and STEPHEN E. BIALKOWSKI Department of Chemistry and Biochemistry, Utah State University, Logan, Utah 84322-0300 (S.E.B.); and Department of Chemistry and Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556 (O.O.D.)

A prototype small photothermal deflection apparatus was constructed on a laboratory bench. Not including excitation laser, data collection computer, and gas pumps, the apparatus fits on a bench space with a footprint of about 10 cm 6 10 cm. The apparatus lends itself to miniaturization in future assemblies. The apparatus was tested relative to the conventional laboratory-scale photothermal apparatus. Digital filtering procedures developed in this laboratory were used to collect and analyze the data. Numerical simulation of photothermal signal is performed to accurately predict the heat flow process as sample volumes are scaled down. The results show that increased sensitivity is possible with small apparatus configurations. The prototype apparatus also exhibits high linear response with pulsed irradiance for linear absorbers. Future improvements could include miniaturization with robust configuration that will allow for portable use in trace analysis monitoring of atmospheric pollutants. Index Headings: Infrared spectroscopy; IR spectroscopy; Photothermal deflection apparatus; Spectrometry; Instrumentation; Miniaturization; Fluorocarbons.

INTRODUCTION Increased awareness of threats posed by hazardous atmospheric species indicates the need for sensitive, portable, and easy-to-use detection methods. Photothermal spectroscopy is among the viable alternatives when very small absorptions are encountered. However, portable, easy-to-use photothermal instruments have not emerged over the last few years despite the inherent sensitivity of photothermal spectroscopy over conventional absorption techniques. From the viewpoint of apparatus design, the main development issue for a minimized apparatus configuration is to develop a short-focused laser beam. Most photothermal apparatuses are composed of multiple lenses with long focal lengths in order to control beam-spot positions and aberration values. Reducing the focal distance significantly increases aberrations, and compatibility has been very hard to achieve. Apparatus design based on reflective optical components is considered an ideal solution for obtaining a considerable reduction in the focal distance with minimal aberration.1,2 Miniaturized off-axis parabolic mirrors are ideal alternatives to refractive lenses. Focusing mirrors feature absence of color aberration because they are free from the dispersion that affects the quality of refractive lenses. The advantages of metal reflectors have led to widespread use of aspheric surfaces (offaxis parabolas) and if properly designed, they will theoretically give diffraction-limited focusing performance. Received 7 May 2010; accepted 3 November 2010. * Author to whom correspondence should be sent. E-mail: odada@nd. edu. DOI: 10.1366/10-06000

Volume 65, Number 2, 2011

The purpose of this work is to investigate the potential of a portable photothermal detector with no refractive lens component for trace gas measurement, without significant loss of sensitivity. To examine the effects on the photothermal signal of sizing down the apparatus, the performance of the smallconfiguration photothermal spectrometer is compared relative to that of a conventional apparatus setup. In this study, the signal amplitude and the time constants of the time-dependent photothermal signal and energy-dependent photothermal signal are compared. The main goal is the evaluation of the compact apparatus for robust and sensitive trace pollutants measurements that are of importance to agriculture and public health. The apparatuses described herein are based on photothermal deflection spectroscopy measurements.3–11 Photothermal deflection is performed by probing the temperature-dependent refractive index gradient produced when optical energy absorbed by a sample is converted to heat via non-radiative relaxation mechanisms. Photothermal deflection apparatuses can either be configured in collinear or transverse geometry. In the collinear geometry, both the excitation and probe lasers propagate in the same direction, while the lasers propagate orthogonally in the transverse geometry. The apparatuses in this work are based on the collinear configuration. Most schemes for obtaining electronic signal from this technique are based on monitoring the change in position of a probe laser beam spot at some distance past the sample region. The probe beam is typically focused at the region of maximum refractive index perturbation. The divergence in the probe beam past the focus implies that the effects of reduced sampleto-detector distance on the signal should be negligible. Hence, scaling down the apparatus should produce the same, if not higher, signal as the conventional setup. The measured signal should be proportional to optical absorption, sample path length, analyte concentration, and the excitation source power.3 The signal properties are also dependent on the thermo-optical properties of the analyte and the buffer.

THEORY The operating principle of photothermal deflection is well known. Theories for most photothermal methods are based on the solution to Eq. 1, which defines the heat diffusion process that follows when the sample is heated with the excitation laser.3–27 qCp ]T=]t þ rðkrTÞ ¼ Qðr; z; tÞ

ð1Þ

Here q (kg m3) is the density, Cp (J kg1 K1) is the specific heat capacity, k (W K1 m1) is the thermal conductivity, Q(r, z, t) (W m3) is the heat source, and T (K) is the temperature. For a TEM00 pulse excitation beam with a Gaussian irradiance, the heat source, Q(r, z, t), due to sample absorption can be

0003-7028/11/6502-0201$2.00/0 Ó 2011 Society for Applied Spectroscopy

APPLIED SPECTROSCOPY

201

defined as: Qðr; z; tÞ ¼

2aP f ðtÞexpðazÞexpð2r 2 =w2e Þ pw2e

ð2Þ

where P (W) is the excitation laser power, we (m) is the excitation laser beam spot radius, a (m1) is the sample absorption coefficient, z (m) is the optical path distance, and r (m) is the radial distance from the center of the excitation laser beam. f(t) is the time function that defines the temporal pulse profile of the excitation laser. Assuming a Gaussian temporal pulse profile, f ðtÞ ¼ exp½2ðt t0 Þ2 =tp2

ð3Þ

where t0 is the pulse center position, and tp is the pulse-width variance parameter. Following optical absorption by the sample, the resulting thermal energy dissipates from the sample through the surrounding medium, causing a spatially dependent temperature change (]T) in the medium. The temperature gradient consequently causes a refractive index change (]n) that produces the photothermal signal. Assuming the absorbed energy is transferred to the solvent shortly after pulsed irradiation and that the signal is detected prior to thermal diffusion, the photothermal deflection signal (PDS) strength is related to the temperature changes that occur through: Z dn r? dTðr; tÞds ð4Þ hðr; tÞ ’ dT path

Here h(r, t) is the probe-beam deflection angle, dn/dT is the temperature-dependent optical coefficient, and s is the optical path through the sample. The differentials are found for the coordinates perpendicular to the propagation direction of the excitation laser. Numerical Simulation. To access an idealized result of down-scaling a photothermal apparatus, numerical simulations are performed using Comsol Multiphysics v 3.4 finite element analysis (FEA) software. The simulation procedure as applicable to photothermal phenomena has been previously reported.16,28–30 Here, the interest is in showing the effect of apparatus size reduction on the deflection signal. Therefore, only the excitation region is used for the simulation. The layout of the computation domain, as shown in Fig. 1a, consists of the sample region with focused excitation and probe laser. It is assumed that the sample cell wall and its optical windows have negligible thermal conduction. The excitation laser propagates along the positive z-axis with its minimum beam waist at the center of the sample cell. The laser and the sample holder have axis symmetry, which makes a simplified axial symmetry solution appropriate for the simulation. For comparison, two configurations were modeled, the small apparatus configuration and a conventional (large) apparatus. The sample cell and the laser beam parameters as used in the simulation are shown in Table I. The probe beam offset of half the excitation beam waist (we /2) from the excitation laser origin was used for the simulations. The transient analysis of heat diffusion that occurs following sample excitation was obtained by using the FEA software to solve Eq. 1 with the boundary conditions at the walls set at initial temperature, T0 (assuming no radiative or convective heat transfer at the walls and optical windows). This way, the

202


FIG. 1. (a) Schematic diagram of the finite element analysis computation domain. (b) Calculated peak-normalized time-dependent photothermal deflection signal for small apparatus (dashed curve) and large apparatus (solid curve).

temperature difference ]T is defined as T T0. The heat source propagating along the z-axis is defined by Eq. 2. After solving the model, the photothermal deflection signal strength was obtained, according to Eq. 4, as proportional to the path integral of the first radial derivative of the temperature change. The path integral was determined by integrating the first derivative of ]T with respect to r over the region that represents the probe beam as in Fig. 1a. Figure 1b shows the simulated photothermal deflection signal for small (dashed curve) and conventional (solid curve) apparatus configurations. The theoretical prediction illustrates two important points: (1) signal amplitude for small apparatus configurations is more than twice that of large apparatus configurations; (2) in the cooling part of the transient signal, the relaxation time for the small configuration is over an order of magnitude shorter than for the large configuration. The simulation shows an overall TABLE I. Sample cell and the laser beam parameters used in the Comsol Multiphysics simulation. Parameters

Large configuration

Small configuration

Laser power (W) Excitation beam waist (lm) Pulse width (ns) Probe beam waist (lm) Absorption coefficient (m1) Optical path length (mm) Sample cell radius (mm)

0.001 150.0 120.0 50.00 5.190 0.100 15.00

0.001 80.00 120.0 30.00 5.190 .0370 15.00

advantage for scaling down the apparatus size. The fast relaxation in the signal for the reduced configuration is very important, especially when considering convective dynamics associated with flowing samples.

EXPERIMENTAL Large Size Apparatus. The conventional photothermal apparatus was constructed with standard Newport (City, CA) optical mounts on a 4’ 3 8’ Modern Optics optical bench. The footprint of this apparatus was about 1 m2, not including the gas-handling equipment, signal-processing electronics, and the excitation laser. As illustrated in Fig. 2a, the pump laser was a Tachisto, linetunable TEA-CO2 laser (9–11 lm). The output of this laser was a TEM00 mode pulse of ’ 120 ns duration. TEM00 operation was obtained by putting a 1 cm intracavity aperture in the laser. The pulse laser energy was attenuated by placing a ‘‘Venetian blind’’ style infrared attenuator in the beam path prior to the focusing optics. The CO2 laser has ;5% (rms) pulse-to-pulse energy variation when using nitrogen in the gas mixture, as was done in these experiments. The focus spot is not affected by the attenuator because the higher-order diffraction spots will be focused off-axis. The pointing stability is good compared to continuous-wave lasers, though this has not been quantified. Excitation laser wavelengths were determined using an Optical Engineering CO2 spectrum analyzer, Model 16-A. The R(16) line of the CO2 transitions at 9.30 lm was used in these measurements. The probe laser was a polarized 4 mW continuous-wave HeNe laser (632.8 nm) Uniphase Model 1205-1. The pump beam and the probe beam were first combined by using a germanium flat as a beamsplitter. The mixed beams were then focused with a 127 mm focal length BaF2 lens into the center of the sample cell. The sample cell radius is 15 mm with a path length of 100 mm. The distance from the pump beam spot position and the detector is about 254 mm. The size of the focused CO2 laser at the spot position was measured with a razor blade edge on a micrometer-driven translation stage. The beam spot diameter calculated from the razor blade excursion resulting in 10% to 90% of the maximum pulse energy was 150 lm. The size of the probe beam at the minimum waist is about 50 lm in diameter. Photothermal deflection signal was monitored using a United Detector Technology Model 301-DIV single-axis positionsensing detector. Signals from the detector were amplified and electronically filtered with a Tektronix model AM-502 differential amplifier. Integrated pulsed excitation laser energies were simultaneously measured with a Laser Precision Model RjP-735 pyroelectric energy monitor and digitized with a 12-bit analog-to-digital converter (A/D). The integrated pulse energy is directly proportional to the intensity because the laser pulse duration and the spatial spot size are constant from pulse to pulse. Data collection and processing were performed on a PC with matched filter smoothing software that was developed in this laboratory.31–33 Small Size Apparatus. The schematic diagram of the small size apparatus is shown in Fig. 2b. The footprint of the small apparatus prototype is about 0.01 m2 not including the excitation laser, the gas-handling equipment, and the signalprocessing electronic device. The excitation source, the detector, and the signal-processing electronics are the same as for the conventional apparatus. The probe beam is a Model 57ICS006/PS/HS Melles Griot fiber-optic diode laser operating

FIG. 2. Schematic diagram of (a) conventional photothermal deflection apparatus, (b) prototype small photothermal deflection apparatus. L: lens, EM: energy monitor, PSD: position sensing detector, BS: beam splitter, BA: beam attenuator, and M: plane mirror.

at a wavelength of 785 nm. The excitation and probe laser beams are first mixed with a 25 mm diameter germanium plate beam splitter, then focused into the center of the sample cell by a 50 mm focal length off-axis parabolic focusing mirror. The sample cell radius is 4.5 mm with a path length of 37 mm. The distance from the center of the sample cell to the detector is about 30 mm. The measured excitation beam spot diameter is about 80 lm and the probe beam diameter at minimum waist is about 30 lm. Samples. The sample used is a 10 ppm trichlorofluoromethane (CFC-11) Matheson standard premixed gas balanced in nitrogen. To achieve lower concentration in pure nitrogen, serial dilution was carefully performed with an ultrahighvacuum dynamic gas mixer.

RESULTS AND DISCUSSION A prototype photothermal deflection spectrometer in a small apparatus configuration is presented. The prototype apparatus was built without refractive optical components. This is achieved by using short focal length aspheric mirrors to direct both excitation and probe beams into the sample cell. The


203

FIG. 3. Experimental peak-normalized time-dependent photothermal deflection signal for small apparatus (dashed curve: CFC-11 in pure nitrogen) and large apparatus (solid curve: CFC-11 in nitrogen) at 9.3 lm excitation wavelength and 0.8 mJ average pulse energy.

FIG. 4. Energy-dependent photothermal signal for conventional photothermal deflection apparatus (open symbols) and prototype small photothermal deflection apparatus (solid symbols). Sample is 1 ppmv CFC-11 buffered in nitrogen at 9.3 lm excitation wavelength and 0.8 mJ pulse energy.

apparatus was tested in comparison with the conventional laboratory-scale photothermal apparatus, which is normally constructed with refractive optical components. The small apparatus features short excitation and probe laser beam spot size. As a result, improvements in photothermal signal amplitude and shorter signal relaxation time constant were obtained. With collinear excitation–probe beam geometry, the magnitude of the deflection angle is a function of the radial distance from the center of the excitation laser beam spot. This relationship requires that the spatial overlap between the excitation and probe laser beams be known or at least reproducible. Inconsistent probe laser beam position relative to the excitation laser beam between experiments will not only result in changes in the maximum detector signal but also in changes in the temporal characteristics of the signal. The inherent wavelength-dependent characteristics associated with refractive optical elements renders achieving these requirements a challenging task in conventional photothermal apparatus setups. The prototype photothermal apparatus uses off-axis parabola focusing mirrors to focus both the excitation and probe laser beams. By this, the longitudinal aberration normally encountered with wavelength-dependent refractive material such as lenses is greatly reduced. Because reflective optics are wavelength independent, spatial overlap enhancement and reproducibility of the excitation and probe laser beams was achieved by focusing both beams with a single mirror. Obtaining maximum sensitivity also requires that the excitation and probe beams are focused to a small beam waist in the sample cell. With the short focal length aspheric mirror, a smaller beam size with a short Raleigh range can be achieved. This produces lower light intensities at the sample cell windows and subsequently less window heating occurs than when using long focal length optics. Data obtained in this work were analyzed primarily in terms of the excitation laser pulse energy-dependent signal and timedependent signal at constant pulse energy. For the timedependent signal, to a great extent, the experimental result (Fig.

3) is in good agreement with the theoretical prediction (Fig. 1b). Both results show that down-scaling the apparatus configuration provides a net advantage over the conventional design. The peak-normalized signal amplitude for the small apparatus is at least twice the peak-normalized signal amplitude of the conventional apparatus. This was also observed in the finite element method (FEM) results. Thus, better sensitivities can be obtained using the small apparatus configuration. The characteristic short cooling time constant for the small apparatus configuration is important when considering samples in a flowing medium. On the other hand, the fast signal relaxation predicted by the FEM analysis is not obvious in the experimental signal. This may be caused by heat conduction across the sample cell wall, which was not accounted for in the FEM simulation. Overall, the results showed shorter relaxation times for the small apparatus. In terms of large-to-small relaxation time constant ratio, the values are not too different for the experimental and the theoretical results. The model predicted a large-to-small relaxation time constant of about 1.5, whereas the experimental large-to-small relaxation time constant is 1.8. The difference in theoretical versus experimental values may be attributed to differences in error sources in both the theoretical calculation and the experimental measurements and to the fact that the experimental excitation–probe beam offset value may not be exactly half the excitation beam waist (we/2). But overall, the numerical simulation provides a good approximation for the transient photothermal deflection signal. The energy-dependent photothermal deflection signals are shown in Fig. 4. This important measurement shows the correlation between the signal magnitude and the excitation pulse energy. Although the basic theory predicts a signal that increases with excitation pulse energy, this is often not the case when using pulsed high irradiance lasers. Pulsed excitation often produces nonlinear absorption effects. In particular, chlorofluorocarbons such as CFC-11 have shown a nonlinear energy dependence at high excitation irradiance.27 The solution is to operate the excitation source at an energy level that falls within the sample’s linear absorption range. This was achieved

204


by placing an adjustable infrared beam attenuator in the excitation beam path to keep the excitation laser power in the linear range of CFC-11. However, the purpose of Fig. 4 (and the study, for that matter) was not to examine the nonlinear absorption behavior but rather to simply present the data obtained. By using our matched filter smoothing regression technique, two separate data files were recorded during the energydependent signal measurement: the excitation laser energy and the signal amplitude per pulse. Upon each pulse of the excitation laser, the signal amplitude and pulse-energy magnitude were estimated with the homemade matched filter smoothing software. The measured signal plotted against the excitation pulse energy is shown in Fig. 4. Some authors have found that the nonlinear effects, as observed with the conventional apparatus (open symbols), are due to the different energies along the path of the focused laser. The shorter effective path length of the compact apparatus may have served to reduce these effects. Another possible explanation is that differences in the optical setup may have resulted in detector nonlinearity in the conventional setup. This could be caused by the small probe laser spot size at the bicell detector. The compact apparatus has higher probe beam divergence and thus a larger spot size at the detector. In general, the signals for the apparatus in the small configuration have a comparable signal– energy relationship with the conventional apparatus. This invariably implies that the sensitivity and linear range of the small apparatus can be further enhanced by increasing the excitation source intensity. Lastly, the noise level in the energy-dependent signal of the small apparatus is about the same as that of the conventional apparatus. This is due to the fact that both apparatuses use the same excitation source, detector, and electronic device. The signal-to-noise ratio for the small apparatus can be improved by using optimized components with appropriate detection and electronics devices.

CONCLUSION This work demonstrates that the design of a compact photothermal spectrometer can be achieved with reflective optical components. The reported prototype apparatus provides faster relaxation dynamics and higher sensitivity, as much as twice that of a conventional photothermal apparatus. Presently, the nature and properties of off-axis parabolic mirrors such as scattering, reflectivity, and asphericity have not yet been considered. The optimum conditions for measurement have to be successfully obtained from experiment. The sensitivity of

the small apparatus will probably increase even further with further refinements. The use of reflective optics also suggests that less cumbersome photothermal detector design is possible for integration with conventional analytical techniques such as high-performance liquid chromatography (HPLC), gas chromatography (GC), and capillary electrophoresis (CE). ACKNOWLEDGMENTS We wish to thank the Utah State University Space Dynamics Laboratory Enabling Technologies Program for generous support of this research.

1. D. Korsch, Reflective Optics (Academic Press, Inc, New York, 1991). 2. D. J. Schroeder, Astronomical Optics (Academic Press, Inc, 1999). 3. S. E. Bialkowski, Photothermal Spectroscopy Methods for Chemical Analysis (Wiley, New York, 1996). 4. L. C. Aamodt and J. C. Murphy, J. Appl. Phys. 52, 4903 (1981). 5. L. C. Aamodt and J. C. Murphy, J. Appl. Phys. 54, 581 (1983). 6. D. Bicanic, S. Kruger, P. Tores, B. Bein, and F. Harren, Appl. Spectrosc. 43, 148 (1989). 7. D. Fournier and A. C. Boccara, in Photothermal Investigations in Solids and Fluids (Academic Press, New York, 1989). 8. W. B. Jackson, N. M. Amer, A. C. Boccara, and D. Fournier, Appl. Opt. 20, 1333 (1981). 9. J. C. Murphy and L. C. Aamodt, J. Appl. Phys. 51, 4580 (1980). 10. S. L. Nickolaisen and S. E. Bialkowski, Anal. Chem. 57, 758 (1985). 11. J. A. Sell, Photothermal Investigation of Solids and Fluids (Academic Press, New York, 1989). 12. D. T. Allen, E. J. Palen, M. I. Haimov, S. V. Hering, and J. R. Young, Aerosol Sci. Technol. 21, 325 (1994). 13. W. P. Arnott, H. Moosmuller, R. E. Abbott, and M. D. Ossofsky, Rev. Sci. Instrum. 66, 4827 (1995). 14. A. Chartier and S. E. Bialkowski, Anal. Chem. 67, 2672 (1995). 15. O. O. Dada and S. E. Bialkowski, Appl. Spectrosc. 62, 1336 (2008). 16. O. O. Dada and S. E. Bialkowski, Appl. Spectrosc. 62, 1326 (2008). 17. A. Harata, Q. Shen, and T. Sawada, Annu. Rev. Phys. Chem. 50, 193 (1999). 18. E. L. Lasalle, F. Lepoutre, and J. P. Roger, J. Appl. Phys. 64, 1 (1988). 19. G. R. Long and S. E. Bialkowski, Anal. Chem. 57, 1079 (1985). 20. G. R. Long and S. E. Bialkowski, Anal. Chem. 58, 80 (1986). 21. A. Mandelis and R. F. Power, Appl. Opt. 27, 3397 (1988). 22. J. F. Power and A. Mandelis, Appl. Opt. 27, 3408 (1988). 23. M. A. Schweitzer and J. F. Power, Appl. Spectrosc. 48, 1054 (1994). 24. M. A. Schweitzer and J. F. Power, Appl. Spectrosc. 48, 1076 (1994). 25. M. A. Shannon, A. A. Rostami, and R. E. Russo, J. Appl. Phys. 71, 53 (1992). 26. M. J. Smith and R. A. Palmer, Appl. Spectrosc. 41, 1106 (1987). 27. S. E. Bialkowski and G. R. Long, Anal. Chem. 59, 873 (1987). 28. O. O. Dada and S. E. Bialkowski, Appl. Spectrosc. 61, 1373 (2007). 29. P. Joshi, O. O. Dada, and S. E. Bialkowski, Appl. Spectrosc. 63 (2009). 30. B. Z. Zuev, P. A. Gorkin, M. A. Proskurnin, and A. A. Zhirkov, J. Anal. Chem. 64, 379 (2009). 31. S. E. Bialkowski, Rev. Sci. Instrum. 58, 687 (1987). 32. S. E. Bialkowski, Anal. Chem. 61, 2479 (1989). 33. S. E. Bialkowski, Anal. Chem. 61, 2483 (1989).


205