2434
Anal. Chem. 1985, 57, 2434-2436
Double-Beam Thermal Lens Spectrometry K. L. Jansen and J. M. Harris* Department of Chemistry, University of Utah, Salt Lake City, Utah 84112
A double-beam optlcal configuration for thermal lens absorp Hon measurements comblnes advantages of both singlebeam and pumpand-probe Instrument designs. Only a slngle laser Is passed through the sample, ellmlnatlng the allgnment dHflcuitles associated with the pumpand-probe conflguratlon which requires that two laser beams be coaxlally allgned In the sample. Continuous modulation and lock-In ampllflcatlon of the signal allow convenlent appllcatbn of the double-beam method to real-time monitoring situations such as flow Injection analysls (FIA) or chromatography.
The thermal lens effect (1) has been developed into a sensitive method of determining weak absorbances (2). The optical configuration for the measurement has evolved along two distinct lines. The simplest of these configurations utilizes a single laser beam which serves both functions of producing the thermal lens and detecting its presence in the sample (I, 3, 4). Its chief advantages are optical simplicity and corresponding ease of optical alignment. Signal averaging with a single-beam configuration is generally accomplished by transient recording and curve-fitting the signal to a thermal diffusion response ( 5 ) . A second optical configuration utilizes two laser beams, one of which, absorbed by the sample, forms the thermal lens, while the second beam probes the lens in the sample (6, 7). This pump-and-probe arrangement requires coaxial alignment of the two beams within the sample, which is a particularly severe constraint when tightly focused beams are used for small volume detection (8). With chopped, continuous-wave excitation or pumping, the modulation on the probe beam can be synchronously detected by a lock-in amplifier. This convenient method of signal averaging is not generally suitable for single-beam detection due to the overwhelming synchronous background a t the chopping frequency. While the single-beam and pump-and-probe configurations yield comparable limits of detection (9), each has its own advantages in terms of optical and signal processing simplicity. Recently, several experimental arrangements have been proposed to combine advantages from each of the configurations. For example, a lock-in amplifier may be used to detect a thermal lens signal from a single-beam experiment a t the second harmonic of the chopping frequency (8). At this frequency, the synchronous background is small since a square wave contains only odd harmonics of the fundamental. This method, however, requires that the lock-in amplifier have outstanding harmonic rejection characteristics. A variation of the pump-and-probe arrangement utilizing only one laser has been reported ( I 0, I I) in which the single laser beam is split into two beams, one of which is chopped for a pumping beam. The plane of polarization of the pump beam is rotated by 90°,and the two beams are recombined in the sample. Before detection, the probe beam is separated with a polarizer. Like the single-beam configuration, this arrangement avoids chromatic aberrations in the optical elements since both beams are the same wavelength. Although the method still requires careful alignment of the two beams, errors due to beam pointing stability are reduced as the two beams move together. 0003-2700/85/0357-2434$01.50/0
In this work, we report a double-beam configuration for thermal lens measurements in which a single laser beam is chopped and split into two beams, one of which is passed through the sample. The average intensities of the two beams are matched and subtracted, producing a signal which contains only modulation due to the thermal lens. This method preserves the benefits of both single-beam and pump-and-probe configurations. Alignment is easy since only a single beam is passed through the sample. Subtraction of the reference beam from the sample beam intensity by the detector circuitry produces a null signal containing no modulation a t the chopping frequency except when a thermal lens is present. This allows the use of a lock-in amplifier to detect the thermal lens signal. The differential detection scheme also reduces the effect of short-term intensity fluctuations from the laser. The double-beam configuration is evaluated in this paper; real-time monitoring of flow-injected peaks is used to demonstrate applicability to detection in flow systems.
EXPERIMENTAL SECTION The double-beam thermal lens instrument is shown in Figure 1. An argon ion laser (Lexel, Model 95) ,is operated at 514.5 nm and 60 mW; the beam is passed through a 30.2-cm focal-length lens (Special Optics) and focused to a waist. An electronic shutter (Uniblitz, Model 225L2A) is controlled by a microcomputer for transient recording of single-beamdata, and a two-bladed chopper is used for continuous modulation in double-beam experiments. Following the chopper, the laser beam is split into sample and reference beams by a 3' wedged plate (Rolyn). The sample beam is passed through a flow cell (described below) and an optical delay consisting of a series of wedged plates to reduce the intensity. The sample beam next passes through a transmission mask and a collecting lens which comprise an optical processor to calculate the second spatial moment of the laser beam (12). The mask has a radially symmetric parabolic transmission profiie and multiplies the intensity of the far-field beam intensity distribution by r2. The intensity transmitted by the mask is integrated by a lens which focuses the radiation onto a single silicon photodetector (Silicon Detector Corp., Photovoltaic Group SD-200-12-12-241). The two-dimensional second moment thus obtained is proportional to the square of the spot size of the laser beam. Since this method utilizes the entire area of the beam rather than a small portion as in a beam-center measurement, uncertainty contributions from spatial noise in the laser beam are averaged out (12). The reference beam is passed through an optical delay similar to that in a sample beam path and then through a variable neutral density filter (Ealing, 35-6808, density range 0.04-1.0) and is focused onto a matched detector. Single-beam data are obtained by blocking the reference-beam detector. The two photovoltaic detectors are isolated from their cases and are connected in a push-pull configuration to the inverting input of a current-tovoltage converter. This circuit subtracts the current of one photodetector from that of the other. The only gain adjustment is made externally with the variable neutral density filter in the reference path. With matched intensities in the two beams, much of the optical noise is effectively canceled. Signal processing for the single-beam configuration has been previously described ( 5 ) . Sets of fifty 200-point transients were acquired at a digitizing rate of 1kHz. Lock-in detection for both static and flow-injected samples was performed at a chopping frequency of 12.5 Hz with a 1.25-stime constant by using an Ithaco Model 393, Dynatrac 3 Lock-in Analyzer. Sets of 40 points were gathered, averaged, and used for comparison to the averaged single-beam results. Samples were iodine (Fisher, resublimed) 0 1985 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 57, NO. 13, NOVEMBER 1985 0.08
a. '
I ' I I
0.05
f
PD
L2
I
I
I
I ,
1
I
2435
I
I
*
VND
Figure 1. Schematic dlagram of doublabeam thermal lens instrument. are plano-convex lenses, SH is an eiectronlc shutter, C Is a
L1-3
chopper, BS are beam splitters, S Is the sample, M is an aluminum mirror, VND is a variable neutral density filter, TM is a paraboilc transmisslon mask, D is the differential detector, and PD is a photodiode. Breaks in optical path indlcate optical delays. in carbon tetrachloride (Burdick and Jackson), filtered through 0.5-prn-poredisposable inline fdten (Millipore)prior to absorption measurements. Fluid pumping for flow-injection studies was provided at a flow rate of 0.45 mL/min by compressed helium regulated above the sealed solvent reservoir (14). The carrier stream was CCll degassed with helium prior to pumping. The 0.5-mm4.d. manifold tubing and fittings were solvent resistant, and a Teflon-lined injection valve (Rheodyne, Model 50) with a 1OO-pL injection loop was used for sample introduction. Following the injection valve and a 63-cm delay loop, samples passed through a 1-cm path-length flow cell (Starna, 41F) for detection and then out to waste. The flow cell propels the carrier transverse to the laser beam, allowing detection close to the point of sample introduction; the resulting dispersion (15)of the flow injection system, determined from the ratio of the detected peak volume to the injected volume, was approximately one.
RESULTS AND DISCUSSION The thermal lens apparatus was first tested in a single-beam configuration for linearity and sensitivity of response by construction of a calibration curve with six samples of I2 in CCll spanning an absorbance range of A 9.3 X lo-' to 9.3 x The results were linear (correlation coefficient, r = 0.992), and the sensitivity enhancement (2)relative to a Beer's law response was determined from the slope to be E = 337 f 17, which compares favorably with the expected value (2, 13),E = 352, based on the laser power, wavelength, and thermooptical properties of the solvent. The double-beam thermal lens spectrometer was evaluated by first testing its impact on the magnitude of the measured intensity noise. For both double-beam differential detection and single-beam detection, the latter obtained by blocking the reference beam, 200-point transient recordings of the laser intensity were made without the presence of a sample. The average values of the intensities were subtracted, and the residuals are plotted in Figure 2. For a single-beam signal level of 4.0 V, the rms noise is 25 and 8 mV for the single-beam and double-beam intensity measurements, respectively, which corresponds to a relative standard deviation of 0.62% and 0.20% for 1.0-ms samples of the intensity. The reduction in noise provided by the differential detection method is most significant for short-term fluctuations in the laser intensity. Longer-term drift in the relative transmission and alignment of the two arms of the double-beam spectrometer was found to arise from dust, air currents, vibration, and temperature or density gradients in the sample. While the parabolic mask measurement of the laser beam spot size (12) averaged the spatial contributions to the noise more effectively than pinhole detection, the residual differences in the intensity of the sample and reference arms were transformed into signals a t the chopping freduency due to the subtraction step. The effect of this drift is apparent in the
t!
0.05 0.03
I
I
- -0.03 -0.05 -0.08
0
40
120 160 iluE~MIusEcoNDs)
200
Figure 2. Intensity noise comparison. Signal mean is subtracted, and the resulting residuals are plotted (a) single-beam measurement; (b) double-beam differentlal measurement.
Table I. Limits of Detection: in Absorbance Units
time scale of measurement
single-beam transient signal averaging
double-beam lock-in detection
short-term (50 s)* long-term (10 min)e
1.5 X 5.1 x 10-7
5.0 x 10-7d 1.2 x 104
"Defined as LD= [2t (one-sided) X S,] for 95% confidence (16). *For single-beam, 50 202-point transients were signal averaged. For double-beam, the lock-in signal was digitized every 1.25 s for 50 s and averaged. Total measurement time was equivalent. Determined from the uncertainty in fitting the average thermal lens transient ( 5 , 1 7 ) . Determined from the standard deviation of the lock-in signal. e Detection limits obtained from the standard deviation of six replicates of the measurements indicated in footnote b. larger detection limits obtained for longer-term thermal lens measurements with the double-beam configuration (see Table I). The double-beam results are also compared to single-beam measurements for which the long-term detection limits are better by a factor of 2.5 since fitting the transient data allows the initial laser intensity to vary between replicates without affecting the magnitude of the thermal lens response inferred from the data. The apparently larger short-term detection limits for single-beam measurements are probably due to an overestimate of the uncertainty in fitting from the nonGaussian distribution of residuals. On the basis of the long-term reproducibility of the results, the short-term limits of detection for the single-beam and double-beam methods are probably comparable in magnitude. While the long-term noise of double-beam thermal lens measurements is somewhat greater than for single-beam transient fitting due to its sensitivity to intensity drift, the convenience of lock-in signal processing with the double-beam method might be an overriding advantage for real-time monitoring applications. An evaluation of double-beam thermal lens detection for flow injection analysis was carried out. The application of flow injection of trace-level samples into laser-based detectors capitalizes on the small sample
2436
Anal. Chern. 1985, 57,2436-2441
Ii \!-I
the analyte from the solvent in which it is injected (20). Double-beam thermal lens spectrometry could readily be adapted to detection in liquid chromatography. The 1.25-s time response is quite adequate for resolving closely spaced peaks, and the base line noise is comparable to previous examples of thermal lens detection in HPLC (8,11,21-24). The use of a single laser beam through the sample, however, eliminates the alignment difficulties of the pump-and-probe configuration, which are particularly severe for detection in microbore (8, 23) or capillary (24) HPLC, and lock-in amplification of the signal overcomes any delay or trade-offs in fitting transient single-beam data. Work is in progress to evaluate the double-beam thermal lens method for this application.
“I
LITERATURE CITED 5 min.
Flgure 3. Double-beam thermal lens detection of flow-Injected peaks: lo-’ M I, in CCI,, 100-mV lock-in scale: (b) 1 X lo-‘ M I, in CCI,, 1 . 0 4 lock-in scale. (a) 1 X
volume requirements of FIA which are matched to the capabilities of the laser, the smaller likelihood of sample contamination during processing, and the reduced carry-over from higher concentration samples due to the rinsing efficiency (18, 19).
The flow injection response of double-beam thermal lens detection is shown for two concentrations of injected iodine solution in Figure 3. A significant blank contribution to the measured response was observed in the amplitude and peak shape for the lower concentration samples. Based on its more asymmetric shape, this blank contribution is probably arising from refractive index gradients in the region of the sample zone due to slight differences in composition or temperature between the injected solvent and the carrier. The blank response was fairly constant, however, producing linear calibration results and detection limits based on the reproducwhich were larger than ibility of the blank, Amin= 1.7 X the detection limits based only on the baseline noise, Amin= 5.5 x To obtain base-line-limited detection and avoid the difficulty of exactly matching the optical properties of the injected solvent and carrier, a short chromatographic column could be added to the FIA manifold to temporally separate
(1) Gordon, J. P.; Leite, R. C. C.; Moore, R. S.; Porto, S. P. S.; Whinnery, J. R. J . Appl. PhyS. 1965, 36, 3-8. (2) Harris, J. M. “Analytical Applications of Lasers”; Piepmeier, E. H., Ed.; Wlley: New York, in press. (3) Whinnery, J. R:Acc. Chem. Res. 1974, 7 , 225-231. (4) Hu, C.; Whinnery, J. R. Appl. Opt. 1973, 72, 72-79. (5) Dovlchi, N. J.; Harris, J. M. Anal. Chem. 1981, 53, 106-109. (6) Grabiner, F. R.; Siebart, D. R.; Fiynn, G. W. Chem. f b y s . Lett. 1972, 77, 189-194. (7) Long, M. E.; Swofford, R. L.; Albrecht, A. C. Science 1976, 197, 183-185. (8) Pang, T. J.; Morris, M. D. Anal. Chem. 1984, 56, 1467-1469. (9) Carter, C. A.; Harris, J. M. Anal. Chem. 1983, 55, 1256-1261. (10) Yang, Y. Anal. Chem. 1964, 56, 2336-2338. (11) Pang, T. J.; Morris, M. D. Appl. Specfrosc. 1985, 3 9 , 90-93. (12) Jansen, K. L.; Harris, J. M. Anal. Chem. 1985, 57, 1698-1703. (13) Carter, C. A.; Harris, J. M. Appl. Opt. 1984, 23, 476-481. (14) Ruzicka, J.; Hansen, E. H.; Ramsing, A. U. Anal. Chim. Acta 1982, 134, 55-71. (15) Ruzicka, J.; Hansen, E. H. “Flow Injection Analysis”; Wiley: New York, 1981; Chapter 2. (16) Currie, L. A. Anal. Chem. 1968, 4 0 , 586-593. (17) Bevlngton, P. R. “Data Reduction and Error Analysis for the Physical Sciences”; McGraw-Hill: New York, 1969; Program 11-5. (18) Harrls, J. M. Anal. Chem. 1982, 5 4 , 2337-2340. (19) Leach, R. A.; Harris, J. M. Anal. Chim. Acta 1984, 764, 91-101. (20) Leach, R. A.; Harris, J. M. Anal. Chem. 1984, 56, 2801-2805. (21) Leach, R. A.; Harris, J. M. J . Chromatogr. 1982, 218, 15-19. (22) Buffett, C. E.; Morris, M. D. Anal. Chem. 1982, 5 4 , 1824-1825. (23) Buffett, C. E.; Morris, M. D. Anal. Chem. 1983, 55, 376-378. (24) Sepaniak, M. J.; Vargo, J. D.; Kettler, C. N.; Maskarinec, M. P. Anal. Chem. W84, 56, 1252-1257.
RECEIVED for review May 31, 1985. Accepted July 8, 1985. This research was supported in part by the National Science Foundation under Grants CHE82-06898add CHE85-06667.
Time-Resolved Thermal Lens Measurements in Flowing Samples Wayne A. Weimer and Norman J. Dovichi” Department of Chemistry, University of Wyoming, Laramie, Wyoming 82071 Thermal lens calorlmetry Is considered for flowing llquld samples. Under conditions of known flow profile, as in lamlnar flow, It is possible to construct an accurate model of the time-resolved thermal lens signal. This model Is useful In regression analysls of data taken In flowing llquld samples. Furthermore, the time-resolved model may be used In the optimization of thermal lens experiments In liquid chromatography detection. The model Is verified for flowlng liquid samples withln a square duct.
Since its development by Whinnery ( 1 4 , thermal lens techniques have been recognized as useful in the analysis of
weakly absorbing species (6-16). Specifically, the technique has been considered as a detector for chromatography and flow injection analysis ( I 7-22). Optimization of thermal lens detection requires the consideration of the effect of flow upon the signal. Further interest in thermal lens behavior within flowing liquid streams in found in convective perturbation of thermal lens signals (2, 6 , I I ) and in the intercavity thermal lens formed within a flowing dye laser stream (23, 24). Unfortunately, there has been little work on the theory of thermal lens calorimetry in flowing samples, although the temperature profile produced by a laser beam moving over a surface has been considered (25). In general, it is recognized that flowing streams produce an additional contribution to
0003-2700/85/0357-2436$.01.50/00 1985 American Chemical Society