2010
Anal. Chem. 1984,56,2010-2013
penetrates deeper into the matrix since sulfur is sparingly soluble in alcohol (7). The dissolved rn-nitroaniline is carried into the matrix with the solvent. Both reach a point where the incident IR radiation does not penetrate due to scattering losses, thus resulting in reduced absorbance in the diffuse reflectance spectrum. One can also conceive the possibility of rn-nitroaniline reacting with the sulfur. If this were to occur, one would expect a change in band positions, shapes, or relative intensities of the rn-nitroaniline spectrum. There was no substantial change in the rn-nitroaniline spectrum observed except for the decrease of the band intensity with time as the solvent evaporated from sulfur. Thus, rn-nitroaniline chemically reacting with sulfur is not considered to be a reasonable explanation for this observation.
solved with modern nebulization techniques. Diamond powder can be cleaned in a plasma furnace, thus making it possible to reuse the powder. This would offset the higher price of the diamond powder. Using diamond powder as a diffuse reflectance matrix may make it possible to interface DRIFTS with HPLC directly. The methylene chloride extraction step in the interface built by Kuehl and Griffiths ( 5 ) could be eliminated. This possibility is currently under study in this laboratory. Registry No, S, 7704-34-9; calcium fluoride, 7789-75-5; magnesium fluoride, 7783-40-6;silver bromide, 7785-23-1;silver chloride, 7783-90-6; diamond, 7782-40-3.
LITERATURE CITED (1) (2) (3) (4) (5)
Fuller, M. P.; Grififths, P. R. Anal. Chem. 1978, 50, 1906. Fuller, M. P.; Griffiths, P. R. Am. Lab. (Fairfield, Conn.) 1978, IO,69. Fuller, M. P.; Grifflths, P. R. Appl. Spectrosc. 1980, 3 4 , 533. Kuehl, D.; Griffiths, P. R. J. Chromatogr. Sci. 1979, 17, 471. Kuehl, D. T.; Griffiths, P. R. A n d . Chem 1980, 52, 1934. (8) Stewart, J. E. "Infrared Spectroscopy: Experimental Methods and Techniques"; Marcel Dekker: New York, 1970; pp 115-116. (7) Weast, Robert C. "CRC Handbook of Chemistry and Physics", 56th ed.; CRC Press: Cleveland, OH, 1975; p B-147. (8) Cotton, F. A.; Wilkinson, G. "Advanced Inorganic Chemistry", 3rd ed.; Interscience: New York, 1972; p 429.
CONCLUSIONS All of the compounds studied can be used as a substrate for diffuse reflectance spectrometry when certain conditions are considered. These conditions include the spectral region of interest and the solubility of the matrix compound in the analyte solvent. The reactivity of the matrix compound should also be considered. For example, both the silver halides and sulfur (8)react with amines and the silver halides are sensitive to visible light. Of the compounds studied in this work, 3000 grit (6 pm particle size) diamond powder was found to be the optimal size for use with polar solvents. The mechanical difficulty of applying solutions to the surface of the diamond may be
RECEIVED for review September 21, 1983. Resubmitted May 7,1984. Accepted May 11,1984. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by the U.S. Environmental Protection Agency.
Thermal Lens Spectrophotometry of Gaseous Hydrocarbon Molecules in the Infrared Region Tatsuji Higashi, Totaro Imasaka, and Nobuhiko Ishibashi* Faculty of Engineering, Kyushu University, Hakozaki, Fukuoka 812, Japan
Thermal lens spectra are measured for varlous organlc vapors In the Infrared region by changlng the emission wavelength of the 12C160,laser. Acetone has no characterlstlc peak, and benzene and toluene have band structures. Xylenes provlde rather complicated spectra. The spectral shape of 0-xylene Is different from the shapes of m - and p-xylenes, but the spectra for the latter two specles are almost Identical. Methanol and ethanol can readlly be resolved on the spectrum from their characterlstlc peaks. The detectlon llmR Is 12 ppb for methanol, whlch corresponds to an absorbance of 2.7 X io-' cm-'.
Ultratrace analysis of organic compounds such as aromatic hydrocarbons is required for monitoring of the working environment and the atmosphere. Infrared spectrometry is quite attractive because of its capability of providing information concerned with molecular vibration. However, the poor sensitivity of conventional infrared absorption spectrometry allows the determination of the samples above parts-permillion levels, and it is not sufficient for trace analysis of the gaseous pollutants at parts-per-billion levels. A thermal lens effect, first reported by Gordon et al. ( I ) , has been shown to be very useful for spectrophotometric
determination of trace species (2-5). This technique is currently used for the determination of the condensed phase sample in the visible region (1-18). Carter et al. have used this technique in the infrared region for the determination of 2,2,4-trimethylpentane in carbon tetrachloride using a He-Ne laser operated a t 3.39 pm (19). The thermal lens technique is useful not only for the sample in the condensed phase but also for the sample in the gaseous phase. In our previous studies, atmospheric nitrogen dioxide had been detected at parts-per-billion levels in the visible region by using a continuous wave argon laser or a pulsed dye laser as an exciting source (20-23). The spectral output of a COz laser occurs in the infrared region, where most of the organic compounds have absorption bands arising from their molecular vibration. Bailey et al. have measured the transient decay curve of the thermal lens signal for SF6and I, in Ne, Ar, Kr, Xe, COz, CHI, CZH6, C3H6,C&, and C4HIoand determined the thermal conductivity coefficients for these gases (24-26). Recently, we have demonstrated the determination of trace benzene in air using the P(30) line (9.639 pm) of the C02 laser as an exciting source (27). The wavelength of the COPlaser is discretely tunable from 9.1 pm to 10.9 wm in the infrared region. It may allow the measurement of the absorption spectrum, and therefore assignment of the molecular species may readily be achieved. If so,
0003-2700/84/0356-2010$01.50/00 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 12, OCTOBER 1984
Flgure 1.
Block diagram of experimental apparatus. Exhaust
Windowless I I
Exhaust II
Laser Beam
Flgure 2.
2011
Sample G a s ' Structure of windowless flow cell.
the concentration of a specific molecule might be determined from the characteristic peak in the spectrum. In this study we developed a windowless flow cell to remove background occurring from window materials. We measured the thermal lens spectrum for acetone, alcohols, and aromatic molecules by using the C02 laser as an exciting source and demonstrated the selective determination of these molecules from the thermal lens spectrum.
EXPERIMENTAL SECTION Apparatus. The experimental arrangement of the dual beam thermal lens system used in this study is shown in Figure 1. A 1zC1602laser (NEC, GLG2048C, 20 W), which can be tuned to any 80 wavelengths between 9.1 and 10.9 pm, is used as an exciting source. Scanning of the emission wavelength is carried out by changing the angle of the intracavity grating, which is driven by a stepping motor and homemade electronics. A He-Ne laser (NEC, GLG5340,632.8 nm, 4 mW) is used as a probe beam. The exciting laser beam is modulated at 10-20 Hz by a beam chopper, and it is focused into a sample cell by a ZnSe plano-convex lens (11-VI,Inc., focal length 76.2 cm). The flow cell (length 100 cm) with a double cylindrical structure, shown in Figure 2, is used in order to remove blank absorption by cell windows. The slightly reflected exciting beam by a wedged CdTe beam splitter 1 (11-VI Inc.), which is placed at Brewster's angle for the COz laser, is monitored by a piezoelectric sensor. The signal is measured by a lock-in amplifier (NF Circuit Design Block Co. Ltd., Model LI-570). The CdTe beam splitters 1and 2 are used for collimation and isolation of the exciting and probe laser beams. The present optical configuration allows efficient extraction of the probe beam with slight power reduction of the exciting beam. A part of the probe laser beam is reflected by a quartz wedge (beam splitter 3), and passes through the chopper. The modulation timing is detected by photodiode 1 (Hamamatsu Photonics, S1190). This signal is re-formed to rectangular pulses by a pulse generator
(Hewlett-Packard, 8013B),and it is used for synchronization of the electronics. The probe laser beam reflected by beam splitter 1passes through a band-pass filter which blocks room light. The signal intensity at the beam center was detected by a pinhole (diameter, 2 mm) and photodiode 2 (Hamamatsu Photonics, S780-8BQ). The lock-in output was converted to frequency by a V-F converter (Kyushu Keisokki, QVF-100) and integrated 100 times by the counter (NF Co., Ltd., PC-545A). The intensities of the thermal lens signal and the COzlaser power were converted to digital values (12 bit) and were recorded by a microcomputer (Sord, M223 Mark 111). This measurement procedure is carried out at intervals of 0.5 s. The data were stored in the minifloppy disk, and the results were displayed by a X-Y plotter (Watanabe Sokki, WX4671). It took about 50 min for scanning the wavelength of the C02 laser and 20 min for data processing and spectrum display. Reagents. The nitrogen for sample dilution was supplied from a gas cylinder. Acetone, benzene, toluene, o-xylene, m-xylene, p-xylene, methanol, and ethanol used as samples were superfine or analytical grades. The xylene (mixture of 0-,m-, and p-xylenes) was obtained from Kishida Chemical Co. These standard gases were generated continuously from a diffusion tube by using a Permeater (Kitazawa, PD-1B). The concentration was adjusted by controlling the temperature of the diffusion tube and the flow rate of nitrogen.
RESULTS AND DISCUSSION Blank. Blank absorption is a major problem in ultratrace analysis, especially the window material which provides a large blank in the infrared region. Thermal lens spectrophotometry does not necessarily require cell windows for the measurement of the sample. The double cylindrical cell allows the sample to flow through the inner tube and to be exhausted from the outer tube. As the flow rate of the exhaust gas was adjusted to be slightly higher than that of the sample gas, a small amount of the room air is exhausted through the outer tube. This mechanism allows the preparation of the standard sample gas without using a closed cell and removes hazardous leakage of the toxic gas into the laboratory room. It is pointed out that background drift may, however, occur by a turbulent mixing of the streams (28). The thermal lens spectrum of diluent air containing no sample gas was measured before the determination of the organic samples. A small random noise was observed from 9.1 pm to 10.9 pm, even when the exciting beam was inter-
2012
ANALYTICAL CHEMISTRY, VOL. 56,NO. 12, OCTOBER 1984 0.08
F
-
i 0.02 -
0.04
L.
0
13 5 0 1 0
9 5
10 3
&tale l e n g t h k
10
3
11 0
!
1n)
0.02 0
0.04 0.02
Thermal lens spectra of acetone (150 ppm). Ordinate is (a) thermal lens signal (Io- I J Z 0 and (b) CO, laser power (watts). Flgure 3.
0
0.06 0.03
0 9.0
9.5
10.0
d-%.A.n-.rL 10.5
11.0
Wavelength (urn)
Figure 5. Thermal lens spectra of xylenes: (a) o-xylene (19 ppm), (b) m-xylene (22 pprn), (c) p-xylene (24 pprn), (d) xylene mixture
(mixture ratio unknown, 24 ppm). 9.c
9.5
10.0
15.5
11.0
l l a v e l e n g t h [ Jm:
Flgure 4. Thermal lens spectra of (a)benzene (22 ppm) and (b) toluene (40 ppm).
cepted. This background signal may arise from light scattering of the probe laser beam by small particulates in the air. The particulates also gave a large blank noise, when the exciting laser beam was introduced into the sample cell. It seems to be originating from light absorption of the COPlaser beam by the particulates. When the sampe flow was stopped, the particulates fell down by sedimentation and the background noise could be greatly reduced. When nitrogen from a gas cylinder was used as a diluent, background could be reduced. Therefore, nitrogen was used as the diluent in this study. No appreciable effect occurring from turbulent mixing could be observed in this study, though the refractive index and heat conductivity are slightly different for nitrogen and air. In practical applications the particulates in the air should be carefully removed by filtration. Acetone. The thermal lens spectrum of acetone (150 ppm) is shown in Figure 3. The abscissa shows the exciting wavelength of the COPlaser, which is tuned to any of -80 wavelengths from 9.1 pm to 10.9 pm. The ordinate of the spectrum (a) indicates the thermal lens signal, (Io- Im)/lo, and that of the spectrum (b) is the COz laser power (watts). The emission spectrum of the COP laser has four maxima around 9.2 pm, 9.5 pm, 10.2 pm, and 10.7 pm, which correspond to two pairs of P and R branches. Since the thermal lens spectrum of acetone is similar to the emission spectrum of the COz laser, acetone has a flat absorption spectrum in this wavelength region. These bands seem to be due to VC-C stretching vibration (29). Acetone has no characteristic band and has low molar absorptivities in this region, so that trace analysis may be difficult by using the present thermal lens spectrophotometry. Aromatic Compounds. Figure 4a and Figure 4b show the thermal lens spectra of benzene and toluene, respectively. The spectrum of benzene has a peak at around 9.639 pm, which corresponds to the P(30) line of the COz laser. On the other hand toluene has a peak at around 9.250 pm corresponding to R(24). Both aromatic compounds have strong absorption bands a t the 9.6 pm band system, and they seems to be due to Sc-H in-plane bending vibration (30). The toluene molecule provides the absorption band at around 9.2 pm, and it may be assigned to SC-H in-plane bending vibration, which is specific for a monosubstituted aromatic compound (9.01-9.35 pm) (29-31). On the other hand benzene gives a very weak band for the 9.2 pm band system. From this fact benzene and toluene can readily be identified on the thermal lens spectrum.
w
4.0
9.5
10.0
11.0
10.5
l$avelength(un)
Figure 6. Thermal lens spectra of alcohols: (a) methanol (2.6 ppm), (b) ethanol (4.3 pprn), (c) methanol (2.5ppm) ethanol (0.84 ppm).
+
The intensity of the thermal lens signal above 10 pm is relatively low, which is consistent with the reported low molar absorptivities in this wavelength region (32). Figure 5 shows the thermal lens spectra of o-xylene, mxylene, p-xylene, and the mixture of these compounds. The absorption band of the xylenes extends from 9.1 pm to 10.9 pm, and the spectral features are more complicated than those of benzene and toluene. The xylene molecules provide the strong absorption band at the 9.6 pm region, which is known to be due to the dC-H in-plane bending vibration for 1,2-, 1,3-, and 1,Cdisubstituted aromatic compounds (31). In the spectrum of o-xylene the band intensity at around 9.2 pm is relatively weak and that at around 10.2 pm is relatively strong in comparison with other xylenes. The present result is consistent with the data obtained by the absorption measurement (32). From the thermal lens spectrum for the mixture sample shown in Figure 5d, it is found that o-xylene is not dominant in the sample since the spectral feature is similar to m- and p-xylenes. Unfortunately, the absorption spectra of m-xylene and p-xylene are almost identical, and the present thermal lens spectrophotometry cannot resolve the contribution of these compounds on the spectrum. Alcohol. The thermal lens spectra of methanol and ethanol are shown in Figure 6a and Figure 6b, respectively. Methanol has a sharp characteristic peak at 9.676 pm, which corresponds to the P(34) line of the C 0 2laser. This result agrees with the data obtained by photoacoustic spectrometry (33). This band may originate from vc-o stretching vibration for a primary alcohol (29, 31). When water was measured as a sample, a strong peak was observed at P(34). But, it was found to be arising from methanol used for rinsing the diffusion tube. Methanol has a high vapor pressure and a large absorption cross section, therefore trace methanol is sometimes detected as an impurity. Ethanol has a relatively strong absorption
ANALYTICAL CHEMISTRY, VOL. 56, NO. 12, OCTOBER 1984
band in the 9.2-pm region, which may originate from U C and ~ uC4 stretching vibrations. The thermal lens spectrum of the mixture of methanol and ethanol is shown in Figure 6c. The presence of methanol can readily be identified from the characteristic peak at P(34), and its concentration is calculated to be 2.4 ppm, which agrees fairly well with a calibrated value of 2.5 ppm. The relative intensity of the band at 9.2 pm for the mixture is larger than that of methanol, and this absorption band seems to be occurring from ethanol. From the signal intensity of this band the concentration of ethanol is calculated to be 0.81 ppm, which is close to a calibrated value of 0.84 ppm. As described, the semiquantitative determination of methanol and ethanol can be easily carried out from the thermal lens spectrum. Other Species. Most of the organic compounds have C-C bonds in their molecule, so that they absorb COzlaser emission between 9.1 and 10.9 pm. It implies that present thermal lens spectrophotometry is useful for many organic molecules and might be used as an universal detector for gas chromatography. It is noted that even the inorganic molecule NH3 has an absorption band in this region (33) and seems to be detected at ultratrace levels. Sensitivity. For methanol a straight analytical line was observed from 0 to 5 ppm, and the background signal corresponded to 100 ppb of methanol. This background seems to originate from light absorption by the CdTe beam splitters and by the particulates in the sample cell. When the value of -6 given for the parabolic lens model (18),in which higher-order terms were taken into account, was plotted against the concentration of methanol, the analytical curve was linear at least to 40 ppm. The parabolic lens model provides a longer analytical curve as previously reported (21, 34). Methanol could be measured at a concentration as low as 47 ppb, the detection limit being 12 ppb ( S I N = 2) by using the P(34) line of the C 0 2 laser with an output power of 7 W. The detection limit corresponds to an absorbance of 2.7 X lo-’ cm-l, the minimum detectability of the thermal lens signal (IoIm)/Iobeing 4.3 X lo4. The present result shows that thermal lens spectrophotometry is a few hundred times more sensitive than conventional spectrophotometry. By taking account of the absorption cross section, the detection limit is estimated to 4 ppb for ethylene and 19 ppb for ethanol. On the other hand the detection limits for the other compounds used in the study are estimated to several hundred parts per billion. It is reported that the detection limit of ethylene is 10 ppb for photothermal deflection spectrophotometry (35). On the other hand 5 ppb of ethylene is detected by photoacoustic spectrometry (33). It is considered that the sensitivity of thermal lens spectrophotometry is similar to these spectrometric methods. The enhancement factor (E),which indicates the relative sensitivity of thermal lens spectrophotometry to conventional absorption spectrometry (6), is given by eq 1 for the aberrant lens model (17, 34). Io and I , are intensities at the beam
Io - I ,
0.524P dn (E)A
= 0.5248 = 2.303- Xk
the assumption in the theory; (ii) the present C02 laser does not have an exact Gaussian mode; (iii) a modulation frequency of 10 Hz is too high to obtain the steady-state intensity. The achieved detection limit is much lower than that estimated from the enhancement factor. This result arises from the fact that the signal (Io) and reference (I,) intensities are succeedingly measured at the modulation frequency so that much smaller signal can be detected by thermal lens spectrophotometry than conventional absorption spectrometry. According to the theoretical estimate, a continuous laser source provides a rather poor detection sensitivity in thermal lens spectrophotometry. On the other hand the thermal lens spectrophotometer with a pulsed laser is very sensitive especially for the gas-phase sample (23). It is emphasized that an enhancement factor of 8 X lo5 is achieved for the system consisting of a C02 laser with an output energy of 1 J (20). The pulsed thermal lens system may be advantageously used for a gas chromatographic detector. By use of pure argon as a carrier gas to reduce background, the organic compounds may be separated by a column from impurities such as water and COz, and they might be determined at ultratrace levels.
LITERATURE CITED
= 2.303EA (1)
IO
center immediately and sufficiently long after introducing the C02 laser beam. P is the power of the COz laser, X the wavelength of the probe laser, k the heat conductivity, (dn/dT) the variation of refractive index with temperature, and A the absorbance of the sample. It is noted that eq 1 is valid only as an approximation because of poor mode-match between the visible He-Ne laser and the infrared C 0 2 laser. The calculated value from the slope of the analytical curve is 14. This value is only 7% of the theoretical one calculated from eq 1. The enhancement factor might be decreased by the following reasons: (i) the sample cell is too long to satisfy
2013
(31) (32) (33) (34) (35)
Gordon, J. P.; Leite, R. C. C.; Moore, R. S.;Potto, S.P. S.; Whinnery, J. R. J . Appl. PhyS. 1985, 36,3-8. Harris, J. M.; Dovichi, N. j. Anal. Chem. 1980, 52,695A-706A. Kiiger, D. S.Acc. Chem. Res. 1980, 13,129-134. Imasaka, T.; Ishibashi, N. Trends Anal. Chem. 1982, 7 , 273-277. Kliger, D. S. “Ultrasensitive Laser Spectroscopy”; Academic Press: New York, 1983; pp 175-232. Dovichi, N. J.; Harris, J. M. Anal. Chem. 1979, 51, 728-731. Dovichi, N. J.; Harris, J. M. Anal. Chem. 1980, 52,2338-2342. Dovichl, N. J.; Harris, J. M. Anal. Chem. 1981, 53, 106-109. Dovichi, N. J.; Harris, J. M. Anal. Chem. 1981, 53,689-692. Carter, C. A.; Harris, J. M. Anal. Chem. 1983, 55, 1256-1261. Imasaka, T.; Miyalshi, K.; Ishibashi, N. Anal. Chim. Acta 1980, 175, 407-4 10. Miyaishi, K.; Imasaka, T . ; Ishibashi, N. Anal. Chim. Acta 1981, 124, 381-389. Mori, K.; Imasaka, T . ; Ishibashi, N. Anal. Chem. 1982, 54, 2034-2038. Miyalshi, K.; Imasaka, T.; Ishibashi, N. Anal. Chem. 1982, 54, 2039-2044. Haushalter, J. P.; Morris, M. D. Appl. Spectrosc. 1980, 34,445-447. Fujiwara, K.; Lei, W.; Uchiki, H.; Shimokoshi, F.; Fuwa, K.; Kobayashi, T. Anal. Chem. 1982, 54,2026-2029. Sheldon, S. J.; Knight, L. V.; Thorne, J. M. Appl. Opt. 1982, 27, 1663- 1669. Hu, C.; Whlnnery, J. R. Appl. Opt. 1973, 12, 72-79. Carter, C. A.; Brady, J. M.; Harris, J. M. Appl. Spectrosc. 1982, 36, 309-314. Mori, K.; Imasaka, T.; Ishibashi, N. Anal. Chem. 1983, 55, 1075-1079. Higashi, T.; Imasaka, T.; Ishibashi, N. Anal. Chem. 1983, 55, 1907-1910. Shimanoe, K.; Imasaka, T . ; Ishibashi, N. J . Chem. Phys. 1983, 79, 320 1-3205. Higashi, T.; Imasaka, T.; Ishibashi, N., Hakozaki, Japan, unpublished work, 1983. Bailey, R. T.; Cruickshank, F. R.; Pugh, D.; Johnstone, W. J . Chem. Soc., Faraday Trans. 2 1980, 76,633-647. Bailey, R. T.; Cruickshank, F. R.; Pugh, D.; Johnstone, W. J . Chem. Soc., Faraday Trans. 2 1981, 77,1387-1397. Bailey, R. T.; Crulckshank, F. R.; Pugh, D.; Mcleod, A,; Johnstone, W. Chem. Phys. 1982, 68,351-357. Higashi, T.; Imasaka, T.; Ishibashi, N. Bunseki Kagako 1982, 37, 680-68 1. Rohde, R. S.;Buser, R. G. Appl. Opt. 1979, 78, 698-704. Colthup, N. B. J . Opt. SOC.Am. 1950, 40,397-400. Silverstein, R. M.; Bassler, 0. C.; Morrill, T. C. “SDectrometric Identification of Organic Compounds”; Wiley: New Yoik, 1981; pp 95-180. Tanaka, S.;Iida, Y. “Kikibunseki”; Syokabo: Tokyo, 1979; pp 66-94. Andersson, P.; Persson, U. Appl. Opt. 1984, 23, 192-193. Kreuzer, L. B.; Kenyon, N. D.; Patel, C. K. N. Science 1972, 177, 347-349. Carter, C. A.; Harris, J. M. Appl. Opt. 1984, 23,476-481. Fournler, D.;Boccara, A. C.; Amer, N. M.; Garlach, R. Appl. Phys. Lett. lg80, 37,519-521.
RECEIVED for review January 16,1984. Accepted May 30,1984. This research was supported by a Grant in Aid for Scientific Research (Grant No. 57430016) from the Ministry of Education of Jdpan and by a Steel Industry Foundation for the Advancement of Environment Protection Technology.
