1260
ANALYTICAL CHEMISTRY, VOL. 50, NO. 9, AUGUST 1978
with those cited in recent publications ( 2 1 , 1 2 ) . The new MIP detection limits are improved in most cases over those obtained with single-shot sample injection and are improved in all cases over continuous nebulization sample injection systems. Significantly, the detection limits obtained on the present single-shot He MIP system were calculated on the basis of the 1-pL sample aliquot actually used, and therefore represent routinely obtainable values. Interferences. MIPs have been noted to be affected by the introduction of relatively large amounts of sample material (-1 Fg absolute). However, the new plasma remained ignited during injection of sample material up to the maximum amount atomizable from the microarc used, about 5 pg absolute. Of course, the relative standard deviation of the measured signal degrades as larger and larger amounts of material are atomized into the plasma; however, the plasma remains ignited and, in that sense, proves to be a significantly more durable source than many previous MIPs. As suggested earlier, ionization interferences are expected in any MIP, and in the new plasma a significant amount of ionization of any easily ionizable element (e.g., Ca, Na, Li) appears to occur. As in most determinations, an ionization suppressant can be employed to overcome such interferences. Further examination of ion emission lines is warranted to determine their analytical utility. Refractory elements, such as A1 and Si, have been found to exert little effect on analyte signals in the plasma. This result is not unexpected in view of the use of the microarc sample atomizer ( 4 ) . However, further investigations of solute vaporization interferences are necessary and are currently under way.
CONCLUSIONS T h e atmospheric pressure, helium microwave-induced plasma generated in the cavity designed by Beenakker ( I ) is a durable, stable, and highly efficient excitation source for emission spectrometry of metallic elements. It is easy to ignite and operate and uses low volumes of inert support gas. Moreover, the cavity does not require cooling. Although
injected material does lead to reduced excitation efficiency and increased instability, this MIP exhibits a significant improvement over other versions of microwave-induced plasma in its tolerance of sample and solvent material. The high temperature of the new plasma leads to increased ionization and population of higher excited states; this result requires careful choice of emission lines to be used for analytical measurement. Dynamic background correction should prove useful with this plasma for the elimination of broadband molecular emissions.
ACKNOWLEDGMENT The authors thank C. I. M. Beenakker for useful discussions concerning the use of his new cavity. Also, we greatly appreciated the assistance of Rodney K. Williams in early stages of the investigation.
LITERATURE CITED (1) C. I. M. Beenakker, Spectrochim. Acta, Part 8, 31, 483 (1976). (2) C. I. M. Beenakker. Paper 305, Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, March 1977. (3) C. I. M. Beenakker, Spectrochim. Acta, Part 8 ,32, 173 (1977). (4) L. R. Layman and G. M. Hieftje, Anal. Chem.. 47, 194 (1975). ( 5 ) S. Ramo and J. R. Whinnery, "Fields and Waves in Modern Radio", 2nd ed., John Wiley & Sons, New York, N.Y., 1953. (6) A. G. Gaydon, "The Spectroscopy of Flames", 2nd ed., John Wiiey & Sons, New York, N.Y., 1974. (7) I. Kopp, R. Lindgren, and B. Rydh, "Table of Band Features of Diatomic Molecules by Wavelength Order", Version A, Insttute of physics, University of Stockholm, Stockholm, Sweden. (8) P. M. Houpt, Anal. Chim. Acta, 88, 129 (1976). (9) K. Fallgatter, V. Svoboda, and J. D. Winefordner, Appl. Spectrosc., 25, 347 (1971). (10) W. L. Wiese, M. W. Smith, and B. M. Glennon, "Atomic Transition Probabilities-Hydrogen Through Neon", NSRDS-National Bureau of Standards, 4, Voi. 1, US. Government Printing Office, Washington, D.C., 1966. (11) R. K. Skogerboe and G. N. Coleman, Anal. Chem., 48, 611A (1976). (12) S. Greenfield, H. McD. McGeachin, and P. 8 . Smith, Talanta, 22, 553 (1975).
RECEIVED for review February 10,1978. Accepted May 5,1978. Supported in part by the Office of Naval Research, by the National Institutes of Health through grant PHS GM 17904-05, and by the National Science Foundation through grant CHE 76-10896.
Quantitative Detection of Trace Impurities in Gases by Infrared Spectrometry of Cryogenic Solutions Samuel M. Freund," William B. Maier 11, Redus F. Holland, and Willard H. Beattie University of California, Los Alamos Scientific Laboratory, P.O. Box 1663, Los Alamos, New Mexico 87545
A technique for considerably improving the sensitivity and specificity of infrared spectrometry as applied to quantitative determination of trace impurities in various carrier or solvent gases Is presented. A gas to be examined for impurities is liquefied and infrared absorption spectra of the liquid are obtained. Spectral simplification and number densities of impurities in the optical path are substantially higher than are obtainable in similar gas-phase analyses. Carbon dioxide impurity (-2 ppm) present in commercial Xe and ppm levels of Freon 12 and vinyl chloride added to liquefied air are used to Illustrate the method.
There are several methods for identifying and determining trace impurities in gases, such as mass spectroscopy, gas 0003-2700/78/0350-1260$01 .OO/O
chromatography, and infrared spectrometry, or a combination of these methods. Infrared spectrometry has the advantage of being a versatile technique, but it suffers from poor sensitivity and occasionally from poor specificity, Le., from difficulty in identifying overlapping bands of different compounds when complex mixtures are to be analyzed. Very long pathlengths may be employed to overcome the poor sensitivity, but with certain mixtures, quantitative and often qualitative analyses are still impossible. In this paper we present a technique for greatly improving the sensitivity and specificity of infrared spectrometry as applied to quantitative determination of some trace impurities in gases by utilizing cryogenic solutions. A gas to be examined for impurities is cooled and pressurized until it becomes a liquid, which is a solvent for the impurities. Solubilities of trace gases in liquefied carrier gases are often 0 1978 American Chemical
Society
ANALYTICAL CHEMISTRY, VOL. 50, NO. 9, AUGUST 1978
1261
0.9 cm-' RESOLUTION
oL 06 cm I 3 cm
2400
RESOLUTION PATHLENGTH
2350
73cc
0 2360
2250
PAVENUMBER icm
Figure 1. Infrared spectrum of 100.3 Torr of CO, vapor at 45 OC (curve I) and that of 2 ppm of CO, impurity in liquid Xe at -1 10 OC (curve 11). In the vapor, the '2C0, band overlaps that of 13C0, at the longer wavelengths
sufficiently high t o make analyses practical. Since molecular densities in liquids are -2 x molecules/ cm3, a 1-ppm level of some impurity translates into ~ 1 0molecules/cm3 ' ~ of impurity in solution, much above the molecular density in the gas phase. Molecular rotation is suppressed in solution, so that the generally complex rotational structure of a solute vibrational band usually collapses to a single, sharp feature having a peak absorbance (at moderate resolutions) higher than that for the gas a t room temperature. Provided that the liquefied carrier gas is essentially free of infrared absorptions a t wavelengths where the impurities absorb, detection of low levels (ppm-ppb) of impurities can be performed. T h e technique is illustrated with four examples. Vinyl chloride (CzH3Cl)and dichlorodifluoromethane (CClPF2)are dissolved at ppm concentrations in liquid air, and commercially available spectroscopic grade Xe is liquefied in order to examine C 0 2 and fluorocarbon impurities. Part-per-million levels of impurities are easily and quantitatively observable in cells with about 1-cm optical pathlengths.
EXPERIMENTAL Infrared spectra were obtained with a Perkin-Elmer Model 180 scanning spectrometer which was continuously flushed with dry nitrogen. Instrumental time constants were of the order of 2 s. The vacuum insulated, copper cryogenic cells used have been described elsewhere ( I ) . Optical pathlengths and cell volumes were 1.3 cm and 2.6 cm3, respectively, for the experiments conducted in liquid Xe (LXe) and 2.6 cm and 5.0 cm3 for those in liquid air (LAir). Calcium fluoride windows were utilized in both cases. All of the gases employed were used as supplied by the manufacturers without further purification. In order to determine the approximate range of sensitivity of the technique, it was necessary to measure the peak absorption cross sections of the dissolved test gases. Two procedures for dissolution were employed. In the first, a solute gas of interest was mixed with the solvent gas and introduced into the empty, room temperature cell. The lower portion of the cell was cooled and the upper portion heated to condense the solvent gas and additive gas only in the bottom of the cell. During the condensation, a source of solvent gas was kept open to the cell. The second procedure commenced with a cell partly filled with the solvent liquid and maintained at the final temperature of interest. The premixed solute and a small amount of solvent gas were then rapidly swept into the cell with additional solvent gas and condensation was allowed to proceed until the viewing volume was filled-this point is determined by direct observation of the liquid level in the cell. In both procedures, the solution was stirred with a Teflon coated magnetic bar driven by an external source. These mixing procedures cannot put more than the measured amount of additive gas into the cell; therefore, the additive levels quoted herein are upper limits to the concentrations.
2350
2340
2330
2320
WIG
WAVENUMBER (cm")
Figure 2. Infrared spectrum of 6 ppm total CO, dissolved in liquid Xe at -1 10°C (curve I) and that of the commercial LXe used, which has about 2 ppm CO, impurity (Curve 11)
RESULTS AND DISCUSSION T h e results from the COS calibration and trace determination experiments are shown in Figures 1 and 2. In Figure 1, we compare the 4-pm vapor phase spectrum of pure COP a t 45 "C with a spectrum of the C 0 2 impurity in the liquefied (-110 "C) commercial spectroscopic grade xenon used in our laboratory. [Substantial ( - 1-100 ppm) fluorocarbon impurities have been found in samples of spectroscopic grade rare gases from several suppliers. Spectra are not given herein. T h e manufacturer's specifications quote K0.5 ppm COz and levels of fluorocarbons "too low to measure".] The absorption in curve I1 of Figure 1 corresponds to about 2 ppm of C 0 2 in the xenon sample. The absorption feature of dissolved l3CO2 in natural abundance is just perceptible a t 2271 cm-' and corresponds to -0.02 ppm. Note the almost complete suppression of rotational structure in the liquid phase. Further, the v3 band of the solvated COSis broader than the individual rotational features resolved in the vapor phase spectrum, although the instrumental resolution for the two traces is identical. T h e width of the band in solution may be partly the result of the C 0 2 molecules being located in a variety of environments in the liquid xenon. In other cryogenic solvents, the width of the spectral features may be different ( 2 ) . Actually, the width of the COPbands in Xe solution is convenient for quantitative analysis, since for accurate measurement of absorbance, the spectrometer pass band needs to be appreciably narrower than the spectral width of the absorption features. This requirement is easily satisfied in the solutions, with the resolutions indicated in Figures 1 and 2 . Considerably higher resolutions than the Model 180 allows would be required to measure accurately the absorbance for the COS gas a t the pressure of Figure 1. Figure 2 shows a spectral scan of approximately 6 ppm total C 0 2 dissolved in LXe and a scan (of LXe containing C 0 2 impurity as in Figure 1. These spectra were obtained at a slower scan speed in order to improve the quantitative determination of the peak heights. T h e substantial signalto-noise ratio observed for these C 0 2 levels suggests that a factor of 100 reduction in C 0 2 concentration could be measured quantitatively with the same apparatus and conditions. Use of a longer optical pathlength and/or sophisticated data averaging and processing techniques would further improve the sensitivity of the method. T h e suppression of rotational structure of solvated molecules is a general phenomenon and can be of great value in the analysis of the spectra of complex mixtures. Figure 3 shows an infrared spectrum in the 10-km region of a mixture of CF2C12(0.3ppm; features a t 917.5 and 886.5 cm-') and CPH3Cl(2 ppm; features a t 943.0 and 899.0 cm-') dissolved in liquefied bottled air a t -188 "C. This spectrum is compared
1262
ANALYTICAL CHEMISTRY, VOL. 50, NO. 9, AUGUST 1978
L
;L
CCC
09 c w RESOLUTION I 2 6 cm DATHLENGTH ! 25-2 1
9%
900
--830
WAVENUMBER (cm"1
Figure 3. Infrared spectrum (curve I) of the mixture of CCI,F, (0.3
ppm; features at 917.5 and 886.5 cm-') and C2H,CI (2 ppm; features at 943.0 and 899.0 cm-') dissolved in liquid air at -188 O C , and spectra of 25 O C mixtures of 10 Torr CCI,F, and 100 Torr of C,H,CI (curve 11) and 10.6 Torr of CCI,F, and 6.8 Torr of C,H& (curve 111)
with those for two gas-phase mixtures of the same compounds. It is apparent that little qualitative or quantitative information can be derived from the vapor phase spectra, whereas the bands of the solvated molecules are completely resolved. One can test for saturated solutions by raising or lowering the temperature of the liquid and observing the effect on the absorption bands. The additive gases investigated are more soluble a t higher temperatures. By maintaining a substantial pressure of solvent gas over the liquid, a broad liquid-phase temperature range is available for a given solvent. Finally, the liquid-phase features are shifted in energy relative to the origins of vapor phase rotation-vibration bands. For example, gaseous CF2C12has band origins at 922 and 882 cm-' ( 3 ) ,whereas the observed transitions in solution are at 917.5 and 886.5 cm-l, respectively. The magnitude of this shift is dependent on which solvent is used, but our data suggest that a more reasonable location for the quoted 882 cm-l feature would be somewhat higher than 886.5 cm-'.
CONCLUSIONS We have demonstrated that conventional infrared spectrometry can be used to analyze trace impurities in some gases
with improved sensitivity if the mixture is liquefied. This work utilizes absorption pathlengths of the order of 1 cm, yet achieves sensitivities comparable to pathlengths of many meters under usual vapor-phase conditions. Increasing the effective pathlength by using a multiple traversal scheme or by actually lengthening the absorption cells would further improve the sensitivity of the method. The widths of the features observed in the cryogenic liquids are broader than the resolved rotational lines in low pressure gas-phase spectra but much narrower than those characteristic of high pressure. Advantages of the technique then are (a) the high densities in the optical path and the higher peak absorption cross sections provide sensitivities which are greatly improved (>lo3) over atmospheric pressure gas-phase methodology for similar path lengths; (b) the higher densities of the trace impurities in the liquids mean that errors due to sorption or reaction at the walls of the spectroscopic cell are not as critical as in gases, where extremely small molar concentrations are typical for trace impurities; (c) once absorptivities are known at the liquid temperature, the method gives absolute concentrations independent of the pressure of the original gaseous sample; (d) chemical reactions of species which might be unstable in the gas phase may be suppressed a t the very low temperatures used; and (e) the analysis of mixtures may be facilitated by the dramatic simplification of the absorption spectra.
ACKNOWLEDGMENT The authors thank Bruce Stewart for his technical help and P. Aldridge for his support of this work. LITERATURE CITED William 6.Maier 11, Samuel M. Freund, Redus F. Holland, and Willard H. Beattie, "Photolytic Separation of D from H in Cryogenic Solutions of Formaldehyde", submitted for publication in J . Chem. Phys., 1978. L. J. Marabella, A p p l . Spectrosc. Rev., 7 , 313 (1973). "Tables of Molecular Vibrational Frequencies", T. Shimanouchi, Ed., fhys. Chem. Ref. Dafa, 3, 269 (1974).
RECEIVED for review February 17, 1978. Accepted May 22, 1978. This work was performed under the auspices of the U S . Department of Energy.
Photoacoustic Spectroscopy Applied to Systems Involving Photoinduced Gas Evolution or Consumption Robert C. Gray and Allen J. Bard* Department of Chemistry, The University of Texas at Austin, Austin, Texas 78772
Photoacoustic spectroscopy (PAS) was used to study photoinduced reactions where PAS signals attributable to gas evolution and consumption have been observed In addition to the usual thermally generated pressure fluctuations. Examples of such PAS studies of oxygen consumption in the photooxidation of rubrene and gas evolution in the heterogeneous photocatalytic oxidation of acetic acid to methane and CO, at a platinized TiO, catalyst are given. The sensitivity of the method and possible further applications are also described.
Recently there has been a resurgence of interest in the theory and applications of the photoacoustic effect and photoacoustic spectroscopy (PAS) (1-19). In the usual mode 0003-2700/78/0350-1262$01 .OO/O
of operation, PAS involves the detection (with a microphone) of a pressure wave induced by thermal changes in a sample upon absorption of light. The sample, enclosed in a leak-tight fixed volume cell, is illuminated with intensity modulated light. If some species in the sample absorbs the light and is promoted to an excited state, relaxation takes place, in part or totally, via radiationless transitions. These radiationless transitions generate heat which diffuses both into the sample and to the sample/gas interface. Heat transferred to the gas at this interface creates a pressure increase in a gas boundary layer at the sample surface which in turn compresses the remaining gas in the fixed volume cell. The amplitude of the pressure wave thus created depends on the incident light intensity, the modulation frequency, the relative thermal properties of the sample and gas, cell geometry, optical ab1978 American Chemical Society
