Evaluation of a Self-contained Linear Diode Array Detector for Rapid

petitive scan storage, can be used for improving steady state measurements. However, the OMA's 5BCD memory range limited the capacity 'for signal aver...
0 downloads 0 Views 626KB Size
petitive scan storage, can be used for improving steady state measurements. However, the OMA's 5BCD memory range limited the capacity 'for signal averaging whenever weak signals were to be measured in a region containing a strong signal. In quantitative terms, a bright signal near detector saturation limited the number of repetitive frame scan accumulations to about 100. Since the improvement in signal-to-noise ratio corresponds to the square root of the number of successive additions, this memory range would limit a low-level signal improvement to a factor of 10. The interfaced system has a memory range which is capable of accumulating lo6 frame scans with high-level signals for a signal-to-noise improvement of 1000. The length of time involved, nine hours for lo6 frame scans, limits the amount of signal averaging that is experimentally feasible. However, intermediate accumulation periods expanding the original system's 3.3-second full scale accumulation limit allow the analyst a wide selection in optimizing the signalto-noise ratio for different applications. The preservation of the complete data spectrum enables sophisticated data manipulation and reduction by the P D P 11/20 system. While off-line systems have the potential to be as sophisticated as on-line in their data treatment, their scope is severely limited by the amount of information the operator can manually transfer from the OMA. Multiple array manipulations, such as the automated spectral stripping routine, must have the complete spectrum in order to perform accurately. Other complex calculation routines can be incorporated to reduce operator input and provide for computer-assisted experimental studies such as detection limits and flame profiles. By employing a command string interpreter structure, the programming has been designed to streamline operator input and facilitate software additions. The flexibility of the vidicon system is maintained by providing complete operator control with keyboard commands which can be learned with a minimal amount of practice. Standard options include the features which make the vidicon instrument attractive to analytical chemists, i.e., multielement analysis, spectral stripping, internal standardization, automatic curve-fits, and others. New functions can be easily added since the interpreter routine was structured for ex-

pansion. In effect, the system was designed primarily for dynamic operation rather than static inflexibility. Several detection system design improvements s h o d d make the vidicon an even more viable analytical tool. Computer controlled tube readout would significantly improve the real-time dynamic range and obviate the use of spectral filters. New spectral data formats as found in echelle spectrometers (12, 13), or split-grating instruments, may be more compatible with the two-dimensional vidicon target structure. New tube developments providing better ultraviolet region sensitivity, larger target areas, and better resolution should make computerized image tube systems the preferred multichannel spectroscopic detectors.

ACKNOWLEDGMENT The authors thank J. H. Boutwell and D. D. Bayse of the Center for Disease control, Atlanta, Ga., for supplying the samples of analyzed bovine serum.

LITERATURE CITED (1) K. W. Busch and G. H. Morrison, Anal. Chem., 45, 712A (1973). (2) D. G. Mitchell, K. W. Jackson, and K. M. Aldous, Anal. Chem., 45, 1215A (1973). (3) M. J. Milano, H. L. Pardue, T. E. Cook, R. E. Santini, D. W. Margerum, and J. M. T. Raycheba, Anal. Chem., 46,374 (1974). (4) Y. Talmi. Anal. Chem., 47, 658A (1975). (5) K. W. Busch, N. G. Howell, and G. H. Morrison, Anal. Chem., 46, 575 (1974). (6) Ref. 5. D 1231 (7) Ref. 5, p 2074. (8)N. G. Howell, J. D. Ganjei, and G. H. Morrison, Anal. Chem., 46, 319 119761 - -, \

(9) N. S. Gill and R. S. Nyholm, J Chem. SOC.,3997 (1959). (10) J. A. Dean and T. C. Rains, "Standard Solutions for Flame Spectrometry." in "Flame Emission and Atomic Absorption Spectrometry," Vol. 2, J. A. Dean and T. C. Rains, Ed., Marcel Dekker, New York, N.Y. 1971, p 327 (11) J R. Sarbeck, P. A St. John, and J. D. Winefordner, Mikrochim. Acta (Wien), 5 5 (1972). (12) A. Danielsson and P. Lindblom, Phys. Scr., 5, 227 (1972). (13) D. L. Wood, A. B. Dargis, and D. L. Nash, Appl. Specfrosc., 4, 310 (1975).

RECEIVEDfor review October 22, 1975. Accepted December 12, 1975. This work was supported by the National Institutes of Health, Grant No. 5 R01 GM 19905-03.

Evaluation of a Self-contained Linear Diode Array Detector for Rapid Scanning Spectrophotometry Dennis A. Yates and Theodore Kuwana" Department of Chemistry, Ohio State University, Columbus, Ohio 432 10

The commercial availability of a linear array solid state diode detector providing large number of elements (1024), spectral sensitivity from near UV to near IR, associated electronics for a self-contained unit, and fast interrogation times (up to 10 MHz), made it attractive for evaluation as a detector for spectral acquisition in a rapid scanning spectrometer (RSS). The unit replaced the exit slit and photomultipliers of our existing RSS. The operational characteristics of the 1024-element linear diode array detector obtained from Reticon Corporation was evaluated and will be discussed. 510

ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976

The basic types of designs and performance characteristics of Rapid Scanning Spectrophotometers (RSS) have been reviewed by Santini, Milano, and Pardue ( 1 ) . Of these, two particular scanning methods have been quite recently discussed and are pertinent to this paper. In one (2-4), the incident angle on a grating surface was changed by an oscillating galvanometer mirror system and the dispersed spectrum was then swept across a stationary exit slit. The detector of the light intensity was considered as a single element, and it was commonly a photomultiplier tube (PMT). In the other, the dispersed spectrum was kept

S

SOURCE

-4

-5vF

TIME

Figure 1. General representation of the relation between ( A t )and

cycle time (A r,)

stationary and a detector consisting of multiple elements was responsive to light impinging on its surface. The response from each element corresponded to a wavelength width dependent on the linear dispersion of the spectrum and the physical size of the element. The detector may have elements arranged in multiple, two-dimensional arrays (Silicon Vidicon tube) or a single row of elements (linear diode array (LDA)). In these array detectors, the elements were interrogated in some sequential mode and the readout was indicative of the spectral intensity distribution. Rapid scanning spectrophotometers based on the galvanometer scanning and Silicon Vidicon detector were recently discussed and performance advantages and limitations compared ( 4 ) . The LDA has seen limited usage. A 256-element LDA was used for determining emission spectra by Horlick ( 5 ) .Dessey (6) has computerized a 128-element LDA for RSS applications. However, to date, a complete, ready-to-operate commercially available LDA unit has not been evaluated in terms of the performance characteristics and spectral properties for rapid scanning of visible absorption spectra. Thus, in this paper, the Reticon 1024-element LDA unit (Reticon Corporation, 450 E. Middlefield Road, Mountain View, Calif. 94043) which included complete operational circuits and required only the addition of two external power supplies for operation will be evaluated using the optical layout of our presently available galvanometer system RSS.

EXPERIMENTAL The Reticon LDA unit (received May 1974) contained 1024 photodiode elements, the signal processing, and the timing logic circuits. The unit was made up of three printed circuit (PC) boards. One PC board, the driver amplifier board, contained the LDA chip and the current-to-voltage converter. The second, the RC400 P C board, contained the clock circuits. The function of the third board, the CASH-lB, was to integrate the ac voltage pulses from the LDA and output a dc voltage. A detailed description of the operation of these components is provided in the manufacturer's literature. The clock frequency, generated by the RC400, determines the frequency of photodiode sampling (one clock pulse per photodiode). Clock rates of up to 10 MHz were possible. The RC400 used in this study had a preset internal clock rate of 100 KHz. The scan time was therefore 1.024 X s. The cycle time was determined by a 12-bit presettable binary down counter mounted on the RC400 P C board. Binary switches on the RC400 P C board determined the initial counter setting. The minimum cycle time was 1028 (4 clock pulses were required to start the scan) clock pulses and the maximum cycle time was 4096 clock pulses. Therefore,

Figure 2. Optical layout of t h e spectrometer

Components: Source: Tungsten lamp or xenon lamp. S: Slit (Harrick Scientific Corp.). MI: Concave mirror, focal length 12.7 cm. GA: Galvanometer (Bell and Howell Corp.). MP: Concave mirror, focal length 15 cm. G: Diffraction grating (Bausch and Lomb, 300 lines/mm blazed 400 nm). M3: Plane mirror. LDA: Linear diode array detector. A k Wavelength width sampled by LDA

repetition rates between 97 scans/sec and 24 scans/sec were possible using the 100-KHz internal clock. Figure 1 represents this sequence. The scan time (AT1) is fixed, dependent on the clock rate. T h e cycle time (ATt) is variable through the use of the 12-bit presettable counter. The final output was a dc signal (as opposed to an ac pulse train) which can be displayed on an oscilloscope or fed to an analog-todigital converter for computer processing. The trigger and clock outputs were available to synchronize a computer or oscilloscope. The optical configuration was similar to a rapid scanning spectrometer (RSS) described elsewhere (2, 3). Several optical modifications have occurred and the present configuration is shown in Figure 2. In this arrangement the entrance slit image, SI, was focused by MI a t a point between galvanometer, GA, and the focal point of Mz. MZfocused the image of GA on the grating, G, and also focused the S1 image on the LDA. G dispersed the light and the LDA samples a portion of the dispersed light. A cylindrical lens may be used to focus the vertical axis of the dispersed light on the LDA, but was not required in this study since the light intensity was sufficient. Note that the LDA was placed a t the exit slit plane and served as 1024 individual slits. This configuration was chosen since it had several advantages which were useful to retain in a LDA optical system. The galvanometer allowed simple variation of the sampled wavelength region and gave adequate resolution for most work (theoretical = 1.2 nm using 300 lines/mm grating a t 630 nm). The galvanometer also allowed a convenient setting of the spectral region which was to be sampled. This configuration gave an adequate wavelength sampling region which can be adjusted to user requirements. A final reason for the choice was the availability of this optical configuration in our laboratory. The diode array (RL1024C-17), clock circuitry (RC400), and signal processing circuitry (CASH-1B) were purchased from Reticon Corporation (Reticon Corporation, 450 E. Middlefield Rd., Mountain View, Calif. 94043). Mounting the CASH-1B circuit board was a problem since it contains 13 trim pots which must be accessible for adjustment to optimize the LDA performance. In our spectrometer, the dispersing optics were mounted on a flat plate about 6 inches above the spectrometer base. The LDA unit was mounted in one half of a Bud box on a three-axis translation stage. The unit was set to the proper optical level by a stem which fits into an optical mount fastened to a triangular optical rail. The unit can also be conveniently removed and replaced if this should be required. While it is possible t o detach the circuit board containing the array from the other two boards (electrical connection made with a 14-wire, flat ribbon cable), undesirable noise was introduced by this change. Optimization of the unit was performed about once a week to maintain the desired performance level. External power supplies (&I5 and +5 V) were purchased from Elasco Power Supply Corporation. The supplies were mounted ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976

511

C

n

B

-D n 450

530 WAVELENGTH

Figure 3.

Typical LDA spectra

(A) Tungsten source base line, (B) Holmium oxide filter, (C) Zero line (light source off). All spectra were signal average 100 times. Cycle time was 4.096 X s

separately from the LDA unit and external to the spectrometer. Cables were used to transfer the power. Co(N03)~ solutions were used to determine the response linearity. These solutions were prepared using Baker Analyzed Reagent Chemicals in 1% H2S04. The absorbances of these solutions were determined using a Cary 15 spectrometer (absorbance measured a t 520 nm). The response linearity spectra were acquired via computer and signal averaged 100 times. The light source for the spectrometer was a 100-W tungsten iodine quartz lamp or a 75-W xenon arc lamp powered by a highly regulated power supply. Alignment of optical components and positioning of the LDA were performed using a TRW Argon Ion Laser No. 71B or a Spectra Physics Model 125 HeNe laser as a light source. Resolution and dispersion were also determined using these lasers. Dispersion was determined by measuring the number of photodiodes between two laser peaks. Resolution was measured by determining the wavelength width at one half-peak height of a single laser line. Dispersion and resolution using a laser were recorded via oscilloscope with Polaroid camera attachment. Computer acquired spectra were taken by a Data General Nova 800 computer with 32K core memory and a Diablo 1.2-million word, cartridge, moving-head disk. A single channel 12-bit Datal ADCN-12B Analog-Digital (ADC) converter was used. This converter has an analog range of &5 V giving a least significant bit resolution of 2.5 mV. All transfers from the ADC were internally hardware-timed and employed the computer’s direct memory access channel. A 200-kHz data conversion rate is possible with this converter. Two 12-bit Datal DAC-VR-12B Digital-to-Analog converters (DAC) were also available to plot data on an oscilloscope or x-y plotter. The trigger pulse from the LDA was used to initiate the computer hardware, and the LDA clock was used to synchronize the ADC conversion. 1024 data points were taken a t 10-ps intervals.

RESULT AND DISCUSSION T o illustrate performance, typical computer acquired transmittance spectra (using a Tungsten source) of a base line, holmium oxide filter spectrum, and background with the light source off are shown in Figure 3. These spectra are a result of 100 averaged scans. The shape of these transmittance spectra is the result of the convolution of the light source wavelength distribution, diffraction grating wavelength efficiency, and LDA photodiode wavelength response. A high pressure xenon arc lamp has also been used with the LDA unit. In general, this lamp provides more light throughout, since it is a better point source, particularly a t shorter wavelengths as compared to the tungsten lamp. Narrower slit widths or shorter integration times can then be used with the xenon arc lamp. The main problem with the xenon arc lamp is the atomic emission lines which appear in the visible region centered about 480 nm. The xenon arc emission lines were a problem because they appear as sharp peaks superimposed on the xenon source continum. The emission lines had two important effects. First, 512

ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976

Figure 5. Oscilloscope trace of HeNe laser peak used in calculating resolution. Each diode covers 0.19 nm (A, = 630 nm)

the LDA detector signal-to-noise ratio was limited by the emission lines (i.e., the slit widths had to be decreased so that the LDA detector was not saturated in the emission line region). Second, in the transmission mode, sample absorbtion peaks whose wavelength distribution overlap the emission line wavelengths had their spectral bandshape distorted (i.e., the sample absorbtion peak had the emission lines superimposed on the absorbtion peak which may cause difficulty in visually interpreting the sample spectrum). The seriousness of this effect was dependent on the sample spectral bandshape, molar absorbtivity, sample concentration, and degree of wavelength overlap between the emission lines and the sample peak. This effect could be eliminated by computer subtraction of a base-line spectrum from a sample transmission spectrum. Figure 4 shows an absorbance spectrum of the same filter taken with a RSS unit using photomultiplier tubes. To determine the maximum resolution, a HeNe laser was used as the light source. Figure 5 shows the spectrum of this laser. The wavelength axis was expanded using the oscilloscope controls to display approximately 11 nm. The resolution (peak width a t one-half height) was measured to be 1.2 nm. This resolution was determined by measuring the number of diodes at one-half peak height (approximately 7 photodiodes) and multiplying the number of photodiodes by 0.19 nm/diode (one photodiode corresponds to a width of 0.19 nm). This resolution was comparable to our RSS using photomultiplier tube detectors.

In this spectrometer, the wavelength width (AX) sampled by the LDA detector was a direct function of the optical gain. The optical gain was determined by a ratio of the distance between the entrance slit image (as formed by MI in Figure 2) and M2 to the distance between Mz and the exit slit image. In this spectrometer, the optical gain was determined t o be 3 and resulted in the entrance slit being magnified by a factor of 3 a t the exit slit, Sz (Figure 2). The width and height of the dispersed beam was also magnified (although resolution remained constant). The AX can be adjusted readily anywhere between 190 to 250 nm. With an optical gain of 3 and a A i of 190 nm, a maximum signal (tungsten source base line) was obtained a t an integration time of 40.96 ms. No additional optical focusing element (i.e., a cylindrical lens) was used to focus the vertical exit slit plane on the photodiode because image distortion would have occurred. The width could have been increased by using a xenon source and increasing the optical gain by adjusting the image distances. Additional light throughput could also be obtained by replacing the galvanometer with a larger flat mirror (the galvanometer was the limiting aperature stop in this spectrometer). A nonlinearity in wavelength also existed in the optical system due to the curved exit slit focal plane. T o calculate the magnitude of this error, simple trigonometric functions were used. The error for a 190 nm AA was 0.03% a t each end of the LDA detector (center of detector was assumed to have zero error). This error was considered insignificant with the present resolution and AX. We will discuss several important parameters which determine the usefulness of a detector. Among these are sigspectral response, response linearnal-to-noise ratio (S/N), ity, and convenience of use. The maximum S/Nwas the ratio between the maximum signal output and the noise a t the LDA unit output. The maximum signal was equivalent to 5.00 V (saturation voltage). The minimum signal was 0.00 & 0.02 V. This 5-V change corresponded to a photodiode element capacitor charge depletion of 4 pC. The noise in the LDA unit arose from four principal sources. First, the fixed pattern noise from the clock circuits was capacitively coupled onto the video lines. The CASH-1B sample and hold circuits partly eliminated this noise since the clock noise integrated to zero. The LDA unit output showed clock frequency noise caused by the switching between the two S H amplifiers on the CASH-1B. As can be seen in Figure 6, which shows a trace of several photodiode elements, spikes appear on the output due to this switching. The second source of noise was dark current. This dark current was charge which leaked from the photodiode charge storage capacitor and was not due to light striking the photodiode element. The dark current was specified by Reticon to be 0.01 pC or 0.25% of the maximum output signal a t 22 "C with an integration time of 10.28 ms. Experimental verification of this figure was done by varying the integration time with no light striking the LDA detector. The third source of noise was the fundamental thermodynamic noise. This was specified by Reticon to be 0.1% of the maximum output. The fourth source of noise was the electronic noise induced on the output by the additional LDA unit processing components (switching transients, drift, etc.) and appeared to be the major noise source in the LDA unit. We experimentally measured the noise value by the computer acquisition of 4096 repetitive spectra. The computer saved the digital value for a known photodiode element during each scan. The 4096 saved points were then plotted. In this manner, the peak to peak and the RMS noise dependence on light intensity, wavelength, and time was ex-

Figure 6. Oscilloscope trace of a nonlinear response area in the LDA detector. Vert = 0.5 V/div

amined as well as obtaining the actual noise figure. The noise did not appear to be wavelength or light intensity dependent. The RMS noise value was measured to be 50 mV. The experimental maximum signal-to-noise value was then calculated to be 1OO:l. This indicated a sensitivity of 1% transmittance. The response linearity was evaluated using known standard solutions of cobalt nitrate. The absorbance was found to be linear from 0.00 to 1.5 & 1%absorbance units. The spectral response of the LDA detector was similar to that of a typical photodiode detector with a maximum response a t ca. 990 nm. I t decreased to about 10%of the maximum response a t 1100 and 250 nm. As the data indicate, the LDA unit has rather limited applications due to its lack of sensitivity. Horlick's data ( 5 ) suggested that the LDA detector was a suitable RSS detector but his signal processing circuitry was different. In comparison to his results, we conclude that the principal noise source must be in the signal processing electronics. We have not yet attempted to modify the circuits to see if the noise could be reduced. Since the Reticon LDA unit requires no additional circuifry for operation (except a h15 and +5 V power supply), it was operational as received. The LDA gave a nonuniform output which is seen in Figure 6. The nonlinear portion of the LDA encompasses a width of about 14 photodiodes and appeared to independent of wavelength and light intensity. This nonlinearity was believed to be due to imperfections (Le., impurities) in the LDA active surface. An important point to consider is what possible maximum scan rate could be achieved with this device. If the upper specified clock rate of 10 MHz is used, the maximum would be 9.7 kHz. The scan rate is higher than any other rapid scanning spectrometers for comparable wavelength scan width and repetition rate. The present limitation of using such high clock rates is to obtain sufficient light throughput and to obtain a means of acquiring data. Computer data acquisition a t 10 MHz (if all 1024 diodes are sampled) is difficult. Special techniques such as transient recorders with large internal memories would be required. The problem of data acquisition and storage is not trivial if analysis of all spectra taken a t high scan rates is desired. There are several major drawbacks associated with the LDA detector which we feel should be mentioned. The manner of the LDA detector operation requires that the entire LDA detector must be sampled. Since in many types of work the entire samples spectrum may not be of interest, time is wasted sampling unwanted data. An addressable readout mode would eliminate this problem, but this type of device is not available, although the technology certainly exists to product it. The device would generally require a computer to take maximum advantage of the device. ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976

513

I t should be mentioned that Reticon indicates that the LDA unit is intended for prototype applications. As such, this unit was probably designed and intended for a broad variety of applications and performance may be improved by designing circuitry for specific applications. An example of an application has been the analysis of SO2 in the wavelength range of 280-315 nm. Sensitivity of 100-2000 ppm of SO:! in a cell of 50 cm length was reported (7). We predict widespread usage of the solid state array detectors because of the attractiveness of being able to electronically (vs. mechanically) control all aspects of the spectral acquisition.

LITERATURE CITED (1) R . E. Santini. M. J. Milano, and H. L. Pardue, Anal. Chem., 45, 915A (1973). (2) J. W. Strojek, G. A. Gruver, and T. Kuwana, Anal. Chem., 41, 487 (1969). (3) F. Hawkridge and T. Kuwana, Anal. Chem., 45, 1021 (1973). (4) R. M. Wightman, R. L. Scott, C. N. Reilley. and Royce W. Murray, Anal. Chem., 46, 1492 (1974). (5) Gary Horlich and Edward G. Codding, Anal. Chem., 45, 1490 (1973). (6) Raymond E. Dessey, "Computers in Chemical and Biochemical Research," Vol. 2, Academic Press, New York. 1974. (7) J. W. Strojek and T. Kuwana, Rocz. Chem., 49, 379 (1975).

RECEIVEDfor review February 3, 1975. Accepted December 4, 1975. We ghtefully acknowledge financial support provided by National Science Foundation Grant No. MPs 7304882 and by USPHS Research Grant GM 19181.

Plasma Emission Detection of Chlorinated Pesticides in Inert Matrices D. J. Runser and C. W. Frank* Department of Chemistry, University of Iowa, Iowa City, Iowa 52242

A self-initiating and continuous-flow electrodeless discharge was used to produce spectra of organic compounds. The discharge was driven by a 300-W, 13.56-MHz radiofrequency generator. Chlorinated pesticides were monitored at the 2780-A CCI radical emission band. Absolute detection limits for DDT in alumina and soil were 5.9 and 8.2 pg, respectively. Interferences from inorganic chlorides (NaCI) and urea at concentrations up to 10 000 ppm were negligible.

The first application of electric plasmas to spectrochemical analyses of pesticides was made by Bache and Lisk ( 1 ) in 1965. Using the method suggested by McCormack et al. ( 2 ) ,an argon microwave plasma was used as a detector for gas chromatographic analysis of organophosphorous pesticides. This work was extended to organic bromine, chlorine, iodine, phosphorous, and sulfur pesticide residues using a helium microwave plasma ( 3 ) . Organic mercury compounds have also been studied ( 4 ) ,where sensitivities in the nanogram range have been reported. However, as with other existing analytical procedures for pesticides, preliminary sample extraction, isolation, and pre-concentration were required prior to detection. Studies in our laboratory have been carried out with a radio-frequency electrodeless discharge cell used for studying the emission spectra of organic compounds ( 5 ) . Using this emission source, an investigation was made to study the feasibility of a one-step method of detection of chlorinated pesticides in inert matrices by monitoring the 2780-8, CC1 radical emission band. The work reported here describes the results of this initial study with DDT in soil and alumina samples.

EXPERIMENTAL An argon radio-frequency electrodeless discharge was used in this study (Figure 1). The discharge cell was a self-initiating and continuous flow cell which produced an intense, stable, and reproducible plasma flame and striated column for spectral excitation. Details of cell construction and performance are described elsewhere (6). 514

ANALYTICAL CHEMISTRY, VOL. 48, N O . 3, MARCH 1976

The discharge was sustained using a Tracerlab RFG-300 radiofrequency generator with an operating frequency of 13.56 MHz. A Tracerlab PA-301 Plasma Activator served as the radio-frequency tuner. Emission spectra were studied using a three-meter grating spectrograph with a modified Eagle mounting and a reciprocal linear dispersion of 5.59 A/mm. Kodak Spectrum Analysis No. 1 Plates were used to record the data. Union Carbide, Linde, high purity dry argon, purity 99.996% was used as the carrier gas. Technical grade DDT recrystallized from isopropyl alcohol with a melting point of 108.5-109.5 "C was used. (Literature m p 108.5-109 "C) ( 7 ) . DDT-alumina and DDT-soil samples were prepared using Fisher Scientific Levigated Alumina and 40-mesh brown soil, respectively. Five percent samples were made by weighing both components to five-place accuracy, dissolving in thiophene free benzene, mixing for 5-7 h, and evaporating the solvent under aspirator vacuum using a Rinco rotating evaporator. The amount of DDT present was confirmed by extraction from the inert matrix with benzene, using a micro Soxhlet extractor, and analyzed by quantitative gas chromatography with a Bendix Chroma-lab, Series 2100, Gas Chromatograph using a 6-ft X %-in. stainless steel column packed with 5% OV-1 on 60/80 mesh chromosorb W and a thermoconductivity detector. Samples were weighed out in a triangular tantalum boat which slid over the tantalum grounding strip already in position in a radio-frequency cell (Figure 2). The sample boat was held in place by tabs A and B. The two sections of the cell (A and B, Figure 1) were joined together and evacuated. Once the argon carrier gas flow was established, the plasma flame and striated discharge were initiated. When the striated discharge arced toward the tantalum ground, it heated and vaporized the sample. Vaporization was complete within 1-2 s. Sample sizes varied from 10-0.1 mg.

RESULTS AND DISCUSSION Using this radio-frequency cell and the sample procedure described, intact samples of DDT in alumina and in soil were studied for the presence of the intense 2780-8, CC1 radical emission. The occurrence of CC1 emission from CC14 in gas discharges has been well documented (8, 9) and the presence of CC1 emission from simple aromatic and nonaromatic chlorinated hydrocarbons including DDT, Aldrin, Endrin, and Heptachlor in the radio-frequency discharge has been observed in our laboratories ( 5 ) . By monitoring the 2777.68, 2778.76, 2786.67 and 2788.39-A transitions of the 2780-A band of the CC1 radical emission, the