weight, and apparatus dead volume for a solute on a given liquid phase. It is assumed that the isothermal temperature dependence of the retention index is negligible over a range of retention temperatures which might be obtained for different programs. This assumption is valid for the vast majority of solutes, as shown by the data of Kovats (8) and Wehrli and Kovats (149, who introduced the use of the retention index for isothermal gas chromatography. Retention indices were obtained for some esters, alcohols, aldehydes, and ketones on each of the two columns used under the conditions of this study. These indices were then compared to those obtained for the same compounds on Carbowax 20M by Van Den Do01 and Kratz (14). These authors used a 20% liquid phase loading, a program rate of 4.6’ C./minute in the range 75’-228’ C., and a carrier gas flow rate of 80 ml./minute. When compared with those values previously reported in the literature, agreement was to within + l % with rare exception, and the values obtained on the columns used in this study agree t o within =t0.5%. Programmed temperature retention indices reported in this work are therefore suitably reproducible retention parameters for use with Carbowax 20M. Retention indices below 600 are somewhat difficult to measure with accuracy, owing to the short retention time for the lower hydrocarbon standards on a Carbowax 20M column. The validity of assuming constant AI values for homologous products was
confirmed by plotting the foregoing calculated retention indices us. chain length for each of the series. The plots are remarkably linear and parallel to each other and to the plot for normal paraffins, which is linear by definition. Some deviation from linearity was noted a t low carbon numbers for only a few of these functional groups. Such deviations for product homologous series would serve to increase the scatter of the resulting AI values for the lower members of the irradiated series.
CONCLUSIONS
The method proposed here is suitable for the qualitative identification of small amounts of pure organic compounds. Since homologous and/or common products are formed within each irradiated series, products to be expected from previously unstudied members of the series may be predicted. Solids may be difficult to study by this technique because of the relatively small absolute energy of a steady-state mercury discharge. This might be overcome through the use of flash photolysis, which would produce high instantaneous radiant energy and reduce the analysis time. Liquid phase photolysis requires samples of the order of 1 mg., whereas pyrolysis-GLC is applicable to samples in the microgram range. Photolytic degradation-GLC is a technique which holds promise as a complementary method for the identification and study of organic compounds.
ACKNOWLEDGMENT
The authors express their appreciation to James M. Miller for many helpful discussions during the course of this work. Thk zuthors also thank MicroTek Instruments, Inc., of Baton Rouge, La., for the loan of the 2500R and DSS 172 D P F F gas chromatographs used in this study. LITERATURE CITED
(1) Ausloos, P., Can. J . Chem. 36, 383
(1958). (2) Ettre, K., Varadi, P. F., ANAL.CHEM. 35, 69 (1963). (3) Janak, J., Nature 185, 684 (1960). (4) Keulemans, A. I. M., Perry, S. G., “Gas Chromatography, 1962” h1. van Swaay, Ed., p. 356, Butterworths, London, 1962. (5) Knight, A. R., Gunning, H. E., Can. J . Chem. 39, 2466 (1961). (6) Knight, A. R., Gunning, H. E., Ibid., 40, 1134 (1962). (7) Knight, A. R., Gunning, H. E., Ibid., 41, 2849 (1963). (8) Kovats, E., Helv. Chim. Acta 41, 1915 (1958). (9j Kuntz, R. R., Mains, G. J., J . Am. Chem. Soc. 85, 2219 (1963). (10) Kuntz, R. R., Mains, G. J., J . Chem. Phys. 68, 408 (1964). (11) O’Seill, H. J., Putscher, R. E., Dynako, A., Boquist, C., J . Gas Chromatoo. 1. (2). 28 (19631. (12) Perry,“% G:, Ibid., ‘2, 5 4 (1964). (13) Phibbs, M. K., Darwent, B. deB., J . Chem. Phys. 18, 679 (1950). (14) Van Den Dool, H., Kratz, P. D., J . Chromatog. 11, 463 (1963). (15) Wehrli, A,, Kovats, E., Helv. Chim. Acta 42, 2709 (1959). RECEIVEDfor review July 22, 1965. Accepted September 7 , 1966. Presented, in part, Division of Analytical Chemistry, Lab-Line Award Symposium, 149th hleeting, ACS, Detroit, Mich., April 1965.
Sensitive Selective Gas Chromatography Detector Based on Emission Spectrometry of Organic Compounds ARTHUR J. McCORMACK, S. C. TONG, and W. D. COOKE Department o f Chemistry, Cornell University, Ithaca,
b A gas chromatography detector has been developed which is based on monitoring the intensity of the electronic emission spectra of the eluted organic compounds in an argon carrier gas. The spectra are excited in the plasma of a 2450-Mc. electrodeless discharge and detected photoelectrically in the ultraviolet-visible region. The detector sensitivity is 2 X gram of hexane per second. BY choosing the wavelength of various atomic lines and molecular bands, a degree of selectivity can b e obtained. 1470
ANALYTICAL CHEMISTRY
N. Y.
The system can b e sensitized to the halogens, phosphorus, and sulfur, as well as permanent gases. The detector has a high sensitivity and selectivity, a small volume, rapid response, and a wide dynamic range.
E
LECTRONIC spectra emitted from
electrical discharges through gases have been studied since the early beginnings of spectrometry. Such plasma sources have been used frequently to obtain information on spectrometric constants and much of the information
on the spectra of diatomic molecules has been obtained from this technique (6). However, very few analytical methods have been developed, and this paper suggests some possible applications. The red band spectrum of nitrogen excited by a spark coil is widely used as a leak detector in vacuum systems. Fay, Mohr, and Cook (2) studied the efficiencies of various types of electrical discharges in exciting the emission spectrum of traces of nitrogen in argon. The intensity of the emitted spectrum could thus be used to determine very low
6
5
4
3
.
2
1
0
5
4
3
2
1
0
5
4
3
2
1
0
MINUTES
Figure 1 . Chromatogram of n-heptane, propyl chloride, and hexafluorobenzene
levels of nitrogen contamination. Gatterer (3) used a radiofrequency discharge in an evacuated tube for the determination of halogens and sulfur in organic compounds. Sauter (8) monitored the OH-band intensity a t 3064 A. to determine traces of water in organic compounds. Since only traces of impurities are sufficient to emit a detectable specific spectrum, the application of the technique to gas chromatographic detection seems to merit investigation. Some work was carried out by burning the organic effluent from a gas chromatographic column into carbon dioxide, which was ultimately introduced into a low pressure 8-;\IC.electrodeless discharge. The band spectrum of CO \vas monitored photoelectrically and the device served as a sensitive detector for organic comlTounds (7'). Johnstone and Douglas ( 4 ) used a band system a t 2536 -\. to detect the elution of polynuclear hydrocarbons and Sternberg and k'oulson (9) monitored the change in total spectral emissivity to detect organic vapors eluted from a column. Initially, the plasma of a hydrogenosygen flame a t the column exit was used as the esciting source to obtain the emission spectrum of organic compounds ( 5 ) . -Although the technique rem l t r d in a sensitive, selective detector, it was abandoned when it was found that substantially higher sensitivities could bf obtained by using an electrical discharge for esciting the spect,ra of organic vapors. A variety of electrical discharges were evaluated for the excitation of spectra of the eluted compounds. High voltage a x . and d.c. glow discharges, a hollow cathode d.c. discharge, and 8and 2450-Me. electrodeless discharges TYCIT all used for the generation of the p l a m a . The intensity of the spectra emitted from the 2450-Mc. source was significantly higher than that of the other Sources. I3y monitoring the intensity of appropriste bands or lines in the spectra emitted from thc microwave-generated plasma, it was liossihle to develop n
detector of high sensitivity with some selectivity. With all the elements examined, it was possible to find spectra which were characteristic of the element. For example, emission at 3852 A. (the Cz system) arises from carbon compounds, a 2536-A. atomic line results from phosphorus-containing molecules, and the 2788-A. or CCl region is sensitive to chlorinated materials. By setting the wavelength of the monochrometer a t one of these three positions it was possible to detect selectively molecules containing each of these elements. Figure 1 gives an example of this procedure. A three-component mixture containing n-heptane, hexafluorobenzene, and propyl chloride was injected into the chromatograph. By setting the monochrometer on the C2 band at 3852 A4.all three components were detected, since they all contained carbon (Figure 1, left). When the 2516-A. line was used, the detector gave a selective response for fluorinated compounds as shown in the second chromatogram, and the single chromatographic peak for hexafluorobenzene \\as detected. Using the CCl band at 2788 A. only the chlorinated hydrocarbon was detected. T o ascertain the optimum wavelength for each element, it u a s necessary to record the microwave-excited emission spectra for compounds containing the elements for which detection was desired. This was done by passing argon continuously over a compound containing the elements n hich have sufficient vapor pressure to give a detectable spectrum. The argon vapor was then passed through the quartz capillary in the microwave cavity. The spectra were recorded with a scanning monochroineter as well as with a 3meter grating instrument with photographic recording. I n exploratory work, photographic recording was chosen because the entire spectrum could be obtained in a short time. Once the wavelength region of interest I$ as accertained, that region could be scanned photoelectrically and the optimum wavelength for detection obtained.
The mechanism of excitation and fragmentation involves electron impact, Free electrons in the discharge regions (initially introduced with a spark coil to initiate the discharge) acquire sufficient kinetic energy from the electrical field to cause fragmentation and excitation. All the spectra recorded which are assignable arise from either diatomic molecules or free atoms, indicating rather complete fragmentation. For example, the peaks recorded for nhexane a.nd benzene were essentially the same. Recombination of atoms is also extensive, as indi-ated by the presence of bands originating from cyanogen for all organic compounds in presence of nitrogen as well as the Csbands in the spectrum of CC14. SELECTIVITY OF DETECTOR
The proposed technique measures the intensity of a particular molecular or atomic line a t a fixed wavelength and thus records the chromatographic peak for a species containing the desired element. If a specific element is being detected-for example, chlorine, using other the CC1 band a t 2788 .\.-any species eluted from the column which gives rise to radiant emission a t this wavelength will also be recorded as a chromatographic peak. This spectral interference may arise from poor resolution of the desired line from a nearby interfering line or more generally from an increase in background radiation in the vicinity of the desired peak. Because of t'he large number of lines resulting from CY, Cz,and CH emission which are distributed over a \ d e range of wavelengths, all organic compounds interfere to some degree with the detection of various elements. The extent of interference, or the specificity of the detector to a parbicular species, can be estimated by determining the selectivity ratio or the volume of hexane required to give a chromatographic peak of the same area as a unit quantity of the species of interest. I n the case of a n iodo compound such as methyl iodide, detected by the 2062-X. atomic line, l o 4 times as much hesane as methyl iodide would be required to produce an equivalent' chromatographic response. For iodine compounds this ratio is very favorable, while for chlorine compounds detected a t 2788 A. the ratio can be as small as 20, which results in a poor selectivity for chlorine. The magnitude of the selectivity ratio depends on many factors. The particular line chosen for detection of the desired element may have serious interference, while another may be relatively free from spectral overlap or background enhancement. The nature of the compound causing the interference also has a pronounced variability. For example, the selectivity ratios for n-hexane and benzene reVOL. 37, NO. 12, NOVEMBER 1965
* 1471
sulted in different values, with hexane generally resulting in poorer selectivity. Other factors affecting the selectivity ratio are the dispersion of the monochrometer and the slit width. The smaller the band pass of the monochrometer the greater the selectivity, but the lower the sensitivity, especially for atomic lines. For example, the selectivity ratio of the 2062-A. line of iodine compared to hexane was lo4 with 50-micron slits and lo7 with 6micron slits. The microwave power level and the pressure in the discharge tube will also greatly affect the selectivity. The experimental conditions will be dictated by the particular problem under consideration. Yo attempt was made in this ork t o find the optimum experimental condition which would result in maximum selectivity. It is hoped that the selectivity of the detector can be increased and work is proceeding along these lines. The major problem in selective chromatographic detection seems to be the increase in spectral background rather than band and line overlap. This problem, while leading to interference in chromatographic detection, would be much less important when the spectrum is scanned to indicate the presence of a particular functional group (Figures 4 t o 10). Under such circumstances a n increase in background would cause little difficulty. There are two possible solutions to the problem of selective chromatographic detection. One would involve a dualchannel spectral analyzer nhich could correct for the increase in background and the other would be a rapid scan monochrometer which could rapidly scan the spectrum of the material emerging from the column. EXPERIMENTAL
Apparatus. The gas chromatography unit in this work was a modified F and M Scientific Co. Model 609. The flame ionization detector was removed and replaced aTith a quartz capillary t h a t projected from the oven into the center of a 2450-Mc. microwave cavity. Quartz was used because of its transparency to ultraviolet radiation and its higher softening temperature. Other types of glass were found to melt under radiation by high microwave power. The stainless steel cover on the column oven was replaced by an asbestos cover t o prevent damage to the oscillator by reflection of microwave radiation back through the cavity to the magnetron. Two types of cavity were used: a tapered cavity with a slot to accept the quartz capillary, and a coaxial cavity with an opening through which the emitted spectrum could be observed. The tapered cavity had a higher sensitivity, whereas the coaxial cavity could accept a larger sample. The coaxial cavity was used when it mas 1472
ANALYTICAL CHEMISTRY
[K1
Figure 2. tector A. B. C. D.
E. F. G. H.
1. 1. K.
Schematic diagram of deCarrier gas inlet Injection pot Column Quartz discharge tube Microwave generator Microwave cavity Quartz lens Monochrometer Multiplier phototube Amplifier Recorder
necessary to obtain the complete spectrum of an organic molecule. The 2450-Mc. power oscillator was a 125watt Raytheon diathermy unit, Model CMD4. The emitted light from the discharge was focused by means of a quartz lens on the entrance slit of a 0.5-meter Jarrell-Ash Model 82000 Ebert-mount scanning monochrometer. The detector system consisted of a 1P28 phototube with associated high voltage power supply, amplifier, and recorder. The apparatus is schematically shown in Figure 2. The discharge had to be initiated by means of a spark coil which was removed once the self-sustaining plasma was formed. The diameter of the quartz tube containing the discharge was an important parameter in detector performance. With discharge pressures of a few millimeters of mercury the minimum diam-
eter was approximately 10 mm. Any smaller diameter tended t o result in an unstable discharge. At atmospheric pressures, smaller internal diameters yielded significantly more intense, stable spectra and a 1-mm. capillary was generally used. The carrier gas emerging from the capillary was passed through a bubbler containing dioctyl phthalate to prevent backdiffusion of air into the capillary. This resulted in a significant decrease in spectral background. I n some cases troublesome carbon deposits accumulated a t the inner surface of the qual*tz. The maximum amount of sample that could be accommodated without extinguishing the discharge was about 1 mg. These difficulties could be eliminated by igniting the discharge only after the solvent peak had been eluted ( I ) . These carbon deposits could be readily burned off by the injection of a few milliliters of air with the microwave power applied. As the sensitivity of the detector was increased, the amount of material which could be detected became so small that the formation of deposits was no longer a problem. Large amounts of sulfur and phosphorus caused the formation of white, insoluble, nonoxidizable deposits. Cnder normal analytical usage no difficulty was experienced with these deposits. I t was not necessary t o supply heat to the capillary to prevent condensation, because the absorption of microwave power by the quartz resulted in a rather high temperature. Carrier Gas. Both helium and argon were used as carrier gases and for sustaining the plasma discharge. I n much of the original work a low pressure discharge was used to excite the emission spectra of the organic molecules. Under such conditions helium was the gas of choice because of the fewer lines in its spectrum compared to argon. It was subsequently discovered that a stable discharge could be obtained a t atmospheric pressure using argon in conjunction with
N
U N
tI
Figure 3.
Emission spectrum of n-hexane
a small diameter capillary. Because of the complexity of the associated vacuum system, argon a t atmospheric pressure v a s used as the medium for spectral excitation.
I 0
TRIETHYL
PHOSPHATE
" C
m
W
n I
(D
*)
n-hexane
CN
RESULTS AND DISCUSS'ON
Carbon Compounds. T h e emission spectra of all organic compounds containcd many comples bands, particularly in t h e region between 3500 a n d 5600 A . T h e most prominent bands were t h e C K , C f ,and CH systems. -1typical low resolution emission spectrum of hesane is shown in Figure 3. Although no nitrogen was purposely added, t h e cyanogen band system yielded the most intense band system of all those encountered. The source of nitrogen to produce these bands was probably nitrogen impurities in the argon and leaks in the system. The most intense line was the band head a t 3883 A. shown in Figure 4 and this pcak was used when the highest detector sensitivity was desired. The detection limit (signal/noise of 2) for n-hexane was 2 x gram per second a t a n argon carrier gas flow rate of 20 mm. per minute and a slit width of 25 microns. This very high sensitivity esceeds most other methods of chromatographic detection. Other prominent spectral features include the C p swan bands at 3852, 5165, and 7852 A,, and the CH region at 4314 -1. If t'he detector is sensitized at 3883 A, all organic compounds emerging from the column are detected and the instrument functions as a highsensitivity , nonselective detector for organic material. The sensitivity of the detector varies with the line chosen as well as with the particular molecule. The relative response factors a t 3852 A. for a variety of organic molecules are listed in Table I. Iodine Compounds. T h e atomic iodine line a t 2062 A. was found t o be very intense and allowed t h e detection of small quantities of iodinated compounds. T h e detect,ion limit, using methyl iodide, was 7 x 10-l4 gram of iodine per second, using an argon flow of 20 ml. per minut,e a t 100OC. and a slit of 25 microns. Because of the lack of other spectral lines and background in this short wavelength region, the selectivity for this element was very high. The ratio of response for methyl iodide a t 2062 A. cornpared to n-hexane was lo4. This means that lo4 times as much hesane was necessary to give a chromatographic peak of equal area as a unit amount of methyl iodide. Sulfur Compounds. Sulfur-containing compounds give rise t o a characteristic emission band spectrum degraded toward t h e red. T h e spect r u m shown in Figure 5 was obtained for carbon disulfide, b u t other sulfur
CN
N
7
CN
A I
Wavelength
Wavelength
Figure 4. Emission spectrum in a cyanogen band region
Table 1.
Figure 5. Band spectrum of CS molecule
Wavelength
Figure 6. Atomic phosphorus lines
Relative Area Response Factors at 3852 A. per Unit Volume
Carbon tetrachloride Chloroform Carbon disulfide n-Hexane 2,3-Dimethylbutane Propyl ether Ethyl ether
n-Hexane 31 40 70 100
103
120
130
compounds such as benzenethiol, propanethiol, ethanethiol, and thioethers produced similar emission. T h e limit of detection with t h e 2576.A. line was 10-9 gram of CS,per second using an argon flow of 50 ml. per minute a t 90°C. and a slit of 25 microns. The assignment of the bands agrees closely with the C-S band system (6). Konsulfur-containing carbon compounds increase the background radiation throughout the 2576-A. region, resulting in a chromatographic selectivity factor of 100. Thus it waq not possible, using this region, to identify unequivocally chromatographic peaks arising from trace amounts of sulfur compounds in organic samples without scanning the spectrum of the eluted compound. Phosphorus Compounds. T h e emission spectra of phosphorus-containing compounds contain four lines originating from atomic phosphorus st 2534.0, 2535.7, 2443.3, and 2554.9 A . Of these, t h e 2535.7 1.line is t h e most intense (Figure 6). hlolecular bands arising from PO fragments appear as very intense peaks superimposed on the cyanogen bands a t 3240 and 3271 A. (Figure 7 ) . Further information on the phosphorus functional group assignment can be obtained from the emission of PS bands in the region 4600 to 4900 -A,,
=
100 Propyl chloride Propyl bromide n-Heptane n-Nonane n-Decane Benzyl trifluoride Dibromoethane
I
130 130 180 190 250 420 450
.. Wavelerqth
Figure 7. Band spectrum of PO molecule superimposed on C2-CN bands
as shown in Figure 8. The latter band system, arising from the PS diatomic molecule, is much less intense than the PO system or the atomic phosphorus lines. The sensitivity of the detector to phosphorus compounds is about 10-11 VOL. 37, NO. 12, NOVEMBER 1965
1473
FLUORINE
‘SF6
V OD
Figure 9. Band spectrum of C t l molecule
Wavelength
Figure 8. molecule
Band spectrum of
PS
gram per second using the atomic line a t 2535.7 A. with argon flow of 20 ml. per minute at 100OC. and a slit of 25 microns. This line has been used for the detection of phosphorus-containing pesticide residues at the nanogram level (1). Because of the specificity of the detector, no prior cleanup procedures are necessary. The selectivity ratio for the phosphorus atomic line a t 2537 compared to hexane was 100. Chlorine Compounds. Chlorinated organic compounds are characterized by a variety of emission peaks. The CCl diatomic molecule (10) giving rise to the band a t 2i88 A. shown in Figure 9 appeared to be the most useful for the detection of chlorine. The sensitivity for detection was high (2 X gram per second using argon flow of 20 ml. per minute a t 100OC. and a slit of 25 microns) but the selectivity ratio compared to hexane was only 20 a t a slit width of 10 microns.
Figure 1 1 . One region of emission spectrum of malathion
1474
ANALYTICAL CHEMISTRY
An emission band in the region of 5000 A. arising from chlorinated materials was more intense than the 2788-*\, band, b u t interference from the Cz swan bands limited its analytical applicability. Other bands and lines originating from chlorine-containing compounds were observed, but the assignment of these peaks to known transitions was unsuccessful. Tn o sharp peaks at 5970 and 7321 A. were observed as well as a structured band between 6625 and 6644 A. Fluorine Compounds. JF7hen fluorinated materials were passed through the microwave field, a series of sharp lines was obtained in t h e 2510-A. t o 2532-8. region, as shown in Figure 10. These lines are assignable to atomic silicon transitions. I t appears t h a t fluorinated materials react with t h e quartz of the capillary to yield a volatile silicon compound (possibly SIR) and excitation of this molecule results in the atomic silicon lines. Other types of molecules, particularly those containing no hydrogen, such as CC14, or small amounts of hydrogen also result in a silicon spectrum. The sensitivity of the 2516-&\. line for fluorine was 5 x gram per second using an argon flow of 20 ml. per minute a t 100’ C. and a slit of 25 microns, but the selectivity ratio compared to hexane Jvas only 20. KO unique lines or bands could be definitely assigned to molecular fragments containing fluorine. However, the presence of fluorine atoms in the plasma had a pronounced effect on the spectra of other molecular species. For example, the C?; bands characteristic of carbon compounds are greatly diminished when fluorine is present in the molecular species. The intense Nz (or NH) bands present in the spectrum of pure argon carrier gas are so decreased in intensity by the presence of fluorinated compounds that negative chromatographic peaks are obtained for fluorinated compounds when the intensity of the 3360-.\. nitrogen band is monitored. However, the intensity of the C B bands does not seem to be greatly affected by fluorine atoms. These observations indicate that the
Wavelength
Figure 10. Atomic spectrum of silicon resulting from fluorinated compounds
fluorine atoms are either inhibiting the excitation or combining with nitrogen atoms to form some stable species xhich removes the nitrogen from its usual molecular compounds. However, no peaks or bands could be found for nitrogen-fluorine combinations. I n addition to the depression of some band intensities by fluorine, it was also discovered that the intensity of some lines could be enhanced by a factor of 10
-._ = E .-9
80
I-
70
-
60
-
\
c
c
50-
6
40-
d
30-
2
eo -
IO
0
-
1
I
a 25
I
50
75
IM)
v\
, 125
~ l c r o w a v ePower (Wottsl
Figure 12. Effect of microwave power output on response
+
1 -Chloropropone, 2 7 8 8 , C-CI Carbon tetrachloride, 2 7 8 8 , C-CI 0 1,6-Dimethylnaphthalene, 3 8 8 3 , C N n-Heptane, 3 8 8 3 , C N A 1 -Chloropropane, 3 8 8 3 , C N Chloroform, 2 7 8 8 , C-CI 0 1,2-Dichlorohexafluoropentene, 1, 3 8 8 3 , C N Ethyl iodide, 2 0 6 2 ,
V
0
I
i
-1.5
I
-2
I
-2.5
Figure 13.
I I I I -3.0 -3.5 -4.0 -4.5 log ~ L o Sample f Injected
Dynamic range
I
-5.0
of detector
to 1000 by the presence of very small amounts of fluorine in the molecule. For example, a line a t 5611 A. of unassigned origin, which generally arises from all organic molecules, is enhanced by a factor of 100 in the presence of fluorine compounds. This line can be used for the detection of fluorine with a sensitivity of 3 x 10-l2 gram per second under the same argon flow rate. A pair of very sharp lines a t 7890 and 7990 A., a group of lines in the 4310- to 4450-A. region, and another group a t 5000 to 5060 A., although apparently not originating from fluorine, are also greatly enhanced by the presence of this element. It is unfortunate that none of the lines that exhibit this increased intensity with fluorine could be assigned to a known molecular species. If such assignments could be made, it would greatly facilitate an understanding of the observed phc'nomena. Miscellaneous Compounds. T h e OH band a t 3000 to 3100 A. might serve for the detection of oxygenated compounds. However, with the apparatus used, the OH band appeared in all spectra recorded, eved those from pure hydrocarbons. Apparently small amounts of oxygen were always available for combination with the fragments of organic compounds, giving rise to the OH molecule. I n order to use this band for oxygen detection, it would be necessary to exclude oxygen rigorously from the carrier gas and eliminate leakage of air into the system. A similar situation exists for nitrogencontaining species. Because of the presence of traces of nitrogen gas in the system, the CN band system usually gave rise to the most intense spectral emission in almost all the recorded spectra. It was therefore not possible, with the present apparatus, to detect nitrogen-containing species of organic compounds selectively. Presumably if nitrogen could be rigorously excluded, the detector could be sensitized to nitrogen compounds by using the CN band system a t 3883 A. I n fact, it
-5.5
0
Figure 14.
IO
20
Effect
seems that the method is admirably suited for the detection of very low levels of either elemental N2 or nitrogen compounds. I n the emission spectrum of pure argon, the Nz band system is by far the most intense peak. I n contrast to other sensitive detectors, small molecules such as C02 and HzO give responses. The 3883-A. CK line can be used for the detection of COz and the 3089-A. OH line can be used to monitor water. The sensitivity of C 0 2 is lo-' gram per second and that of H20 is 2 x 10-6 gram per second, using argon flow of 20 ml. per minute a t room temperature with a slit of 25 microns. A line a t 2985 A. could be used for the detection of bromine-containing compounds, but the selectivity ratio, compared t o hexane, was only 10. The interference was not caused by line overlap but by a general increase in background caused by hexane in this region. Structural Applications. The presence of diatomic molecular fragments as evidenced by their recorded emission spectra may serve a useful purpose in the elucidation of structure. An example of this application is indicated for the emission spectrum from malathion in the vapor phase shown in Figure 11. The presence of phosphorus is indicated by the presence of the 2536-A. atomic phosphorus line. The CS band system arises from the sulfur in the molecule and the characteristic PS band (not shown in the figure) appears in the 4700-A. region. These spectra can be obtained from very small quantities of material approaching the nanogram region. A major complication with emission spectrometry of organic molecules is the possibility of atomic recombination giving rise to combinations that do not exist in the original molecule. An attempt to construct a functional group correlation table is being pursued, and other analytical applications of the technique are being sought.
30 40 50 Flow Rate mp/min.
60
70
of carrier gas flow on sensitivity
Effect of Microwave Power Output. The intensity of the emitted spectra is greatly dependent on the microwave power supplied t o the cavity, as shown in Figure 12. With the available equipment, the output power of the oscillator could be varied from Oto 125 watts b u t below 20 watts it was difficult to generate a self-sustaining discharge. With some lines the intensity increased with increasing power, while with others a decrease was observed. With some carbon compounds such as n-hexane the intensity of the CY bands is proportional to the microwave power output, while with others the intensity levels off a t higher levels. The intensity variation with chlorinated materials is particularly interesting. The intensity of the C-C1 band at 2794 A. from 1-chloropropane is essentially independent of microwave power level, while the same peak from chloroform increases sharply. Possibly one complicating factor is the thermal fragmentation of the molecules a t the hot quartz surface of the capillary. The temperature of the surface increases with increasing power and the thermal fragmentation process would probably vary from molecule to molecule, resulting in changes in the spectral pattern. It is obvious from Figure 12 that relative response factors are dependent on power levels and some advantage can be taken of the observation. If the problem was concerned with the detection of chlorinated material, a low power level would enhance the response for chlorine while repressing the intensity of the cyanogen bands. With phosphorous compounds enhanced detection was also observed a t lower power levels. Other Detector Properties. The linearity of the detector response t o n-nonane was checked over a range of 20,000 (Figure 13). T h e wide range of sample sizes was obtained by using various concentrations of n-nonane in hexane, different injection volumes, and a variety of splitters. VOL. 37, NO. 12, NOVEMBER 1965
1475
Table II.
Element Carbon
Compound
Fluorine
CsF6
Chlorine Bromine Iodine Phosphorus Sulfur
CHC13 CHBr3 CHaI ( C2H,0)3P0
cs2
Table 111.
Gas
Detection limit,
WaveAssignlength, A. ment 3883 CY 5165 C* 5166 ? 2516 Si 2788 CC1 2985 1 2062 I P 2555 cs 2575
Selectivity ratio us. n-hexane
g./sec.
2 x 10-16 3 x 10-14 3 x 10-12 5 x 10-10 8 x 10-lo 2 x 10-7 7 x 10-14 1 1
...
x 10-11 ( 1 ) x 10-9
Wavelength, A. 3833 3089 3364 3364 2897 (neg. peak) 6037 (neg. peak) 5876 4861
i
10 20 20 10 104-1 07 100 100
Assignment CN OH 9 H or N2 XH or Nz CO +
co He H
Detection limit, g./sec. 60.
1 x 10-7 2 x 10-6
9 x 10-10 6X 1 x 10-7 4 x 10-9 1 x 10-8 2 x 10-9
For reasons that are not understood the area response of the detector increases with column temperature (Figure 15). The temperature of the quartz capillary is higher than the column temperature in all case. and it is difficult to understand hon- the column temperature could have such a large effect. X summary of detection limits and selectivity ratios is presented in Table 11. The detection limit for some permanent gases in argon is sholyn in Table 111. LITERATURE CITED
(1) Bache, C. A., Lisk, 1). J., ANAL. CHEIM. 37, 1477 (1965). (2) Fay, Homer, hlohr, P. H., Cook, G. A., Ibid., 34, 1254 (1962).
END OF SYMPOSIUM
ANALYTICAL CHEMISTRY
Response vs, Column Temperature for n-heptane. Argon Flow R a t e Z O m l / m i n.
Detection of Gases in Argon
The response of the detector to nnonane is greatly dependent on carrier gas flow and is closely approximated by a logarithmic function (Figure 14). The mechgnism of this dependence is as yet unknown. It appears that the detector would be very useful with capillary columns and their associated loiv volunietric carrier gas flow rates. The free volume of the detector-i.e., the portion of plasma focused on the entrance slit of the monochrometer-is of the order of 1 mm. resulting from the small diameter of the quartz capillary. The response time was limited by the electronic system used. With faster electronics it is estimated that the response time is a t least a millisecond and perhaps substantially less.
1476
10 1
Summary of Sensitivity and Selectivity Data
90.
120.
150.
180.'
Column Temperature ('C)
Figure 15. Area response vs. column temperature for n-heptane at 3883 A.
(3) Gatterer, A , , Frodl, V., Ric. Spettroscopiche Lab. Astrofs. Specola Vaticina 1, 201 (1946). (4) Johnstoiie, R. A. W.,I>ouglas, A . G., Chem. Ind. (London) 1959, 154. (5) AIcCormack, A . J., AIS. thesis, Cornell University, 1963. ( 6 1 Pearse, R. W. R., Gaydon, A. G., 'Ideiitification of hIolecular Spectra," JVilev. Sew Tork. 1963. ( 7 ) Robnison, I). if'., 11 R. the>i5, Cornell Univerzity, 1959. (8) Sauter, E , Z. n'aturJorsch. 3a, 392 (1948). ( 9 ) Steriiberg, J. C., Poulson, R. E., J . Chromatog. 3,306 (1O6O). (10) T'erma. R. D.. JIulliken. R. S.. i.M o l . Spectry. 6 , 419 (19Gl). ~
RECEIVEDfor review April 19, 1965. Accepted July 8, 1965. Division of Analytical Chemi\try, Lab-Line Award Sympo~ium, 149th Aleeting, ACS, Uetroit, llich., April 19G5, Work supported by the ru'ational Science Foundation under Grant GP 1818 and the National Iiisdute5 of Health under Grant ESR 14-65.