432 Analyst, June, 1973, Vol. 98, $9. 432-442 A Quantitative Tunable Element-selective Detector for Gas Chromatography BY W. R. McLEAN, D. L. STANTON AND G. E. PENKETH (Imfierial Chemical Industries Limited, Petrochemicals Division, Billingham, Teesside, TS23 1 JB) A detector based on the atomic-emission spectra that result when organic compounds are decomposed in a low-pressure, microwave-sustained helium plasma is described. All of the non-metallic elements normally found in organic compounds can be sensitively and selectively detected in a linearly proportional and quantitative manner by means of conventional diffraction grating spectrometer equipment. A controlled amount of a scavenger gas is used to prevent carbon deposition inside the plasma tube. The chromato- graphic column outflow is split between the element-selective detector and a non-selective flame-ionisation detector.The latter acts as a reference for interpreting element-selective detector results and assists with the determina- tion of atomic ratios and the empirical formulae of organic compounds. THE gas-chromatographic detection of organic compounds by emission spectroscopy with a microwave-powered plasma was first reported by McCormack, Tong and Cookel and subsequently developed by Bache and Lisk2s3 and others4s5 The organic compounds emanat- ing from the gas chromatograph are fragmented in the high-energy plasma to produce emission spectra, which are then monitored with a suitable spectrophotometric detection system, In principle, the technique is applicable to a very wide range of elements, but most work to date has centred around the elements of interest in pesticide analysis, for example, sulphur, phosphorus and chlorine.Our objectives were to extend this range and to improve the quantitative characteristics of the system, and because of the latter objective we preferred to use a low-pressure helium plasma to produce the atomic-emission lines. The lower energy of an argon plasma is insufficient to prevent the association of atoms into pairs and this leads to the production of complex band-emission spectra, which reduces the spectral selectivity on atomic lines. The association of atoms into pairs also reduces the population of free atoms, and for quantitative work this effect adds a complicating mass-action influence to atomic emission and gives compound-specific effects.The higher energy of the helium plasma greatly inhibits formation of atom pairs, to the obvious benefit of the atomic-emission characteristics. In our early work, both the qualitative and the quantitative performance were hampered by persistent deposition of carbon on the inner walls of the plasma tube, but a dramatic improvement resulted when a small amount of air was continuously bled in. Subsequently, it was found that either oxygen or nitrogen would act as a carbon scavenger, and this discovery enabled either element to be included in the range of the technique by using the other as scavenger. APPARATUS- The apparatus is shown diagrammatically in Fig. 1. Gas chromatograph--We normally used a Pye 104 gas chromatograph but other makes have been used with equal success.Any column with low bleed characteristics can be used; the conditions for the chromatograms shown in this paper were- Column . . . . Temperature . . .. . . 112 "C (pre-heater 200 "C) Pressures .. .. . . Head of column, 170 kN m-2; tail of column, 35 kN m-2 Detector splitting ratio . . 1 : 1 Scavenger gas . . .. @ SAC and the authors. .. .. 3 m x 2.5 mm, packed with 10 per ccnt. Apiezon L on 00 to 80-mesh DCMS-treated Chromosorb W . . Oxygen (or nitrogen) : 0.1 to 1 per cent. of total plasma gas according to demandMCLEAN, STANTON AND PENKETH 433 Strip chart Strip chart recorder recorder I. 2. 3. 4. Microwave Nitrogen I Hydrogen Amplifier Amplifier Air (FID) I (MPD) Power Power supply handling (FID) (MPD) systems Sample - FID supply - Microwave cavity 5.Diffraction grating Plasma discharge Monochromator entrance slit Exit slit and photomultiplier 6. Scavenger gas supply 7. Air - hydrogen supply 8. Helium supply t o chromatograph Fig. 1. Block diagram of the apparatus blasma detector (MPDb-The Dlasma emission is contained in a thick-walled quartz capillary'tube of 10 mm 0.d. and 1 mm i.d., with an over-all length of 15 cm. This tube is mounted vertically in an assembly that also holds the microwave cavity in position. The pressure within the plasma is adjustable from an arbitrary 0.25 torr to higher pressures by means of a vacuum pressure regulator in the line connecting the plasma to the vacuum pump (Fig. 2). Microwave power, generated by a 0 to 200-W generator (Electro-Medical Supplies Limited, London) a t 2.45 GHz from 100 to 200 W, is supplied to the tuned 214L cavity via a reflected power meter and flexible wave guide.The tuning stubs on the cavity are adjusted until the reflected power is the minimum attainable. With this type of cavity, the power reflected can be made as low as 1 per cent. The plasma is initiated by means of a Tesla coil and gives an intense plasma discharge about 10 cm long. Gas su$+Zies-High-pressure sources of the following gases were used- Helium . . .. .. . . Grade A Carrier gas Nitrogen . . .. . . White spot grade Air . . .. .. . . Electrolytic grade : ~~~~~~~ ::$: detector support gases Oxygen . . .. .. Hydrogen . . .. Grade A helium was purified by passing it through a B.O.C.helium purifier, in which oxygen and nitrogen were removed by the hot titanium sponge and hydrocarbons were oxidised to carbon dioxide and water by hot copper(I1) oxide; these products were then removed in a modified 5A molecular sieve unit placed externally to the purifier and maintained at -80 "C. The scavenger gases were dried by passing them through a cold trap at -80 "C. Because of the relative amounts of each gas used, the purity of helium is approximately one hundred times more critical than that of the scavenger gases, so for the more common elements carbon, nitrogen, hydrogen and oxygen, the greater the spectral purity of the helium, the higher is the sensitivity. I t was realised at an early stage that the value of element-selective results was much enhanced by the simultaneous recording of non-selective results, e.g., those from a flame- ionisation detector.This realisation led us to develop an interface system that enabled a434 McLEAN, STANTON AND PENKETH: A QUANTITATIVE TUNABLE [Analyst, vol. 98 Sample Head pressure Tail pressure I I I Gas I chromatograph: I column I system I I I I I 7 I I I I I I 1 I I I I I I I I I i I Scavenger I L--- _ _ _ _ _ _ _ _ _ J bleed h L I MPD detector Plasma con t ro I Fig. 2. The chromatograph and the inter- facing system to two detectors gas-chromatographic system to be coupled with two or more detectors that were operated at atmospheric or lower pressures. The principle is illustrated in Fig. 2. The exit of the column system is maintained at a pressure in excess of that of the atmosphere by a separate supply of carrier gas; the extra supply then merges with the column exit flow and splits into equal parts across specially designed flow restrictors to the detectors.By allowing the total flow-rate to the detectors to be more than the optimum flow-rate through the gas-chromatographic column, no additional constraints are imposed upon the chromatography. 1-485.99 77 D-656. 7 . H-656.281 -486.1 33 rT. Ah = 0.134 nm Ah = 0.181 nm Fig. 3. Resolution of the hydro- gen - deuterium atomic-emission lines (wavelengths in nanometres)June, 19731 ELEMENT-SELECTIVE DETECTOR FOR GAS CHROMATOGRAPHY 435 Spectrophotometer-The monochromator used was a Hilger and Watts Monospek 1000 with a 102 x 102-mm diffraction grating of 1200 linesmm-l (blazed at 300-Onm) to give a dispersion of 0.88 nm mm-l. The original IP28 side-window photomultiplier was replaced with a similarly designed Hamamatsu R446 photomultiplier, which extended the optical range available from 190.0 nm to greater than 900.0 nm with an excellent degree of sensitivity. SPECTRAL CHARACTERISTICS- The spectroscopic system gave good line spectra for all of the elements examined with very little evidence of molecular band emission.As examples, Fig. 3 shows the resolution of the hydrogen and deuterium lines and Fig. 4 the resolution of the triplets of oxygen and nitrogen. The preferred wavelengths for the elements examined are shown in Table I. The technique should be equally applicable to boron and mercury' or, indeed, any element that can be introduced into the plasma.Energy levels involved in the emission of atomic spectra from non-metals are included in Table I and are illustrative of the high energy of the helium plasma. Nitrogen 746.879 I 744-256 742.388 I Oxygen 777.1 93 777.41 4 777.543 I I 740.0 750.0 760.0 770.0 780.0 Fig. 4. metres) Atomic-emission spectra for nitrogen and oxygen (wavelengths in nano- QUALITATIVE ELEMENT-SELECTIVE DETECTION- An artificial mixture containing most elements of interest was made up and run through the system to illustrate the element selectivity (Fig. 5 ) . To compare the selectivity for the various elements, n-heptane was used as a standard carbon and hydrogen containing com- pound, and in Table I1 the selectivity factor given is the ratio of the mass flow-rate of heptane to the mass flow-rate of the element required to give equal signals at the element-selective emission wavelengths.Inter-element effects are dealt with later. TABLE I ELEMENT-SELECTIVE EMISSION WAVELENGTHS AND EXCITATION ENERGIES Element C H D F c1 Br I S P N 0 He Emission wavelengthlnm 247.857 486.133 656.100 6854302 479.454 470.486 516.119 645.388 253.665 '746.879 777.193 587.662 Ionisation state 1 0 0 1 2 2 2 Energy levels/eV - E, + El 12.69 7.69 16.29 12.74 13-98 12.09 16.31 14-60 18.54 16-96 17.03 14.4 13.51 12.11 18.21 16.94 12.08 7.2 12-34 10.73 23.06 26-17 - -436 ,u MCLEAN, STANTON AND PENKETH: A QUANTITATIVE TUNABLE [Analyst, Vol. 98 14 ‘1 i; 1 I, Fig. 5 . Element-selective traces on a test sample for C(b), H(c), D ( d ) , O(e), N(f), F(g), CI(h), Br(i), I ( k ) and S(Z) versus a flame-ionisation detector reference (a).Emission wavelengths are: C, 247.867 nm; H, 656.281 nm; D, 656.100 nm; 0, 777.193 nm; N, 746.879 nm; I;, 686.602 nm; C1,1479.454 nm; Br, 470.486 nm;June, 19731 ELEMENT-SELECTIVE DETECTOR FOR GAS CHROMATOGRAPHY 437 11 7 Fig. 5, continued I, 516.119 nm; and S, 545.388 nm. Throughout, peaks are: 1, deuteroacetone; 2, nitroethane; 3, fluoro- benzene; 4, toluene; 5, n-butyl iodide; 6, n-nonane; 7, chlorocyclohexane ; 8, anisole; 9, diethyl disulphide; 10, octan-2-one ; 11, bromobenzene; 12, o-dichlorobenzene ; 13, o-bromotoluene ; 14, n-undecane438 MCLEAN, STANTON AND PENKETH: A QUANTITATIVE TUNABLE [Analyst, Vol. 98 TABLE I1 DETECTION LIMITS, SPECTRAL BACKGROUND LEVELS AND SELECTIVITY IN DETECTION Element C H D F c1 Br I S P N 0 Detection limit*/ng s-l 0.08 0-03 0.09 0.06 0.06 (0-06) 0.05 (0.05) 0.09 (0-05) 2.9 3.0 0.091 (0.02) - (0.009) Total background as element/ng s-l 0.8 1 0.22 0.17 0.091 0.46 0.72 0.86 1.1 113-0 98.0 - Selectivity ratio versus n-heptane* - - 880 2300 510 (44) 1300 (38) 400 (38) 390 (22) - (1000) - - * Figures in parentheses are values obtained by Bache and LiskZ for detection limits and selectivity ratio vemus phenanthrene.QUANTITATIVE ELEMENT-SELECTIVE DETECTION- Linearity of dual detection system-The flow to the two detectors maintained a constant splitting ratio irrespective of sample size. This property is illustrated in Fig. 6 and demon- strates the linear emission characteristics of the element detector relative to the flame- ionisat ion detector .150 E E 2 100 \ 0, a) JZ Y Q) P .- n 5 50 / I I I K / ' Acetone I FID peak heighthm 150 E -5 100 + J= 0) aJ t Y a L1 .- n 50 n 5 Benzene 0 I I 1 - 0 50 100 FID peak height/mrn Fig. 6. Linear response of the dual detection system on H ( a ) , C(b) and O(c) versws a Emission wavelengths are : flame-ionisation detector reference for acetone and benzene. C, 247.8 nm; H, 486.1 nm; and 0, 771.1 nm Sensitivity and dynamic range-The linear dynamic range is subject to an upper limit, which occurs when the concentration of a component is too large and either the scavenger is insufficient to prevent deposition of carbon or a quenching effect occurs, which perturbs the linear emission characteristics. At the lower end of the range, the limits of detection are subject to the values of the background signal levels at the element-selective wavelengths.Within these limits, the linear dynamic range for fluorine, for example, covers four decades. The best detectable limits and background levels so far obtained are shown in Table 11. The detectable limits given are twice the noise level observed on the background. The background is composed of photo-tube dark current, stray light and spectral contamination due to impurities, and the levels stated are calculated in terms of mass flow-rate for purposes of comparison.June, 19731 ELEMENT-SELECTIVE DETECTOR FOR GAS CHROMATOGRAPHY 439 Determination of atomic ratio and empirical formulae-With a manually tuned, single- channel spectrometer it is convenient to use the element-selective emission as a peak height ratio with the flame-ionisation detector signal.Effectively, this ratio is equivalent to the slopes of the graphs shown in Fig. 6. Let actual slopes be defined as- aC aH 8 0 a T D P a m # K D T ’ * 9 etc* For compound X, where FMpD - is the flow-rate (splitting) ratio to the two detectors, R, is the microwave plasma detector response per gram-atom of carbon, nc is the number of carbon atoms per molecule, n, is the number of gram-moles of compound X and (RF)X is the response factor per gram-mole of compound X on the flame-ionisation detector. FFID Similarly, and, on division- (E) =- n KO x 2 = constant x oxygen to carbon atomic ratio. x Kc nc 0.6 0.5 0.4 -1- 0.3 0 0 0.2 0.1 0 0.1 0.2 0.3 0.4 0.5 0 1 2 3 4 5 6 7 8 9 1 0 Atomic ratio carbon : oxygen Atomic ratio oxygen : carbon I Fig.7. Calibration graphs for oxygen to carbon and carbon to oxygen ratios The constant Ko/Kc is independent of the compound-specific response factor of the flame-ionisation detector and can be evaluated by reference to any known oxygen-containing compound. The independence of the emission signal from molecular properties is illustrated in Fig. 7, where the signal ratio - is a linear function of the known oxygen to carbon ratios in a variety of compounds. The inverse ratio, - , also expresses the carbon number if there is only one oxygen atom per molecule. Further results for hydrogen to carbon ratios, obtained by using ethylbenzene as a single reference standard, are shown in Table 111.Similar behaviour has been noted for the other elements detected selectively by this technique. ao aC X 8 0440 MCLEAN, STANTON AND PENKETH: A QUANTITATIVE TUNABLE [Analyst, Vol. 98 The highly selective detection of the elements carbon, hydrogen, deuterium, oxygen, nitrogen, fluorine, chlorine, bromine, iodine, sulphur and phosphorus in a linearly propor- tional and quantitative manner with good sensitivity has been a continuing feature of this work, and it is now possible to use the technique in order to obtain the empirical formulae of organic compounds separated by gas chromatography as a step towards the identification of compounds. Undoubtedly, the major factor leading to quantitative element-selective detection has been the use of a scavenger gas.As carbon is not appreciably volatile below 3500 "C (boiling-point 4200 "C) and silica melts at 1700 "C (boiling-point 2200 "C), it is to be expected that elemental carbon will plate out on the relatively cold walls. When organic compounds are pyrolysed in the plasma, the action of the scavenger is to hold the carbon in its volatile elemental state- DISCUSSION C +O---+CO . . . . . . * * (1) (ii) co - c + 0 . . .. - - (2) solid, ex pyrolysis gas (4 plasma gas energy frecatom when the necessary oxygen (or nitrogen) scavenging atoms are produced as a result of the plasma discharge. The bond strength of C-0 (256.7 kcal mol-1) is equivalent to 11.1 eV (C-N is equivalent to 7.8 eV) and virtually complete dissociation into separate atoms, as in equation (2), is readily achieved by the plasma energy available. An indication of the energy of the plasma is given in Table I, where energies of 12 to 19 eV are required in order to produce atomic emission from the non-metallic elements.In this work, oxygen (and nitrogen) scavenger gas levels in the plasma gas were kept in the 0.1 to 1.0 per cent. V/V range. Below 0.1 per cent., deposition of carbon was a problem. Above 1.0 per cent., deposition of carbon was not a problcrn, but the amount of carbonaceous material that could be tolerated could exceed that required to overload the linear atomic- emission characteristics of the plasma. This effect marks tlie upper limit of the dynamic range of tlie technique. TABLE I11 HYDROGEN TO CARBON ATOMIC RATIOS FOUND IN HYDROCARBONS H to C ratio found Theoretical H to C ratio Cyclopentane .. .. .. 1.990 2.000 Cyclohexane . . . . .. 2.020 2.000 Cyclooctane . . .. .. 2-023 2.000 Methylcyclohexane . , .. 2415 2.000 Dimethylcyclohexane . . .. 2.012 2.000 Trimethylcyclohexane . . .. 2.019 2.000 Isopropylcyclohexane . . .. 2.008 2.000 Cyclohexene . . .. .. 1.652 1-667 Pent-1 -ene .. .. .. 1.997 2.000 Hex- l-ene .. .. .. 2.052 2.000 Hept-3-ene .. .. .. 2.047 2.000 Oct-l-ene . . .. .. .. 2.027 2.000 Dec-l-ene . . .. .. .. 2.041 2.000 n-Hexane .. .. .. .. 2.347 2.333 n-Heptane.. .. .. .. 2-335 2.286 n-Octane . . * . .. * . 2.300 2.250 n-Nonane . . a . .. .. 2.266 2.222 n-Decane . . .. .. .. 2.249 2.200 n-Undecane . . .. .. 2.251 2.1 82 Benzene .. .. .. .. 0-982 1.000 Toluene .. .. .. .. 1.142 1.143 Ethylbenzene (reference standard) o-Xylene .. .. .. .. %3 %50 The lower end of the dynamic range is set by the noise level of the background signal at the various element-selective wavelengths. While some elements are detected more sensitively than others, in all instances the level of the background signal has a major effectJune, 19731 ELEMENT-SELECTIVE DETECTOR FOR GAS CHROMATOGRAPHY 441 on the sensitivity. This effect is illustrated by the data shown in Fig. 8 and shows good correlation between detection limits and background. It is significant that the least sensitive elements are oxygen and nitrogen, followed by hydrogen and carbon; attention to the following details can noticeably improve performance. The vacuum lines and joints should be tested very carefully for leaks, the gas-chromatographic columns should be pre-conditioned in situ, and the helium gas supply should be dried once more between the pressure regulators and the gas chromatograph by means of tubes containing phosphorus pentoxide.Also of great importance is the provision of very smooth and stable power supplies to the microwave generator and phototube. While the detection limits for some elements can be limited by contamination levels in the plasma, the background limitations on the less common elements are due to continuous radiation (plasma emission and stray light) phototube properties, e.g., dark current, spectral range and sensitivity, monochromator resolution and the optics of the light collection and filtration system. The shape of the emission source is optimised in the form of a cheaply replaceable, l-mm bore, thick walled, clear silica tube.10 t .- I-’ CI % ’ -a L)- I-’ .- 0.1 .- -I 0 4 1 0.001 Background Fig. 8. Effect of background signals on the limit of detection. (These data show that the limit of detection is more a function o f spectral background than element identity.) Basic units are gram- atoms x per second Interference effects are restricted to spurious spectral band emission when the plasma is overloaded but this is instantly recognisable from the magnitude of the flame-ionisation detector response. A chemical effect peculiar to fluorine and chlorine gives rise to phantom oxygen emission through the possible reaction scheme- plasma SiO, + F - SiF + 2 0 wall energy gas plasma - oxygen emission SiF - S + F - silica emission gas atomic This effect makes the detection of oxygen in polyfluoro and polychloro compounds very difficult by this technique. Equally, it inhibits the selective detection of silicon.An alterna- tive plasma tube material to silica would be useful in solving this particular problem.442 MCLEAN, STANTON AND PENKETH The original idea of adding a second, non-selective, detector to act as a reference for comparing the element-selective data has been extended to assist in the quantitative inter- pretation of the data into atomic ratios. Determination of the necessary atomic ratios for an evaluation of empirical formulae can be seen to be a very laborious process if the single- channel form of this technique described here is used. It is very much more efficient to use a multi-channel spectrometer with simultaneous detection of many elements and work is proceeding on the construction of an instrument of this type. Modern techniques of data handling could lend themselves to an automatic print-out of the empirical formulae of organic compounds eluted from a gas chromatograph. Because the amount of a compound is the sum of its atomic parts, a means of quantitative analysis is made available which is not compound-specific in its response and which does not require to be calibrated by use of the compounds being analysed. The authors thank Harry Fraser for his considerable practical contributions to this work. REFERENCES 1. 2. 3. 4. 6. 6. Imperial Chemical Industries Limited, British Patent Applications 20366/70, 41246/70 and 7. Received July 12th, 1972 Accepted January 24th, 1973 McCormack, A. J., Tong, S. S. C., and Cooke, W. D., Analyt. Chem., 1966, 37, 1470. Bache, C. A., and Lisk, D. J., Ibid., 1967, 39, 786. -,- , J . Ass. Off. Analyt. Chem., 1967, 50, 1246. Braman, R. S., and Dynako, A., Analyt. Chem., 1968, 40, 96. Dagnall, R. M., Pratt, S. J., West, T. S., and Deans, D. R., Talanta, 1970, 17, 1009. Bache, C. A., and Lisk, D. J., Avzalyt. Chem., 1971, 43, 960. 41960/70, 1970.