Studies in gas chromtography with special reference to vapour detectors

An essential part of a gas chromatographic apparatus is a vapour detector capable of monitoring vapour concentrations in the gas stream emerging from the column. When this work was started (1958), there were three "high-sensitivity" detectors known, (capable of detecting less than 10-6 gms.) They were: 1) The flame ionisation detector; Vapour enters a small flame of hydrogen burning in air where it is thermally ionised. A polarising voltage is applied between two electrodes, one of which is usually the metal burner. The ion concentration in the flame allows a current to flow through an external circuit where it is made to drive a recorder. 2) The discharge tube detector: The potential drop across a low pressure gas discharge is very sensitive to small concentrations of impurities. Very small amounts of vapour can therefore be detected by passing the carrier gas into a continuously pumped discharge tube, and using the potential drop across the tube to drive a recorder. 3) The triode ionisation gauge: This piece of apparatus is normally used for low-pressure measurement. Molecules of gas or vapour in the gauge are ionised by electrons from a hot filament, which are accelerated by a positive potential on a suitable electrode. The ions are attracted to a negatively charged collector electrode. When applied to gas-chromatography, part of the carrier gas is fed into a continuously pumped ionisation gauge and the accelerating potential is adjusted so that it is insufficient to ionise the carrier gas molecules, but high enough to ionise molecules of vapour. The ion current constitutes the output from the detector. Detectors of each of these three types have been built and tested. This was done with the object of assessing their relative merits in various applications, in this laboratory and elsewhere, for which less sensitive detectors had been found inadequate. The parameters compared include simplicity of construction and operation, reliability under operating conditions, noise levels with, and suitability for, various types of sample, and sensitivities, both of one detector to a range of compounds, and of different detectors to the same compound. The flame ionisation detector has the merits of simple and rugged construction (although for the highest sensitivity the detector and the electronic "impedance converter" must be carefully made). It is highly sensitive, particularly to hydrocarbons, and has a low (potentially very low) noise level. Various detectors were built, differing in details of construction, and the merits of the different arrangements compared. The final form adopted incorporated a "concentric double burner" in which the sample in a stream of nitrogen was fed directly into the flame by a narrow tube inside the actual burner. This was to avoid the possibility of reaction between hydrogen and some of the samples, which might have occurred if nitrogen and hydrogen were mixed before the flame, as is the usual practice. The use of separate gas flows for flame and carrier allows them to be adjusted independently. With certain samples, in particular those, such as compounds of tin and silicon, which give involatile oxides on combustion or pyrolysis, there was considerable noise of a "spiky" kind. This could largely be eliminated by reversing the customary polarity of the high polarising voltage (the usual polarity is 'jet negative') and by using an unusual type of electrode - a wire ring or metal foil cylinder. This detector is not damaged by overload, and is completely insensitive to water: two properties which make it very suitable for biological problems. This has been shown by some work, on the scents of moths; the compounds of interest being present in trace amounts in excess of water. The response of the flame ionisation detector to the foll- owing compounds has been checked, and in most cases values for the sensitivity have been obtained: hydrocarbons, carbon tetrachloride, and many other organic compounds, mono-, di-, tri-, and hexa-silanes, silicones, phosphorus chloronitriles, silicon tetrachloride, stannic chloride, chlorine, bromine, iodine, hydrogen chloride, metal acetyl- acetonates, carbon disulphide, ammonia. For theoretical work on the flame ionisation detector a knowledge of the flame temperature under various conditions is needed. Measurements were made on the detector used in the rest of this work by the sodium line reversal method, and by small thermocouples. The value obtained was about 1380°C at a large burner and about 1700°C at a small one. (H2 flow rate: 50cc./min.). The triode ionisation gauge is potentially very sensitive indeed to organic compounds and the noise level is low. However the filaments are very susceptible to "poisoning", and the requirements in vacuum apparatus are exacting. The discharge tube detector is very sensitive, but the noise level is high. The minimum detectable sample is roughly the same as that for the flame ionisation detector, though the absolute sensitivity is much greater. The detector overloads easily and the electrodes become contaminated and the discharge is extinguished, but samples smaller than 10-4 gms. have no adverse effect. After the work had begun a fourth high sensitivity detector was announced, the argon ionisation detector. Here vapour molecules carried in a stream of argon are ionised by collision with excited argon atoms. The excited argon atoms are produced by bombardment with electrons, accelerated by a potential of 1000-2000 volts. Radiation from a radioactive source within the detector provides the electrons by ionisation of argon atoms. The ion current is monitored. Some work was done on this detector, and it was seen to be capable of very high sensitivity along with a very low noise level, but to be liable to erratic behaviour if the samples were too large. It is the only one of the four detectors which is non- destructive (although the triode ionisation gauge usually take only a portion of the sample. The approximate response of the detector to some organic and inorganic compounds has been determined. They include mono- and di-silane, monogermane, ammonia, phosphine, and hydrogen sulphide. Two variant forms of this detector were also briefly investigated: a commercial all-glass detector, and an all-glass detector with the source of radiation outside the ionisation chamber, both of which may be useful for corrosive samples. Various methods of injection of known amounts of sample have been used: a capillary of known volume filled with sample, injection of vapour from a trap of known volume at a known pressure, injection from sample tubes filled with vapour at a known temperature and pressure and sealed, and continuous saturation of the gas stream with vapour from a sample at a known temperature. The last method was the most satisfactory where it was applicable.