10 GREENFIELD, MOULE AND PERRY : CONDUCTIMETRIC [Analyst, 'C'Ol. 91 The Conductimetric Determination of Microgram Amounts of Phosphine in Air* BY S. GREENFIELD, H. A. MOULE AND R. PERRY (Albright & Wilson (Mfg.) Ltd., Research Department, Oldbury, Birmingham) A conductance cell is proposed as a sensitive means of detecting and determining phosphine in air at the part-per-million level. The contaminated air is passed through mercuric chloride solution in the cell, where reaction occurs between the phosphine and the mercuric chloride, liberating hydrogen chloride. This causes a rise in conductance proportional to the amount of hydrogen chloride, and hence to the amount of phosphine. Suggestions are made for the determination of phosphine in discrete samples of air, and for the continuous monitoring of contaminated air.VARIOUS methods have been proposed for detecting and determining phosphine. For example, a known volume of air can be passed through a standard silver nitrate solution, the residual silver nitrate then being titrated with standard potassium thi0cyanate.l For determining phosphine in acetylene, the gas has been passed through a solution containing mercuric sulphate and potassium chloride. The acetylene is passed until the solution is exhausted, as indicated by a strip of silver nitrate paper a t the gas exit; the phosphine content can then be calculated.2 A sensitive detector consists of a glass tube containing either a paper strip impregnated with silver 11itrate,~9* or silica-gel particles, with silver nitrate,5 a copper salt and a mercury complex,6 or a mercury salt and auric chloride' as impregnant. A known volume of con- taminated air is pumped through the tube, and the length of stain is compared with standards.It is claimed that such tubes will detect phosphine down to 0.01 p.p.m. I t has also been claimed that phosphine can be determined spectrophotometrically by the colour produced by reaction with silver diethyldithiocarbamate.8 In any process in which phosphine is used there is always present the possibility of a leakage, with the accompanying toxicity hazard, and it was considered that some form of continuous monitoring of the atmosphere near the plant was desirable. Kone of the above methods lends itself to automation, or to continuous recording, whereas the conductance cell to be described can be used, not only to examine discrete samples of air, but also, in conjunction with a suitable self-balancing bridge and recorder, for monitoring purposes.The cell contains mercuric chloride solution through which the air is passed. Hydrogen chloride is liberated by reaction of the phosphine with the mercuric chloride, causing a sharp rise in conductance, proportional to the amount of phosphine absorbed. EXPERIMENTAL THE CELL- The design of the conductance cell is similar to those already describedgi10711 for the determination of carbon and hydrogen in organic compounds, except that the electrodes are much larger, and the working volume is 11 ml. The electrode system consists of a cylindrical inner electrode of platinum gauze, surrounded by two half-cylindrical outer electrodes, also of platinum gauze.Possible leakage round the sealed-in wires is prevented by a coating of Araldite resin. A steady flow of air is essential in order to give a uniform circulation of solution and hence a steady conductance reading. (The conductance of a flowing solution is not the same as that of a stationary one,) To achieve this steady flow the jet size must be carefully chosen, since if it is too large the bubbles will emerge in bursts. This not only causes inter- mittent circulation of the solution, but also allows the solution to rise up inside the jet. This must be avoided, since this solution is absorbing phosphine but not contributing to the conductance. On the other hand, too small a jet can cause considerable back-pressure in the air-inlet system, which can give rise to loss of phosphine by leakage through the joints.The cell is illustrated in Fig. 1. * Presented a t the meeting of the Society on Wednesday, November 4th, 1964.January, 19661 DETERMINATION OF MICROGRAM AMOUNTS OF PHOSPHINE IN AIR 11 The optimum size of jet is best found by trial and error, but as a guide the jet used in these investigations was approximately 0.1 mm in diameter. The easiest way to obtain such a jet is to draw a capillary with a slow taper and then to cut or grind this back until the desired results are obtained. n Fig. 1. Conductance cell THE MEASURING APPARATUS AKD TEMPERATURE COMPENSATION- The conductance meter was made to a Guest, Keen and Nettlefold design,l2?l3 and is similar to those already de~cribed.~ *lo *l1 Temperature compensation is again by thermistor, but since mercuric chloride solutions, being non-electrolytes, do not have the necessary temperature characteristics, a small amount of hydrochloric acid solution has to be added.Fig. 2 shows the degree of compensation achieved for an unused solution of mercuric chloride with 0-00046 per cent. of added hydrogen chloride, and for a solution with 0-00053 per cent. of added hydrogen chloride, with a particular value of shunt resistor. The graphs are of the percentage change in conductance as the temperature is changed by 0.5" C above and below the mean temperature of 25" C. I I U 3 - V s " -0.10- I 1 I I I I I -0.5 -0.3 -0. I 0. I 0.3 0 Temperature change, @C (Mean temperature, 25@C) 5 Fig.2. Degree of colnpcnsation for part-used (curve A) and fresh (curve B) solutions12 GREENFIELD, MOULE AND PERRY : CONDUCTIMETRIC [AndySt, VOl. 91 Since the thermostatically-controlled bath holds the temperature constant to -t 0.03" C, reference to Fig. 2 will show that the variation in meter reading over this range is +0.0007 per cent. ( + O . O l ohm in 1500) for unused solution and +0.006 per cent. (k0.09 ohm in 1500) for partially used solution. Both these are obviously within the limits of accuracy of reading the meter. CHOICE OF ELECTROLYTE- The requirements of a satisfactory electrolyte are ( a ) that it shall absorb all the phosphine from the sample; (b) that it shall react with the phosphine to give a product that will change the conductance of the solution, and (c) that this change shall be as large as possible.A number of possible reagents, including copper sulphate , cuprous chloride, silver nitrate, mercuric and mercurous nitrate solutions, has been tried, but each failed to satisfy one or more of these requirements. Mercuric chloride, however, absorbs phosphine well, even at low concentrations, e.g., 0.1 per cent. At the same time, being virtually un-ionised, it has a very low conductance. This entails the use of large electrodes to give conductances within the optimum working range of the meter, which in turn gives high sensitivity. In addition, the reaction with phosphine yields hydrogen and chloride ions in solution, which gives a large change in conductance for a small amount of phosphine.A disadvantage of mercuric chloride is that solutions of it do not have the desired negative coefficient of resistance, but this can be conferred by addition of a trace of hydro- chloric acid to make, say, a 0.00048 per cent. solution. Three reactions at least are possible, all giving rise to hydrogen Chloride, but one giving phosphorous acid in addition- The question arises of the reaction between phosphine and mercuric chloride. PH3 + 3HgC12 -+ P(HgC1)S + 3HCl . . . . * * (1) 2PH3 + 3HgC1, -+ P2Hg3 + 6HC1 . . . . * (2) * - (3) PH3 + 6HgC12 + 3H2O -+ HcJPO~ + 3Hg2C12 + 6HCl Reactions (1) and (2) both have the same yield of hydrogen chloride per mole of phos- phine, but reaction (3) not only gives twice as much hydrogen chloride, but also gives 1 mole of phosphorous acid.The conductance under these conditions would be much greater than for reactions (1) or (2). Small increments of 0-002 N hydrochloric acid were added to the cell, and the conductance change was noted each time. A graph was plotted of conductance change against weight of added hydrogen chloride, after correction for dilution of the absorbing solution. A sample of phosphine was diluted suitably with air and a known volume of this was injected by means of a gas-tight syringe into the carrier gas at such a rate that the concentration of phosphine in the gas entering the cell was 2.5 p.p.m. The conductance change was noted and compared with the calibration graph to give the weight of hydrogen chloride released in the solution. This was then converted to weight of phosphine according to each of the three equations.By this means the recovery of phosphine was 88 per cent. if equation (1) or (2) held, but only 33 per cent. if equation (3) held. From this it was concluded that equation (1) held, since the mercuric chloride was present in large excess. From previous experience, 88 per cent. purity was regarded as a reasonable figure; it does not represent low recovery, since a silver nitrate detector on the exit tube of the cell indicated no loss of phosphine over many determinations. The reaction that in fact takes place was found by the following technique. CALIBRATION PROCEDURE- Since the reaction between phosphine and mercuric chloride gives only hydrogen chloride , the cell can be calibrated by adding small known amounts of a suitable dilute hydrochloric acid solution and measuring the resulting conductance changes.It is obviously a simpler and more accurate procedure to add hydrochloric acid for calibration purposes than to use phosphine. A graph is then drawn of conductance change against weight of hydrogen chloride used; this should be a straight line. This permits a factor to be calculated for weight of hydrogen chloride in pg per ohm and hence weight of phosphine in pg per ohm from the equation- PH, + 3HgC12 -+ P(HgCl), + 3HCl.January, 19661 DETERMINATION OF MICROGRAM AMOUKTS OF PHOSPHINE I N AIR 13 It must be noted that the addition of hydrochloric acid not only increases the concen- tration of the hydrogen chloride, but also increases the volume of the solution.In practice, when phosphine is being determined, the solution volume is not increased, so the effective addition of hydrogen chloride, with respect to the original solution volume, must be calculated for each addition. It is this effective addition- where x = initial volume of solution in ml; Cy = sum of the increments of hydrochloric acid solutions in ml; and C = concentration of the added hydrochloric acid solution in pg per ml, that is plotted against conductance meter-readings. hydrogen chloride. DRIFT- As substantially dry gas (e.g., air or nitrogen) is passed through the cell, the conductance is found to increase at a constant rate. Most of the time this conductance drift is attributable to the slow reduction in volume of the solution as water is removed by the gas, but there remain occasions when the drift is somewhat higher owing to factors at present unknown.Correction is made for the drift by noting the change in conductance over a known period, with air or nitrogen passing through the cell at the normal rate (20 ml per minute). This permits a figure for drift in ohms per minute to be calculated, usually less than 0.1 ohm per minute. SENSITIVITY- The factors that might be expected to influence the sensitivity include (a) the initiaI concentration of the hydrogen chloride in the mercuric chloride solution, ( b ) the volume of the solution, (c) the electrode configuration, and (d) the measuring apparatus. The initial concentration of the hydrogen chloride can be shown theoretically to have no effect on the conductance change for a given amount of phosphine.Since, however, the measuring apparatus is at its most sensitive when working between 1000 and 3000 ohms, the initial concentration of hydrogen chloride must be chosen, having regard also to the electrode configuration, so as to give an initial conductance within this range. The volume of solution is, on the other hand, critically important, since, for a given amount of phosphine and a given electrode assembly, the conductance change is greater the smaller the solution volume. I t follows, therefore, that the initial volume of solution must always be measured as accurately as possible, so as to be identical with the volume used in the calibration. As regards the electrode configuration, the electrodes should be as large as possible, and as close together as possible, since this gives the maximum conductance change for a given amount of phosphine.The steady drift at any time can be calculated from the conductance-bridge readings over a suitable period. These readings have an uncertainty at the 95 per cent. confidence level of 50.3 ohm, which corresponds in our particular cell to 0.005 pg of phosphine. Since an upward displacement by 0.3 ohm of the drift due to phosphine can be detected, it follows that an air sample containing, say, 0.05 p.p.m. of phosphine could be analysed by taking a discrete sample of 66.6 ml. A continuously monitored sample of the same phosphine level would be detected, at 20ml per minute, after 3 minutes 20 seconds. INTERFERENCES FROM OTHER GASES- Any gas that reacts with mercuric chloride to give hydrogen chloride, or which dissolves in water to give ions, or which reacts with the hydrogen chloride, will give a conductance change and thereby interfere.For instance, ammonia causes a drop in conductance, but can be removed from the sample gas by passing it through granular calcium chloride. Sulphur dioxide, hydrogen sulphide, arsine and stibine cause increases in conductance; all these can be removed, without affecting the phosphine, by passage through a tube of potassium hydroxide pellets. The graph should be a straight line, and has been found to be so up to 12 pg of added Some slight curvature may be apparent at higher concentrations. By timing the sample determinations, corrections for drift can be made.14 GREENFIELD, MOULE AND PERRY [Analyst, Vol.91 METHOD The first time the cell is used, set the temperature compensation as previously described.9 Also linearise the conductance meter by the following technique. Substitute a decade box for the electrode assembly, and note conductance readings, S, for various values of resistance, R, on the decade box. Plot R values against 1/S values, when it will be found that the graph cuts the R axis close to, and below, the origin, at a point that may be called RL. This repre- sents the resistance of a choke in the cell arm of the bridge, which can be compensated for by a suitable shunt, S’, across the measuring arm of the bridge. To calculate the value of S’, substitute the value of RL in the formula: S’ = PQ/RL, where P and Q are the fixed arms of the bridge.Empty the cell and wash it well with water until the conductance is less than 50 ohms. Again empty the cell and blow it dry with filtered air; do not attempt to dry it with organic solvents. Fill the cell with exactly 11 ml of absorbing solution, with nitrogen flowing at 20ml per minute. Note the conductance when the temperature has stabilised, as denoted by a steady drift. Add successive 20-4 portions of 0.002 N hydrochloric acid, preferably by means of a Microcap disposable capillary, and note the steady conductance readings after each addition. Correct the conductance readings for drift and plot the conductance change for each effective addition against the weight of hydrogen chloride added. From the graph calculate a factor for the conductance change per pg of phosphine from the equation- PH, + 3HgC1, -+ P(HgCl), + 3HC1.The cell is now ready for use. USE OF THE CELL- Discrete samples of air containing small amounts of phosphine-A pressure sample con- tainer, as used by the National Coal Board, is suitable. Ensure that the two absorber tubes are not exhausted. Pass purified nitrogen at 20 ml per minute and note conductances over a period until a steady drift is indicated. Calculate the value of the drift in ohms per minute. Pass the contaminated air through the cell via a pressure-reducing valve and a flowmeter at 20 ml per minute for a timed period and note the conductance change. Correct for drift and find the weight of phosphine from the calculated factor.Discrete samples of air containing higher concentrations of phosphine-Fit a T-tube, having a serum cap on the leg of the T, into the line before the two absorber tubes. Pass purified nitrogen through the cell at 20 ml per minute and note conductances over a period until a steady drift is indicated. Calculate the value of the drift in ohms per minute. Inject a suitable volume of sample in small increments from a gas-tight syringe into the nitrogen stream via the serum cap, and note the conductance change. Correct for drift and find the weight of phosphine from the calculated factor as before. Continuous monitoring of contaminated air-The cell can be used as it stands for this purpose, but the conductivity bridge needs constant attention. One of us (S.G.), however, has used the bridge in a continuously recording form, which permits the cell to be used for continuous-monitoring purposes, although the cell must be recharged periodically with absorbing solution. Alternatively, a flow-through cell can be designed, which will give the phosphine content of the sample as a conductance reading, rather than as a change in con- ductance. Such a cell, however, would have a reduced sensitivity, and calibration with hydrochloric acid would be difficult. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. REFERENCES Filz, W., Mitt. chem. Forsch.Inst. Wirt. (?st., 1954, 8, 61. Strizhevskii, I. I., and Zaitseva, V. P., Zau. Lab., 1956, 22, 546. Lugg, G. A., Commonzeiealth of Australia Departwent of Supply, Defence Standards Laboratory, Hughes, J . G., and Jones, 4. T., Amer. Ind. Hyg. Assoc. J . , 1963, 24, 164. Nelson, J. P., and Milun, A. J., Analyt. Chew., 1957, 29, 1665. Kitagawa, T., and Ogawa, T., .J. Electrochem. Soc. Japan, 1951, 19, 258. German Patent 1,129,731, 1962; Chew. Abstr., 1962, 57, 3142. VaSAk, V., Chemicke‘ Jisty, 1956, 50, 1116. Greenfield, S., Analyst, 1960, 85, 486. Greenfield, S., and Smith, R. D., Ibid., 1962, 87, 875. -- , Ibid., 1963, 88, 886. Moneypenny, H. K., J . Scient. Instrum., 1949, 26, 10. -, G.K.N. Group Research Report No. 261. Report No. 258 (1962). Received February 12th, 1965