24 STOCKDALE: THE MOLYBDATE METHOD FOR THE DETERMINATION OF [VOl. 83 The Molybdate Method folr the Determination of Phosphorus, Particularly in Basic Slag and in Steel BY D. STOCKDALE (The University Chemical Laboratory, Cambridge) The yellow precipitate of molybdopliosphate obtained under controlled conditions has the empirical formula (3XH4),H[P(Mo,0,,),] .H,O, and it is transformed into the triammonium salt, (NH4)3[P(Mo3010)4], by washing it with a dilute solution of ammonium nitrate. Conditions for the complete removal of phosphoric acid from solution as the former salt have been delimited and the results have been applied to the determination of the phosphorus in basic slag and in steel. When the precipitate is formed in the presence of substantial concen- trations of sulphates or of ferric salts it IS too heavy, and the method used to obtain a correct result is to dissolve the first precipitate in ammonia solution, to remove the molybdenum as molybdenum sulphide and to re- precipitate the phosphorus as molybdoplhosphate.Volumetric methods, with and without the use of formaldehyde, have been examined. Some pH curves are given and the use of ammonium paramolybdate as a volumetric standard is described. INVESTIGATION OF THE CONDITIONS FOR PRECIPITATING MOLYBDOPHOSPHATE COMPOSITION OF THE PRECIPITATE- There is some doubt about the composition of the salt obtained by adding a solution of ammonium molybdate to a solution of an orthophosphate containing large excesses of nitric acid and ammonium nitrate. The formula of the salt prepared under the conditions to be described later and after the precipitate had been washed with dilute nitric acid and dried at 140" C was (NH,),H[P(Mo,O,,,),].H,O. It was found that 1877.5 g of precipitate (1 gram-formula-weight) were equivalent to 24 g ram-equivalents of sodium hydroxide ; the neutral solution required a further 2 gram-equivalents of sodium hydroxide after treatment with formaldehyde.In addition, the ammonia evolved on distilling the precipitate after treatment with sodium hydroxide was found, on collection in standard acid, to be equivalent to 34.8 g per gram-formula-weight (required by formula, 34.0 g) ; also, on heating the precipi- tate and passing the gases over copper oxide, the water collected in traps containing anhydrone was equivalent to 98.6 g per gram-formula-weight (required by formula, 99.1 g).There are many reports in the literature that the precipitate as first prepared is the triammonium salt, with either one or two molecules of nitric acid in addition, e.g., (NH,), [P(Mo,O!,,),] .HN0,.2H20. If the precipitates obtained in these experiments were originally of this composition, heating them to 140" C would entail the loss of ammonium nitrate. To get more information on this point, precipitates were prepared in the usual way and were kept at room temperature after ithey had been washed. One sample was dried in a desiccator over pellets of potassium hydroxide and another was air-dried on a porous plate, it being found later that there was no sulbstantial difference between the two.As 1877 g of sample were found to be equivalent to 22.51 gram-equivalents of sodium hydroxide and to 24.48 gram-equivalents after the addition of formaldehyde, there would appear to be only two ammonium groups present. Other portions of the precipitates were dried at 140" C to constant weight and the above-mentioned equivalents were corrected to a "dry" basis, becoming 24.1 9 and 26.30, respectively. As the yellow precipitate is hygroscopic, these results are consistent with the view that the original material was the diammonium salt of the formula given above, but damp and contaminated by a small quantity of nitric acid. The pure compound appears to be (NH4)2H[P(Mo30),].H20. When washed with a dilute solution of ammonium nitrate, the diammonium salt passes easily into the triammonium salt, (NH,)3[P(Mo,0,,),], with a decrease in the pH of the solution used for washing. It was found that 1876.5 g of precipitate (1 gram-formula-weight) were equivalent to 23 gram-equivalents of sodium hydroxide ; the neutral solution required approximately a further 3 gram-equivalents of sodium hydroxide after treatment with formaldehyde.January, 19581 PHOSPHORUS, PARTICULARLY IN BASIC SLAG AND IN STEEL 25 Note that the molecular weights of these salts are so nearly the same that it is a matter of indifference in gravimetric analysis whether one or other or a mixture of the two is weighed.The gravimetric factors for the diammonium and triammonium salts are, respectively: for P, 0.016506 and 0.016515; for P,O,, 0.03781 and 0.03783. CONDITIONS FOR PRECIPITATION- Stirring-Local supersaturation can persist for long periods in a stagnant solution even in the presence of the precipitate, no matter what the composition of the solution.Stirring, therefore, is essential if erratic results are to be avoided. When borderline conditions are being examined, the stirring must be done in some standard way, and in this work it was done with a stout glass rod by hand during the addition of the precipitant and thereafter for 1 minute every 15 minutes, four times in all during the hour allowed for precipitation. C 01 1 0 10 20 30 1 Volume of nitric acid, sp. gr. 1.42, present, rnl Fig. 1. Conditions for the precipitation of molybdophosphate in 250 ml of solution at 70" C: x and y , the boundaries between zones A and B and B and C, respectively, are for 12.5 g of ammonium nitrate in the solution; similarly x1 and yr are for 2 6 g of ammonium nitrate in solution.Point P indicates the standard conditions under which the work described in the paper was carried out. The numerical values indicate the difference, in mg, between the calculated and deter- mined weights of precipitate E'ect ofmolybdic oxide and aitric a c i d 4 f the many factors that affect the formation of the precipitate, the excess of molybdic oxide used and the concentration of the nitric acid are the most important. These have been investigated, the precipitates being formed in 1 hour at 70" C from a final volume of 250 ml containing 25 g of ammonium nitrate, stirring being as described above. A weight of AnalaR potassium dihydrogen orthophosphate to give a precipitate of about 300 mg was taken and the solution was brought to 70" C and so adjusted that after the molybdate had been added the final volume was 250ml with the concentration of the reagents as desired.The molybdic oxide was added as a solution of ammonium paramolybdate, in small part from a burette to make up the weight of molybdic oxide required, but the bulk, itself at 70" C, from a free-running pipette, this being the last adjustment to be made to the solution. The results are shown in Fig. 1, the numbers on the diagram being the "found less calculated" weights in mg and refer only to solutions con- taining 25 g of ammonium nitrate (full lines). 'With too little molybdic oxide and too much nitric acid, some phosphate remains in solution, but when the proportion of these reagents26 STOCKDALE: THE MOLYBDATE METHOD FOR THE DETERMINATION OF [Vol.83 is reversed, the precipitates are too heavy. For this condition there is a time effect : in many experiments the particles of yellow precipitate were coarse and settled out quickly, leaving a clear solution, which became turbid or threw down a white precipitate during the last few minutes of the hour. The position of the boundary between zones A and B, therefore, depends to some extent on the time allowed for precipitation and also, presumably, on the temperature. There is, however, a very wide irange of conditions, zone B, under which precipitates of the correct weight can be obtained. These precipitates have been washed with untreated nitric acid, and all are therefore slightly light.The following concentrations have been standardised for much of the further work described in this paper: 1.8 g of molybdic oxide in excess, 35 ml of nitric acid, sp.gr. 1.42, and 25 g of ammonium nitrate in a final volume of 250 ml (point P in Fig. 1). Efect of ammonium nitrate-The broken lilies in Fig. 1 show the effect of reducing the concentration of ammonium nitrate from 25 to 12.5 g in 250 ml. This is unimportant except with high concentrations of nitric acid. E$ect of temperature-This was investigated by using potassium dihydrogen phosphate and the standard mixture, with a precipitation time of 1 hour. Results for precipitation between 30" and 70" C were satisfactory, but those at 80" C were about 2 per cent.too high. Clearly a temperature of 70" C is about the maxiinum at which the pure molybdophosphate can be obtained, and it was decided for greater certainty to adopt 60" C as the standard temperature for precipitation. This has proved satisfactory for basic slag and for other determinations of phosphorus, except in the presence of a large concentration of iron, as for phosphorus in steel. In this last analysis precipitates obtained at 60" C were too heavy, but those got at 40" C were of approximately the correct weight (see Table 111, p. 32). Attempts were made to throw down the precipitate at room temperature (15"C), by leaving the solutions overnight with clean air bubbling through them, but about 10 per cent. of the phosphate was not precipitated.Washing the precipitate-If the yellow precipitate is washed with pure water, there is a serious risk of it breaking up and passing through the filter as very fine white particles. If the water contains a high enough concentration of some suitable electrolyte, this does not happen. Electrolytes often used are nitric acid, ammonium nitrate and potassium nitrate, the last only when the yellow precipitate is not to be weighed directly. In rare instances, 1 per cent. solutions have not prevented break-up, and the use of solutions of double this con- centration is recommended, Losses during washing may be serious. With 2 per cent. ammonium nitrate, 350 mg of clean precipitate in a Gooch crucible lost about 1 mg per 100 ml of solution added at the rate of 10 ml per minute.With 2 per cent. nitric acid under the same conditions the losses were 15 to 20 mg. Somewhat crude attempts to measure the solubilities at room temperature were made, it being found that the solubility of the molybdo- phosphate in 2 per cent. ammonium nitrate was 55 mg per litre and somewhat more in a 5 per cent. solution. Two per cent. nitric acid reacts slowly with the precipitate, so that the true solubility in it was not found. The solubility after 24 hours was 250 mg per litre, but the solid had completely disappeared after a further 5 days, indicating a solubility greater than 400 mg per litre. The effect of potassium nitrate was examined only super- ficially, but it also seems to react with the precipitate, losses in the crucible being rather smaller than with nitric acid.Clearly 2 per cent. ammonium nitrate is a suitable liquid for washing, provided care is taken not to over-wash. Its use, however, demands that in exact work the precipitates should be dried at 280" C to eliminate the ammonium nitrate. Acceptable results have been obtained by using 2 per cent. nitric acid that had been shaken with molybdophosphate, either specially made or old precipitates contaminated -with asbestos, leaving the residues in the nitric acid overnight and filtering through No. 42 Whatman filter-paper before use. The beakers were cleaned with two 20-ml volumes of pure 2 per cent. acid and the crucibles were washed with five 10-ml volumes of the treated acid. The precipitates were dried at 140" C. This seems to be no more than a trick whereby the loss in weight of the precipitate through dissolution in the acid is made up by the weight of the solids that the treated acid leaves behind.AccidentaE introduction of phosphorus-It must be remembered that in the determination of phosphorus relatively very large amounts of reagents are used. I t follows that what appear to be quite insignificant traces of phosphorus in a reagent may have an important effect on a result. In accurate work, therefore, it is essential either to use a pure phosphate asJanuary, 19581 PHOSPHORUS, PARTICULARLY IN BASIC SLAG AND IN STEEL 27 a control or to have reagents free from phosphorus. In this connection most samples of ammonium nitrate have been found to contain significant amounts of phosphate.Precipitant-The precipitates have been produced by adding enough nitric acid and concentrated solution of ammonium nitrate to the phosphate solution to give the desired final concentration, and then by adding molybdic oxide as ammonium paramolybdate; a solution often used contained 50 g of the salt per litre to which small amounts of ammonia had been added as a stabiliser. ,4 50-ml portion of this solution contained 2 g of molybdic oxide, which will usually give the required excess in 250 ml. A precipitant suitable for general use and free from phosphorus was prepared as described below under “Reagent.” REAGENT- Ammonium molybdate reagent-Dissolve 250 g of ammonium nitrate in 250 ml of water in a flask and add 350 ml of nitric acid, sp.gr. 1.42. Dissolve 25 g of ammonium para- molybdate in 150 ml of water and slowly add this solution to the nitrate solution, shaking the flask well during the addition.Dilute to 1 litre and add 1 mg of potassium dihydrogen phosphate in solution. Put the flask into a water bath at 60” C and leave it there for 6 hours, shaking it occasionally. Allow the solution to cool with the bath overnight and filter it next day through a No. 42 Whatman filter-paper, without washing the filter-paper. No precipitate formed in a sample of the reagent that was kept for 3 months. PROCEDURE- Dilute the approximately neutral solution, which should not contain more phosphate than will give a precipitate of 400 mg, to 150 ml. Raise its temperature to 60” C, preferably by using a water-bath with crude thermostatic control, and run into it 100 ml of ammonium molybdate reagent, itself at 60” C.Use a fast-flowing pipette and stir to mix the two solutions thoroughly. Maintain the temperature for 1 hour, well stirring the contents of the beaker from time to time. Collect the precipitate in a Gooch or sintered-glass crucible, using two 20-ml portions of 2 per cent. ammonium nitrate solution to transfer it from the beaker, and rubbing over the glass with a rubber-tipped glass rod. Wash the precipitate in the crucible with five 10-ml volumes of the ammonium nitrate solution. Dry the precipitate a t 280” C. METHOD FOR DETERMINING PHOSPHATE RE s u LT s For 12 determinations on potassium dihydrogen phosphate made in four groups by the procedure described above, on different days, the results were: found, 343.9mg, with a standard deviation of k0.8 mg; required, 34343 mg.With precipitates washed with nitric acid, for 16 determinations on 5 days: found, 343-8 & 0.4 mg; required, 344.0 mg. There is no reason for thinking that these precipitates are other than stoicheiometric. Although there is an element of adjusting the technique to give the desired result when weighing as the diammonium salt, this is absent for the other salt, when the precipitates were washed until they were only just clean, the test being made by adding a small amount of ammonium chloride to the solution before filtration and looking for chloride in the washings. A sample from one of the sources was recrystallised from water and also from dilute hydrochloric acid.No difference between any of the five samples could be detected and therefore it seems probable that all were of high purity. The samples were dried at 110” C for 1 hour before use. Additional evidence about the purity of the phosphate is given on p. 29. The potassium dihydrogen phosphate used came from three sources. GRAVIMETRIC DETERMINATION OF TOTAL PHOSPHATE IN BASIC SLAG The slag analysed was a sample of Basic Slag A from the British Chemical Standards series, used in the “as received” condition. This slag had been examined by 16 analysts by a variety of methods. For gravimetric and volumetric molybdate procedures only (15 results), the median result was 12.88 per cent. and the mean 12.96 per cent., with a standard deviation of 0.21 per cent. In the analysis of slag, the reagent used to get the phosphate into solution must not interfere in the precipitation of the molybdophosphate.It should be efficient in that no Their mean result was 12.92 per cent. of P,O,.28 STOCKDALE: THE MOLYBDATE METHOI) FOR THE DETERMINATION OF [Vol. 83 further treatment of any insoluble residue should be necessary, and the solutions should be quick to filter. The efficiencies of hydrochloric, nitric, perchloric and sulphuric acids have been examined for these requirements, the usual. methods for extraction being employed. At some stage, and more particularly during the final stages of washing, filtration was always slow, but from this point of view the treatments with hydrochloric and perchloric acid were of about equal efficiency and markedly superior to the others.All the acids left coloured residues containing phosphorus, which were got into solution by fusion with sodium bicar- bonate after digestion with hydrofluoric acid. In this respect hydrochloric acid was the least efficient. It was found, with 2-g samples, that about 1 per cent. of the phosphate was retained in the residue after the treatment with hydrochloric acid, while the phosphate left behind by each of the other acids was about one-third of this. It would seem, therefore, that for the most accurate results the first residues cannot be neglected, no matter what the treatment, and that there is nothing to be gained by departing from the well tried preliminary opening with hydrochloric acid. In work of a less exact nature, perchloric acid might be used and the residues neglected.The determinations were made on aliquot portions representing about 100 mg of slag and giving about 350mg of molybdophosphate, at least two solutions of the phosphate prepared by each of the four methods being used. Either the diammonium or the tri- ammonium salt was produced and the determinations were made with and without controls of pure potassium dihydrogen phosphate. As is tlo be expected in view of the results given earlier, the effect of controls on the final result was either very small or negligible. All the determinations were made in accordance with the procedure described on p. 27, and the results are shown in Table I. TABLE I ANALYSIS OF BASIC SLAG Standard deviation, Number of Range of P,:O, Mean PpOb P,O, per 100 parts Treatment determinations content, content, of slag Hydrochloric acid .. 8 12.63 to 12.73 12.68 f 0.036 Nitric acid . . .. 10 12.68 to 12-76 12.73 f 0.028 Perchloric acid . . 14 12-65 to 12.80 12-74 & 0.046 % % Sulphuric acid . . 6 13.28 to 13.79 13.49 f 0.18 From these results it is seen that, the anomidlous result with sulphuric acid apart, the- means do not seem to be significantly different and give a final result of 12.72 per cent. of P,O,. This is about 1 in 50 lower than the stated P,O, content. INFLUENCE OF SULPHATE AND IRON- Precipitates produced in the presence of substantial concentrations of sulphuric acid or of ferric salts are always heavier than expected, sometimes to the extent of 20 mg in 300 mg. The traces of sulphate or iron they contain are in no way sufficient to account for this increase in weight.It is thought that in the presence of these ions a precipitate of a molybdophosphate richer in molybdenum than the duodeca salt is formed in part. If this view is correct, it would seem that to determine the molybdenum in an impure precipitate is no way out of the difficulty. To substantiate this view, precipitates made from potassium dihydrogen phos- phate were heated to 470" C and weighed as P20,.24MoO,, it being expected that any occluded sulphate would be eliminated at this temperature. The results were as follows: required, 366.8 mg; found, in absence of sulphate, 366.1 mg; found, after precipitation from 250 ml containing 5 g of ammonium sulphate, 388.3 mg. In another experiment, precipitates were made with and without the addition of ammonium sulphate. The ratio of their mean weights was 1.052.These were then determined volumelrically, when the ratio of their sodium hydroxide values was found to be 1.048. The interfering substance must neutralise sodium hydroxide, and must, I think, be molybdenum trioxide. INFLUENCE OF PERCHLORATE- The general similarity between the sulphate and perchlorate ions gave rise to the expectation that the precipitates would be too heavy when thrown down in the presence of perchlorates. In one experiment, in which pure potassium dihydrogen phosphate and 18 gJanuary, 19581 PHOSPHORUS, PARTICULARLY IN BASIC SLAG AND IN STEEL 29 per 250 ml of perchloric acid were used, the mean weight of the precipitates was 390.7 mg, the calculated weight being 383.0 mg, and the sodium hydroxide ratio was 1.024.The standard deviations were large, as with sulphate. The effect, therefore, exists, but it is unlikely to prove of importance in practical analysis, as the concentration of acid used in these experiments was many times that likely to be found in a solution from the extraction of a phosphate. It also seemed that it might be possible to lose phosphoric acid by volatilisation when fuming with perchloric acid, as with sulphuric acid. Therefore, some potassium dihydrogen phosphate was fumed with l o g of 60 per cent. perchloric acid, the operation being more prolonged than would have been necessary in the extraction of a slag. The first effect should have caused the weight of the precipitate to be rather high; this second effect would cause it to be low.It would seem, therefore, that this second effect also exists, but again to an extent unlikely to prove of importance. PURIFICATION OF THE PRECIPITATE- Many unsuccessful attempts have been made to purify molybdophosphate precipitates by simple dissolution in an alkaline liquid followed by re-precipitation from a solution con- taining nitric acid. Stoicheiometric precipitates made from potassium dihydrogen phosphate were almost invariably heavier after such treatment. The reason for this has not been found, but the only successful method of purifying precipitates that were heavy because they had been made in the presence of sulphate or iron was to remove the molybdenum and to make a fresh start with the solution of orthophosphoric acid (see p.30). THE MAGNESIUM PYROPHOSPHATE METHOD- The purity of one of the samples of potassium dihydrogen phosphate after being dried, but without recrystallisation, was tested by the method of Schmitz,l the precipitates weighed being either magnesium pyrophosphate or magnesium ammonium phosphate. The salt appeared to be about 99.7 per cent. pure. This low result was accounted for in part by the dissolution of the magnesium ammonium phosphate during washing. No phosphate could be found in the solution from which the salt had been precipitated, but the later washings after concentration always gave a yellow precipitate by the molybdate test, The final result for the purity of the phosphate after correcting for this loss was 99.83 & 0.10 per cent.In fact, about exactly the calculated weight was obtained. DETERMINATION OF PHOSPHORUS IN STEEL As is well known, the molybdophosphate will not form quantitatively in a solution containing a high concentration of ferric ions unless the concentration of molybdic oxide is correspondingly increased. If, however, the standard conditions are modified in this way, the results are unsatisfactory, because the weight of precipitate obtained depends on the amount of precipitant used. This is shown in Fig. 2. The points X near the curve for the steel containing 0.076 per cent. of phosphorus refer to experiments in which the nitrate ion was replaced by the perchlorate ion-there is no iniprovement. There is, however, some improvement if the precipitate is formed at 40" C instead of at 60" C, and indeed it is possible to get reasonable results at the lower temperature by a single precipitation, followed by either gravimetric or volumetric analysis.Some representative results are given later. The horizontal lines, x, in Fig. 2 represent the limiting conditions under which a quantita- tive result can be obtained with a solution of pure potassium dihydrogen phosphate, and show the marked retardation caused by iron. The vertical lines represent estimates of the weights of precipitate that should have been obtained from the steel used. It might be thought that it would be possible to devise a method in which only a limited excess of oxide is used, as shown by the intersection of these lines with the curves (about 2-4 g in Fig.2), but attempts to do this gave unsatisfactory results. Under this condition the weight of precipitate obtained is sensitive to small changes in experimental detail, and the results lacked precision. If the earlier findings of this paper are correct, namely that the increase in weight of the precipitate is due to excess of molybdenum oxide and that this cannot be removed by simple dissolution and reprecipitation and also that the molybdophosphates are suitable gravirnetric compounds in that they are completely insoluble in the mother liquors and normally stoicheiometric, it would seem reasonable to isolate the phosphate by using a large30 STOCKDALE: THE MOLYBDATE METHOD FOR THE DETERMINATION OF [VOl. 83 excess of molybdic oxide to be quite certain of complete precipitation, to remove the niolyb- denum from this precipitate and to precipitate the phosphate anew as molybdophosphate from what would be essentially a solution of oirthophosphoric acid free from interfering substances.A method on these lines is described below. I I Fig. 2. Determination of phosphorus in various steels: curve A, 3 g of a steel containing 0.028 per cent. of phosphorus at 60" C; curve B, 2 g of a steel containing 0.076 per cent. of phosphorus a t 4.0" C; curve C, 3 g of a steel contain- ing 0.076 per cent. of phosphorus a t 60°C. Points marked X near curve C represent an experiment in which nitrate ion.5 were replaced by perchlorate ions. Ordinates y , y1 and yz represent the theoretical weight of precipitate; the abscissa x represents the limiting conditions when potassium dihydrogen phosphate is used in place of the steel MOLYBDENUM SULPHIDE METHOD REAGENTS- Concentrated ammonium molybdate reagent-,% solution containing 40 g of ammonium molybdate, 250 g of ammonium nitrate and 250 ml of nitric acid, sp.gr.1.42, per litre, made as described on p. 27. As this solution is slightly unstable, after a week or so a small deposit forms. Dilute ammonium molybdate reagent-A solution containing 20 g of ammonium molybdate, 200 g of ammonium nitrate and 250 ml of nitric acid, sp.gr. 1.42, per litre. There was no precipitate after the solution had been kept for 3 months. PROCEDURE- To 2 g of the steel and 40 ml of water in a covered 400-ml beaker add 20 ml of nitric acid, sp.gr.1.42, in such a way that the reaction does not get out of control. (Note 1.) Gently boil the solution on a hot-plate until most of the nitrous fumes have been driven out. Add 2ml of 2.5 per cent. w/v potassium permanganate solution, stir and continue to boil for 3 minutes. Dissolve the manganese dioxide by adding water saturated with sulphur dioxide drop by drop to the stirred solution (about 1.5 ml may be required) and continue the gentle boiling for 2 minutes longer. Dilute the solution to 150ml with water, and heat it to 60" C. Add 100 ml of concentrated ammonium molybdate reagent, itself at 60" C, to the The solution should be decanted from this before use.January, 19581 PHOSPHORUS, PARTICULARLY IN BASIC SLAG AND IN STEEL 31 stirred solutions, delivering from a free-running pipette.Stir the contents of the beaker from time to time and filter through a 9-cm Whatman No. 40 filter-paper after 1 hour. Rinse the beaker twice with 2 per cent. ammonium nitrate solution at room temperature, using 20-ml portions. Discard the filtrate. Drip 5 ml of diluted ammonia solution (1 + 2) on to the paper and collect the filtrate in the original beaker. When all traces of the yellow precipitate have disappeared, wash the paper with six 10-ml portions of near-boiling water. Bring the volume of the filtrate to 100 ml and pass in hydrogen sulphide until there is no further change of colour. Dilute 10 ml of diluted hydrochloric acid (1 + 1) to 50 ml in a 400-ml beaker. Pour the acid into the sulphide solution and raise the temperature nearly to the boiling-point.Filter the solution through a 9-cm Whatman No. 40 filter-paper, collecting the filtrate in the beaker used for the hydrochloric acid. Wash the beaker with three 20-ml portions of near-boiling 1 per cent. hydrochloric acid, using a rubber-tipped glass rod to remove the last traces of precipitate, and then wash the precipitate on the paper seven times with the hot acid (see Note 2). Dissolve any precipitate by adding nitric acid drop by drop, and then increase the pH by adding diluted ammonia solution (1 + 2) until methyl orange just turns yellow. Bring the volume to 50 ml and the tem- perature to 60" C. Add 50 ml of dilute ammonium molybdate reagent, itself at 60" C, and continue the precipitation as before. Collect the precipitate in a Gooch or sintered-glass crucible, washing it as before, but using five portions of the ammonium nitrate when washing it in the crucible.Dry the precipitate at 280" C. NOTES- Wash the paper four times with 10-ml portions of this solution. Cover the funnel with a clock-glass. Reduce the volume to about 30 ml by evaporation. The use of a rubber-tipped glass rod is essential. 1. As about 5 ml of nitric acid are used per g of steel, about 10 ml of nitric acid will be left. 2. As phosphate may be retained by the molybdenum sulphide, the precipitate must be washed thoroughly, and the correctness of the washing technique used should be checked from time t o time by dissolving the molybdenum sulphide and testing for phosphate in the solution. Almost certainly some molybdenum sulphide will pass into solution during the washing and will appear as a brown scum when the filtrate is evaporated.This scum is dissolved by nitric acid and the presence of this molybdenum does not affect the final result. METHOD FOR DETERMINATION BY A SINGLE PRECIPITATION Prepare the solution of 2 g of steel as described above under the molybdenum sulphide method. Add 100 ml of con- centrated ammonium molybdate reagent, itself at 40" C and refiltered if necessary. Stir the solution from time to time. After 1 hour, collect, wash, dry and weigh the precipitate, as described above for the molybdenum sulphide method. THE ANALYSIS OF STEELS These methods have been tested against the British Standard method2 and against controls made by adding potassium dihydrogen phosphate to a solution of high-purity iron.In the British Standard method the phosphorus is precipitated as molybdophosphate after the removal of silicon by baking and of arsenic by means of bromine. The phosphate is then converted into lead molybdate and is weighed as such. The iron used was A.H.R.4, supplied by the courtesy of the British Iron and Steel Research Association and declared to contain less than 0.001 per cent. of phosphorus. Somewhat uncertainly, by a difficult analysis, it was found to contain 8 p.p.m. of phosphorus. They were B.C.S. Nos. 240, 232 and 152 and were declared to contain about 0.028, 0.076 and 0.083 per cent. of phosphorus, respectively. There was nothing unusual about their content of other elements, except that No.152 contained 0.24 per cent. of sulphur and 0.015 per cent. of arsenic. The steels had been analysed by groups of independent analysts by the British Standard method for the most part, except for No, 152, an older standard, for which a direct molybdate procedure had been used. Bring the volume to 150 ml and the temperature to 40" C. The steels analysed were from the British Chemical Standards series.32 STOCKDALE: THE MOLYBDATE METHOD FOR THE DETERMINATION OF [VOl. 83 Usually at least four determinations were made for each steel by each method (see last column of Table 111). These were made in groups, each group consisting of two samples of a steel and two controls containing a similar weight of phosphorus made by adding known weights of potassium dihydrogen phosphate to high-purity iron.The phosphate was added to the control at the earliest possible stage and the two pairs were treated as nearly as possible in the same way. It was expected that day-to-day variations in technique would cause all the results of a group of four to have errors of the same sign. It was found, however, that the errors were nearly random and that the final results would have been almost the same had the controls and the steels been treated apart. The results were in small groups and the standard deviation for each was calculated. If the deviation for a result was more than twice the standard deviation, that result was rejected on the ground that a “wild” result might have been caused by an abnormal error: in any event its inclusion in such a small population would no doubt have an undue effect.Of 94 determinations in 15 groups, five, of which four were for steels by the British Standard method, were rejected in this way. TABLE 11: The results for the controls are given in Table 11. DETERMINATION OF PHOSPHORUS IN S’TEEL: RESULTS OF CONTROLS Phosphorus Method found (F), % British Standard . . . . 0.0340 Molybdenum sulphide . . 0.0319 Single precipitation . . . . 0.0320 British Standard . . . . 0-0759 Molybdenum sulphide . . 0.0764 Single precipitation . . . . 0.0768 Phosphorus sought (S), 0.0308 0.Q308 0.0308 0.8757 0.0757 04757 Yo F-S, p.p.m. + 32 + 11 + 12 + 2 + 7 + 11 Number of determinations 6 6 4 11 10 8 Points t o note in Table I1 are the exceptionally high value got for low-phosphorus steels by the British Standard method and that the “found” values are always high.I t may be that all the methods have a bias in the high direction, but it is more probable that phosphorus has been accidentally introduced, because the differences do not depend on the amount of phosphorus sought. The appearance of 0.02 mg of phosphorus from the reagents or elsewhere would account for this discrepancy. The results for the steels are given in Table 111. The found-less-calculated values were obtained for the controls and were applied to the absolute results got for the steels. The results after adjustment in this way are given in the fourth column of the Table. TABLE 111: DETERMINATION OF PHOSPHORUS IN STEEL Steel Method British Standard . . B.C.S. No. 240 Molybdenum sulphide Single precipitation .. B.C.S. value . . . . British Standard . . B.C.S. No. 232 Molybdenum sulphide Single precipitation . . B.C.S. value . . .. British Standard . . Single precipitation . . B.C.S. value . . . . I I Molybdenum sulphide .. .. .. .. .. .. .. .. .. .. .. .. Absolute result. 0.0296 04,0270 0-0273 0-0280 0.0749 0.0741 0.0753 0.0760 0-0814 0.0794 0.0806 0.0830 % o f P Controlled result, 0-0268 0.0261 0.0263 yo of P 0-0746 0.0739 0.0742 0.0808 0.0785 0.0796 Number of determinations 4 4 6 5 8 4 9 4 4 In an attempt to obtain some idea of the prelcisions of the three methods, the results for the low-phosphorus controls and steel were collected together to give reasonably large groups, and so also for the high-phosphorus determinations. The grouped standard deviations are shown in Table IV.There is some indication that the British Standard method is the least precise, at least for steels containing substantial quantities of phosphorus.January, 19581 PHOSPHORUS, PARTICULARLY IN BASIC SLAG AND IN STEEL 33 The results given in the Tables have been examined statistically at some length, but little has emerged. It seems probable that there is no significant difference between any of the controlled mean results got by the three methods, except that the molybdenum sulphide method may have given a lower result for B.C.S. No. 152. This is the steel that contains 0.015 per cent. of arsenic, and it may be that the sulphide method eliminates arsenic or some other interfering element more thoroughly than do the other methods.The inter- ference, if any, is very small, and is equivalent to about 1 part per hundred thousand of phosphorus. The method of additions showed that small quantities of chromium, silicon, titanium and vanadium do not interfere in the molybdenum sulphide method. TABLE Iv DETERMINATIOK OF PHOSPHORUS IN STEEL : PRECISION OF RESULTS (a) f \ Standard Standard deviations for deviations for low-phosphorus high-phosphorus Method results, results, p,p.m. p.p.m. British Standard . . . . +6 (10) &I0 (25) Molybdenum sulphide . . &2 (10) f 5 (22) Single precipitation . . f7 (10) f 4 (16) (b) I \ Standard Standard deviations for deviations for low-phosphorus high-phosphorus results, results, p.p.m. p.p.m. h 1 6 (11) f17 (28) f 2 (10) * 6 (23) k 7 (10) f 4 (16) Columns (a) refer to results after the rejection of five determinations (see p.32); columns (b) refer to all results. The numbers of determinations are shown in brackets. COMMENTS ON THE BRITISH STANDARD METHOD- In this method the use of a rather greater concentration of molybdic oxide in the initial precipitation of the molybdophosphate than is employed here is recommended. I should, therefore, expect the results given by the method to be rather high. Co-precipitation of lead orthophosphate with the lead molybdate might also cause the results to be high. On the other hand, they might well be low, because the specification does not insist on vigorous and repeated stirring, but says only “shake the solution until the precipitate forms and allow to stand on the bench for 20 min.” These effects might quite well lead to a correct final result, but a rather large deviation is perhaps to be expected.Collected precipitates of lead molybdate have been examined from time to time and phosphate has always been found in them, but only to a trifling extent. My estimate is that the co-precipitation of phosphate causes the final result to be high by 1 in 200. Phosphate, however, has been found in the washings in important amount. Table V shows results got by working with low and high phosphorus controls and with two of the steels. In these experiments the specification for the precipitation as molybdophosphate was followed as closely as possible. The mother liquor was then diluted and the phosphate in it was precipitated as suggested earlier in this paper, 1 hour being allowed for the salt to form.The final weighing was as lead molybdate. Each experimental result in the Table is the mean of four determinations that differed rather widely among themselves. TABLE V PHOSPHORUS IN RESIDUES BY THE BRITISH STANDARD METHOD 2-g samples of steel were taken in each experiment Lead molybdate in main Lead molybdate Total Lead molybdate precipitate, in residue, lead molybdate, sought, g g g g 0.0876 Low-phosphorus control 0.0964 0.0075 0.1039 B.C.S. steel No. 240 . . 0.0835 0.0069 0.0904 - High-phosphorus control 0.2270 0.0152 0.2322 0.2153 B.C.S. Steel No. 152 . . 0-2289 0.0163 0.2442 - The calculated weights for the steels are 0-075g of lead molybdate for B.C.S. No. 240 at 0.026 per cent.of phosphorus and 0.225 g of lead molybdate for B.C.S. No. 152 at 0.079 per cent. of phosphorus. These results confirrn those of Table 111, namely that without controls34 STOCKDALE: THE MOLYBDATE METHOD FOR THE DETERMINATION OF [vole 83 high results were got for the steel low in phosphorus and that those for the other were nearly correct. There is always, however, an important fraction of the phosphorus left in solution. There must be compensating errors, and hence the experimental conditions must be varied as little as possible if a high precision is to be obtained. It will be seen from Table V that, if a control is applied at either stage, the effect will be to give approximately correct results for both steels. It is suggested that the reason for the relatively large precipitates got for the low-phos- phorus steel is that when the phosphorus content is low, the precipitate is somewhat slow to form, its grain size is small and it settles slowly, so leading to less supersaturation.CONCLUSIONS- I t should be noted that, although the results got for the controls were either nearly correct or slightly high, the percentages of phosphorus both in the slag and in the steel as now determined are definitely lower than the accepted values, by about 1 in 50 for the slag and by about 0.002 per cent. for the steel. VOLUMETRIC FINISH FOR DETERMINING PHOSPHATE Possibly the volumetric methods for the determination of phosphorus that are now most used are based on the method of P e m b e r t ~ n . ~ ? ~ Pemberton washed his precipitates with water, dissolved them in standard potassium hydroxide and titrated the excess of hydroxide with nitric acid, with phenolphthalein as indicator.He standardised his acid against sodium carbonate. He stated that his precipitates were the triammonium salt and that 23 gram- equivalents of base were equivalent to 1 gram-akom of phosphorus. R,[P(Mo,O,,),] + 23ROH -+R2HP0, + 12R2Mo04 + 11H20, with R = K or NH,. Pemberton's method was followed in general outline in the present work. After the gravimetric analysis, dry precipitates of known weight were dissolved in an excess of sodium hydroxide that had been standardised against AnalaR potassium hydrogen phthalate, and the excess was titrated with hydrochloric acid, with phenolphthalein as indicator. It was found advantageous to have some of the hydroxide in a second burette, to overshoot the titration with the acid and to finish with the solution a very faint pink, at a pH of about 8.There have been many suggestions that better results are to be obtained by the elimina- tion of the ammonia before the completion of the titration. This can be done either by heating the precipitate to about 450" C by boiling the solution in sodium hydroxide, or through the use of formaldehyde. The last of these methods is the most convenient and it has been tested extensively. In addition to the Pemberton equation, the relevant equations are- R,H[P(Mo,O,,),].H,O + 24ROH -+ R,HPO, + 12R2Mo0, + 13H,O H,[P(Mo,O~,)~] + 26ROH += R2HP0, + 12R2Mo04 + 14H20 It was found for the diammonium salt that without and with formaldehyde 1 gram- formula-weight was equivalent, respectively, to 23-96 * 0.16 and 25-98 * 0-13 gram- equivalents of sodium hydroxide, and similarly for the triammonium to 23.02 * 0.07 and 25.73 & 0.09 gram-equivalents of sodium hydroxicde.The number of titrations in each group was about 20, about 350 mg of precipitate being used for each titration, with approximately 0.25 M hydrochloric acid and sodium hydroxide. The deviations given are the standard deviations. Reasons for the low result for the triammonium salt after the elimination of the ammonia, and for the greater precision of the work with this salt, have not been found. In routine analysis it will usually be undesirable to eliminate by heat a reagent used in washing and therefore the work on the diammonium salt is of academic interest only.If, however, the precipitate is collected on a pad in a filter crucible and washed with a dilute solution of ammonium nitrate, in accordance with the technique described on p. 27, with a single final wash with 20ml of absolute ethanol, and dissolved straightway in sodium hydroxide, good results can be obtained relatively quickly. For precipitates made from a solution of potassium dihydrogen phosphate, and collected on asbestos pads, after the removal of the ammonia with formaldehyde the results were as follows: required, 18-81 ml of the standard hydroxide; found, after correction, 18.87 5 0-03 ml. The correction for the nitrate retained by the pad was only 0.03 ml of 0-25 M hydroxide. The conclusion is that in general outline the method of Pemberton is quite satisfactory, and entirely so if the acidJanuary, 19581 PHOSPHORUS, PARTICULARLY IN BASIC SLAG AND IN STEEL 35 and alkali are standardised against precipitates made from pure potassium dihydrogen phosphate.If, however, the precipitates are washed with water, as advised by Pemberton, there is serious risk that they will break up and pass through the filter. Phenolphthalein is satisfactory as indicator, and there is little to be gained, and indeed there may sometimes be a small loss of accuracy, through the use of formaldehyde. The commoner practice is to wash the precipitate with a solution of potassium nitrate instead of ammonium nitrate. The potassium ions replace the protons as before and a neutral molybdophosphate is obtained.This method has not been examined in detail, but it seems to be satisfactory even though the yellow precipitate is substantially more soluble in the 2 per cent. potassium salt than it is in 2 per cent. ammonium nitrate, as an examination of the used washing solutions will show. The cationic exchange, however, does not stop with the replacement of the protons, but one of the ammonium groups is replaced also to give (NH,) K,[P(Mo,O,,),]. This exchange, which takes place fairly easily, has little effect on the result of an ordinary titration, but 24 gram-equivalents, somewhat indefinitely, of sodium hydroxide are required per gram-atom of phosphorus if formaldehyde is used. If this salt is kept under a dilute solution of potassium nitrate, the solid remains of constant composition, but it dissolves slowly and it may be that there is further slow displacement to give a soluble salt. Fig.3. Titration of unused 0-25 M sodium hydroxide after addition of: curve A, (NH4)3[P(Mo,0,0),] ; curve B, “synthetic” (NH,),H [P(Mo,O,,),] ; curve C, (NH,), H[P(MO,O,~),].H,O; curve D, (NH,),H [P(Mo,O,,),] .H,O, after treatment with formaldehyde. The theoretical end-points are shown on each curve Some of the pH curves plotted during this work are reproduced in Fig. 3; a Cambridge pH meter was used. Usually about 350 mg of the salt were dissolved in 25 ml of 0.25 M sodium hydroxide and the solution was then titrated with 0.25 M hydrochloric acid, so that 1 ml of acid represents about 1.4 gram-equivalents per gram-formula-weight. The calcu- lated end-points, on the basis of the equations, are shown on the curves.An indicator that changes near pH 8.0 is suitable, and phenolphthalein would seem to be the best in view of the ease with which the first appearance of red can be seen. As none of the curves for the ammonium salts shows an abrupt change of slope, it is somewhat surprising that the precision to be obtained in an ordinary titration is as great as it is. The destruction of the ammonia with formaldehyde approximately doubles the rate of change of the pH near the end-point for the triammonium salt, but the greater complication introduced by the use36 STOCKDALE [Vol. 83 of this reagent counter-balances this advantage and does not lead to an increase in precision.The curve for the “synthetic” diammonium salt was obtained by dissolving suitable weights of diammonium hydrogen phosphate and molybdic oxide in 25 ml of standard sodium hydroxide, and the similarity of curves B and C, Fig. 3, is further proof of the formula Although the Pemberton method is so satisfactory that an attempt to improve on it is perhaps redundant, the matter has been pursued further by introducing ammonium para- molybdate, (NH,),(Mo,O,,) .4H,O, as an analytical standard. Such samples of the AnalaR grade of this salt as I have handled have been without exception of very high purity, and a preliminary check of the purity can be made by heating the salt to 300” C. Most samples then lose a trifle more weight than the theoretical and acquire a slight bluish tinge. The weight is regained and the blue tinge lost on moisteiiing with nitric acid and reheating. Two 6-g samples lost 1.1079 and 1.1081 g (required, 1.1083 g). Had a sample contained 1 part per thousand of non-volatile impurity the loss in weight would have been 1-1072 g. The powdered salt is stable when kept in a desiccator with technical calcium chloride. As the end-point with phenolphthalein is very poor when the salt is titrated with sodium hydroxide, the removal of the ammonia with fornialclehyde is essential. Portions of the paramolybdate and the yellow precipitate are dissolved separately in an excess of sodium hydroxide and neutral formaldehyde is added; the solutions are titrated with hydrochloric acid after 15 minutes, with slight over-shooting, and the end-point is reached by adding a few drops of hydroxide. As titrations can be carried out in exact parallel, uncertainties about the complete destruction of the ammonia are partly eliminated. In these experiments the spent solutions were kept and the end-points were again determined after 3 hours and again next day, the additional sodium hydroxide required a.fter 3 hours being 0.2 or 0-3 ml in 50 ml, and 0-05ml next day. The results obtained after 3 hours were slightly better than those obtained after 15 minutes, but nothing was gaincd by putting aside the flasks for a longer time. When about 350 mg of weighed, heated diammonium molybdophosphate and 500 mg of paramolybdate were used, the ratios by volume of 0.25M sodium hydroxide for equal weights of the salts were 1.226 0.004 after 15 minutes and 1.223 & 0.004 after 3 hours (required, 1.223). Starting from potassium dihydrogen phosphate without intermediate weighing, the ratio of volumes for 0.1 g of phosphate to 1 g of paramolybdate were 1.689 4 0.003 and 1.686 rfi 0.004, respectively (required, 1.686). The results after 3 hours are slightly more accurate and rather more precise than those obtained by using sodium hydroxide standardised against potassium hydrogen ph thalate. Guidance in the statistical analysis was given by the late Mr. J. Wishart. I thank the British Iron and Steel Research Association for the gift of the high-purity iron. (NH*),H [P(Mo,O,,),l* REFERENCES 1. 2. British Standard 1121 : Part I : 1943. 3. 4. Schmitz, B., 2. anal. Chem., 1924, 65, 46. Pemberton, H., J . Amer. Chem. SOC., 1893, 15, 382. - , Ibid., 1894, 16, 278. Received June 8112, 1966