Feb., 19511 MATERIALS FOR AGRICULTURAL PURPOSES 65 The Accurate Determination of “ Phosphoric Anhydride ” by Means of Quinoline Phosphomolybdate BY H. N. WILSON (Pyesented at the meeting of the Society on Wednesday, October 4th,, 1950) Present methods of determining P,O, are critically surveyed, and the requirements of a new method considered; it should be as accurate as the “official” method or better, and much quicker. As it appears that the phosphomolybdate reaction is the most suitable, previous work on this subject is summarised, and the reaction discussed in detail. It is then shown that by precipitating quinoline phosphomolybdate instead of the ammonium salt, very accurate results can be obtained by a volumetric method. Few sub- stances present in fertilisers interfere except ammonium salts, which can be destroyed; a large excess of sulphuric acid also interferes.Results by the “official” and the new method are compared and the respective standard deviations are 0.065 and 0.024 per cent. of P,O,. Full working details are given for the accurate determination of P,O, in fertilisers. THERE are two reactions, one or both of which enter into most analytical methods for the determination of phosphate- (1) The precipitation of ammonium magnesium phosphate, which is subsequently converted to magnesium pyrophosphate and weighed. (2) The precipitation of ammonium phosphomolybdate, which may be weighed as such, titrated, or converted to ammonium magnesium phosphate. All procedures involving (l), as almost all accurate or official methods do, are slow, and none are of the highest order of accuracy.There are several reasons for this. The precipitation of magnesium ammonium phosphate is slow, and is not a simple reaction; the precipitate produced may be contaminated with Mg3(P04)2, Mg(NH4),(P0,),, or other substances.l No substances other than MgNH,P04 will be quantitatively converted to the pyrophosphate on ignition, and it is only by purifying the original precipitate by solution and re-precipitation in presence of a small excess of reagent that absence of unconverted salts can be assured. The work of Epperson2 conclusively confirms that double precipitation is essential if accurate results are to be obtained. Accurate or referee methods in which the phosphate is first isolated as ammonium phosphomolybdate invariably continue by conversion of the isolated phosphate to magnesium[Vol. 76 66 WILSON : THE ACCURATE DETERMINATION OF “PHOSPHORIC ammonium phosphate? Prior precipitation as phosphomolybdate from acid solution has the great advantage of separating the phosphate from almost all substances that would interfere with the subsequent application of the magnesia process.Many of these “combined” procedures do not include a re-precipitation of the magnesium ammonium phosphate, and it is probable that a compensation of errors is involved. Many attempts have been made to avoid the use of molybdates by precipitation of magnesium ammonium phosphate from a solution containing citric acid or ammonium citrate, which by formation of complexes prevents interference from calcium, iron or aluminium.It was stated by Epperson2 that presence of citrate caused results to be slightly low and erratic, though only three results are quoted. The “citric magnesia’’ method is accurate, however. This is substantiated by Hoffmann and Lundel14 who recommend it as a method for referee analysis. It has never been popular in the fertiliser trade, perhaps under the influence of the Fertiliser and Feeding Stuffs Act regulations, which prescribe molybdate precipitation and one subsequent precipitation as magnesium ammonium phosphate, and perhaps also because it tends to give lower results. Numerous modifications of the volumetric rnolybdate procedure have been suggested, but none are sufficiently accurate for referee analyses, and all depend on an empirical standardisation of acid and alkali solutions using some supposedly pure phosphate as a standard substance.The “factor” obtained is neither stoicheiometric, nor constant. It is usually also considered that the weighing of ammonium phosphomolybdate is only suitable for small amounts of phosphate, as the errors caused by the variable composition of the precipitate are too great to allow the application of the method to major amounts. Never- theless, it appeared to the author, following his experience with silicomolybdate,5 that a phosphomolybdic acid method could be evolved that would be at least as accurate as the “official” method, and would take only a few hours. REQUIREMENTS OF A NEW METHOD To add to the vast and unnecessary number of variants on the themes noted above, to produce yet another method of analysis for so familiar a substance as “P205”, requires some justification, and the only real justificaticln would be that the new method should have clear advantages over its predecessors.(1) At least as accurate as the “standard” methods. (2) Applicable to a large variety of samples without much modification, and free from interferences. (3) Reasonably simple in manipulation. (4) Quicker than “standard” methods. (6) Truly stoicheiometric, and not dependent on empirical factors. Requirement (4) at once cuts out any procedure in which there is a slow precipitation or re-precipitation, i.e., magnesium ammonium phosphate is inadmissible. Requirement (2) also means that precipitation of magnesium ammonium phosphate is not suitable, as all substances which are precipitated as hydroxides or phosphates in ammoniacal solution must be absent.Requirement (3) suggests tha.t double precipitation or the washing of precipitates with solvents other than water is undesirable. The only known reaction which fulfils requirement (2) is the phosphomolybdate reaction, and by use of quinoline as precipitant, requirements (1) and (5) have also been met. THE PHOSPHOMOLYBDATE METHOD As a t present carried out volumetrically, the phosphomolybdate method has not the qualities of a standard method; the composition of the precipitate varies with the temperature of precipitation, the concentration of ions other than phosphate and the amount precipitated. Usually standardisation is carried out against potassium di-hydrogen phosphate. This is not desirable; because of the influence of other ions present in the sample and not in the standard a bias is introduced (results tend to be high), the factors are empirical and vary with conditions, and the end-point is not very good.It. should be- The formula for ammonium phosphomolybdate is approximately- (NH,) [PO,. 12MoOJ .uzH20, but the precipitate tends to drag down and tenaciously hold other ions, or even acids. The older literature abounds in formulae for this salt, and it is significant that such formulae6Feb., 1951 J ANHYDRIDE” BY MEANS OF QUINOLINE PHOSPHOMOLYBDATE 67 as (NH,),P0,.12Mo0,.2HN0,.H20 are common. Sulphate ions have been said to have a very marked effect, and ammonium sulphate in particular influences the composition of the precipitate.Falk and Sugiura’ considered that a definite compound was formed, to which they gave the formula (NH4),,(P0,),S0,.53M00,, which is unlikely, but indicates the extent of contamination under their conditions. Johnsons comments on the variable quality of the precipitate produced by molybdic acid reagents from different sources, which again points to variation in composition. Allen and Gaults found high results by the volumetric method for samples that had been digested with sulphuric acid to destroy organic matter. Lunge and KeanelO summarise various modifications of procedure that lead to volumetric “factors” which, calculated in terms of 0.5 N sodium hydroxide solution, range from 1.421 to 1640mg of P20,.“As regards fertilisers . . . which contain large percentages of phosphate, the liability of the precipitate to variation under different circumstances appears to the writer to make any modification of the (volumetric method) inadequate when really accurate results are required” (Bernard Dyer, Zoc. &.lo). The rather poor end-point is due to the presence of free ammonia in the solution, according to the equation- (NH,),[P0,.12Mo03] + 26NaOH = 12Na,MoO, + N+HPO, + 3NH3 + 14H,O. Molybdic acid is a weak acid, and ammonia is a sufficiently strong base to affect the indicator used (phenolphthalein) : the extent of interference will vary with temperature and concentra- tion. That the precipitate is a non-stoicheiometric compound is far less likely than that it is impure, as the conditions under which it is formed, viz., simultaneous formation of the complex acid and precipitation of the salt from a strongly acid solution containing relatively high concentrations of other ions, are those most likely to yield an impure precipitate.Entangled nitric acid or other acid would be difficult to wash out, and the carrying down of excess molybdic acid would be even more likely. Unskilled operation can produce a precipitate visibly contaminated with molybdic acid, but who can say how often small but significant amounts of “free” molybdic oxide are carried down, and perhaps “allowed for” in the method of standardisation? A recent report by Bourdon and Cottell states that the ammonium phosphomolybdate does not contain nitric acid, but that on precipitation in presence of excess of ammonium molybdate, the compound is always contaminated by molybdic acid, presumably adsorbed.Under standard conditions the extent of adsorption is more or less constant. Finally, one must consider the solubility of the precipitate: probably even when it is allowed to stand for some time in contact with the mother liquor so that the crystals can grow, some may be lost in washing. This solubility is twofold, (a) the usual physical solubility in aqueous fluids and (b) the decomposition of the precipitate by hydroxyl ions. In gravimetric analysis this is prevented by washing the precipitate with very dilute nitric acid,3 but in volumetric analysis a neutral wash liquid must be used. Potassium nitrate solution is popular for this purpose, but is probably no better than water.So a method can only be based on the phosphomolybdate complex, with its favourable equivalent weight and freedom from interference, if a precipitate can be produced that is (a) less soluble than ammonium phosphomolybdate, (b) of constant composition, free from adsorbed or occluded impurities, and (c) free from cations that will interfere in the titration, if a volumetric process be used. The recent work5 on silica indicates how these desiderata may be attained. The complex acid should first be formed in solution, not formed and precipitated simultaneously. This means that the solution must be free or practically free from ammonium salts before the molybdate reagent is added, and that ammonium molybdate must not be used as reagent.After formation, the phosphomolybdic acid must be slowly precipitated as a salt. Quinoline seems a suitable base ; quinoline phosphomolybdate has a very low solubility and quinoline is a sufficiently weak base not to interfere in the titration. THE QUINOLINE PHOSPHOMOLYBDATE METHOD It was found that quinoline phosphomolybdate has almost all the desired properties. It contains exactly 12 moles of molybdic anhydride per mole of PO4”‘, it is very insoluble, and a volumetric method of very great accuracy has been worked out. This does not depend on empirical standardisation but simply on the reaction- (C,H,N),H3[P0,.12MoO3] + 26NaOH = Na,HPO, + 12Na2Mo0, + 3C9H,N + 14H20, which is quantitatively achieved, so that in this reaction the equivalent weight of P20, is These undesirable features are explicable.68 WILSON : THE ACCURATE DETERMINATION OF “PHOSPHORIC [Vol.76 142.08/52 = 2.732. The solubility of the quinoline phosphomolybdate is negligible and, as will be seen below, the whole procedure is of satisfactory accuracy. The quinoline phosphomolybdate produced is not absolutely pure ; attempts to work out a gravimetric method were not quite successful. Whatever the contaminant, it is neutral and has no effect on the volumetric method; it is probably water retained in the crystal lattice, or it may be a minute amount of adscrbed sodium mollybdate. For work on the usual analytical scale, the volumetric process is so good that there is little cause to regret the failure of the gravimetric process.The reactions are completely quantitative ; by the weighing of aliquots of solutions instead of measuring them a considerable potential source of error has been avoided, and by recourse to weight burettes it is likely that a further advance would be made, but so far that has not been necessary. EXPERIMENTAL The description of the experimental work is divided into three parts, (a) the establishment of the quantitative nature of the proposed method, (b) attempts to establish a gravimetric method and (c) the effect of other elements and anions, and the application to fertiliser analysis. THE QUANTITATIVE NATURE OF THE REACTION-. The best analytical methods are only quantitative within a limited range of conditions; even as simple a matter as the precipitation of sulphate ion as BaSO, is only quantitative when properly carried out and hence it is to be expected that the method underinvestigation will only be accurate if the best conditions for precipitation, etc., can be found.A preliminary experiment was performed by precipitating the phosphate from a known amount of AnalaR potassium di-hydrogen phosphate under conditions similar to those used to precipitate quinoline silicomolybdate. On dissolving the precipitate in standard sodium hydroxide solution, titrating back and calculating the PdO, present from the factor given above, 50.26 mg of P,O, were found, 50.0 mg having been present. This preliminary result was regarded as most encouraging. A series of experiments was then made to ascertain the best conditions for complex formation, precipitation, etc., and a procedure was tentatively specified.The next step was the preparation of standard solutions of phosphate, whose phosphate content should be known with the greatest possible accuracy. Two sources of P,O, seemed most suitable, either dilute orthophosphoric acid or potassium di-hydrogen phosphate. Neither is ideal: phosphoric acid is not a very strong acid, hence its reference to standard substances such as sodium carbonate is not easy; potassium di-hydrogen phosphate is not easy to analyse for all possible impurities, and there is some evidencez for the existence of a more acid compound, KH,PO,.H,PO,, that would be very difficult to detect. On the other hand it has been much used by Sarensen and others as a standard salt for preparing buffer solutions.It was therefore decided to use both these chemicals as standards and to check the accuracy of the method against both. Preparation of 0.5 M phosphoric acid solution-Dilute 56 ml of phosphoric acid (sp. gr. 1.740, and passing the AnalaR specification in every particular) to about 600ml and boil gently for half an hour to ensure that any other phosphoric acids present shall be converted to orthophosphoric acid. Standardisation-The “end-point” for the neutralisation of the first hydrogen ion of H,PO, occurs at pH 4.5, and is much sharper than the second end-point at pH 9 (see, e.g., Smith13). Experiments with indicators showed that the most definite colour change at this pH was given by a mixed indicator containing 6 parts of bromo-cresol green solution (0.4 g per litre) and 1 part of methyl red solution (0.5 g per litre).This indicator is pink at pH 4.4, green at 4.6 and blue at pH 5. Partingtons (Zoc. cit., p. 623) states that the end-point is sharper at 55” C, so titrations were made at that temperature. As comparison solution an approximately 0.5 M solution of potassium diihydrogen phosphate, containing the same amount of mixed indicator as the sample solution, was used. The standard substance was sodium carbonate, freshly prepared by heating AnalaR sodium bicarbonate to 300” C in a platinum dish until constant in weight, and cooling it in a desiccator over fresh phosphoric anhydride; between 1.26 and 1.31-g portions were weighed out, giving titrations of about 50 ml. The usual precautions were taken to remove carbon dioxide at the end-point, and a calibrated chamber burette was used for the titrations.Cool and dilute the solution to 2 litres.Feb., 19511 ANHYDRIDE" BY MEASS OF QUINOLINE PHOSPHOMOLYBDATE 69 The results were 47.40, 47.37, 47.39 and 47.34 g of per litre; average47.375 g of H3P0, per litre. So that weight aliquots could be used as conveniently as volume aliquots, the density of this solution at 20" C was carefully determined by means of a Regnault's pyknometer, of approximately 135-ml capacity, very carefully calibrated at 20" C, a similar vessel being used as counterpoise throughout. As any weighings of the solution would be made in air, the density determined is apparent (i.e., not corrected for the buoyancy of the air).The mean result of two very closely agreeing determinations was- Density at 20/4" = 1.0237. PuriJication of potassium di-hyd~~gen phosphate-According to Pascal14 this salt can be purified completely except for traces of aluminium, iron and calcium by recrystallisation from a boiling saturated solution. Accordingly, 200g of AnalaR salt were dissolved in 400ml of boiling distilled water, and the solution kept on a boiling water-bath for several hours. The clear colourless solution was rapidly cooled with constant stirring, and allowed to stand overnight. The crystals were filtered with suction on a hardened filter-paper, washed twice with ice-cold water and once with 50 per cent. alcohol, and dried in an electric oven at 105" C. The dry salt was lightly ground in a Wedgewood mortar and dried for 8 hours at 105" C with occasional stirring. It was cooled over P,O, and then regarded as pure.Even so, a 0.5 M solution after a few days deposited a few minute flocs of some white insoluble substance, possibly aluminium phosphate (compare Pascal, Zoc. cit., p. 288). This substance was readily dispersed again by shaking, and must have been present in exceedingly small amounts. For most of the work, this pure salt was weighed as such on a semi-micro balance, so avoiding any errors of volumetric measurement. The correct pH for the end-point of the titration-In the volumetric molybdate method one is titrating excess of sodium hydroxide and also hydroxyl ions from the dissociation of tri-sodium phosphate. The end-point for the phosphate part of the titration (Na,P04 to Na,HPO,) is about pH 9; from the data of Travers and Malaprade15 it can be seen that the end-point for the titration of molybdic acid is about pH 8, or rather higher.As there are 12 moles of molybdic acid present per mole of phosphoric acid, an indicator changing at about pH 8.5 is desirable. Thymol blue (pH range 8 to 9-6) is very satisfactory; it is found easier to avoid over-running the end-point by adding some phenolphthalein, which decolorises at about pH 9 and indicates the approach of neutralisation. TABLE I A faint turbidity appeared, so the solution was filtered through a pulp pad. DETERMINATION OF KNOWN AMOUNTS OF PHOSPHATE Phosphoric acid Potassium di-hydrogen phosphate r P,O, present, mg 8.15 13-76 20.67 27-64 34-72 17.13 30.68 47.44 7 I ~- P,O, found, P,O, present, mg mg 8.00 16-65 13.73 31.87 20-50 53-81 27.52 26.02 34-63 42.34 17-20 30.65 47.45 v P,O, found, mg 16.72 31-96 53.53 26-00 42-11 NoTEs-(~) In all the above experiments the amount present was unknown to the analyst.(2) The first five results are low by an average amount of 0.1 mg. This was found to be due to a blank that was too high. In subsequent tests the volumetric solutions used for the blank were one-fifth as strong as those used in the test: it can be seen that the bias in subsequent results is negligible (-0.02 mg) and that the average error is $-0-07 mg. Conditions of fvecipitation-By measuring known volumes of the 0.5 M phosphoric acid solution, precipitating under various conditions, washing and titrating the precipitate, the best acidity and temperature of precipitation were found.Acetic acid was shown to be unnecessary. The exact amount of hydrochloric acid present does not seem to be very important, as long as enough is present to prevent the precipitation of quinoline molybdate; 40 ml of diluted hydrochloric acid (1 + 1) per 100 ml of original solution is very satisfactory, and 50ml will do no harm, By addition of 50ml of concentrated acid to 100ml of the70 WILSON : THE ACCURATE DETERMINATION OF “PHOSPHORIC [Vol. 76 original solution prior to addition of the quinoline, results are somewhat low (e.g., 24.04 mg of P,O, recovered from 25 mg). The phosphomolybdate complex is readily formed at the concentration of 20 ml of concentrated hydrochloric acid per 100 ml of solution, especially when warm.Precipitation of the quinoline salt should take place slowly from boiling solution (see below under “Procedure” for details). The “blank,” which is probably due to silica from the glassware, is important. The aliquots of 0-5 M phosphoric acid solution were weighed in stoppered weighing bottles on an analytical balance, the aliquots of di-hydrogen phosphate on a semi-micro balance. It is considered that these results completely establish the validity of the volumetric method. Reszclts-Table I summarises the results on known amounts of phosphate. ATTEMPTED GRAVIMETRIC METHOD- Numerous attempts to establish a gravimetric method were made, as this would avoid all errors associated with volumetric measurements. The insoluble salt was precipitated as in the volumetric process, washed several times with 1.5 N hydrochloric acid, then washed free from acid with cold water and dried to constant weight. It was ascertained that the precipitate was slightly hygroscopic, but could ‘be dried to constant weight in 2 hours at 105” C ; further drying at 120” C did not result in any further loss of weight.The results were rather disappointing; the precipitates were not quite pure enough, and results calculated on the theoretical molecular weight were high. In Table I1 the weight of precipitate obtained from various weights of P,O, is given, also the “factor” (P,O, per gram of precipitate) and the apparent molecular weight of each precipitate. By comparing this with the theoretical molecular weight, the degree of contamination can readily be seen.TABLE I1 GRAVIMETRIC RESULTS WITH QUINOLINE PHOSPHOMOLYBDATE Weight of P,05 taken, mg 11.31 22.41 31.76 13.99 20.54 14*78* 29.46* 10.51 17-21 27.51 15*06* Average Theoretical . . . . Weight of precipitate obtained, mg 353.6 668.7 992.4 440-0 642.7 462.8 917.8 332.3 540.3 864.8 472.0 . . . . .. .. .. ‘ ‘Factor” 0.03199 0.03202 0.03200 0.03176 0.03 196 0.03194 0.03210 0.03163 0.03185 0.031 81 0.03 191 0.03 19 1 0.03210 Apparent molecular weight of precipitate 2219.8 2218.2 2219.1 2233.6 3222.1, 22 2 3.8 321 3.4 22445 2229.6 2232.5 2226.0 2225.6 2212.8 * In these experiments, the source of P,O, was potassium di-hydrogen phosphate, in the remainder it was 0.5 M phosphoric acid. The nature of the contaminant was not discovered; it cannot be quiiioline molybdate or molybdic acid, or the volumetric method would have been in error ; it is not sodium chloride or hydrochloric acid as chloride could not be detected in the precipitates.It may be water very obstinately retained, or sodium molybdate, but the difficulties of analysing accurately for excess molybdate are very great. Preliminary experiments have shown that the gravimetric method can be used in micro- chemical analysis; up to 4 mg of P,O, the precipitates are of the correct composition. It is of interest that the precipitate is very readily soluble in acetone, but not in ether. ATTEMPTS TO PREPARE CINCHONINE PHOSPHOMOLYBDATE- By using the same technique, but substituting cinchonine for quinoline, a creamy- yellow precipitate is readily obtained.It is precipitated in a very finely divided state, filters badly and is difficult to remove from the sides of the beaker. The molecular weight of 3(C,,H22N,0)H,P04. 12Mo0, is 2708-2 ; in our experiments the molecular weight foundFeb., 19511 ANHYDRIDE” BY MEANS OF QUINOLINE PHOSPHOMOLYBDATE 71 was 2352,2365 and 2353. It is suggested that because of the peculiar shape of the cinchonine molecule, the crystal lattice may be unable to accommodate the theoretical amount of the base. Pyridine was also unsatisfactory as a precipitant because pyridine phosphomolybdate is not sufficiently insoluble. APPLICATIONS OF THE VOLUMETRIC METHOD- Only those ions likely to be important in fertiliser analyses were investigated, viz., (a) lime, (b) fluorine, (c) ammonia, ( d ) iron, (e) magnesia, (f) alkali salts, (g) citric acid, (h) ammonium citrate, (i) nitric acid and ( j ) sulphuric acid.(a) Lime-The presence of 1 g of lime has no effect on the determination (50 mg of P,05 present). (b) Fluorine-In the presence of 2 ml of hydrofluoric acid, results were somewhat high (e.g., 51.5 mg instead of 50 mg). This is due to attack on the beaker, bringing silica into solution, and the effect can be avoided by adding boric acid. For example, to 50 mg of P205 were added 3.5 g of calcium carbonate, 5 g of boric acid and then 2 ml of 40 per cent. hydrofluoric acid. Hydrochloric acid was then added until the calcium carbonate dissolved and the determination was continued as usual; 49.96 mg of P205 were found in one experiment and with a larger quantity of calcium carbonate, 50.08 mg.(c) A mmonia-Ammonia, as might be expected, interferes. For example, in presence of 1 g of ammonium sulphate, only 47.83 mg of P205 were found. The ammonia can readily be destroyed by sodium nitrite, aqua regia or hypobromite. With sodium nitrite, a result of 49-96 mg was obtained; with aqua regia, 49.84mg. In the method finally adopted, hypobromite was used because of convenience and speed. One gram of pure iron was dissolved in a few milli- litres of nitric acid and added to 25 ml of a solution containing 50 mg of P20,; the determined amount of P205 was unaltered. (e) Magnesia-Two grams of MgS0,.7H20 had no effect on the determination of 50 mg (f) Alkali salts-Five grams of potassium chloride and l o g of sodium sulphate added to a 50-mg aliquot of P,05 have no effect on the determination.(g) Citric acid and ammonium citrate-These salts are important in fertiliser analysis. Half a gram of each was dissolved in a 25-ml aliquot (that is equivalent to a 2 per cent. solution) so that it would represent the solution obtained by extraction of a sample with 2 per cent. citric acid or ammonium citrate solution. The 2 5 4 aliquot (of potassium di-hydrogen phosphate solution) contained 49.64 mg of P20,, and the titrations of the solution, the citric acid and the ammonium citrate solutions were identical. (h) Nitric acid-Substitution of an equivalent amount of nitric acid for hydrochloric acid in the analysis made no difference to the results. ( j ) SuZ$huric acid-With this present instead of an equivalent amount of hydrochloric acid results are erratic and high (cj.other workers7s9). Besides the normal yellow quinoline phosphomolybdate, a white solid is also precipitated. This was not analysed, as it was impossible to separate it from the yellow precipitate. Even in the absence of quinoline and phosphate, molybdic acid is slowly precipitated when sodium molybdate solution is heated with dilute sulphuric acid; this does not occur with hydrochloric acid, at least in the concentra- tions used in this analytical method. Precipitation occurs much more readily if some ammonium sulphate is also present. The amounts of SO,” and NH,’ in the precipitate are quite small, and adsorption seems more likely than formation of compounds. If hydrochloric acid is present in amount slightly in excess of the sulphuric acid, the interference is prevented, but the total acidity should not be greatly in excess of 2 N.It is preferable to avoid large excesses of sulphuric acid, and until further investigation has been made, the volumetric molybdate method should not be applied, e.g., to determining phosphorus present in organic substances that have been oxidised by Kjeldahl’s method. COMPARISON BETWEEN THE NEW METHOD AND OTHER METHODS Before using the method it is necessary to be sure that it is free from interference. (d) Iroiz-Iron is without effect. of P,O,. Various samples have been analysed by the new and older methods. Table I11 Thirty-four samples of mixed fertiliser were analysed for water-soluble P20, by two summarises the results.[Vol.76 72 analysts working independently (but using the sarne volumetric solutions). The mean differ- ence between the duplicates was 0.027, corresponding to a standard deviation of 0-024. This may be compared with the standard deviation of the molybdate magnesia method of the Act; two analysts working independently analysed 27 samples of a similar mixture for water-soluble P,O,, and the standard deviaticln was found to be 0.065. WILSON : THE ACCURATE DETERMINATION OF “PHOSPHORIC TABLE I11 COMPARISON BETWEEN METHODS OF ANALYSIS r-__-_-_--h- - Fertiliser and Quinoline Method Sample Feeding Stuffs Act, pliosphomolybdate, Keniarks P206, % P2°5, 9’0 Morocco rock 1 . . . . 33.1 33-29 Total P& 33.29 Morocco rock 2 .. . . Superphosyhate . . . . Mixed fertiliser sample No. 1 Mixed fertiliser sample Solution containing, N, Ditto . . . . . . No. 2 P205 and K,O Calcium phosphate . . 33.0 20.15 20.12 10.72 10.72 10.67 10.69 11.61 11-64 4.88 4.89 5.60 5.27 5.57 5.28 5-56 5.29 5.59 542 44.48 44.51 44.45 44.38 44.52 44.38 32.87 20.29 20.23 10.66 10.66 11.52 11.49 4.81 4-82 5-56 5.57 44.24 44.24 44.24 Total P,O, Total PzO, 10.64 by F. & F.S. 10-66)citrate method Water-soluble P,O, Water-soluble P,O, 5.60 per cent. of P,O, present Total P,@, Finally, several samples of fertiliser were specially prepared by grinding to pass a 200-mesh sieve. Three analysts each received two portions of each sample, and made one analysis of each, with results as shown in Table IV. TABLE I:V Each was well mixed, and divided into six portions.TESTS BY VARIOUS ANALYSTS, USING QUINOLINE PHOSPHOMOLYBDATE METHOD Analyst Sample Mixed fertiliser X . . .. . . Mixed fertiliser I’ . . . . . . Mixed fertiliser 2 . . .. . . Phosphate rock . . . . . . H. N. R. P,!O,, ‘X 11:23 11-19 15.47 15.39 18.62 18.63 33-20 33.15 J . S. p2°5@ % 11.23 11.18 15.35 15.37 18.60 18-65 33-38 33.39 11,-1:3 11.14 15.38 15.34 18.59 18.60 33-23 33.17 METHOD Hydrochloric acid-Concentrated, and diluted to 1 + 9, 0.5 N and 0.1 N solutions. Sodium hydroxide-0-6 N and 0.1 N solutions, both free from carbon dioxide, made from AnalaR pellets. Sodium molybdate solution-A 15 per cent. solution of the hydrated salt in water. (See note 1 below.) Quinoline hydrochloride solution-Add 20 ml of redistilled quinoline (synthetic quinoline has been found preferable to the usual coal tar product) to 800 ml of hot water acidified with 25 ml of concentrated hydrochloric acid, and stir well. Cool to room temperature, add paper REAGENTS-Feb., 19511 ANHYDRIDE” BY MEANS OF QUINOLINE PHOSPHOMOLYBDATE 73 pulp and again stir well.Filter with suction through a pulp pad, but do not wash. Dilute t o 1 litre with water. Boric acid-Pure. Bromine water-Distilled water saturated with bromine at room temperature. Perchloric acid-AnalaR, 60 per cent. Mixed indicator solution-Mix two volumes of 0.1 per cent. phenolphthalein solution with three volumes of 0.1 per cent. thymol blue solution (both in alcohol). PROCED WRE- A. Total P20, in basic slag, phosphate rock, szcperphosphate, etc.(not containing ammonia)- (1) For samples of normal P,O, content, weigh 2.500 g into a 150-ml beaker, add 20 ml of water and 10ml of perchloric acid. (See note 2.) Warm until most of the sample is dissolved, then evaporate until fuming, and continue to heat while gently fuming with the beaker covered for a t least 15 minutes, until attack is complete. Allow to cool, rinse the cover with a little distilled water, allowing the rinsings to run into the beaker, add 20 ml of diluted hydrochloric acid (1 + 9), warm carefully and then boil until all salts are in solution. Filter through a 9-cm Whatman No. 30 filter-paper in a 2-inch funnel into a stoppered conical 125-ml flask, which has been weighed to the nearest 0.5 mg, preferably with a similar flask as a counterpoise.Transfer the contents of the beaker to the filter with the minimum of water, using a wash bottle with a fine jet and a succession of small washes, allowing each t o run through before adding the next to the filter. Wash the filter thoroughly with small washes of warm water in the same way. The total volume of washes and filtrate will be from 100 to 110ml. Discard the filter-paper. Carefully dry the outside, and weigh to the nearest 0.5 mg. Call the weight of solution A grams. Weigh a dry stoppered weighing bottle, 6 x 3 cm diameter. Transfer to the weighing bottle, by means of a pipette, an aliquot that should contain about 50 mg of P20, (but not more than 60 mg), e.g., if the material contains about 20 per cent. of P205, an aliquot of about 10 ml is desirable.Stopper the weighing bottle and weigh again, to the nearest 0.5 mg. Call the weight of the aliquot B grams; then the weight of the sample taken for analysis is 2-5 x B / A g. (3) Wash the stopper of the weighing bottle, collecting the washings in a 500-ml conical flask (this is precautionary as the stopper should not have become wet) and then quantitatively transfer the aliquot to the flask, washing the weighing bottle with about 90ml of cold water. Add 20ml of concentrated hydrochloric acid, then 30ml of sodium molybdate solution. Raise the temperature to boiling, and from a burette with a coarse jet add a few drops of quinoline solution. Swirl the solution in the flask during the addition, again heat to boiling and add quinoline solution drop by drop with constant swirling until 1 or 2 ml have been added. Again boil, and to the gently boiling solution add the reagent a few millilitres at a time, with swirling, until 60 ml in all have been added.In this way a coarsely crystalline precipitate with good filtering properties is produced. Allow the solution to stand in a bath of boiling water or on the edge of a hot plate for 15 minutes, then cool to room temperature. (4) Prepare a paper-pulp filter in a funnel fitted with a porcelain cone, and tamp well down. Decant the clear solution through the filter, and wash the precipitate twice by decantation with about 20 ml of hydrochloric acid (1 + 9). (See note 4.) Transfer the precipitate to the pad with cold water, washing the flask well, and wash the filter and precipitate with cold water, with small washes of about 25 to 30m1, letting each wash run through before applying the next, until the washings are acid-free.(Test for acidity with litmus paper; 6 washes are usually sufficient). Transfer the pad and precipitate back to the original flask (now acid-free). Insert the funnel in the flask and wash it well with water to make sure that all traces of precipitate are transferred; use about 50ml of water. Shake the flask well so that filter-paper and precipitate are completely broken up. From a calibrated burette or pipette run in exactly 50.0 ml of 0.5 N sodium hydroxide solution, swirling the flask during the addition. Shake until the precipitate is completely dissolved. Add a few drops of indicator solution and titrate with 0-5 N hydrochloric acid.The end-point is very sharp; the solution becomes pale green and at the end-point suddenly changes to pale yellow. Record the volume of hydrochloric acid used to within 0.03 ml, and subtract from the volume of 0.5 N sodium hydroxide solution (50.0 ml). (2) Cool the flask, stopper it and shake to mix the contents thoroughly. This need not be accurately measured. (See note 3.)WILSON : THE ACCURATE DETERMINATION OF “PHOSPHORIC [Vol. 76 74 Run a blank on all reagents, excluding only the aliquot of sample solution, but use 0.1 N acid and alkali solutions for the titration a i d calculate it to 0.5 N sodium hydroxide. Subtract this blank from the volume neutralised by the original precipitate (see note 5). 1 ml of 0.5 N sodium hydroxide soluticn = 1.366 mg of P205 and B.Total P,O, in mixed fertilisers containing ammonium salts- (1) Weigh 2.500 g of sample into a 150-ml beaker, add 1 g of boric acid and 20 ml of water. Warm until the boric acid is dissolved. Add 10ml of concentrated hydrochloric acid (see note 6), evaporate to dryness and bake gently for 15 minutes. Allow to cool, moisten the residue with 1 to 2 ml of concentrated hydrochloric acid and add 15 ml of hot water. Warm until the salts are dissolved and proceed as under A (1) from “Filter through a 9-cm Whatman No. 30 filter-paper . . . .” (2) Proceed as A (2) above. (3) Transfer the aliquot to a 500-ml flask a!; under A (3), but only use 70 ml of water. Add 2 g of sodium hydroxide (about 10 pellets) and swirl the solution until the pellets have dissolved.This amount of sodium hydroxide will normally be sufficient; enough should be present to liberate all the ammonia from its salts and leave an excess. Add from 20 to 40 ml of bromine water, according to the quantity of ammonia present (10 ml of bromine water destroys about 20 mg of ammonia). Mix the solution, allow it to stand for 5 minutes, acidify with concentrated hydrochloric acid added dropwise (normally about 6 ml of acid is required) and boil gently to remove excess of bromine. If necessary, adjust the volume to 100 ml, add 20 ml of concentrated hydrochloric: acid, and from this point proceed as under A (3) above. C . “ Water-soluble P205,” ammonium salts absent.- (1) Weigh 10 g of sample into a 500-ml volumetric flask (class A calibration) and add 400 ml of water a t 20” C.(See note 7.) Dilute to the mark with water, mix well and filter through a dry Whatman No. 31 filter-paper. Reject the first 20 to 30 ml and then collect the filtrate. Adjust the temperature of the filtrate to 20” C. With a calibrated pipette (25 or 50 ml, according to the amount of soluble P,O, expected), transfer a suitable sized aliquot to a 500-ml conical flask. Dilute to about 100 ml with water. (2) Add 20 ml of concentrated hydrochloric acid and from this point proceed as under A (3) above. D. LLWater-soluble P205,” ammonium salts @ese&- (1) Proceed as C (1) above. (2) Proceed to the destruction of ammonium salts as in B (3) above, and complete Shake on a machine for exactly 30 minutes. the determination as in A (3) above.NOTES ON THE PROCEDURE- 1. The sodium molybdate solution should not be kept too long, as it tends to dissolve silica from the glass bottles. 2. Some samples of phosphatic minerals have been found to be very resistant to attack by hydrochloric acid, but so far we have come across no samples that did not yield all their phosphate to an attack with perchloric acid. This procedure has also the advantage of completely removing fluorine, hence precautions to prevent later attack on beakers (addition of boric acid) are not necessary. 3. Volumetric measurements cannot be as accurate as weighing, so for work of great accuracy it seems desirable to eliminate measurement of volume as far as possible. It would be possible to weigh an aliquot of 100 mg or so direct on a semi-micro balance, but in non- homogeneous materials there is always some doubt as to how truly representative a very small sample may be.With a modern aperiodic balance taking a 200-g load, the procedure outlined here is quite speedy and reduces any error in taking the aliquot to negligible pro- portions. For referee work, it is probable that use of weight burettes for all “volumetric” work, the solutions being standardised on a kilogram basis instead of a litre basis, wouldFeb., 19511 ANHYDRIDE” BY MEANS OF QUINOLINE PHOSPHOMOLYBDATE 75 result in a further gain in accuracy at the cost of a very small loss in time and convenience. Errors due to changes in temperature would vanish. 4. The preliminary washes with acid remove most of the excess of quinoline and molybdate, and prevent errors through the precipitation of quinoline molybdate or molybdic acid in the pad or precipitate.5. The “blank” is important; it is mostly due to silica, and must be carefully determined. It should not be more than about 0-4 or 0-5ml; about 0.2ml comes from the sodium molybdate and the remainder from the sodium hydroxide and the glass apparatus. Soft soda or potash glass must not be used. Flasks that become scratched or etched must be discarded; occasionally a flask that has been satisfactory may begin to yield appreciable quantities of silica. 6. This treatment is satisfactory in rendering silica insoluble; perchloric acid is not used, as the low solubility of ammonium perchlorate in water (and its rather unstable nature) may give rise to trouble.The addition of boric acid is essential if fluorides are present, and should precede acidification with hydrochloric acid. 7. “Water-soluble P205” has only one meaning in fertiliser analysis-as defined in the Fertiliser and Feeding Stuffs Act regulations. The regulations specify 20g of sample and final dilution to 1 litre. Use of 10 g, as described, is satisfactory, and 500-ml flasks are easier to accommodate in shaking machines. The time of shaking must be adhered to, and filtration must take place immediately, because of possible slow reactions between the solid phase and the solution. We have not found that silica in appreciable quantities passes into solution during this treatment, but if a fertiliser of alkaline reaction were examined, silica might be dissolved.Acknowledgments are made to Mr. H. N. Redman, who carried out most of the experi- mental work. REFERENCES 1. Bassett, H., “Theory of Quantitative Analysis,” pp. 81-88. 2. Epperson, A. W., J . Amer. Chem. SOC., 1528 50, 321. 3. Fertiliser and Feeding Stuffs Act Regulations, 1932. 4. Hoffmann, J. I., and Lundell, G. E. F., J . Res. Nut. Bur. Stand., 1937, 19, 69. 5. Wilson, H. N., Analyst, 1949, 74, 243, 6. Partington, J. R., “Inorganic Chemistry,” 4th Ed., p. 940. 7. Falk, K. G., and Sugiura, K., J . Amey. Cliem. SOC., 1915, 37, 1507. 8. Johnson, C. M., “Chemical Analysis of Special Steels,” 4th Ed., p. 315. 9. Allen, H. R., and Gault, L., J . Ass. Off. Agric. Chem., 1947, 30, 135. 10. Lunge, G., and Keane, C. .4., “Technical Methods of Chemical Analysis,” 2nd Ed., Vol.111, 1931, 11. Bourdon, D., and Cotte, J., Bul. SOC. Chim. Fv., 1949, 16, 429. 12. Parker, E. G., J . Phys. Chem., 1914, 18, 653; Chem. Abst., 1915, 9, 168. 13. Smith, T. B., “Analytical Processes,” 2nd Ed., p. 183. 14. Pascal, P., “Traiie’ de Chimie Minkvale,” Vol. 6, 1934, p. 287. 15. Travers, A, and Malaprade, Bul. SOC. Chim. Fr., 1926, 39, 1413. pp. 536-538. IMPERIAL CHEMICAL INDUSTRIES LIMITED RESEARCH DEPARTMENT BILLINGHAM, Co. DURHAM DISCUSSION MR. L. GANELLIN asked whether the author had had any experience in the determination of phosphate in basic slag by the quinoline method, and in particular, with citric acid soluble P,O,. It was his experience that silica soluble in the citric acid extract interfered in molybdic precipitates and had t o be removed by the standard precipitation method.MR. WILSON replied that he had not had any experience with citric-soluble P,O, in basic slag by this method, but he thought that silica, if extracted, might interfere. This interference could probably be avoided. The yH range over which silicomolybdic acid was formed was narrow and, a t the acidity prescribed, conversion of silica to silicomolybdic acid would be far from complete. Moreover, it was only from fresh unpolymerised silicic acid that silicomolybdic acid was formed. If to the aliquot of citric acid solution taken for analysis, 20 ml of concentrated hydrochloric acid were added, as in the usual procedure, and the mixture was gently boiled for half an hour and then diluted to the proper volume before addition of the sodium molybdate solution, i t was improbable that silica would interfere.The author had pointed out the high standard that he had set himself, in particular in that he hoped the method would not embody any empirical factors. With this in mind, would i t not have been more desirable to offer the volumetric procedure merely as a quicker alternative to the gravimetric procedure; if SO, was MR. R. C . CHIRNSIDE complimented the author on a lucid account of an elegant piece of work.76 WILSON [Vol. 76 the constitution of the quinoline phosphomolybdate precipitate clearly known ? Had the author considered the possible application of X-ray diffraction methods to i3 study of this precipitate? It might well be that a critical examination of the patterns so obtained would throw light on the nature of the attachment of the water that the author believed the precipitate contained.It would show whether the water was in the crystal lattice or was merely firmly adsorbed, and so enable conditions for its complete removal to be found. Similarly, some light might be thrown on the nature of the cinchonine precipitates that the author mentioned earlier in the paper. MR. WILSON replied that i t was likely that X-ray diffraction methods would prove valuable in the investigation of the persistent moisture in the precipitake. He agreed that a gravimetric method would have been the most desirable but, as was clearly shown in Table 11, the gravimetric results were not quite good enough, whereas the volumetric method gave acceptable results.Whilst i t was still possible that a Favimetric procedure might be evolved, he thought t h a t the investigation of weight titration was more likely to lead to a method of the highest possible accuracy. MR. J. G. SHERRATT enquired whether the method was applicable to organic fertilisers and feeding stuffs, such as animal residues containing bone, and if so, what procedure was recommended for the destruc- tion of organic matter. Hitherto he had invariably used sulphuric acid to oxidise organic matter prior to the determination of P,O,, but this involved the su'bsequent presence of ammonium sulphate and an excess of sulphuric acid in the final solution, both of which interfered with the proposed new method. MR. WILSON replied that i t was preferable not to use sulphuric acid, but if i t were used as much as possible should be evaporated, and the remainder neu-tralised by sodium carbonate before the addition of hydrochloric acid.Although there might be hesitation about the use of perchloric acid with organic matter, this mixture had been used in the wet oxidation of organic compounds (see, e.g., J . A s s . 08. Agric. Chem., 1943, 22, 182, or Shirley Institute Memoirs, April, 1949, Vol. 23). Hamlin, in the latter of these papers, described the successful destruction of up to 5 g of cotton yarn with a mixture of 14 ml of perchloric acid, 20 ml of nitric acid and only 1 ml of sulphuric acid. MR. C . G. DAUBNEY asked, in view of the reliance placed on AnalaR chemicals, whether the author would disclose the difference between AnalaR phosphate and the recrystallised phosphate used as his standard material to prepare the standard phosphate solution.MR. WILSON said that in this investigation he had wished to neglect no possible precaution, and there- fore had purified the material. DR. J. H. HAMENCE said that workers in his laboratory had spent many years trying to evolve a really efficient volumetric method for the determination of P,O,, but without any real success, and he therefore would like to congratulate Mr. Wilson on his new method. In the absence of interfering substances, the St. Gobain method had been found to give very satisfactory results. The St. Gobain method was based on the well-known Pemberton method, with the difference that neutral formaldehyde was added for the final alkali titration; this gave a very much sharper end-point, since i t removed the ammonia radical. This procedure gave excellent recoveries with simple phosphates and with water solubles from super- phosphate, but unfortunately the process broke down in the presence of fluorine and silica. Although many modifications of the method had been tried, he had never been successful in avoiding interference from silica other than by the somewhat lengthy process of removing silica prior to solution in acid. In view of this difference, he would be pleased if the author would amplify the procedure that he employed to avoid interference by fluorine and silica, particularly in substances like basic slag and apatite phosphate. MR. WILSON said that, for compounds like apatites that were soluble only with difficulty, he thought the treatment with perchloric acid was best. He had come across no instance in which fuming a finely ground sample with perchloric acid did not bring into solution the whole of the P,O, in reasonable time. This treatment also removed hydrofluoric acid. If trea.tment with hydrochloric acid was preferred, boric acid should first be added to combine with the fluorine. Mr. Wilson added that he thought the relative freedom from interference of his method as compared with the ammonium phosphomolybdate method was explained in part by the difference in crystal structure of the precipitates; whereas ammonium phospho- molybdate was cubic, the quinoline compound was of a much lower class of symmetry] and i t might well be due to the different crystal lattice that the precipitate crystallised pure, without including or absorbing foreign ions from the solution. It was possible that a mixture of nitric and perchloric acids could be used.