124 SCHOELLER AND POWELL : ON THE ALKALIMETRIC ESTIMATION OF CERTAIN DIVALENT METALS IN THE FORM OF TERTIARY PHOSPHATES, WITH ESPECIAL REFERENCE TO THE VOLUMETRIC DETERMINATION OF COBALT AND NICKEL. BY W. R. SCHOELLER, PH.D., AND A. R. POWELL. (Read at the Meetiiag, April 5, 1916.) THE development of technical inorganic analysis has resulted in the gradual replace- ment of gravimetric by volumetric methods; and if precipitation methods can be superseded by the niore scientific processes of oxidation, reduction, or saturation, a further advance will have been accomplished.External indicators and unstable solutions are the rule in the first, the exception in the latter, group of methods. The former are generally empirical, hence assay and standard must be titrated under identical conditions, whereas the latter are amenable to stoichiometric calculation, which allows a variety of elements or compounds to be estimated on the basis of a common standard.This applies especially to saturation methods, the most con- venient primary standard for which is sodium carbonate. This mode of analysis has not been utilised to any extent for the determination of metals other than alkalis and alkaline earths.Recently, however, it mas applied by Miles ( J . Chem. SOC., 1915, 107, 988) to the determination of lead. Stolba’s (Zeitsch. anal. Chem., 1877,16, 100) method for the estimation of phosphoric acid, based upon the conversion of magnesium ammonium phosphate into acid phosphate in presence of an indicator of the methyl orange group, as expressed by the equation SMgNH,PO, + 2H,SO, = MgH,(PO,),+ QSO, + (NH,),SO,, can be used for the determination of magnesium, and the method has been extended to zinc in zinc ammonium phosphate.We have investigated the alkalimetric determination of some other metals capable of yielding tertiary phosphates of the general type M”NH,PO,, which are obtained as crystalline precipitates, and there- fore easily filtered off and washed.The results obtained are given below, together with brief notes on the published methods for magnesium and zinc. Magrtesium.-Magnesium is the only metal that is always separated as the double phosphate of the type here discussed. The volumetric process obviates the bedious ignition to pyrophosphate. Instead of washing the precipitate with dilute ammonia, and partially drying the filter to get rid of the excess of the latter (Low, “ Technical Methods of Ore Analysis,” 5th edition, 1911, 159), we prefer to displace the ammoniacal washing solution with 50 per cent.alcohol (Classen, “ Theorie und Praxis der Massanalyse,” 1912, 214), after removing the filtrate and washings from under the funnel and substituting a flask containing acidulated water, so as to prevent, re-absorption of ammonia vapour by the wet filter.The precipitate may be rinsed back and titrated direct with &-or acid in presence of cochineal, or it may be dis- solved in a, measured quantity of the acid and the excess of the latter titrated back with alkali, adding methyl orange, or preferably cochineal.THE VOLUMETRIC DETERMINATION O F COBALT AND NICKEL 125 Zinc.-We need not enumerate the drawbacks of the ferrocyanide and sodium sulphide methods, both typical examples of precipitation analysis.The alkalimetric determination of zinc has been described by Walker (Sutton, “Volumetric Analysis,” 9th edition, 1904, 359) ; and though the precipitation as phosphate (Treadwell-Hall, (( Analytical Chemistry,” Part II., 3rd edition, 1912, 140) involves a sIightly more elaborate preliminary treatment than the ferrocyanide method, the superiority of the subsequent titration over the ordinary zinc assay will, we trust, lead to its more general adoption.Cad?nium.-What appears to be the only volumetric method so far published is the permanganate titration of the ferrous salt produced by the action of an acid solution of ferric sulphate upon cadmium sulphide (Sutton, Zoc.cit., 167). I n the writers’ experience this type of titration, applicable to several metals, has not found general favour in practical metallurgical analysis. In gravimetric work cadmium is often weighed as sulphate (Low, Zoc. cit., 61). By precipitating a neutral solution of the sulphate with a large excess of di-ammonium phosphate, cadmium ammonium phosphate is obtained as a silky crystalline precipitate, which may be dried at 1 0 5 O C.and weighed as CdNH,PO,+H,O, or ignited to pyrophosphate (Miller and Page, School of Miiaes Qunyterly, 1901, 22, 301.) On account of the high atomic weight of cadmium, we used A acid (1 C.C. = 0.00562 grm. Cd.) to dissolve the phosphate precipitate, the excess being titrated back with FG alkali in presence of cochineal, because this indicator is more sensitive to decinormal solutions than methyl orange (Classen, loc.cit., 216), and its colour increases iu intensity as the end-point (tie., alkalinity) is reached. (For the same reason we give preference to methyl orange when titrating to acid reaction.) The results obtained are giveu in Table I.: TABLE I. Number. 1. 2. 3. 4. 5. 6. 7. e. 9. 10. Cadmiurn - Take 11. 0.0074 0.0263 0.0295 0.0527 0.0637 0-0779 0.0790 0.1053 0.1147 0.1216 Found. 0.0075 0.0258 0.0298 0.0529 0.0638 0.0780 0.0787 0.1051 0.1160 0.1234 Error. + 0~0001 - 0.0005 + 0.0003 + 0~0002 + 0~0001 + 0*0001 - 0*0003 - 0*0002 + 0.0013 + 0.001.8 The figures are in close agreement, except Nos.9 and 10. This is accounted for by the progressive decolorising effect of increasing quantities of cadmium on cochineal, which renders the end-point less distinct; in presence of about 0.1 grm. of126 SCHOELLER AND POWELL : cadmium in 100 to 150 C.C. bulk, the solution, instead of beeoming violet, turns to a pale claret colour. With smaller quantities of cadmium the end-point is easily per- ceptible, hence it is advisable to have less than 0.1 grm.present in the assay. It is interesting to note that, unlike cadmium, a neutral solution of zinc sulphate strikes a deep pink coIonr on addition of a few drops of cochineal tincture, Manganese.-The two permanganate methods (Low, Zoc. cit., 162, 167) for manganese are satisfactory, although as regards Volhard’s method the standardi- sation of the permanganate solution has not yet been quite satisfactorily decided (Classen, Zoc.cit., 344). The precipitation as manganese ammonium phosphate (Treadwell-Hall, Zoc. cit., 126), though not advantageous in the cases of ores, is useful in the analysis of manganese salts. By treating the precipitate as described under cadmium we obtained results shown in Table 11.TABLE 11. Manganese- Error. Taken. Found. Number. 1. 2. 3. 4. 5. 0.0124 0-0310 0.0620 0.1240 0.1860 0.0131 0.0307 0.0616 0.1229 0.1846 + 0-0007 - 0.0003 - 0*0004 -- 0*0011 - 0*0014 The results are a trifle low, which we believe to be due to a, slight loss in the washing. Cobalt.-The precipitation of cobd t as cobalt ammonium phosphate was first described by Dirvell (C‘hem.News, 40, 268). Directions for carrying out the process have been published by Clarke (Chem. News, 48, 262) and by Hope (Beringer, “Textbook of Assaying,” 12th edition, 1910, 442). Though for full details these references should be consulted, a brief comparison of the two methods appears desirable, as we found the properties of cobalt ammonium phosphate to vary according to the procedure followed.Clarke adds an amount of ammonium phosphate equal to five times the combined weight of cobalt and nickel, and five parts of hydrochloric acid to one of ammonium phosphate. Ammonia is cautiously added to the boiling solution till the precipitate redissolves, and the liquid vigorously stirred to precipitate the cobalt.Hope directs adding nearly the same excess of ammonium phosphate, but no hydrochloric acid or ammonium chloride. The boiling solution is treated drop by drop with dilute ammonia whilst continually stirring until the blue amorphous precipitate at first produced becomes pink, The essential difference between the two methods lies in the quantity of ammonia added, while the properties of the resulting precipitates differ as shown below :THE VOLUMETRIC DETERMINATION OF COBALT AND NICKEL 127 Clarke’s Method.Coarsely granular. Deep lilac. Precipitate generally nickel-free, but fil- trate usually contains a little cobalt. Precipitate slowly soluble in acid, which must be added in excess and titrated back (indirect titration). Hope’s Method. Silky crystalline.Pale lilac. Filtrate cobalt-free, but precipitate usually Precipitate instantly soluble in acid; carries a little nickel. suitable for direct titration. Indirect Titration.-The titration with acid and alkali did not prove a com- plete success. In presence of more than 0-05 grm. of cobalt, methyl orange was useless, the pink colour of the solution masking the end-point ; nor was the appear- ance of the bluish precipitate that marks the end of the reaction sufficiently pronounced to serve as an indicator.Nickel sulphate as a decoloriser proved dis- appointing, the liquor assuming a dirty-grey tinge, through which the delicate transition-point was difficultly perceptible. With lacmoid the change of colour could not be determined with sufficient accuracy. Cochineal was found to be the most serviceable indicator, but the results were always slightly high, as shown in Table 111.TABLE 111. Number. 1. 2. 3. * 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Indicator. Cobalt- Taken. 0.0202 0.0280 0.0493 0.0250 0.0409 0.0479 0.0499 0,0562 0.0759 0.0784 0.1048 0.1124 0.1406 0.1687 Found. 0*0210 0.0285 0.0497 0.0256 0.0414 0.0490 0.0511 0.0575 0.0755 0.0799 0.1065 0.1135 0.1424 0.1699 Error.+ 0.0008 + 0*0005 + 0-0004 + 0.0006 + 0.0005 + 0*0011 + 0.0012 + 0.0013 - 0.0004 + 0.0015 + 0.0017 + 0.0011 + Oo0O18 + 0.0012 We ascribe this positive error to the action of cobalt on the indicator, having made the observation that the addition of a few drops of a neutral cobalt solution to distilled water tinted with cochineal produces a deep pink colour which requires * Mean of four titrations, error varying from nil to +0*0006.128 SCHOELLER AND POWELL : a drop of Hence in titrating the excess of acid the colour changes to purple before the solution is actually alkaline. Although, roughly speaking, the error tends to increase with the amount of cobalt, we were unable to establish a definite relation between the magnitude of the error and the quantities of metal or indicator used.For this reason, and because the end-point is not easily perceptible in presence of a large amount of cobalt, we prefer the direct method described below. Direct Titratio%.-The filter containing the precipitate is spread against the side of the beaker, rinsed down with water followed by a few drops of the acid used in the titration, and again with water to displace the acid, taking care not to use more than about 30 C.C.in all. The filter is now discarded, and the addition of acid con- tinued until the precipitate disappears. No indicator is required, for as long as any but a, small quantity of precipitate remains it imparts a lilac colour to the liquid, while the last trace is sufficiently conspicuous by the light reflected on the glittering scales to determine its disappearance within 0.1 c.c., even in presence oJf suspended filter fibres.The advantage of this method over the preceding one lies in the fact that large quantities of cobalt can be thus determined, provided the cobalt is precipitated in the modification instantly soluble in acid ; and as this is easily cou- taminated with nickel if present to any extent, double precipitation must be resorted to.acid to discharge. Table IV. shows results obtained by the direct method. TABLE IV. Number. 1. 2. 3. 4. 5. 6. 7. 8. 9. Cobalt- Taken. 0.0402 0.0954 0.1159 0.1164 0.1447 091616 0.1818 0.1902 0.3022 Fouiid. 0.0405 0,0963 0.1160 0.1149 0.1445 0.1598 0.1808 0.1901 0-3023 i- 0.0006 + 0.0009 + 0~0001 - 0.0015 - 0-0002 - 0*0018 - 0~0010 - 0~0001 + 0~0001 NickeZ.-A method for separating nickel and cobalt, based upon the conversion of cobalt into a cobaltarnine salt and precipitation of the nickel as nickel ammonium, phosphate, has been worked out by Clarke (J.Xoc. Chem. Ind., 1896,25,866). We did not investigate the application of the alkalimetric method to nickel ammonium phosphate, because the precipitate is less insoluble than the other phosphates discussed in this paper, and the accuracy of Moore’s cyanide titration (Sutton, Zoc.cit., 251, 252) could not be exceeded ; but we desire to call attention to the factTHE VOLUMETRIC DETERMINATION O F COBALT AND NICKEL 129 that by precipitating cobalt as cobalt ammonium phosphate a feebly ammoniacal filtrate is obtained in which the nickel can be titrated by cyanide without further manipulation.This advantage does not seem to have been hitherto realised, for Mellor (“ Treatise on Quantitative Inorganic Analysis,” 1913, 400) thus sums up the present position of the volumetric determination of the two metah : u The methods which have been suggested up to the present for the determination of cobalt volumetrically are not satisfactory except under special conditions, and hence no volumetric method for cobalt has won a place in general practice.Volumetric methods for nickel also fail in the presence of cobalt, although the cyanide process is quite good when cobalt is absent, or only present in minute quantities.” We verified the accuracy of the nickel titration after separating cobalt as the double phosphate, and obtained very satisfactory results (see Table V.).The nickel filtrate, after being left to cool, should be titrated without undue delay, as it gradually deposits a pale green transparent crystalline precipitate, practically insoluble in ammonia, and containing nickel and phosphoric acid. Once formed, it requires a, fair excess of hydrochloric acid for its solution, after which the liquid may be made slightly amrnoniacal, cooled, and titrated in the usual manner.TABLE V. Number. 1. 2. 3. 4. Nickel Taken. 0.0356 0-0630 0.0856 0-1196 Cobalt Added. 0.0799 0.0983 0.0574 0.0843 Nickel Found. 0,0357 0.0630 0.0855 0.1191 Error. + 0~0001 0~0000 - 0~0001 - 0.0005 The superiority of the phosphate process over all the other methods of separating nickel and cobalt lies in the fact that it allows a separate volumetric estimation of the two metals to be made, and that both are obtained in the form directly suitable for titration. If a small quantity of cobalt has escaped precipitation, due to the use of an excess of ammonia, it is at once detected by the darkening of the liquid in the nickel titration, while the colour of the cobalt precipitate is a sure indication of con- tamination with nickel : if bright lilac, the separation has been properly carried out ; but if of a dirty pink colour, too little ammonia has been added, and it will be necessary to dissolve in a little acid and repeat the precipitation, which can be done in a few minutes.The atomic weights being practically the same, nickel may be reckoned as cobalt and vice versa.Table VI. gives a few results obtained by the process; in carrying out the separation we usually add 5 grms. of ammonium chloride, and always reprecipitate the cobalt, unless present in small quantity, adding more ammonia in the first than in €he second precipitation. In this manner a good separation is effected in the first operation, while the second yields a precipitate which can be titrated with acid alone, no indicator being required.130 SCHOELLER AND POWELL : TABLE VI.Number. 1. 2. 3. 4. 5. 6. Cobalt- Taken. 0.0130 0.0180 0.0520 0.0520 0.0624 0-1043 Found. 0.0135 0.0177 0.0524 0,0524 0.0636 0.1059 Error. Nickel- Error. Taken. Found. + 0.0005 + 0.0004 + 0.0004 f 0.0012 + 0.0016 - 0.0003 0,0175 0.1168 0.0519 0.0519 0.0554 0.0088 0.01 78 0.1 175 0.0520 0.0517 0.0554 0.0092 + 0-0003 + 0.0007 f 0~0001 - 0.0002 0~0000 + 0-0004 The greater part of the experimental work recorded above was carried out in the metallurgical laboratory of the Sir John Cass Technical Institute.DISCUSSION. Mr. G. N. HUNTLY said that, although the phosphate method was of considerable interest, it seemed not to have come into general use.In the determination of magnesia the phosphate precipitate, after being washed with ammonia, required very thorough washing with alcohol, and provided this was duly carried out the method was quite satisfactory. Like the present authors, he had found cochineal to be the best indicator for the purpose.For the determination of zinc the method was exten- sively used in America, where, however, it was the practice to weigh the precipitate. The zinc, was separated first as sulphide, and redissolved, being then thrown down as the double phosphate, which was weighed, either after drying or (which was more satisfactory) after ignition. If the phosphate precipitate could be readily and accu- rately titrated, he thought this would be found more satisfactory than weighing the precipitate, but there was some little difficulty in connection with the question of its composition.In America the theoretical factor based on atomic weights was used, while Bell and Ibbotson recommended an appreciably lower factor, and he had found that the lower factor seemed for some reason to give the more correct result.For the determination of manganese it was also the practice in America to weigh the precipitate, but he believed that ice-water had to beused for washing it, which intro- duced rather a complication. The bismuthate method for medium quantities and the persulphate method for small quantities were so rapid and accurate that even this method hardly seemed likely to displace them.With regard to cobalt and nickel, it was to be noted that the authors had started with pure solutions; one would like to know in what form the cobalt and nickel would be separated from a, mixture before the phosphate method was applied, as any other metals that might be present would also be thrown down by the phosphate. For nickel the cyanide titration method was largely used, but the glyoxime method was so easy and accurate that it would probably be dificult to displace it.Dr. J. T. DUNN said that he had tried the phosphate method some years ago forTHE VOLUMETRIC DETERMINATION OF COBALT AND NICKEL 131 the determination of zinc, but had given it up on account of the difficulty of seeing the end-point. With regard to nickel and cobalt, he should, like Mr.Huntly, be interested to hear whether the authors had done more than merely apply the method to pure salts or mixtures thereof, and how, in the determination of nickel and cobalt in, for instance, steel, they would prepare for the application of the phosphate method. The PRESIDENT remarked that the question of separation of nickel and cobalt seemed to be a particularly difficult one.I t seemed likely that in the future an im- portant part would be played, not only by nickel steel, but also by cobalt steel, and in that event a simple volumetric method for the determination of cobalt would be a great desideratum. Dr. SCHOELLER, in reply, said that the manganese precipitate was slightly soluble in water, and to a much smaller extent in ammonium salts. A neutral ammonium salt like ammonium nitrate must, therefore, be used for washing, in default of which the result would be slightly low. I t might be mentioned that the application of the method to cadmium was as new as its application to cobalt. He did not think that any other volumetric method except the sulphide titration had yet been proposed in the case of cadmium. The separation of nickel and cobalt from other metals, especially manganese, zinc and iron, was certainly a difficult problem. Clark and Hope had started with pure nickel and cobalt salts, and the present in- vestigation had been commenced at the point where these chemists left off, the volumetric method having been applied to solutions containing nickel and cobalt salts only. In Hope’s method the iron was separated as basic acetate, and the nickel and cobalt precipitated as sulphides in a solution weakly acid with acetic acid. The sulphides were dissolved in uquaregia, and this solution was used for the phosphate precipitation. That method of separation was not, however, very satisfactory. Mr. Powell and he had been experimenting for some time with a new method of separating cobalt and nickel, but they were not yet able to record any definite results.