2 24
CHAS. B. D U D L E Y AKIJ F . N. PEASE.
results obtaitied on the three steels by the two methods, as follows : >‘t aiid A s s i s t a i l t Chelniit, Pennsylvania Railroad. .kltoona, 1%
0
N E of the most coninion methods of separating phosphorus
from iron, either in ores, pig iron, wrought iron or steel. is by means of molybdic acid, the separation giving rise to the well-known yellow precipitate of ammonium phosplioi7iolybdate. The subsequent treatment of this )-ellow precipitate is very varied. Some chemists prefer to dissolve the precipitate in ammonia, and then determine the phosphoric acid by means of inagnesia mixture. Others prefer to weigh the yellow precipitate just as obtained, and some even to measnre the volume of 1
Read a t the Baltirnore ineetiilg. Dectiiiher
25, ’by3
DETERMINATION OF PHOSPHORUS IN STEEL.
2 2j
this yellow precipitate in a specially graduated apparatus. Perhaps however, by far the largest number of chemists, prefer to treat the yellow precipitate by some volumetric method, either what may be called the alkali method, described by Handy in the Proceedings of the Engineer’s Society of Western P e m a . , 8, 78, or the perhaps better known permanganate method, described by Emnierton, published in the Trans. A . I. dkf. E., 15, 93, and subsequently modified by Wood, Shinier, Drown, Jones, and others. W e use the perniangaiiate method. This method as we prefer to use it, and with all the precautions which we deem essential to secure uniform and accurate results, except as to the interference produced by arsenic, has been published in the /. A n a l . APpZ. Cheni., 7, 108, and also in the /. Am. Chem. SOC.15, 519. What follows is a discussion of some points suggested by the work done preliminary to drawing up the method above referred to. In order that what is given may be clearly understood, it may not be amiss to state that by this method, the steel is dissolved in nitric acid of 1.13 sp. gr., the solution boiled a few seconds. and then treated with potassium permanganate to destroy carbonaceous matter, and possibly secure complete oxidation of the phosphorus ; the separated manganese oxide, reduced by a few crumbs of ferrous sulphate, the solution then heated to a definite temperature, and a measured volume of molybdate solution, which is likewise at a definite temperature, added, the vessel enclosed to prevent loss of heat, and shaken vigorously for five minutes, allowed to stand a few minutes for the precipitate to settle, and then filtered. After thorough washing with acid amiiioniuni sulphate wash water, the yellow precipitate is dissolved in ammonia, the solution treated with a definite amount of sulphuric acid, and then passed through the reductor, and finally titrated with potassium permanganate of known strength, the amount of phosphorus being ultimately obtained by niultiplying the number of cubic centimeters of permanganate used by the proper factor. I t is obvious that the accuracy of the determination b\- the method briefly outlined above, depends on several conditions : I. Is all the phosphorus i n the steel taken to start with
actually iii tlie y e l l o ~precipitate ohtainect, or i n other \ y o r ~ l s . can phosphorus lit. coiiipletel!. separate(1 iroiii iron 1)~. iiioly1)tlic :wid? 1-p” this point we have done i i o \voi-k, :inti caiiiiot express : ~ i i opiiiioii. Tlie late J . I.,a\vi-c.iice Sniitli i i m i to cl:iiiii tlint the cmiiplcte separation of \.e!-! ,siiiall riijioLiiit.4 oi’ 11110sl~lioriisfroiii lnrgc aiiiouiitb of iroii 11!. iiicaiis oi niolJ-lxiic :icitl \\’as :it 1c:ist tloulJtfu1. a r i d t h a t tlic iiicist i-clia!ile pr(~ce(1~irt: twiisistecl in iirst conceiitratiiig the i)iiospliorus into : vel-) siiiall :iinouiit of iron, aiitl tlieii ,separatiiig it iroiii this iroii ! I > , 1iic;1iis of ~ ~ i o l > ~ h ncitl. tlic Assuming. lio\\.e~.cr.tiiat all tlic pli(is~)liorus i n tlie steel is i i i the ellow precipitate. he1 era1 further qiiestions :iris?. 11. Is the !.ellon precipitnte of coiistniit conipositioii, that is, i,% it iiit!el)entlelit of the conditiolis wider 11.liich it is formed ? It is oi)\.ioiis tlint this is :I very iniportant poiiit for all those iiietliotls \\.liicli either weigh or iiieasure the yello~v1xecipitate. :iiitl also for tlie permnrigaiiate volumetric iiiethotl. \vliich iii realit!. actuall!- iiieasures not tlic pliospliorii~or plii).;phoric aciti in tlie yelloiv precipitate hut rather tlic niol>,lidic ncitl. tlie ~ ~ l i o s pliorus k i n g determined from its relatioil to the iiiol!.btlic acid. If now the yellow precipitate is not ili c~JllStnlltconiposition, or iiiore clefinitel!. i f tlie relatioii hetiveeii tlic. Ijliospliorus a i i t l iiiolybdic acid in this preci1)itate is iiot a coiibt:iiit (.)iic, l)ut is affected 1). tlie conditiolis uiitier ivliicli the precipitate i b foriiicd~ it is clear that these coriditioiis must lie tlefiiiitel!. clefiied, aiitl aIwa!.s closely followed, or there will be great uiicertaiiity i n tlie results. \Vhat then arc tlic fact3 i n tlie r a s e ? It is \\.ell kiio~viithat almost froni the l q i i i i i i i i g of the iise of i n o l ~ ~ l i dic acid i n phosphorus tleteriiiiiiatioiis there has heeii iiiore or less uncertainty as to the constitution of tlie ycllou. precipitate. ant1 especiall!. that tlependiiig on the coiiditioiis, it is very liable to be coiitaminated with free mol\.btlic acid. Furtlieriiiore, Eiiiniertoii has apparently deiiioiistrnted i n tlie place cited abo\.e, that the temperature of prccipitation has :i ver!. important influence on tlie final result. H e also c1:iiiiis tliat t h e aniouiit of free acid in the solution. the coiicentmtioii of the solution, and the length of time allowed for precipitation also affect tlie final result. Other \vorkers, especially \Vood, in the ~
D E T E R X I X h T I O N O F PHOSPHORUS IN STEEL.
227
J . AmzZ. Chcm., 1, 138, Shinier, Tmm. A . I. M. E., 17, 100, and Drowi, Tmm. A . Z, A% E . , 18, go, have put a good deal of study on the method. S o n e of the determinations given that we have been able to study, show that the final result is independent of the coiiditions under which the yellow precipitate is obtained. We ourselves have made a good many experiments, varying tlie conditioiis under which the yellow precipitate is obtained, and feel safe in saying that almost any variation in the conditioiis will produce some difference in the final result. Much of our ii-ork simply confirnis that of other operators. Perhaps it would be worth while to give a single illustration. Two sets of oiie grain each of three samples of steel containing different amounts of carbon and phosphorus were dissolved in seventy-five cc. of 1.13 sp. gr. nitric acid, and further treated exactly alike according to our standard method above referred to, except that before adding the molybdate solution, ten cc. of strong ammonia was added to each flask of one set. This procedure would as is evident, result in the foriiiation of the yellow precipitate in one set in a nienstruuni containing more ammonium nitrate, aiid less free nitric acid than the other contained. T h e results obtaine:i are as follows, the figures being the phosphorus in tlie three steels :
n'ithout aiiiiiioiiia
IYitli
aiiiiiionia
...
, I
s o .I steel. Per c e n t .
I
:.... ...... 0.036
.......... 0.037 1 3 ........... 0,037 2
I 2
j
........... 0.047
........... 0.043 ........... 0.045
NO. 2
steel. Per cent.
0.055
so. 3 steel. Per cent. 0.029
0.055
0.029
0.055
0.030
0.064
0.036
0.062
0.033 0.039
0.067
11 Tlie difference in tlie results is quite marked. Dinii free acid and increase in aiiiiiiotiiuni nitrate apparently leads to higher figures aiid less iiiiiforiiiity iii duplicate determinations. Sonie of our evperimeiits indicate that even a much less change in the proportions of free acid and ainnioiiiuni nitrate than given above affects the final result. Also that in addition to the variables mentioiieci above, n different result is obtained when sugar or other organic substance is used to reduce the separated nianganese oxide than is given when ferrous sulphate
is used. Filially there are quite strong indications. that the total amount of iiiolybdic acid present i n the solutioii and eve11its +tate of oxidatioti have a11 influence oii the fiiial figures ohtaiiic(l. If it be allowed then that the coiitiitioiis. under \vliicii the >-ellow precipitate is foriiietl :i!Yect its coiistitutioli. or iiiort: clefinitel!.. affect tlie ratio lxtuecii tlir iiiol>.l)deriuiii niid pliosphorus i i i this precipitate. or ivhnt aiiioutits to tlie same thing. affect tlie fiiial result, tivo tliiiigs are evident. I . I n order to secure accurate and concordant results, it is iiecessary that the coiiditioiis under \\.hicli tlic yellon. pi-ecipitate is obtniiietl shall he very clearly and sharply tlefinecl, niid t!iat on iiiakiiig the ntial!ws, these coiitiitioiis slid1 be \'t'r). rigidly iiiaiiitaiiietl. 7 . It ]vi11 iiot do to use ratios I ~ s e don the analysis of tlie >.elIo.$vprecipitate obtained under one set of coiiditioiis as n ~ i e a n sof calculation for precipitates obtained under aiiother xiid cliffereiit set of coiiditioiis. -1s a11 esaiiiple of this second stateinelit, it seems clear, if the 1)recetliIig views are correct. that if a cheiiiist obtains a quaiitity of tlie yellow precipitate froiii sodium phosphate rvitli no iron present, and the coriditioiis entirely different from those which preI.ni1 i i i the deteriiiiiiatioii of phosphorus in steel, :mti then i i x k e s an aiial!.sis of this yello\v precipitate to dcteriiiiiie the ratio bet\veeii niolybdic acid ant1 phosphorus, although the figures h e obtains may lie quite accurate, he is still iiot at liberty to use these figures in deteriiiiiiiiig phosphorus iii iroii aiid steel aiid claiiii that they give accurate results. It is of course conceded that they m a y be accurate but there is no certainty of this. I t should be stated that the aiiiouiit of phosphorus iii the yellow precipitate is so siiiall, aiid tlie amount of iiiolybdic acid is so large, that there is room for considerable 1-ariatioii i n ratios without giiririg rise to very serious differences iii results, especially in steels low iii phosphorus. This inct iiiay help to sccouiit for tlie claiiii so generally made b y different cheinists that their results agree within reasoiiabie limits of error with those of other good clieiiiists. But if it is desired to obtain fairly accurate and coiicordaiit results, we see no way of escaping froiii the two stateiiieiits giveii above.
DETEK31ISATION OF PHOSPHORUS IS STEEL.
229
If these points be conceded it is obvious that a still further very important question arises, ziz.: 111. How shall the proper factor for use with the permanganate of potash be obtained? It is clear, since the perniangallate of potash solution is standardized against metallic iron. that we need two ratios, first, the ratio between molybdic acid and iron, and second, the ratio between molybdic acid and phosphorus in the yellow precipitate obtained under definite conditions. What then are these ratios? On looking over the literature on the subject, we find two or three serious uncertainties. T h e ratio between iron and molybdic acid is evidently affected by the change which takes place in the niolybdic acid when it passes through the reductor, and we do not fiiid that there is agreement between those n-ho have worked on this point. Furthermore, Jones has apparently sliown in the Tram. A . 1.ill. E., 18, joj. that the use of the reductor gives different results than are obtained when the molybdic acid in the yellow precipitate is reduced by granulated zinc in a beaker as originally proposed by Enimerton. Again the published ratios between molybdic acid and phosphorus, vary between IOO : I .54 and 100 . 1.pif we may use our own figures for illustration. It is clear therefore that there are two difficulties in the way of using any of the published ratios, first, which ratios shall be used, arid second, the point already discussed quite at length, that these ratios vary, or at least the filial results vary according to the conditions under which the yellow precipitate is obtained. TKOmethods of procedure are clearly open to us. First, obtain a quantity of the yellow precipitate under the conditions which we deem essential, and then tilake independent deterniinations of the phosphorus and molybdic acid in this precipitate. This would give the ratio between these two coiistituents. Then pass a quantity of the same precipitate through the reductor and titrate with permanganate. This will give the ratio between molybdic acid and iron. This we have not done yet, partly for lack of time and partly froni the difficulties of the analysis. In our experience it is not easy to dry and weigh the yellow precipitate without its undergoing change in the process, and our experience with methods of determining
230
D E T E K 3 I I N d T I O N O F PHOSPHOKUS I S STEEI,.
iiiolybdic acid is also meager. There is need of sorile good and careful work 011 this point. I n view of these uncertainties. we chose the other method of procedure, viz.: Secoiid, determiiie the phosphorus i n several samples of steel b y the iiiost accurate knoivn gravimetric iiiethod, then make ail analysis for phosphorus in these same steels by the voluiiietric method, and then use such a factor with tlie potassium periiiaiigallate as will bring these results together. The work done 011 the steels chosen for this purpose is given iii our paper ' ( On ; ~ i i Attempt to Find the Xiiiourit of Phosphorus in Three Saniplei: of Steel." I t is evident that this procedure, while it enables u s to get a factor to use with the potassiuiii perniaiigaiiate' in the phosphorus deterinillations by the volumetric method as we use it, does iiot throw any light on the ratios between iron and molybdic acid. niiti between niolybdic acid aiicl phosphorus in the yellolv precipitate as \ye obtain it. But if we assume that the ratio of iron to iiiolybdic acid is 100 : 90.76 it follows that the ratio of iiiol!.bdic acid to plios1)horus in tlie yellow precipitate as we obtaiii it is 100 : 1.90 a11d this is the calculatioii ~vhich\ve have used i i i our published volumetric niethod above referred to. One further question ma). arise, w".: 111x-ieiv of the apparent variability of the >.ello\vprecipitate xccortliiig to the conditiolis unckr which it is obtained, I V . If the conditions are made constant, vi11 tlie results obtained be uniform and agree with each other ? In reply to this ive may say that so far as our experience has gone, the agreement between duplicate tleteriiiiiiatioiis 011 the same sample of steel where the coiirlitioiis of tlie inethocl as published are rigidly adliered to, rarely exceeds more t11311 n couple of thousandths of 3 per cent., also cliffexwit operators using the published method 011 the same saiiiple of steel, after a little experience is obtained with it, rarely disagree more than three or four thousandths.
