104 RILEY ON THE ESTIMATION OF XVL-On the Estiqnatiorh of Phosphorus in Iron and Steel. By EDWARD RILEY. THE accurate determiriat,ion of the amount of phosphorus present in iron and steel, more especially since the introduction of the Bessemer and Siemens-Martin processes, now so largely used, has become of the greatest technical impor tance. The method most generally employed at iron and steel works isPHOSPHORUS IN IRON AND STEEL. I05 that known as the molybdate method, the other is known as the magnesia method; the latter is not so much used, as the results cannot be obtained so quickly. I need hardly say that in iron and steel works, where so many determinations are required, time is a most impor tan t consideration. It is not my intention, in this communication, to enter into very minute details as to ths method employed in precipitating phosphorus by molybdic acid, or by molybdate of ammonia; as it has been so frequently described in various works, and I find practically that almost every operator varies in the exact method he employs.The percentage of phosphorus is as a rule so small in the pig iron used and the steel made, that to determine it accurately requires much practice and some amount of analytical skill. Speaking from my own experience of more than 25 years, and from my experience and observation as to the results obtained by others who use the molybdate method, I must say that as a practical fact the results obtained by many chemists by it are not satisfactory. I have hesitat,ed to adopt it in my laboratory, although I must admit I have on some occasions obtained very satisfactory resuits ; still I have always felt a certain amount of doubt about the process, whereas witb the magnesia method I have no doubt whatever.I must however admit that grave errors may be made by the magnesia method, and that great care is required in using it. It is quite the exception to find either iron or steel without phos- phorus, only some very few Swedish and foreign irons are free from it. I have, however, very frequently samples and analyses sent to me in which no phosphorus is given, whereas on examining the samples I find in most cases a very appreciable and weighable amount. I am speaking now of analyses made by chemists who have some expe- rience in iron and steel analysis ; by others it is not a t all unfrequent for -10 to -20 per cent.to be overlooked. The cause of this low estimation, or failing to detect the phosphorus, is, I believe, in the case of the molybdate method, due to having the solution too acid, or in other cases not adding a sufficient amount of molybdate solution. In the case of high estimates, which are not unfre- quent, the excess is due to molybd-ic acid being thrown down together with the phosphorus precipitate. The great importance of the accurate determinat'ion of phosphorus, induced me to have a series of experiments made by various operators, so as to ascertain how nearly the practical results obtained in iron and steel works, and in various metallurgical laboratories, actually agreed with the absolute amount of phosphorus present. For this purpose pure peroxide of iron mas prepared, ordinary iron borings were dissolved in dilute sulphuric acid, the solution filtered VOL.XXXIII. K106 RILEY ON THE ESTIMATION OF from the residue, and the sulphate of iron crystallised out ; the mother- liquor was separated and the cryst,als washed with distilled water, re- dissolved in distilled water and recrystallised. The crystals having again been washed, were then dissolved in distilled -water, the solution peroxidised with pure nitric acid, and the peroxide of iron precipi- tated from the diluted solution with ammonia. After washing four times by decantation, the precipitate was redissolved in pure hydro- chloric acid, and reprecipitated by ammonia; the oxide of iron was washed until it was practically free from ammoniacal salts, then detached from the calico filter used, dried and ignited at a dull red heat in a muffle.Before using the peroxide it was thought desirable to ascertain if it was absolutely free from phosphorus. It was tested by the magnesia method usually employed by me. The peroxide of iron was dissolved in strong hydrochloric acid, then reduced, after diluting the solution with sodium sulphite, and the excess of sulphurous acid was expelled by boding ; the solution was nearly neutralised with ammonia, acetate of ammonia added ; the phosphoric acid was then precipitated on boiling with n small amount of peroxide of i'ron, as basic acetate and phosphate. When only minute quantities of phosphorus are present, there is usually enough peroxide of iron in the solnt'ion to precipitate the phosphorus ; if not, a drop or two of a dilute solution of perchloride of iron is added, or a few drops of bromine-water, enough to make the precipitate of a red tinge.The solution after boiling is filtered, the precipitate dissolved in strong hydrochloric acid (it is not necessary to wash the precipitate before dissolving it), and the filter thoroughly washed. The solution of the phosphoric acid with some iron is then evapo- rated to a small bulk, about half an ounce (15 c.c.) ; to this is added about 200 grams of citric acid (13 grms.) ; the solution neutralised with -960 ammonia ; some 20 to 30 drops of magnesia solution added, then a moderate amount of -880 ammonia, so as to make it strongly alka- line.The whole bulk of liquid should not exceed 16 to 2 ounces (50 c.c.). No immediate precipitate is formed unless the phosphorus amounts to -10 or more per cent. ; on standing, however, for R night, the characteristic small star-shaped crystals of the ammonia-mag- nesium phosphate are found adhering to the sides of the beaker and on the stirrer. The solution is strongly agitated the next day with the stirrer, either at once or after the lapse of a little time, whereupon a further precipitate of the granular characteristic ammo- nia phosphate is formed. The solution is allowed to stand for another night, and filtered off on the following day; and the residue is mashed with ammonia-water, dried, ignited, and weighed. The precipitate, after ignition, is either perfectly white or colouredPHOSPHORUS IN IRON AND STEEL.107 only with a little carbon. If any difficulty is experienced in burning this over the gas, it may be heated with a few drops of fuming nitric acid with the cover over the crucible; it then becomes perfectly white. In no case do I find that any iron is precipitated with the magnesia salt. The precautions necessary are to have not too much iron present. I have found that the usual quantity is from 5 to 6 grains of iron from a phosphorus determination on 150 grains of steel. A large excess of citric acid must be used, and also of ammonia. The soh- tions should be of a paleish yellow-green colour ; if it is very dis- tinctly red, good results are not obtained ; if it is dark-red, the results are altogether fallacious.The precipitate obtained by magnesia is so characteristic that there cannot possibly be any mistake about it. The peroxide of iron was found to contain a distinct amount of phosphorus ; 206.595 grains gave Mg2Pz0;*0135 grain. Equal to phosphorus per cent. in the Fe,O,. ........... .0018. 7, 7 7 in iron or steel.. ........ *OQ26. Some pure Mg2Pz0, was prepared by taking about +-pint of mag- nesia solution, made by adding a considerable amount of ammonia and chloride of ammonium to a saturated solution of ordinary sul- phate of magnesia, and filtering off the precipitate and insoluble matter; this solution on standing generally yields a deposit. The clear solution is used for phosphorus determinations. To the above solution, after dilution, was added phosphate of soda, the precipitate formed was allowed to settle, the supernatant liquid decanted off; and the precipitate washed four times with distilled water ; it was then redissolved in hydrochloric acid, reprecipitated with ammonia, filtered, and washed until no appreciable residue was left on evaporating the washings to dryness.The dried precipitate was ignit,ed in a muffle until it. was quite white. In carrying out the experiments, each of my assistants ignited from 213 grams to 218 gramsof the pure peroxide of iron, this quantity re- presenting about 150 grams of metallic iron, the usual quantity taken for phosphorus determinations in steel. The above quantity was dissolved in strong hydrochloric acid. I then very carefully myself weighed out the phosphorus salt, giving each a different quantity.The solution of the peroxide was poured on the magnesium phosphate in a small beaker, and warmed until i t was completely dissolved. The phosphorus waB then determined by my ordinary method, each K 2108 RILEY ON THE ESTIMATION OF assistant working on a solution of iron containing a quantity of phos- phorus unknown to him. The following are the results that were obtained by each :- Peroxide of iron Mg2P,07 Mg,P,O, Mg2P20i taken. added. found. reprecipitated. I. W. H, H e r d s m a n . . 218.62 -46 *475 *460 11. A. E. Tucker ...... 213.45 -48 ,500 -500 111. G. A. J a r v i s ...... 215.26 -35 *380 -365 IV. F. J. Bolt . . . . . . . .214-255 *72 -775 ,745 V. A. L. H u g h e s . . .. 215.215 -71 * 725 - Converting the peroxide of iron into metallic iron, and calculating the percentage of phosphorus present in the iron (after deducting the small amount of phosphorus found in the peroxide), the results are as follows :- Peroxide of iron to Phosphorus p.C. Phosphorus p. c. Difference p. c. in metallic iron. added. found. phosphorus. I. 153.034 -0839 *0840 .0001 11. 149.415 -0897 -0908 -0011 111. 150.682 *0648 -06 78 ,0030 IV. 149.978 -1340 -141 7 -0077 V. 150-6.50 ,1316 *1336 *0020 or a mean error in excess of *0028 per cent. I n all cases the calculation is made on the first precipitation and weighing of the magnesium phosphate. The first four, as my assistants, have had considerable practice in determining phosphorus. No. 5 determination was made by one of my pupils who has had but little practice, not having made more than three or four determinations of phosphorus in iron and steel.The nbo-re were the only experiments made, and it will be seen that practi- cally the exact amount of phosphorus was obtained that I weighed in. In reprecipitating the MgJ'ZO, after weighing, it was dissolved in strong hydrochloric acid, a very small crystal of citric acid added, and excess of ammonia ; and the precipitate after standing for two nights was filtered and weighed as before. An experiment was made on some of the phosphate of magnesia to see what was the actual loss on reprecipitation. Mg,P,O;. Ignited oTer Bunsen burncr. Ignited in muffle, 2.46 gave 2*445* 2.42 The experiment was made with the same bulk of liquid as in an ordinary phosphorus determination, and in precisely the same way, except that no citric acid or magnesia salt was added.* Not quite white.PHOSPHORUS IN IROX AKD STEEL. 10:) I shculd perhaps mention that when there is much phosphorus pre- sent, as in ordinary pig iron (not Bessemer pig), I operate on smaller quantities, and if there is much precipitate I use a rather larger bulk of liquid. From the above experiment, and also from the results obtained in the previous experiments, it will be seen, as I have always found it, that there is no appreciable loss on reprecipitating the Mg2P207, thus proving the practical purity of the precipitate obtained on the first precipitation. I n order to compare the results obtained by the molybdnte method, 1,000 grains of the Fe203, weighed after ignition, mere dissolved in hydrochloric acid ; and t o this 2.496 grams of ignited Mg2P207 were added ; after solution the whole was reprecipitated by ammonia, washed with distilled water, dried, ignited, and weighed.The above quantity of magnesia salt represents in the iron calculated from the Fe203- Phosphorus per cent. .......... *0994 Phosphorus in iron ............ ,0026 *I020 This gives the actual quantity of phosphorus present. I_ To be q u i k certain that no mistake was made, Mr. Herdsman determined the phosphorus by the magnesia method. Grams of Fe203. Mg,P,O,. Phosphorus per cent. in iwn. 213.27 gave -52 equal -09 72 or difference - -0048 Various portions, from 80 to 150 grains, mere sent t o three of t h e largest steel works in the country, where phosphorus determinatioiiv are made daily, by some of our most expert analysts of iron and steel; two other portions were sent, one to a well known chemist with a mxy large experience in iron a,nd steel analysis, and another to a chemist, also with a large experience, but not so well known.The following are the results returned :- Phospho~*us per cent. by Molybda fe Metiiod. I. Phosphorus per cent. .... -0780 -0821 11. ? 9 .... { } Mean, 00785 1, 111. ,9 >9 .... -0640 11. .:0460 IV. .... -0700 v. .... -0760 obviously low, the mean of the five results is- 7, 9 7 ?, 9 9 Kot taking into consideration the secortd result of No. 111, which is110 RILEY ON THE ESTIMATION OF Phosphorus per cent. .................. *0733 Error in determinations per cent......... '0287 All the results are very concordant, except perhaps No. 111, and all low, thus showing that phosphorus determinations as now carried out by our best operators are too low. The above experiments were partly undertaken in consequence of a statement that my determinations of phosphorus in a steel rail were too high, and also that my results were always higher than those given by the Sheffield Steel Works. Take the rail in question : the phosphorus was determined three times. A. 150 grains of steel gave phosphorus per cent. .... ,094 C. 150 ,, ,, .... -096 B. 200 ,, ,, .... *090 71 ,, 9 9 ?, All the analyses were made by different operators. The molybdate gave, analyses made at Steel Works- ,0602 -0586 A. Phosphorus per cent. ..................{ } *0597 99 .................. -0976 B. ,> C. Y ? Mean by magnesia method. Mean by molybdate. Difference. -0933 -0 722 *0211 thus showing a difference between the results very closely approxi- mating the difference in the previous case. Take another case of same sample bars of steel. Two of my assistants (one now at a steel works) made- I. 11. Magnesia method. A Phosphorus per cent. ................ -17'2 -180 B ,, >, ................ *162 -164 B Repeated by same operator.. .......... *165 -170 Molgbdate method. C Chemist at Sheffield Steel Works returned -867 -273 This high result differing so much from the above, the analyses were repeated, and the following results sent by same chemist :- Molpbdate method. C' Phosphorus per cent. ................ -142 0166 C" ................a154 -154 D Another ironworks chemist .......... -145 ,158 thus clearly showing that the magnesia results are always higher. 17 9 ,PHOSPHORUS IN IRON AND STEEL. 111 I have had a very lengthy correspondence with chemists who have had considerable experience in the determination of phosphorus, mostly with chemists at iron and steel works. The opinion of several is that the two methods may be made to give the same results. This I think is quite possible, as the magnesia process may readily be worked to give slightly lower results. In the case of the molybdate method, I find frequently that dupli- cate analyses are made on the same quantity of steel. Now it is quite possible that corresponding results may be obtained, and that they may be quite wrong, as I regard this as no check OD an analysis. The better way is to take twice the quantity in one case to that taken in the other.Before instituting the above experiments, I have most thoroughly tested the magnesia method, and had every confidence in its accuracy when carefully performed. I have used five times as much steel in one case as another, and obtained corresponding results, and numerous cases double, treble, and so on. The experiments given, made by so many different operators, clearly shew, I think, that as the molybdate process is now carried out, the results are too low. The method is, however, much more rapid tban the magnesia pro- cess, and from experiments I have made, I think the molybdate process may be improved. The method I propose is t'o dissolve the steel or iron in 1-20 nitric acid, avoiding m much excess as possible ; to the solution, diluted to rather more than half a pint, sodium sulphite is added, so as to reduce the iron to FeO after boiling off the sulphite (a large excess should be avoided) ; the phosphorus is precipitated with some peroxide as basic acetate and phosphate, precisely as in the magnesia method ; the preci- pitate is dissolved in hydrochloric acid; the solution made alkaline with ammonia; and the precipitate formed dissolved in 1.40 nitric acid.The solution is then precipitated with molybdate of ammonia in the ordinary way. In an experiment made by Mr. Herdsman it was found that the phospho-molybdate came down at once ; in twenty minutes the preci- pitate had settled ; and in less than an hour after adding the molyb- date, the liquid was filtered, and the precipitate ready to dry. The result obtained was phosphorus -077 per cent. in a steel rail containing by the magnesia method -093 ; it was found that some fnrther precipi- tate of phospho-molybdate was formed on heating the filtrate. Although this experiment was not successful, still I am inclined to think that by allowing a little more time, and with more practice, it may be made to succeed, and to give results more accurate than the112 PICICERING ON THE ACTION OF ordinary process, and in about the same time. I propose carrying out a series of experiments in this direction. I have made no remarks on the method adopted by some chemish of dissolving the molybdate precipitate in ammonia and precipitating the phosphorus by a magnesia salt ; this, in my opinion, is no improvement on the ordinary magnesia mcthod (in iron or steel analjses). I n submitting the above results to the Society, I am under great obligations to the chemists who so readily assisted me in making th& various phosphorus determinations, and also to the aid I have received from my assistants, more especially bj- my senior assistant, Mr. Herdsman, who has in fact carried out the larger nhrnber of analyses I have given.