624 SHORT AND WILLIAMS: DETERMINATION OF SMALL AMOUNTS OF [Vol. 83 Determination of Small Amounts of Silicon in High-purity Iron BY H. G. SHORT AND A. I. WILLIAMS (National Physical Laboratory, Teddington, Middlesex) An activation method has been developed for the determination of amounts of silicon in high-purity iron clown to 0.0001 per cent. The results agree with those found for total silicon (acid-soluble plus insoluble) by an absorptiometric method !depending on formation of a molybdo- silicate complex and reduction of this to molybdenum blue. Some minor points in the chemical procedure have been re-investigated. FOR some time past in this laboratory, work on the effects of impurities on the properties of high-purity iron has necessitated the determination of amounts of silicon down to 0.001 per cent.The determinations have been carried c u t by an absorptiometric method based on the procedure described in British Standard 11211 : Part 19 : 1951. This depends on the forma- tion of a molybdosilicate complex, which is reduced to molybdenum blue by ferrous sulphate in the presence of oxalic acid. This procedure can be used only to determine silicon present in such a form as to be soluble in the 5 per cent. v/v sulphuric acid used for dissolution of the sample, and, to obtain the total silicon content, any insoluble matter must be fused with sodium carbonate fusion mixture and the silicon determined in a solution of the resulting melt by a similar absorptiometric procedure. On the whole, this technique has given acceptable results, but, from time to time, abnormally high results have been obtained, which were suspect because they could not be accounted for by the technique of manufacture. These high results were, however, repro- ducible and have been confirmed by the activation procedure.Other disadvantages are (a) the fact that at 0-001 per cent. of silicon the method is being extended to the limit of sensitivity, since this silicon content corresponds to 0.02 absorptio- meter-drum divisions and the blank value for the reagents is usually 0.03 divisions, and ( b ) the difficulty in accurately determining the blank value, which is equivalent to 0.0017 per cent. of silicon under the conditions of the method. The development of activation analysis made it possible to check silicon determinations by an independent technique.It must be emFhasised that use of the activation procedure gives the total silicon content of the sample. ABSORPTIOMETRIC METHOD- Blank determinatiovz-The determination of the blank value cannot be made by carrying out the procedure with omission of the iron sample, owing to interference by reduced molybdenum a t the colour-development stage. On the other hand, iron completely free from silicon, which could be used in a blank determination, is not so far available. This difficulty has in the past been overcome by making two determinations on each sample solution, one aliquot, A, having half the volume of other, E:. Then, if x and y are the absorptiometer readings corresponding to the silicon in the sample and the reagent blank, respectively, the reading obtained for aliquot A is (x + y ) and that for aliquot B is (2% + y).The blank value is then found by subtracting the absorptiomeler reading of aliquot B from twice that of aliquot A. When this difference procedure was used, a mean blank value of 0.042 absorptiometer- drum divisions (equivalent to 0-0024 per cent. of silicon) was obtained over a large number of tests, but the variation was considerable (standard deviation i 0.019 drum divisions). Experiments to discover the cause of this variation indicated that the potassium perman- ganate used to oxidise the sample solution was responsible. This was therefore replaced by hydrogen peroxide in a second series of tests; the blank value was then lower (0.034 drum divisions) and less variable (standard deviation & 0.008 drum divisions).Use of hydrogen peroxide for oxidation was therefore adopted as standard. In the course of the experiments it was found that potassium permanganate solutions stored in glass became contaminated with appreciable amounts of silicon in less than 1 week. EXPERIMENTALNov., 19581 SILICON IN HIGH-PURITY IRON 625 Two grams of iron, of the lowest available silicon content, were dissolved in acid, oxidised, and the solution was extracted with diethyl ether. The aqueous layer was rejected and the ethereal layer was extracted with water; this gave an aqueous iron solution free from silicon. This solution was evaporated in a platinum dish with sulphuric acid to remove hydrochloric acid, and the resulting solution was diluted to such a volume that the concentrations of iron and sulphuric acid were the same as in a normal test.The blank value obtained in the subse- quent colorimetric test was 0.03 drum divisions, which was in agreement with that found when the difference procedure was used. The reagent blank value for the absorptiometric method is therefore thought, with some confidence, to be 0.03 absorptiometer-drum divisions. This should be checked from time to time, when fresh batches of reagents are brought into use, by the difference procedure. Contamination by glassware-Since only a few micrograms of silicon participate in the final colour reaction, some suspicion has always been attached to the use of glass apparatus, particularly for dissolution of samples, for which it is necessary to heat acid solutions. Tests on one particular sample of average silicon content (0.003 per cent.) showed no detectable difference in results when the sample was dissolved and the solution prepared for colour development in either platinum dishes or Pyrex-glass beakers.Similarly, blank values determined by the difference procedure on AnalaR ferrous sulphate that had been dissolved in acid and treated as a sample were identical whether the solutions were prepared in platinum or glass. Portions of the 5 per cent. v/v sulphuric acid used for dissolution of the samples were boiled for several hours in Pyrex-glass and Hysil-glass beakers. The solution from the Pyrex-glass beaker was found to contain 11.5 pg of silicon (equivalent to 0.0008 per cent.on a 1.5-g sample) and that from the Hysil-glass beaker was found to contain 20 pg of silicon (equivalent to 0.0013 per cent. on a 1.5-g sample). The evidence is, therefore, that contamination by glassware is negligible under normal operating conditions ; nevertheless, determinations on samples known to be extremely low in silicon are now carried out in platinum dishes to avoid any possibility of error from this source. The blank value was also determined by an alternative procedure, as follows. ACTIVATION METHOD- Owing to the short half-life (2.65 hours) of the active isotope used (silicon-31), chemical treatment of the sample after activation had to be such that counting could be carried out on the same working day that the sample was received from Harwell.For this reason it was considered that what might be called the classical method for separating silicon- dehydration and subsequent filtration-was too lengthy. Trials were therefore made with a procedure in which silicon was separated by diethyl ether extraction of a complex molybdosilicic acid from acid solution. It was found, however, that the resulting compound was not of sufficient radiochemical purity to avoid previous scavenging steps, contamination with tungsten being particularly noticeable. The same disadvantages were found to apply t o precipitation as potassium fluorosilicate. Experience also showed that the established dehydration method could be carried out almost as quickly, if care were taken to keep the volume of solution to be evaporated to a minimum.The silicon was prepared for counting by collecting the separated silicic acid on a small circle of filter-paper, which was partly dried by washing with ethanol and diethyl ether. The paper was then placed on a tray for counting, and was afterwards ignited to determine the percentage recovery of added carrier silicon. To a certain extent, radio- chemical purity has been sacrificed to speed of operation. A scavenging step for tungsten was found to be necessary, but the silica isolated for counting always contained a small amount of long-lived active impurity, although the first part of the decay curve was to all intents and purposes parallel to that of the standard. The true silicon count was therefore obtained as follows. Immediately after preparation, the precipitate was counted five or six times at intervals of 30 minutes in order to obtain the slope of the main silicon-decay curve, which was then extrapolated to zero time to give uncorrected silicon counts.In the course of the next day, the precipitate was counted five or six times a t intervals of 1 hour to give the residual long-lived activity, which was also extrapolated to zero time and sub- tracted from the uncorrected silicon activity. The magnitude of this correction varied, but The procedure finally developed is described on p. 627.626 SHORT AND WILLIAMS: DETERMINATION OF SMALL AMOUNTS OF [Vol. 83 in general it formed a small proportion of the total counts; this is shown by the results in Table I. TABLE I CORRECTION OF COUNTING RATE .4mount of Uncorrected activity, Residual long-lived activity, silicon present, counts per minute counts per minute Correction, % Yo 0.0002 0~0020 0.0018 0.0050 0.0002 050015 0.0040 185 7.5 410 30 534 174 1540 40 134 6 340 23 1180 60 4 7 33 2.5 4.5 7 5 METHOD REAGEXTS- and cool the solution.Sulphuric acid, dilute (1 + 19)-To 950 ml of water add 50 ml of sulphuric acid, spgr. 1.84, Hydrogen peroxide, 20-volume-Analytical-reagent grade. Ferrous sulphate solution, 2 per cent.-Prepare a 2 per cent. w/v solution of ferrous sulphate, FeS0,.7H20, in 0.1 per cent. sulphuric acid. Ammonium molybdate solution, 10 per cent. :w/v. Oxalic acid solution, 10 per cent. w / v . Silicon carrier solution-Prepare a solution of sodium silicate containing 5 mg of silicon Manganese carrier solution-Prepare a so'lution of manganous sulphate containing Tungsten carrier solution-Prepare a solution of sodium tungstate containing 10 mg per ml.10mg of manganese per ml. of tungsten per ml. ABSORPTIOMETRIC PROCEDURE- Dissolve a 1.5-g sample of iron in 100 ml of (dilute sulphuric acid (1 + 19) in a platinum dish. Filter the solution through a pulp filter, transfer any insoluble matter to the filter and wash with water. Transfer the filtrate to a platinum dish, warm, and add 15 ml of 20-volume hydrogen peroxide. Warm until effervescence has ceased and finally maintain near the boiling-point for 10 minutes. Cool, transfer to a 250-ml calibrated flask, dilute to the mark, and mix. By pipette, place two 20-ml portions of the solution in clean dry beakers. To one portion add 10 ml of 10 per cent.w/'v ammonium molybdate solution, mix, and set aside for 15 minutes a t 20" i. 2" C. Add 20 ml of 10 per cent. oxalic acid solution and 5 ml of 2 per cent. ferrous sulphate solution. Mix well and set aside for 10 minutes. This is solution A. To the other portion add 20 ml of 10 per cent. oxalic acid solution, 10 ml of 10 per cent. w/v ammonium molybdate solution and 5 ml of 2 per cent. ferrous sulphate solution. Mix well and set aside for 10 minutes. Measure the optical-density difference between solutions A and B in 4-cm cells with a Spekker absorptiometer and Ilford No. 606 filters. Determine the amount of soluble silicon by reference to a standard graph prepared by adding known amounts of sodium silicate solution to pure iron and then carrying out the above-described procedure.Normally, allow a blank value of 0.03 absorptiometer-drum divisions, but, when a new batch of reagents is used, check this in the following manner. In addition to the two aliquots for the silicon determination, measure two 10-ml aliquots, add 10 ml of 2 per cent. v/v sulphuric acid to each and proceed as before. If the absorptiometer-drum readings of these four solutions are A , B, A' and B', respectively, the reagent blank value is given by- Reserve the precipitate for the determination of insoluble silicon. This is solution B. 2(A' - B') - ( A - B) To determine insoluble silicon, ignite the residual insoluble matter from the dissolution Dissolve the of the sample in a platinum crucible and fuse with 0.2g of fusion mixture.Nov., 19581 SILICON I N HIGH-PURITY IRON 627 melt in water, add 2.1 ml of 50 per cent. v/v sulphuric acid and dilute the solution to 100 ml.Take two 20-ml aliquots and proceed as for the determination of soluble silicon, but add 1 ml of the ferrous sulphate solution instead of 5 ml. Carry out a blank determination on a 0.2-g portion of fusion mixture in the same manner. ACTIVATION PROCEDVRE- Irradiate 1 g of iron a t pile factor 10 for 6 hours. Dissolve it in a mixture of 15 ml of hydrochloric acid, 3 ml of nitric acid and 5 ml of silicon carrier solution. Evaporate the solution to dryness, dissolve the residue in 5 N hydrochloric acid and spin the solution in a centrifuge. Discard the solution and wash the impure silicic acid with hot 5 N hydrochloric acid until the iron colour has been removed.Fuse with 40 mg of sodium tungstate and 0.3 g of sodium carbonate, and dissolve the melt in 10 ml of hot water. Add 1 ml of manganese carrier solution dropwise, stir, spin in a centrifuge and discard the precipitate. Just neutralise the solution with dilute nitric acid (methyl red as indicator), add a slight excess of saturated mercurous nitrate solution, spin in a centrifuge and discard the precipitate. Add 1 ml of tungsten carrier solution dropwise, stir, spin in a centrifuge and discard the precipitate. Precipitate the excess of mercury with hydrochloric acid, spin in a centrifuge and discard the precipitate. Add 5 ml of hydrochloric acid to the solution and evaporate to dryness. Dissolve the residue in 10 ml of 5 N hydrochloric acid, spin in a centrifuge and discard the solution. Wash the silicic acid twice with warm 5 N hydrochloric acid and transfer to a filter. Wash with water, ethanol and diethyl ether, mount, and count a t intervals of 30 minutes. Repeat the counting next day at intervals of 1 hour. Determine the percentage recovery by igniting the precipitate and weighing the silica produced. Filter, wash with water, and ignite. RESULTS The reproducibility of the absorptiometric method is shown by the results of seven determinations of soluble silicon in a particular sample; the maximum found was 0,0075 per cent., the minimum found was 0.0071 per cent. and the standard deviation was F0.00013 per cent. The amounts of total silicon in various iron samples were determined by both methods. The results were as follows- Sample No. . . . . . . . . . . &I7251 M640 111535 M554 M585 M633 Silicon by absorptiometric method, yo . . 0.0019 0.0094 0.0049 0.0020 0.0044 <0.0005 Silicon by activation method, Yo . . . . 0.0019 0.0091 0.0044 0.0015 0.0049 0.0002 The work described was carried out as part of the general research programme of the National Physical Laboratory and is published by permission of the Director of the Laboratory. Received May 'ith, 1958