November, 19631 SHORT PAPERS 891 SHORT PAPERS The Determination of Silica in Rocks and Minerals by Isotope Dilution with Silicon-31 BY R. H. FILBY AND T. K. BALL (Mineralogical - Geological Museum, University of Oslo, Norway) PRECISE and accurate methods are necessary for determining silica in rocks and minerals, particu- larly for purposes of rock classification. The classical gravimetric method for determining silica is time-consuming and has been shown to be subject to several sources of error.' Rapid spectro- photometric methods based on measurement of the blue colour of the reduced molybdosilicate complex2 are also subject to interferences from certain ions in solution, particularly arsenate and ph~sphate.~ In the past few years, radiometric methods of chemical analysis have become common, and both neutron-activation and isotope-dilution methods have been applied to rock analysis.We have developed an extremely rapid and simple method for determining silica, which is accurate, reason ably precise and is not subject to most of the interferences affecting the gravimetric method. This method has been used for the routine determination of silica in silicates at the Mineralogical - Geological Museum, and is suitable for use wherever reactor facilities are available. EXPERIMENTAL A small amount of 2.6-hour silicon-31 (as active silica) is added to the powdered rock sample (ratio of active silica to silica in the sample, approximately 1 to 30) ; the mixture is fused with sodium hydroxide a t 800" C and the cooled melt dissolved in water. The solution is acidified with hydrochloric acid and evaporated until most of the silica has precipitated.Evaporation to dryness must be carefully avoided as this increases the impurity content of the final silica. The silicic acid is removed by spinning in a centrifuge, washed thoroughly and ignited a t 1000" C for 2 hours. A weighed portion of the silica is mounted on a planchet and counted with an end-window Geiger - Muller counter connected to a scaler. Standards of the active silica are also counted. The silica content of the rock is calculated from the expression- The principle of the method is described below. - a] 100 [" Y ; x 2o Percentage of silica in sample = -~ W Where, w = weight of sample taken, m = weight of silica in sample, Y = count rate of silica in sample a t time t, Y , = count rate of standard at time t and a = amount of active silica added.Irradiated Johnson Matthey 'Specpure' silica exhibited gamma activity, due to sodium-24, in addition to some longer lived beta activity. After irradiation the silica is fused with sodium hydroxide, the melt dissolved in water and the solution scavenged for impurities by adding ferric chloride; the resulting ferric hydroxide is removed by centrifugation. It was found that most of the foreign beta activity could be removed in this manner. The interference due to sodium-24 was eliminated by plotting the decay curves for the standards and correcting for the residual sodium-24 activity. The scavenged solution was used to prepare standards and to "spike" the samples.In preparing the silica for counting it was found that not more than 30 mg of silica could be used, otherwise self-absorption of the beta radiation occurs. It was established experimentally that self-absorption is negligible if not more than 20 mg of silica are used. Difficulty was found in obtaining sufficiently pure silica for irradiation. NUCLEAR AND IRRADIATION DATA Thermal neutrovt flux-approximately 10l2 neutrons per sq. cm per second. p- (1.48 MeV) ReactioN- 30Si(n, ~ ) ~ l S i -+ 31P. Ti = 2.62 h Cross section of reaction-0.1 10 barns.892 SHORT PAPERS [Analyst, Vol. 88 METHOD APPARATUS- tion. circle. REAGENTS- Thin end-window Geiger - Muller tube and scaler-Obtainable from the Nuclear Chicago Corpora- Circular stainless-steel planchets--31 mm in diameter and having a 25-mm diameter inscribed Silica-"Specpure," obtainable from Johnson Matthey & Co.Ltd. Sodium hydroxide pellets-AnalaR grade. Ferric chloride solution, 10 per cent. w/v, aqueous. Silver nitrate solution, 1 per cent. wIv, aqueous. Hydrochloric acid, sp.gr. 1-18-AnalaR grade. Dilute hydrochloric acid (1 + 9). PROCEDURE- Weigh out 0.5 g of "Specpure" silica (dried at 110" C) into a small polythene tube, and heat-seal the ends. Irradiate the capsule in a nuclear reactor with a thermal neutron flux of 1012 neutrons per sq. cm per second for a period of 2 days. Set the material aside for several hours after irradia- tion to allow any short-lived activity to decay. Open the capsule, and transfer the contents quantitatively to a nickel crucible.Add 5 g of sodium hydroxide pellets, and heat the covered crucible at 800" C for 8 minutes. Allow the crucible to cool, then place it in a 200-ml beaker, add 100 ml of water, and heat on a steam-bath until the contents of the crucible have dissolved. Remove the crucible from the beaker, wash it, add 10 drops of ferric chloride solution, and heat on a steam-bath to complete precipitation of ferric hydroxide. Allow the solution to cool, filter, wash the filter-paper, and make the solution up to 250 ml in a calibrated flask. Transfer 0.250 ml of scavenged silica solution by pipette on to a 25-mm diameter circle of tissue paper placed on a 31-mm diameter stainless-steel planchet. Weigh out 0-5 g of the sample into a nickel crucible, and by pipette add 5 ml of scavenged silica solution.Place under an infrared lamp, and evaporate slowly to dryness. Add 5 g of sodium hydroxide pellets, and fuse at 800" C for 10 minutes. Cool, add 15 ml of water, and heat the crucible gently to dissolve the cake. Transfer the contents to a 500-ml nickel evaporating dish, and wash the crucible thoroughly with dilute hydrochloric acid. Add 20 ml of concentrated hydrochloric acid to the contents of the evaporating dish, and place the dish on a steam-bath until the solution has evaporated to approximately 10ml. (Most of the silica will have now precipitated.) Transfer the precipitate to a 15-ml centrifuge tube, and spin it. Wash the pre- cipitate with 10 per cent. hydrochloric acid and then with water until the filtrate gives no positive test for chloride with a 1 per cent.solution of silver nitrate. Transfer the precipitate to a 15-ml quartz crucible, and ignite for 2 hours at 1000" C. Cool, and grind the silica to a fine powder in an agate mortar. Weigh out accurately 15 to 20 mg of the silica on to a steel planchet, and with water make an even deposit in the inscribed area. Dry under an infrared lamp. Prepare 10 such samples for counting. COUNTING- ImmFdiately after their preparation, count the standards (at least 10,000 counts), and recount at frequent intervals over a period of 12 hours. Plot a curve of log,, activity versus time. Apply a correction for the longer-lived activity (24Na) in the original silica solution. This correc- tion can be calculated from the deviation of the curve from linearity.The samples should be counted as soon as possible after preparation, and not less than 4000 counts should be recorded. Apply a correction for decay of silicon-31 during counting and for coincidence, if significant. Read the count rate of the standard, at the time of counting the sample, from the decay curve of the standard. RESULTS Prepare five standards for counting by the procedure described below. Evaporate the solution to dryness under an infrared lamp. Calculate the silica content of the sample from the expression given on p. 891. Listed in Table I are results for silica in several standard reference samples together with I t can be seen that No reaqon for this lack of agreement accepted values. good agreement is obtained for all samples except N.B.S.70. was found. Relative deviations are also listed to indicate precision.November, 19631 SHORT PAPERS 893 Precision is within 2 per cent., and this will be satisfactory for many geological purpcs2s in view of the simplicity and rapidity of the method. Statistical evaluation of the results for G-1 TABLE I RESULTS FOR THE DETERMINATION OF SILICA IN STANDARD SAMPLES Standard Number of Mean silica Relative Accepted value for sample determinations content, yo deviation silica content, % G-1 . . . . 39 72.56 1.70 72.65% w-l . . . . 22 52.77 1-50 52.64% N.B.S. 70 . . 18 67.49 1-90 66.63t N.B.S. 99 . . 11 68.53 - 68-66t Sulphide ore . . 2 34.28 - 34-63 * Values given in the “Second Report on a Co-operative Investigation of the Composition t National Bureau of Standards Certificate value. $ Value given in the “Report of the Nonmetallic Standards Committee,” Canadian Asso- of Two Silicate Rocks.”l ciation of Applied Spectroscopy. show that no bias is found in the distribution of the values, as occurs in the results obtained by the normal gravimetric meth0d.l REFERENCES 1. Stevens, R. E., Fleischer, M., Xiles, W. W., Chodos, A. A., Filby, R. H., Leininger, R. K., Ahrens, L. H., and Flanagan, F. J., “Second Report on a Co-operative Investigation of the Composition of Two Silicate Rocks,” U.S. Geological Survey Bulletin No. 1113, Washington, D.C., 1960. 2. 3. Riley, J. P., Anal. Chim. Acia, 1958, 19, 413. Anderson, L. H., Ark. Kemi, 1962, 19, 257. Received March Tth, 1963