Analyst, May, 1973, Vol. 98, f@. 335-342 335 A Method for the Determination of Silver in Ores and Mineral Products by Atomic-absorption Spectroscopy* BY GEORGE WALTON (Department of Physical Science, Western New Mexico University, Silver City, New iVexico 88061, U..S.il.) An atomic-absorption spectroscopic method for the determination of silver in some ores and mineral products is reported. The effects of sorption of silver on the container walls, cleanliness of equipment, stability of standards and sample solutions, interference from silicate ions and background absorp- tion have been studied. The sample is treated with mixed acids (including hydrofluoric) and evaporated to dryness; the acid extract of the residue is then made strongly ammoniacal and the solution nebulised. Recoveries are, on average, about 10 per cent.higher than those obtained by using fire assay. Silver values obtained by the proposed procedure are almost quantitative for sulphide ores, but with other ores they are about 3 per cent. low, unless the final residue is re-examined. THE determination of silver by atoniic-absorption spectroscopy in water,l solution^,^^^ soil and rocks,*+ alloys,6 sulphide minerals,' mine and mill products, bullions and other materials has been reported. The studies reported have either been of low precision or have related to narrow and specific problems. The present study involves the use of atomic-absorption spectroscopy for the determination of silver in ores and mineral products, particularly in siliceous or sulphide ores in the range from 0.0 to 25 troy oz per net ton, and is based on the use of pure silver solutions as standards.For many years, the accepted and official method for the determination of silver in ores has been fire assay. The rapidity and simplicity usually associated with atomic-absorption measurements has aroused considerable interest in applying this technique to the determina- tion of silver. While its application to simple solutions is possible, substantial interferences have been noted in the determination of silver in ores. In order to obtain results that are comparable with those obtained by use of the fire-assay method, the method of simulation or standard addition had been used; in some instances, the standards used for atomic absorp- tion have been made from ore samples, the silver content of which was based on fire-assay values.The fire-assay technique has also been applied to the determination of gold by atomic-absorption ~pectroscopy.~ Results obtained by the atomic-absorption method are compared with fire-assay results obtained on the same samples. UNIFORMITY OF SAMPLE- A discussion of this subject can be found in the paper by Thompson, Nakagawa and Van Sickle.lo Because the distribution of silver (or any other element) in natural ores and mine products is rarely completely homogeneous, one of the advantages of the fire-assay method is the large sample size taken for analysis (from 7.29 to 29-17 g, or even larger), which tends to reduce the effect of sample inhomogeneity. For atomic-absorption spectro- scopy, the sample size should be chosen so as to give a read-out between 20 and 80 per cent.transmittance. With rich samples, this principle might involve taking such a small sample that inhomogeneity could become important, in which event the use of larger amounts of sample and final solution volumes would overcome the problem. The effect of sample inhomogeneity, if it occurs, would appear in the variation of results for a given sample. Currently, the fire-assay method requires that the sample pulp pass an 80-mesh Tyler screen, and we have based our method on a similar requirement. Extra- fine grinding as a means of providing a more homogeneous sample is not desirable, and may even cause plating-out of precious metals on the grinding surfaces. * Presented at a meeting of the New Mexico Academy of Science, October 13th and 14th, 1972, at Portales, New Mexico, U.S.A.@ SAC and the author.336 WALTON: A METHOD FOR THE DETERMINATION OF SILVER IN ORES AND [Analyst, Vol. 98 SOURCE OF ORE SAMPLES AND FIRE-ASSAY RESULTS- Ore samples used in this study (and two samples of blister copper) were obtained from mining operations in Texas, New Mexico and Arizona. Both oxide and sulphide ores are represented. The fire-assay values given in this paper are referee values; they therefore represent the best value taken from at least six determinations carried out by three different labora- tories and by experienced personnel. All samples were obtained as pulp, ground to pass a Tyler No. 80 sieve (U.S. Standard Sieve Series, W. S. Tyler Co., 0-0070-inch opening), and were examined without drying.DISSOLUTION OF SAMPLE- Several of the sample dissolution procedures reported in the literature were tried, but all yielded extremely low results for silver. We then examined the effect of introducing a fusion procedure before the determination, as described by Huff man, Mensik and Riley.ll Obviously, this procedure resulted in complete dissolution of the ores, but consistently low results were still obtained, which suggested that an interference effect was present. The procedure finally adopted for the dissolution of the samples involved the use of a mixture of acids; however, a re-dissolution procedure, included in order to prevent interference by silicate ions, was found to be essential before using the sample for atomic-absorption spectroscopy. We have found that 100 per cent.recovery of the silver present in the ores could not be obtained by a single dissolution step. As discussed later, a re-examination will facilitate the recovery of an additional 1.2 to 5.7 per cent. of silver and is essential for the highest accuracy to be obtained. PREVENTION OF INTERFERENCE BY SILICATE IONS- Rubeska, Sulcek and Moldan' also reported that sulphuric acid depressed the silver signal at 328.1 nm, which was confirmed by Belcher, Dagnall and West.12 Our work confirmed these observations, but in addition we noted a synergistic effect by silicate ions in the presence of strong acids (Table I); silicate ions alone reduced the silver signal to about 93 per cent., but when 10 per cent.V/V of concentrated nitric acid was present the signal was reduced to 17.3 per cent., with 10 per cent. V/V of concentrated sulphuric acid to 46.4 per cent. and with 10 per cent. V/V of concentrated hydrochloric acid to 72.4 per cent. The results in Table I were obtained for solutions containing 2.50 p.p.m. of silver (as the nitrate). Silicate ions were incorporated, where indicated, by adding 3.0 per cent. V/V of sodium silicate solution (40 to 42 .Be). The presence of sodium ions was shown to have no effect on the signal. In the absence of silicate ions, the strong acids exert a much smaller depressant effect on the silver signal, sulphuric acid at the same concentration as above reducing the signal to about 82 per cent. of its original value. Our method overcomes this interference effect by introducing a high concentration of ammonia in the final solutions.The results in Table I1 demonstrate the improved recovery of silver from solutions containing silicate ions plus 5 ml of concentrated hydrochloric acid and 30 ml of concentrated ammonia solution per 100 ml of solution. TABLE I EFFECT OF SILICATE IONS AND STRONG ACIDS ON THE ATOMIC-ABSORPTION Interference by silicate ions was noted by West, West and Ramakrishna.1 SIGNAL AT 328.1 nm FOR 2-50 p.p.m. OF SILVER Transmittance, per cent. Control (no silicate or acid) . . .. .. . . . . 63.5 10 per cent. V / V of Concentrated nitric acid . . . . . . 64.0 10 per cent. V / V of concentrated hydrochloric acid . . . . 63.4 10 per cent. V / V of concentrated sulphuric acid .. . . 69.0 Silicate ions added* . . . . .. . . . . . . 65-5 Silicate* + 10 per cent. V / V of concentrated nitric acid . . 92.4 Silicate* + 10 per cent. V/V of concentrated hydrochloric acid 72.0 Silicate* -b 10 per ccnt. V / V of concentrated sulphuric acid . . 81.0 * 3 per cent. V / V of sodium silicate solution (40 to 42 Absorbance 0.1973 0.1939 0.1979 0.161 1 0.1838 0.0342 0.1428 0.09 16 .Be). Per cent. of control 100.0 98.3 100.3 81.7 93.1 17.3 72.4 46.4May, 19731 MINERAL PRODUCTS BY ATOMIC-ABSORPTION SPECTROSCOPY TABLE I1 RECOVERY OF SILVER BY ATOMIC-ABSORPTION SPECTROSCOPY* FROM SOLUTIONS CONTAINING 3.50 p.p.m. OF SILVER IONS AND 3.0 PER CENT. V/V OF SODIUM SILICATE SOLUTION1 Run No. Absorbance spectroscopy, p.p.m. Ag+ by atomic-absorption Control (no silicate) 0-2660 3-50 1 0.2668 3.51 2 0.2690 3.54 3 0.2683 3.53 4 0.2660 3.50 5 0.2661 3.50 6 0-2645 3-48 7 0.2658 3.50 8 0.2652 3.49 9 0.2653 3-49 10 0.2659 3.50 Average .. .. . . 3.504 Standard deviation, per cent. 0.51 Range . . . . , . . . 0.060 * Silver signal a t 328.1 nm. 7 Including 5 ml of concentrated hydrochloric acid and 30 ml of concentrated ammonia solution per 100 ml of solution. EXPERIMENTAL REAGENTS AND REAGENT BLANKS- 337 Ammonia solution, concentrated-Mallinckrodt analytical-reagent grade, 58 per cent. m/m. Hydrochloric acid, concentrated-Du Pont reagent grade, 37.5 per cent. m/m. Nitric acid, concentrated-Du Pont reagent grade, 70.5 per cent. m/m. Perchloric acid, concentrated-Mallinckrodt analytical-reagent grade, 70 per cent. m/m. HydroJEuoric acid, concentrated-Mallinckrodt analytical-reagent grade, 48 per cent.m/m. Silver-Wire, 99.95 per cent. pure, D. F. Goldsmith Chemical and Metal Corp., Evanston, Sodium silicate solution40 to 42 "B6, City Chemical Corp., New York, U.S.A. The suggested dissolution procedures were carried out without any sample material but sufficient of the reagents were used for a 10-g sample. No detectable atomic-absorption signal was obtained at 328.1 nm, giving a zero reagent blank and establishing that the reagents were sufficiently pure for the intended purpose. IKSTRUMENTAL- The atomic-absorption spectrophotometer used was a Techtron, Model AA-100, instru- ment with a 4-inch solid stainless-steel burner, Type AB41, having a slit width of 0.020 inch and flame path length of 10.0 cm, and in which air at 15 p.s.i.g.and acetylene at 9 p.s.i.g. pressure were used. The acetylene was metered to the burner at a flow-rate of 5 on the panel-mounted flow meter. The burner height was set at 5 on the height adjustment. The lamp used was made by Atomic Spectral Lamps, Melbourne, Australia, and was operated at 4 mA with a coarse gain setting of 4. The 328.1-nm line of silver was used for all measure- ments. The air supply was cleaned and regulated with a Hymatic filter-regulator (Varian Techtron Corp., California, U.S.A.). All of the glassware used was Pyrex. CLEANLINESS OF GLASSWARE- In order to prevent the possibility of silver sorbed on the container walls being carried over into subsequent analyses, Nakagawa and Lakin4 advise a final rinse with dilute cyanide solution.The final sample solutions in our method contain 30 ml of concentrated ammonia solution per 100 ml, and silver is apparently not sorbed from such solutions on to the container walls to a significant extent. We found that one rinse with detergent solution, followed by three rinses with tap water and three with distilled water, is sufficient to remove all silver ions. Glassware cleaned in this manner and refilled with a solution containing 5 per cent. of con- centrated hydrochloric acid and 30 per cent. of concentrated ammonia solution repeatedly gave no detectable silver signal. Ill., U.S.A.338 WALTON: A METHOD FOR THE DETERMINATION OF SILVER IN ORES AND [AnaZyst, Vol. 98 PREPARATION, USE AND STABILITY OF STANDARDS- Standards with silver values similar to those of the samples were run on the atomic- absorption equipment immediately before and after the samples themselves.The standards were made by dissolving silver wire of known high purity in the minimum amount of nitric acid to make 1 litre of concentrated stock solution containing 0.9986 g 1-1 of silver. From this solution, a dilute stock solution was prepared, which contained 0.09986 g 1-1 of silver (0.09986 mg ml-l). The working standards were prepared from this dilute silver stock solution. Except when in actual use, all standards and stock silver solutions were kept in darkness. To prepare the working standards, a measured amount of the dilute silver stock solution was placed in a 100-ml calibrated flask together with 5 ml of concentrated hydrochloric acid; about 50 ml of water were added, then 30 ml of concentrated ammonia solution, and the solution was cooled to room temperature.Finally, the volume was made up to the mark with water and the solution was shaken to ensure homogeneity. Standards covering the range 0.0 to 10.0 p.p.m. of silver were prepared in this manner. A series of standards prepared by including the addition of acid and evaporation, as for actual ore samples, gave identical results. As found by Lockyer and Hames3 and others, the graphs of absorbance ueysus concen- tration also gave straight-line calibration graphs. However, sample results were calculated from the atomic-absorption spectroscopic results obtained by running suitable standards simultaneously with the samples, thus negating any changes in the absorbance to concen- tration ratio due to burner adjustments, etc.The standards prepared for this study containing 5 per cent. of concentrated hydrochloric acid and 30 per cent. of concentrated ammonia solution showed no measurable instability over a 7-month period. Fresh standards prepared at intervals up to 210 days later gave identical atomic-absorption spectroscopic results within the limit of the reading error involved ( 5 0 . 3 per cent. transmittance). In contrast, standards prepared for two solvent-extraction methods (dithizone in ethyl propionate, and triisooctyl phosphorothioate in benzene) gave substantially lower atomic-absorption signals within 4 hours. The excellent stability of the working standards prepared from pure silver is one of the main advantages of our method.PROCEDURE- For each 2 g of ore used, add, in a 260 or 400-ml beaker, 5 ml of concentrated hydrofluoric acid, 10 ml of concentrated hydrochloric acid, 10 ml of concentrated nitric acid and 5 ml of concentrated perchloric acid (see preca~tionsl~). Evaporate the mixture to dryness on a hot-plate, covering the samples with a watch-glass for the first 30 minutes. Add 5ml of concentrated hydrochloric acid plus 30 ml of water (a portion of this water can be used to rinse any residue that remains on the watch-glass into the beaker). Warm the mixture for 5 minutes on a hot-plate, then decant off the liquid into a 100-ml calibrated flask. Rinse the beaker and the residue with 5ml of water, allow the solid to settle and decant off the liquid into the flask.Rinse the beaker and the residue with 30 ml of concentrated ammonia solution, allow the solid to settle and decant off the liquid into the flask. Rinse the beaker and residue with one or two more 5-ml portions of water, allow the solid to settle and decant off the liquid into the flask. Mix the contents of the flask, cool to room temperature, make the liquid in the flask up to the 100-ml mark with water, stopper the flask and shake it. Allow any residue to settle. Adjust the instrument settings to read 100 per cent. transmittance on a solution con- taining 5 ml of concentrated hydrochloric acid and 30 ml of concentrated ammonia solution per 100 ml. Standardise the instrument at 328-1 nm by using standards of known silver content in the appropriate range of the sample values and containing 5 ml of concentrated hydrochloric acid and 30 ml of concentrated ammonia solution per 100 ml of final solution.Aspirate the clear supernatant sample solution directly into the flame of the atomic- absorption instrument, with a wavelength setting of 328.1 nm. (If desired, a small portion of the sample solution can be filtered and the filtrate aspirated into the flame.) The time required for this method is about 3 hours, most of which is devoted to evaporat- ing the solutions to dryness.May, 19731 NINERAL PRODUCTS BY ATOMIC-ABSORPTION SPECTROSCOPY 339 LOSS OF SILVER BY SORPTION- The possible loss of silver by sorption on the walls of the containing vessels has been considered by West, West and Ramakrishnal and by Rubeska, Sulcek and Moldan,’ who recommend the use of mercury(I1) ion in order to form complexes, thus inhibiting the sorption of silver.This method would have the greatest application when extremely small silver values are to be considered, as for example, in the determination of silver in natural waters. To several of the samples we added mercury(I1) ions in the recommended amount (30 mg) before the dissolution procedure. No difference was found in the silver values obtained with and without mercury(I1) ions, and we conclude that the sorption of silver from the sample solutions is negligible. The excellent long-term stability of the standards also supports this view. DETERMINATION OF SILVER BY ATOMIC-ABSORPTION SPECTROSCOPY- The atomic-absorption signal of many metals can be enhanced considerably by the use of organic solvent^.^^^^^-^^ The use of these solvents may be desirable or even necessary in certain instances (very low metal values, for example), but it is not essential to our method.On the contrary, the use of solvents as a concentration device or to enhance the signal intro- duces additional complications that are best avoided in a precision method. In our method, we always obtained straight-line relationships between the absorbance and the silver concentration in the range 0 to 10 p.p.m. of silver at 328.1 nm. In the region of 50 per cent. transmittance, our estimated sensitivity of 50.3 scale division corresponds to about &O-10 troy oz per net ton for a 2-g ore sample when using a final dilution to 100 ml.This sensitivity corresponds to a reading error of about h1.5 per cent. in the determination of silver, for an average sample. As the precision of the fire-assay method is usually con- sidered to be about 2.5 per cent. or more, our method has a comparable precision, even without the use of solvent-concentration or signal-enhancement techniques. The high concentrations of ammonia in our final sample solutions are preferable to high concentrations of acid in terms of equipment corrosion problems. In order to show that the atomic-absorption signal obtained a t 328.1 nm was derived only from silver ions, the final solutions from several samples (made 8 N with respect to nitric acid) were extracted with a 30 per cent. solution of triisooctyl phosphorothioate in benzene.According to Nakagawa and Lakir~,~ based on work by Handley and Dean,lG this extractant is not specific for silver, but is highly selective [only mercury(II), tantalum, vanadium and silver can be extracted under the conditions used]. After four successive extractioi~;, the sample solutions (aqueous phase) gave readings of 100 per cent. transmittance at 328-1 nm, showing that the element that generated the signal obtained at this wavelength was com- pletely extracted by the triisooctyl phosphorothioate in benzene. Significantly high con- centrations (1000-fold molar concentrations based on the silver originally present) of mer- cury(II), tantalum and vanadium ions added to the original sample solutions failed to generate a visible signal at this wavelength.We therefore conclude that the signal originally obtained from the sample solutions at this wavelength was due only to the presence of silver ions. In order to prove that further additions of silver to our sample solutions would give absorbance read-outs that are linearly proportional to the concentrations of silver added, we added known Concentrations of standard silver stock solutions to aliquots of the final sample solutions, then re-measured the absorbances of the solutions (the “method of standard addition”). The results obtained show that the “recovery” of the added silver by the atomic- absorption procedure in our sample solutions was linear, and very close to that expected by comparison with standards based on the pure metal.In twelve such experiments, a total of 39-179 p.p.m. of silver was added and a total of 39.060 p.p.m. of silver was “recovered” by the atomic-absorption procedure. The apparent net loss of 0.3 per cent. is within the error of measurement. We therefore conclude that any other substances present in the final solutions exerted neither an enhancing nor a depressing effect upon the atomic-absorption signal for silver under the conditions used. The fact that we were able to extract the silver from the sample solutions so as to give a reading of 100 per cent. transmittance, and the fact that we were able to obtain a linear recovery of the silver added to the sample solutions, is strong evidence that background absorption is not a significant factor in our method.340 WALTON: A METHOD FOR THE DETERMINATION OF SILVER IN ORES AND [Analyst, Vol.98 RE-EXAMINATION PROCEDURES- One of the major difficulties encountered in this study was that of obtaining complete dissolution of the silver present in the ores. In view of this problem, eight of the oxide ores used in this study were submitted to a re-examination procedure. In every instance, an additional amount of silver, ranging from 2.3 to 5.7 per cent. of the amount found in the first determination, was recovered (see Table 111). TABLE 111 COMPARISON OF RECOVERIES OF SILVER FROM ORES BY ATOMIC-ABSORPTION SPECTROSCOPY FROM FIRST DETERMINATION, RE-EXAMINATION AND SECOND RE-EXAMINATION PROCEDURES Ore Oxide ores- No. 1 No. 6 No. 9 No. 11 No. 13 No. 17 No. 18 No. 20 No. 2 No.12 No. 15 No. 16 Sulphide ores- Average silver content (troy oz per net ton) in 7 first first determination re-examination 12.90 16-05 8-12 2.19 4-13 7.35 4.39 3.65 2.87 2-56 5-22 9.9 1 Sulphide concentrate- No. 19 1-33 0.42 0.74 0.28 0.05 0.10 0.23 0.25 0.15 0-07 0.03 0.09 0.17 0.02 Per cent. of silver recovered on first re-examination 3.3 4.6 3.4 2.3 2.4 3.1 5-7 4.1 2.4 1.2 1.7 1-7 1.5 Average silver content (troy oz per net ton) in second re-examination 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 A second re-examination procedure performed on the same samples did not yield any further signals for silver. It therefore appears that for the highest accuracy to be obtained, it is essential to add the results from one re-examination procedure to those from the original determination, but that additional re-examinations are not necessary.Four sulphide ores gave additional silver values of 1-2 to 1.7 per cent. of the amount found in the first determination on re-examination (Table 111). In general, it is probable that this type of ore, which contains silver in association with galena, is more soluble than the siliceous oxide ores. Mill products, such as blister copper, which are completely dissolved by the method given, would of course require no re-examination. RESULTS We examined twenty samples, consisting of oxide and sulphide ores, sulphide concen- trate and blister copper, by the proposed method, with six to eight atomic-absorption spectro- scopic determinations being made on each sample. The results are presented in Table IV, which also gives the average and range of values and the percentage standard deviation.The results show an average silver content obtained by atomic-absorption spectroscopy that is 10.9 per cent. higher than that obtained by the fire-assay method. The average relative standard deviation is 2.2 per cent. DISCUSSION A substantial amount of recent information has shown that the fire-assay method for determining precious metals is subject to serious negative errors due to loss of the metals at various stages of the determination. For ruthenium, losses of up to 30 per cent. have been found,17 for rhodium up to 12 per cent.,ls iridium up to 63 per cent.19 and platinum as high as 89 per cent. (for neutral slags).20& . y , 19731 MINERAL PRODUCTS BY ATOMIC-ABSORPTION SPECTROSCOPY TABLE IV RESULTS FOR THE DETERMINATION OF SILVER IN ORES AND OTHER MINERAL PRODUCTS BY FIRE-ASSAY AND ATOMIC-ABSORPTION METHODS 341 Sample No.Type 1 Oxide ore 2 Sulphide ore 3 Oxide ore 4 Sulphide 5 Oxide ore 6 Oxide ore 7 Blister copper 8 Oxide ore 9 Oxide ore 10 Oxide ore 11 Oxideore 12 Sulphide ore 13 Oxide ore 14 Blister copper 15 Sulphide ore 16 Sulphide ore 17 Oxide ore 18 Oxide ore 19 Sulphide 20 Oxideore concentrate concentrate Average Best values for silver (troy oz per Standard Range (troy oz per net ton) deviation of atomic- value for silver net ton) by atomic- of atomic- absorption Difference, Percentage No. of by fire absorption absorption results B - A difference, deter- assay spectro- results, (troy oz per (troy oz B - A x 100 8 12.02 13.32 1.1 0-45 1-30 + 10.8 6 2.80 2-94 3.1 0.3 1 0.14 + 5.0 minations ( A ) scopy (R) per cent.net ton) per net ton) 7 8 6-73 7.14 2.1 0.70 0.41 + 6.1 6 8 8 8 7 8 6 6 6 6 6 7 7 6 6 0-64 1.55 16.20 3-2 1 14-6 8-00 8-85 2-10 2-40 3-87 2-04 5.02 9.25 7-32 4.05 0.80 1.64 16.78 3-56 19.8 8.40 9.79 2.24 2.5D 4.23 2-38 5.30 10.08 7-58 4-64 2.8 2.7 1.7 4.1 4.8 4-6 0.9 2.1 0.5 3-7 1.1 1.5 1.7 0.1 1.6 0.09 0.19 1.37 0.46 3.03 1-49 0.55 0-17 0.1 1 0-54 0.15 0.41 0.75 0.22 0.33 0-16 0.09 0-58 0.35 5.2 0.40 0.94 0.14 0.19 0.36 0.34 0.28 0-83 0.26 0.59 + 25.0 + 5.8 + 3.6 + 10.9 + 35.6 -/- 5.0 -+ 10.6 + 6.7 + 8.0 + 9.3 + 16.7 + 5.6 + 9.0 + 3.6 + 14.6 6 1.16 1.35 2.7 0.12 0.19 + 16.4 6 3.48 3.80 1-4 0-24 0-32 + 9.2 Average . . 2.2 Average .. +lorn9 A recent radiochemical study of the fire-assay method for determining silver was reported by Faye and InmanP2l who showed that the major loss of this metal occurs in the cupellation step, approximately 2.5 per cent. being lost to the cupel even at 890 to 900 "C. They noted that at 1000 "C this loss is doubled, and also mentioned that "under certain conditions often used in practice the total error due to losses may be as high as 5 to 10 per cent." They also mentioned that the loss of silver to the cupel is about 25 per cent. higher for bone-ash than for magnesia cupels, and indicated that buttons of less than 25 g will result in incomplete collection of the silver. In another radiochemical study, by Nakamura and Fukami,22 losses of silver (again, primarily to the cupel) from 5 to 30 per cent.were reported. In the above studies, loss of silver by volatilisation has been for the most part neglected, as it is small compared with other sources of error. However, even in 1911, F ~ l t o n ~ ~ reported losses of silver by volatilisation of about 5 to 6 per cent. from 750 to 1000 "C, and losses of up to 29 per cent. by volatilisation of silver in the presence of gold (silver to gold ratio of 2: 1). Fulton also quoted extensive work by Eager and Welch, Godshall, Kaufman and Hillebrand, and Allen on the loss of silver under various conditions of cupellation. Although these two processes appear to cause the greatest losses of silver during the fire assay, there are others. Losses of silver during crucible fusion and scorification have also been noted.The above studies support the view that fire-assay methods, despite their extensive usage and historical background, are obviously subject to losses of metal that can substantially affect their accuracy. Except for the operations of weighing the pulp and the final bead, which operations could be subject to either positive or negative errors, all other major errors inherent in the fire-assay method are negative. In view of this and of the foregoing studies on the various possible sources of error, it is hardly surprising that our atomic-absorption spectroscopic342 WALTON results yield silver values that are consistently higher than those obtained by the fire-assay method. Rawling, Greaves and Amos,24 working with lead concentrates, reported results obtained by an atomic-absorption method for the determination of silver that were, on average, about 0.5 troy oz per net ton higher than those obtained by the fire-assay method in the range 16 to 18 troy oz per net ton. CONCLUSIONS The proposed atomic-absorption method gives an average silver value in the ores and mineral products studied that is 10.9 per cent.higher than that given by fire assay. We have attempted to show that the signal obtained from the samples is due to silver alone, and that the presence of silver in the sample solutions gives a linear signal response for absorbance versus concentration. Under these circumstances, it can be concluded that there is actually a higher silver content in these samples, and that the bias of negative errors inherent in the fire-assay method is responsible for the comparatively low results obtained by that method.Much of the economics of mining is dependent upon the ultimate accuracy of determina- tions of precious metal values in the ore, and we suggest that more accurate results for such metal values are obtained by the proposed atomic-absorption method. Also, procedures for the recovery of metals are aimed primarily at the recovery of metal values as predicted from assays. If there is actually a higher silver content than is indicated by the fire-assay method, this fact must be recognised before serious attempts can be made to win the higher metal values from the ore by metallurgical procedures. We gratefully acknowledge assistance and special supplies from Jesse F. Bingaman and David M.Dennis of Western New Mexico University, Silver City, New Mexico; Harold E. Richard of Hawley & Hawley (Tuscon, Arizona) ; Dickenson Laboratories (El Paso, Texas) ; Walter Cox and the Hurley Laboratories of the Kennecott Copper Corp. (Hurley, New Mexico) ; W. R. Hein and W. C. Lashley of the Silver City Testing Laboratories; and J. Wilson and Ken Wallin of the Gila Analytic and Research Laboratory (Silver City). This research was financed in part by a grant from the Research Committee of Western New Mexico University. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. REFERENCES West, F. K., West, P. W., and Ramakrishna, T. V., Environ. Sci. Technol., 1967, 1, 717. Greaves, M. C., Nature, Lond., 1963, 199, 552. Lockyer, I<., and Hamcs, G. E., Analyst, 1059, 84, 385. Nakagawa, H. M., and Lakin, H. W., Prof. Pap. U.S. Geol. Surv., No. 525-C, 1965, C172. Huffman, C., jun., Mensik, J . D., and Rader, L. F., Ibid., No. 550-B, 1966, €3189. Wilson, L., Analylica Chim. Acta, 1964, 30, 377. Rubeska, I., Sulcelr, Z., and Moldan, B., Ibid., 1967, 37, 27. Hickey, L. G., Ibid., 1968, 41, 546. Olsen, A. M., Atom. Absovption Nezusl., 1965, 4, 278. Thompson, C. E., Nakagawa, H. M., and Van Sickle, G. H., Prof. Pap. U.S. Geol. Surv., No. 600-J3 Huffman, C., jun., Mensik, J . D., and Riley, L. B., Circ. U.S. Geol. Surv., No. 544, 1967, 1. Belcher, R., Dagnall, R. El., and West, T. S., Talanta, 1964, 11, 1257. Muse, L. A., J . Chem. Educ., 1972, 49, A463. Allen, J. E., Spectrochim. Acta, 1961, 17, 467. Lockyer, R., Scott, J. E., and Slade, S., Nature, Lond., 1961, 189, 830. Handley, T. H., and Dean, J . A., Analyt. Chem., 1960, 32, 1878. Thiers, R., Graydon, W., and Beamish, F. E., Ibid., 1948, 20, 831. Allen, W. F., and Beamish, F. E., Ibid., 1950, 22, 431. Barefoot, R. R., and Beamish, F. E., Ibid., 1952, 24, 840. Hoffman, I., and Beamish, F. E., Ibid., 1956, 28, 1188. Faye, G. H., and Inman, W. R., Ibid., 1959, 31, 1072. Nakamura, Y., and Fukami, K., Japan Analyst, 1957, 6, 687. Fulton, C. H., “A Manual of Fire Assaying,” 2nd Edition, McGraw-Hill, New York, 1911. Rawling, B. S., Greaves, M. C., and Amos, M. D., Nature, Lond., 1960, 188, 137. 1968, B130. Received November 14th, 1972 Accepted January M A , 1973