468 I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY Vol. 14, No. 6

tion of the-total of copper and iodate. Table X gives the results.

The titrations of Table X have a sulfuric acid concentra- tion of 0.15 formal and, where copper is present, no consistent error is noted. Since this acid concentration is somewhat lower than that usually specified for the bromate-iodide re- action ($, 4) especially in view of the results of experiment 60, i t is suggested that copper may catalyze this reaction.

TITRATION OF TOTAL CHLORATE AND COPPER WITH THIO- SULFATE. The chlorate-iodide reaction requires such a high hydrochloric acid concentration that the iodometric determina- tion of copper under these conditions was not found feasible.

Summary Iodate, permanganate, bromate, and chlorate in the pres-

ence of cupric salts can be determined by adding an excess of

a standard iodide solution to neutral solutions of the mixtures, acidifying, and back-titrating with standard oxidizing solu- tions. An alternative method, investigated in the case of iodate-copper mixtures, consists in titrating a standard iodide solution with the mixture, all solutions having been saturated with carbon dioxide previous to titration.

The total amount of iodate and copper, of copper and per- manganate, and of copper and bromate can be determined iodometrically with thiosulfate.

Literature Cited (1) Foote and Vance, J. Am. Chem. SOC., 57, 845 (1935). (2) Kapur and Verma, IND. ENG. CHEM., ANAL. ED., 13, 338 (1941). (3) Kolthoff, Pharm. Weekblad., 56, 391-3 (1919). (4) Kolthoff and Furman, “Volumetric Analysis”, p. 387, New York,

(5 ) Swift, J. Am. Chem. SOC., 52 , 894 (1930). John Wiley & Sons, 1929.

Determination of Small Amounts of Gold with Stannous Chloride

COLIN G. FINK AND GARTH L. PUTNAM, Columbia University, Kew York, N. Y.

Previously known modifications of the widely accepted test for gold, the stannous chloride test, are shown to be qualitatively unreliable because of failure to take into account the factor of acid con- centration. So-called colloidal gold may exist in two distinct forms: the yellow form produced in solutions of low acidity, and the purple form pro- duced in solutions of high acidity. The low acid- stannous chloride test developed by the authors is more reproducible, sensitive, and reliable than the high acid test.

HE purpose of this paper is to present a reliable method T for the qualitative detection and estimation of gold. Notwithstanding the fact that of all metals gold is second only to iron in economic importance ( I l ) , experience has indicated that a rapid and reliable chemical method for the qualitative detection and colorimetric determination of traces of gold has not yet been reported in the standard analytical texts. Here- tofore, the most widely accepted colorimetric test for gold has been the stannous chloride “purple of Cassius” test (3, 6, 7 , 10, I,$’), which was discovered by Andreas Cassius in 1663 (8, 9). The authors find that previous descriptions of the test are qualitatively erroneous, as the factor of acid concentra- tion has not been taken into account, and report below a re- liable method for the detection and estimation of traces of gold.

The following reagents for the detection of gold have been considered, tried, and rejected: hydrogen peroxide, hydrogen peroxide-potassium hydroxide, potassium iodide, potassium iodide-potassium hydroxide, potassium iodide-potassium mercuric iodide-potassium hydroxide, resorcinol, resorcinol- potassium hydroxide, formaldehyde, formaldehyde-potassium hydroxide, benzaldehyde, benzaldehyde-potassium hydrox- ide, potassium formate-potassium hydroxide, acetylene- potassium hydroxide, sodium thiosulfate, ferrous sulfate, sodium bisulfite, sodium sulfite-potassium hydroxide, hydro- gen sulfide, and glucose-potassium hydroxide. When used under the proper conditions, stannous chloride was found to give a more sensitive and more reproducible test than any of the above reagents.

Factors Influencing Stannous Chloride Test

The alleged variation in tint of color with variation in the gold concentration is the basis of the colorimetric determination as given in standard analytical texts (3, 6, 7 , 10, la ) .

TABLE I.

EFFECT OF CONCENTR-QTION OF ACID ON T I N T OF COLOR.

EFFECT OF ACID CONCENTRATION ON TINT OF COLOR

N 70 1 0.002 20 10.6 Yellow 2 0.08 10 8.4 Light brown 3 0.16 20 12.9 Tan with faint purple tinge 4 0.32 10 10.2 Tan with distinct purple tinge 5 0.64 10 13.6 Purple

a Includes acid present in stannous chloride reagent. b These values have little bearing on the reproducibility of the test under

ordinar analytical conditions, as i t is very di5cul t to prepare standards which gave exactly the kame ages and compositions as the unknowns. Values are included to show order of magnitude of experimental error.

In order to determine the effect of acid concentration on the tint of color, 40 ml. of a solution containing 0.20 mg. of gold, as bromide, were added to each of ten 50-ml. Nessler tubes, and the solutions were made up to the mark with distilled water and h - drochloric acid. The solutions therefore contained 4 mg. of goyd per liter. The Nessler tubes were closed with a rubber stopper (that had been soaked in chlorine water), and inverted twice. To each tube waa then added 1 ml. of 0.25 M stannous chloride (see Reagents), and the tube was again closed with a rubber stop- per and inverted twice. One or two groups of ten trials each were made for each acid concentration. After standing for 15 minutes, one of the tubes of each series was taken as the standard and the rest of the tubes were compared with it by the balancing tube method.

The data of Table I prove that the acid concentration is an important factor influencing the tint of color in the stannous chloride test. No previous investigator has recognized this (3, 6, 7 , 1 0 , l Z ) . Gold cannot be determined by the stannous chloride method unless the acid concentration of the colori- metric medium is taken into account.

EFFECT OF ACID CONCENTRATION ON VELOCITY OF COLOR DEVELOPMENT. As the velocity of color development has an important bearing on the precision and reproducibility of the test, the effect of acid concentration was studied from this

The results are given in Table I.

June 15, 1942 A N A L Y T I C A L E D I T I O N 469

standpoint. Color development takes place more slowly at the lower gold concentrations, and therefore only 0.10 mg. of gold was used in each trial. I n other respects, the method of preparing the solutions was similar to the method used in de- termining the effect of acid concentration on the tint of color. The time of starting each trial was taken as the time when the Kessler tube was first inverted, after addition of the stannous chloride reagent. After standing for 10 minutes, 2.5 ml. of solution were withdrawn from the Kessler tube, a second trial, identical in composition to the previous trial, was started, and the time required for the two trials to match was noted. Three experiments (six trials) were made with each acid concentration. The results are recorded in Table 11.

TABLE 11. EFFECT OF ACID COSCENTRATION o s VELOCITY OF COLOR DEVELOPMENT

N Sec. 1 0.002 32 Yellow 2 0.08 73 Yellow 3 0.16 400 Light brown 4 0 .64 820 6 5 . 0 1420

0 Includes hydrochloric acid added with stannous chloride reagent. b Average of three experiments (six trials). c With 0.64 N and 5.0 N hydrochloric acid solutions there was a distinct

difference in t int of color between trials 20 minutes and 30 minutes old. d .\fore than 180 seconds required for development of visible coloration.

E f % l u e d

As shown by Table 11, color development takes place more rapidly in the weakly acid solutions. Similar results (not given in Table 11) were obtained vcith gold chloride in place of gold bromide. Moreover, when a limited time is available for the development of the color, the test is more sensitive at the lower acid concentrations.

EFFECT OF GOLD CONCENTRATION ox COLOR INTENSITY. Except for the differences in technique indicated below, the procedure used in determining the effect of gold concentration on color intensity was similar to the procedure used in deter- mining the effect of acid concentration on tint of color. The gold was present as bromide and 0.61 ml. of stannous chloride was used for each trial. After standing for 10 minutes, the colorimetric comparisons were begun (see Comparison of Colors below). One of the trials containing 0.50 mg. of gold was used as the standard for the other trials of Table 111. Approximately 25 minutes from the time of starting the trials were required to complete the tests of each series.

TABLE 111. EFFECT OF GOLD CONCENTRATION ON I~VTENSITY OF COLOR

Average

er Liter of from Series E o E Z r 2 8 olorimetric ' Mean of

NO. Medium Medium0 Series

Gold Added Gold Found Deviation

M g .

oxides or oxychlorides caused coprecipitation of the gold con- tent. The purple form of colloidal gold, in marked contrast to the yellow form, is not stable and invariably precipitates within a few hours from the strongly acid solutions.

I n connection with the effect of acid concentration, a very interesting fact was observed which to the authors' knowledge has never been recorded in the literature: On adding hydro- chloric acid to the yellow solutions, in amount sufficient to bring the concentration to within the range 2 N to B N , the color changes to blue or purple. This color change has been ob- served with solutions containing up to 300 mg. of gold per liter. With gold concentrations of 2 to 20 mg. per liter, the addition of hydrochloric acid causes the yellow solutions to become colorless for a period ranging from a fraction of a second to several minutes, after which a blue or purple color develops.

Regarding the stannous chloride concentration, a series of 24 trials with 0.25 mg. of gold as bromide dissolved in 50 ml. of solution, indicated that within the range of 0.0014 M to 0.14 M stannous chloride in the colorimetric medium, the color intensity is not a function of the stannous chloride con- centration when the low acid test is used. Since stannic halides hydrolyze more readily than the stannous compounds, the amounts of excess oxidizing agents in the gold halide solu- tions should not be much larger than necessary to maintain the gold in solution. Sodium chloride and many other salts accelerate the formation of turbidity in dilute stannous chlo- ride solutions, and for this reason the salt concentrations should be reduced to the lowest practicable values.

Reagents In a 30-ml. test tube were

placed 22.6 grams of c. P. stannous chloride crystals (SnC12.2Hz0) and 2.33 ml. of 11.2 M hydrochloric acid. The mixture was heated and maintained at the boiling point fo: 2 or 3 minutes, and the clear solution was cooled to 40" to 50 , oured into 300 ml. of distilled water, and made up to a final voLme of 400 ml. Turbid reagents, or reagents more than about 8 hours old, were not used.

In a covered 250-ml. beaker were placed 125 mg. of c. P. precipitated gold powder, which did not have a metallic luster, and 5 mi. of satu- rated bromine water. Agglomerated particles were crushed with a stirring rod and 10 ml. of bromine water added, 0.5 to 2 ml. at a time; after each addition the beaker was kept covered with a watch glass, to prevent loss of bromine, and swirled. The gold bromide solution was warmed to about 50" in order to re- move most of the excess bromine and then poured into a 250-ml. volumetric flask. The undissolved gold remaining in the beaker was treated with 2-ml. portions of bromine water until the gold dissolved, less than 20 ml. of bromine water being required. The gold bromide solution was warmed to 50", transferred to the 250-ml. volumetric flask, and made up to the mark with distilled water. This reagent, containing 500 mg. of gold per liter, is stable for several months.

The aqua regia solution of gold chloride was prepared according to Treadwell and Hall (IO). The aurous cyanide solution was prepared by aspirating air through a sus- pension of precipitated gold in dilute sodium cyanide solution. The chlorine gas used for the preparation of the chlorine water was generated by the action of hydrochloric acid on calcium hypochlorite. The chlorine concentration in the chlorine water was determined by the potassium iodide method (6).

STANNOUS CHLORIDE (0.25 M ) .

AURIC BROMIDE (500 MG. OF GOLD PER LITER).

OTHER REAGENTS. '

4 Data are mean of several trials a t each gold concentration. One of Sessler tubes of series 3 (0.60 mg. of gold) was used as standard for other trials of table.

Table I11 indicates the possibility of estimating the gold content of an unknown within about 12 per cent, even though the standard has a gold content very different from that of the unknown.

OTHER CHARACTERISTICS OF THE LOW ACID-STANNOUS CHLORIDE TEST. Regarding the stability of the yellow solu- tions, the authors have never observed any precipitation except in those cases where a flocculent precipitate of stannic

Recommended Procedure Colors were compared with matched

50-ml. Nessler tubes having a maximum length variation of about 2.5 per cent. The standard and unknown test solutions were diluted until the color intensities of 50-ml. portions were a proxi- mately equal. In the cases of the trials of Table IS', the dht ions were made in portions of 20 ml.; in all other experiments the dilutions were made in portions of 50 ml. Final comparison of colors was made by pouring the more highly colored solution from its Xessler tube until the colors matched and measuring the depths of the solutions. Thus a combination of the dilution and balancing tube methods was used.

COMPARISON OF COLORS.

470 I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y Vol. 14, No. 6

PROCEDURE FOR DETERMISATION OF GOLD IN CYANIDE SOLD TIOXS. The procedure used was that of Bugbee (3), modified as indicated:

One liter of the cyanide solution containing gold was poured into a 4-liter beaker, 0.12 ml. of saturated lead acetate solution added, and the resulting precipitate dissolved by stirring. To the solution was added 0.75 gram of zinc dust, and the solution was swirled. After addition of 20 ml. of saturated sodium cyanide and 20 ml. of concentrated ammonia solution, the solution was vigorously stirred for 5 minutes. The lead sponge precipitate was allowed to settle and all but about 50 ml. of the solution de- canted. The precipitate was transferred to a filter containing an 11-cm. Whatman No. 2 paper and washed to the apex of the filter paper. After thoroughly washing with distilled water, dilute (10 to 25 per cent) nitric acid was cautiously added until all the precipitate, with the exception of the black gold particles, had dissolved. The filter paper was washed with distilled water, and the black gold particles were transferred to the apex. A 100-ml. beaker was then placed under the funnel, and 10 ml. of saturated bromine water were passed through the filter paper, 1 ml. at a time, over a period of 15 minutes, the funnel being kept covered to prevent loss of bromine. If black particles remain on the filter paper after this treatment, it is generally an indication of the presence of carbon in the zinc dust rather than of incom- plete solution of the gold.

The filter paper was washed with 8 ml. of water, and the fil- trate containing the gold heated to about 45” on a hot plate. The beaker was removed from the hot plate and allowed to stand for about 40 minutes, during which time most of the bromine es- caped from the solution. The solution was placed in a 50-ml. Xessler tube and made u nith distilled water to a volume of 18 ml., and 2 ml. of 0.25 .@?stannous chloride reagent were added. Immediately after the stannous chloride addition, the tube was closed with a rubber stopper that had been washed with chlorine water, and inverted twice. The gold standards, before addition of the stannous chloride, had a gold concentration of 1.73 mg. per liter of gold as bromide, or 0.0311 mg. of gold in 18 ml. of standard solution. After standing for 10 minutes or more, the relative color intensities were compared as indicated above.

Experimental Evidence Experiments with the low acid-stannous chloride method

having demonstrated its superiority to the standard test method in which the acid concentration is both high and un- controlled, confirmatory trials were made with a synthetic leaching soluticn. The concentrations of the various constitu- ents were (grams per liter): free sodium cyanide, 0.5; iron, 0.025; copper, 0.025; and calcium oxide, 0.6.

The iron mas added in the form of potassium ferrocyanide. Copper was added in the form of sodium cuprocyanide, which was prepared by adding sodium cyanide to copper sulfate solution until the precipitate which first formed had just dissolved. Variable amounts of aurocyanide solution (see Reagents) were also added. The synthetic leaching solution is believed to be similar to those encountered in cyanidation practice. Some results obtained with this procedure as ap- plied to this solution are given in Table IV.

TABLE IV. ESTIMATION OF GOLD IN CYANIDE SOLUTIONS BY THE LOW ACID-STANNOUS CHLORIDE TEST

Trial N o . Gold Added0 Gold Foundb Error M V , .I1 g, R

1 0.00 0.004 . . 2 0.00 0 .000 . .

Mean 0 .00 0,002 . .

3 0.0311 0 , 0 2 6 - 16 4 0.0311 0.022 - 29

Mean 0.0311 0,024 -22 5

5 0.0467 0.048 + 3 6 6 0.0467 0.043 - 7 . 9

Mean 0.0467 0.04% - 5 , s

7 0.0622 0 .076 +22 8 0.0622 0.064 f 3

Mean 0.0622 0.070 + 1 2 . 5

9 0.0934 0.090 - 3 . 6 10 0,0934 0.102 + 9 . 2

Mean 0.0934 0.096 + 6 . 4

11 0.124 0.112 - 9 . 9 12 0.124 0.112 - 9 . 9

Mean 0 .124 0.112 - 9.9 0 Dissolved in 1 liter of synthetic leaching solution, b Corrected for blank (trials 1 and 2).

Discussion The tints of color obtained in the experiments of Table I V

(acid concentration below about 0.05 N ) varied from light yellow to light brown; in no instance was a purple or blue color obtained. Thus, for the first time, the authors have been able to make the stannous chloride test under such conditioiis that a constant tint of color is obtained, regardless of the gold concentration

With respect to reproducibility, precision, and accuracy, the method is superior to the stannous chloride test given in the standard texts. Following directions given in standard texts, the authors have found it impossible to obtain the stipu- lated qualitative and colorimetric results (4). The tint of color obtained in the stannous chloride test for gold is primar- ily a function of the acid concentration rather than of the gold concentration, as shonn by Table I.

Table IV indicates that by controlling the acidity, the stan- nous chloride test can be placed on a semiquantitative basis. The maximum errors were of the order of 0.01 mg. There are no cupellation losses with this method, and special appara- tus, such as muffle furnaces and assay balances, is dispensed with. The errors in weighing with an assay balance are generally of the order of 0.01 mg. ( I ) , corresponding to 33 per cent error with 0.03 mg., and the partial substitution of filtra- tion for decantation probably reduces the errors due to loss of gold when separating the gold particles froin solution ( 2 ) . When less than 0.04 mg. of gold is to be determined, the low acid-stannous chloride test is believed to be superior to the gravimetric assay procedure with respect to accuracy and speed.

When a yellow or brown solution, pi-epared by the low acid- stannous chloride method and containing gold in the colloid state, is treated with sufficient hydrochloric acid to bring the acid concentration to within the range 2 N to 6 N , the solution changes from the relatively stable yellow form to the unstable purple form. With gold concentrations of 2 to 20 mg. per liter, there is a distinct time interval during which the solu- tion becomes colorless before development of the purple color.

The time required for development of the color to full inten- sity is much less a t low than a t high concentrations of hydro- chloric acid, and the test is therefore probably more repro- ducible a t low concentrations of hydrochloric acid.

to 0.14 iM stannous chloride, the tint and intensity of color with the low acid-stannous chloride test are not appreciably affected by change in the stannous chloride concentration.

Literature Cited

Within the range of 0.0014

Bugbee, E. E., “Textbook of Fire Assaying”, 3rd ed., p. 74, New

I h i d . , p. 131.

Fink, C. G., and Putnam, G. L., unpublished work. Scott, W. W., and Furman, N. H., “Standard Methods of

Chemical .4nalysis”, 5th ed., Vol. 1, p. 273, New York, D. Van Nostrand Co., 1939.

York, John Wiley &Sons, 1940.

Ib id . , pp. 245-6.

Ib id . , p. 439. Snell, F. D., and Snell, C. T., “Colorimetric Methods of Analy-

sis”, 2nd ed., Val. 1, pp. 402-5, New York, D. Van Noetrand Co., 1936.

Thomas, A. W., “Colloid Chemistry”, p. 116, New York, Me- Graw-Hill Book Co., 1934.

Ih id . , p. 144. Treadwell, F. P., and Hall, W. T., “Analytical Chemistry”. 8th

U. S. Bur. Mines, “Minerals Yearbook 1940”, Washington,

Yoe, J. H., “Photometric Chemical .4nalysis”, Yol. 1 , pp. 191-7,

ed., Val. 1, p. 280, New York, John Wiley & Sons, 1932.

D. C., Government Printing Office, 1941.

S e w York, John Wiley & Sons, 1928. AB~TRACTED from a chapter of the dissertation submitted by Garth L. Putnam in partial fulfillment of the requirements for the degree of doctor of philosophy. This research was made possible through the Weston Fellow- ship.