ANALYTICAL CHEMISTRY
840 glycol a t constant mole fraction water. Figures 3 and 4 are similar to 1 and 2 except that density instead of refractive index is the property plotted. Figure 5 is from the data of Table 11, and is a plot of refractive index against mole fraction water at constant mole fraction methanol. Figure 6 was obtained from Figure 5 and is a plot of refractive index against mole fraction methanol at constant mole fraction water. Figures 7 and 8 are similar to Figures 5 and 6 except that density instead of refractive index is plotted. The triangular plot, Figure 9, w&sobtained from the information on the previous plots. If the density and refractive index of an unknown mixture of methanol, glycol, and water are known, the concentrat,ions of the components can be obtained by reference to Figure 9. For practical use an enlarged plot of Figure 9, measuring about 16 inches (40.64 cm.) on theside, was constructed
0.040 DOTTED CURVES ARE CONSTANT DENSITY
SOUD CURVES A R ~ CONSTANT REFRACTIVE INDEX
and found to be satisfactory. Two 1.0 0.0 pairs of proportional dividers are a 0.0 0.1 0.2 0.3 0.4 05 0.6 0.7 08 0.9 1.0 MOLE FRACTION GLYCOL great aid in interpolation. Figure 9. Density and Refractive Index of All Methanol-Glyeol-Water Mixtures The accuracy of the data and chart was checked by preparing four samples (4) Heilbron, I. M., ”Dictionary of Organic Compounds,” Vol. 11, Of various mole fractions of the components. Their refractive p. 31, New York, Oxford University Press, 1936. indexes and densities were measured and the analyses of the (5) “International Critical Tables,” Vol. 111, p. 27, New York, samples mere taken from the plot (Figure 9). The results are McGraw-Hill Book Co., 1928. compared in Table 111. (6) Lange, N. A., “Handbook of Chemistry.” 6th ed., p. 928, Sandusky, Ohio, Handbook Publishers, 1940. (7) Lawrie, J. W., “Glycerol and the Glycols.” AMERICANCHEMICAL LITERATURE CITED SOCIETY Monograph 44, pp. 331-80, New York, Chemical (1) Conrad, F. H., Hill, E. F., and Ballman, E. A., Ind. Eng. Chern., Catalog Co., 1928. 32, 542 (1940). (8) Spangler, J., and Davies, E., IND. EXQ. CHEM.,Ax.41,. ED., (2) Dunstan, A. E., 2. physik Chem., 51, 732 (1905). 15, 96-8 (1943). (3) Gallaupher, A., and Hibbert, H., J . Am. Chmn. Sac., 58, 818 (1936). RECEIYED for review April 16, 19.51. Accelited Februaiy 6 , 1952.
Fire Assay for Iridium R. R. BAREFOOT AND F. E. BEAMISTI Uniuersity of Toronto, Toronto, On turio, Canudu
M
ETHODS for the deterniination of precious metals in ores and other materials by fire assaying have been in use for many years. Few attempts have been made to check the accuracy of these methods and very little work has been done on the accurate determination of iridium in ores. This paper deals with the efficiency of the collection of iridium in fire assays and is one of a series of papers describing the work done in this 1:al)oratory on the analytical chemistry of the platinum metals. In a fire assay, iridium does not alloy with lead but is collected as a suspension in the molten metal. In order to avoid mechanical losses during the pouring of melts containing iridium, Davis ( 4 ) recommended that the fusion be cooled in the pot. Ilowever, even greater losses may occur in attempting to break t’he pot away sufficiently to free the pool of lead for analysis. In t’extbooks on fire assaying ( 3 , so), procedures are outlined for t’he analysis of the lead button by cupellation, in which the platinum metals are concentrated in silver and gold. Iridium does not alloy with silver, and as a result serious mechanical losses occur in addition to any losses to the slag. Plaksin and Marenkov
(15) reported that almost 4% of a 5-mg. sample of iridium was lost from a 75-mg. silver bead obtained by cupellation. Gilchrist (6) modified the Deville-Stas method for the determination of iridium in platinum alloys. The lead button obtained by fusing the alloy in lead !vas treated with nitric acid and then aqua regia, leaving the iridium undissolved. Iron, if present, contaminated the residue and was difficult to remove. This method could not be applied directly to lead buttons obtained from fire assays because of the insoluble slag ivhicti can never he conipletely removed from the lead. Impurities-e.g., iron and nickel-in the lead button may alfio increase the error, Iridium can be converted quantitatively to soluble salts by dry chlorination in the presence of sodium chloride ( 1 2 , 14). Procedures for the removal of base metals from solutions of platinum metals have been described ( 2 , 9, 10, I S , 1 7 ) , in which the base metals are precipit,ated at a convenient p H while the noble metals are held in sohtion as complex nitrites. Iridium map then be determined gravimet.rically by precipitation ~ t sthe Iiydi.:ited oxide ( 7 , 8, 11).
V O L U M E 2 4 , NO. 5, M A Y 1 9 5 2
841
The work was undertaken in order to determine the conditions under which iridium may be quantitatively extracted from materials by means of a fire assay arid to provide a more accurate and rapid method of determining iridium than those hitherto reported. Significant losses of iridium to basic slags were noted, although satisfactory collections of iridium could be made when neutral or acid slags were used. Two methods for the analysis of lead-iridium buttons were outlined. Losses of iridium during the cupellation of lead-iridium buttons were confirmed. A detailed examination of the efficiency of a fire assay for the extraction of iridium from ores and the sources of error are reported.
EXPERIMENTAL
Reagents. A solution of sodium chloroiridate wa5 prepaicd by the dry chlorination of iridium metal sponge in the presence of sodium chloride (12) and dissolving the salts in 0.1 N hydrochloric acid. The starting material was examined spectrographically and was found to contain a trace of iron. The solution was standardized gravimetrically by the hydrolytic precipitation of iridium as its hydrated oxide. After the oxide was reduced to the metal in hydrogen, the samples were leached with hot 2 N nitric acid, ignited, reduced again, and weighed. A blank --as subtracted. The filtrates were checked for the presence of iridium by a visual colorimetric method (16). For three successive 10.01-ml. volumes of solution, the weights of iridium were 5.048, 5.026, and 5.042 mg. The concentration of the solution was thus taken t o be 0.504 mg. of iridium per ml. A second solution was prepared by diluting a portion of the first standard solution to ten times its volume with 0.1 iV hydrochloric acid; it was standardized and was found to contain 0.0502 mg. of iridium per ml. All chemicals other than those used in the preparation of the fluxes were of reagent grade, Fire Assay Procedure. Williams and Wilson 25-cycle, 15-kw. Globar-type furnace was used for the assay work. Twenty-gram fire clay crucibles which had been glazed on the inside with a neutral flux were used for the fusions. The salting of the fluxes and ores with iridium was carried out by placing the ore, which had been mixed with the required amount of flux, in the pot, "hollowing out" the center, and then transferring the iridium solution slowly from a pipet. The mixture was then placed in an oven a t 110' C. for 4 hours or overnight. After drying, the contents of the crucible were poured into a large mortar, where the lumps a e r e broken up and the contents were thoroughly mixed. I t was found in a few experiments near the beginning of the work that some of the iridium solution was absorbed by the walls of the pots through cracks in the glaze before the contents were dried. For this reason, an inside lining of cellophane was used a s an additional precaution; enough flour was withdrawn from the mixture t o keep the size of the lead button about 30 grams. The pots were placed in the furnace a t 815" C. and the temperature was raised as quickly as possible to 1150' t o 1200' C. The time required was usually over 1.5 hours, The melts were poured into iron molds, and after cooling the slags were broken away and the lead buttons were cleansed of as much of the adhering slag as possible by gentle tapping with a pair of forceps. All the slag
Constituents S_iOz
2
Neutral
Very acid
-
80
,
per assay, grams Wt. of ore
a
1
A.T.
-
100
...
assay ton.
107
...
from a single assay was collected, reground to ,pass it 10 45 standard sieve, and reassayed in the same pot with added fiour and litharge. Composition of Fluxes. The fluxes used in this restwuh are recorded in Table I. Some of them were used in previous v ork on other platinum metals ( 1 , 11). ANALYSIS OF LEAD-IRIDIUM BUTTONS
Nitric Acid Parting and Chlorination of Residue. The leadiridium button obtained from a fire assay and weighing about 30 grams was placed in a 400-ml. beaker. A volume of 5 ml. of a 1 to 4 nitric acid solution was added for each gram of lead. The sample was heated on a steam bath until all the lead aqpeared to be dissolved. The solution was diluted with one half Its volume of hot wat,er and filtered through a double filter which consisted of a Whatman No. 44, 5.5-em. filter paper superimposed on a 7-cIll. paper of the same grade. The residue was transferred t o the filter and was washed with 50 ml. of a hot 20% solution of ammonium acetate and 50 ml. of hot water. The filter was dried and placed in a small, clean porcelain crucible. The sample wae ignited in an electric muffle a t a temperature not higher than 650" C. and then reduced in.hydrogen; 0.05 gram of sodium chloride was placed in the crucible and mixed with the residue by rolling the crucible when it was tipped at a small angle. The sample was placed in a silica tube inside a combustion furnace, h slow current of chlorine (2 bubhles per second) wag passed through the tube and the teniperature was maintained a t 650" to 700' C. for 8 hours, as in the procedure described by Hill and Beaniish ( l a ) . About 6 inches of the exit) part of the tube projected beyond the furnace and acted as a condenser for the iridium compounds, which were volatile a t the temperature inside the furnace. The sample was cooled in chlorine; the crucible and contents were placed in a 250-nil. beaker and t,he silica tube was washed out thoroughly with hot 0.1 zV hydrochloric acid, the washings being collected in the beaker. The final volume of liquid was about 100 ml. After a heatin period of 1 hour on the steam bath, the beaker was removed ancfthe crucible rinsed out,. The 6 N hydrochloric acid in the exit trap of the chlorination apparatus was evaporated t o a small volume and examined colorimetrically for the presence of iridium. \{%en the procedure outlined above was followed, no iridium passed into the trap. Iridium was then separated from the silica and other insolubles and from any base metals in solution by a modification of methods recorded in the literature. A volume of 5 ml. of a 10% solution of sodium nitrite was added Table I. Fluxes for Iridium tothehot sample, following which Flux No. a 1 AVsodium hydroxide solution 3 4 5 6 7 8 9 10 was added from a buret until the Siih_Rnh-_ sample was neutral. At this point Niter Niter silicate sihoate Very MonoMonoexcess monoexceas excess Bi2ml. of a 0.1% solution of dibasic silicate borate litharge silicate litharge litharge silicate sodium hydrogen phosphate werc addedslowlyand then the samplc P a r t s b y Weight was made just basic to phenol8.0 12 ... 7 6 11 ... 5 phthalein indicator. The sample 22 ... 4 ... ... 5 mas boiled for 0.5 hour to coagulate the precipitates. The rest of the procedure for the recoverv and the gravimetric determination of iridium was the same as that described in a previous 114 122 126 141 114 123 198 80 publication (9). A double pre... ... ... 1/4ChalcoA.T." '/aA.T. '/hA.T. 1 / 2 A.T. 1 / a A.T. Chalco- Roasted Nickel Iron cipitation of the base metals pyrite pyrite chalcooxide oxide was found t o be necessary in ore ore pyrite ore, ore, 40% ore 13% N i Fer08 order t o obtain a clean separation.
A N A L Y T I C A L CHEMISTRY
842
In the hydrolytic precipitation of iridium, a variation was found t o eliminate small losses t o the filtrate and im rove the groperties of the precipitate. Ten milliliters of a 108 sodium romate solution were added to the hot acidic iridium solution and the sample was boiled without the addition of a base until the color was a dark green; the time required was between 20 and 30 minutes. At this point 1 or 2 drops of a 10% sodium hydrogen carbonate solution were sufficient to make the sample basic to bromocresol purple indicator. The mixture was boiled for 30 minutes, and then digested for 4 hours on a steam bath. Before weighin the sample as metal, a 15-minute leach with hot 3 N nitric acifwas found necessary to remove a small amount of iron and nickel which had been coprecipitated with iridium. This led to a loss of iridium in some samples; the loss did not exceed 0.01 mg. of iridium. A blank from the fire assay was carried throu h the entire procedure; its value was substracted from the weigtt of each sample. A Sartorius projection-reading, air-damped microbalance was used for all the weighings. A a ctrographic examination of the recovered iridium showed that t g r e was a small contamination due to iron, but the assay blank accounted for this. The method was checked by treating synthetic lead-iridium buttone prepared by wrappin a known quantity of powdered iridium sponge in 30 g r a m of kad foil and compressing in a steel mold to a small volume. The buttons were treated as outlined above. The results are shown in Table 11, Nos. 1 to 3. A4blank wa? determined and subtracted. The results of a number of assays for iridium using different fluxes are shown in Table 111, Nos. 1 to 9. The amounts of iridium collected in the fourth buttons were estimated colorimetrically. Perchloric Acid Parting. Perchloric acid parting of leadiridium buttons was investigated because of the advantages of this acid over nitric acid in the analysis of lead-ruthenium buttons (el ). Lead-iridium buttons obtained from fire assays were placed in 250-ml. b a k e r s and 1.5-mL of i o to 72% perchloric acid were added for each gram of lead. The beakers were heated carefully Table 11. .4nalysis of Lead-Iridium H u ttons NO
Iridium Taken, hfg.
Parting Acid
Iridium Recovered, hig.
Error, hfg.
over gas burners until all the lead had been dissolved. The mixtures were cooled and diluted t o twice their volunie with hot water, which was used to wash down the cover glass and the sides of the beaker. After a digestion period of 2 hours on a steam bath, the mixtures were filtered and treated in the same mannvr as in nitric. acid parting. It is important that the filters be washed carefullv in order to remove soluble perchlorates before ignition.
It was suspected that some iridium might dissolve in the strongly oxidizing perchloric acid. A procedure for the precipitation of iridium from solutions of its salts by addition of zinc has been reported (18), and a similar method has been used SUCCCSBfully in the analysis of lead-rhodium buttons. To the hot filtrates from the partings of the first and second buttons, 150 mg. of zinc dust in aqueous suspension were added and the mixtures were digested overnight on a steam bath. The zinc, which had formed a spongy mass, would not dissolve even after a 1-hour boiling eriod, yet plenty of free perchloric acid remained in the sampre. The mixtures were filtered and the residue was chlorinated and examined colorimetrically for the presence of iridium. None could be detected, although the presence of the large quantity of zinc salts reduced the sensitivity of the test t o 0.05 mg. of iridium per sample. The results of a number of determinations are recorded in Table 111, Nos. 10 to 12. The average recovery was loner than with nitric acid parting. Selective Extraction of Iridium Residue after Parting. .4much shorter procedure for the gravimetric determination of iridium in lead-iridium buttons was developed. The lead button was parted with either nitric or perchloric acids, and the residue was filtered off and washed free of lead salta as before. The dried filter was placed in a small, tared latinum crucible and ignited a t a temperature of 600' to 650" After cooling, the residue was treated with 1 drop of 12 N sulfuric acid and 3 or 4 drops of 48y0 hydrofluoric acid. The crucible was placed on an asbestos pad on a hot plate and heated cautiously until all the hydrofluoric acid had been evaporated. The treatment with hydrofluoric acid was repeated three times. The sulfuric acid was fumed off and the sample was ignited a t 300" C. for 0.5 hour. The residue was then leached with hot 3 N nitric acid for 15 minutes, filtered, i nited, and then leached again with hot 1 N hydrochloric acid in t f e same manner. After an ignition in air a t 650" C., the samples were reduced in hydrogen and weighed.
8.
The results of some analyses of synthetic lead-iridium buttons are shown in Table 11, Nos. 4 and 5. Lead-iridium buttons from fire assays and salted with variTahle 111. Distribution of Iridium in Fire Assays ous quantities of iridium were Button Weight, Iridium Recovered, Iridium analyzed; the results are recorded Parting Grams Mg. Taken, Error, Flux in Table 111, Nos. 13 to 24. In Acid Z o . '1st 2nd 3rd .Ith 1st 2nd 3rd 4tha Total Mg. hlg. NO. all cases an assay blank was subNitric 1 32 26 39 25 4.023 0.590 0.389 0.01 5.00 5 04 -0.04 2 2 30 30 39 ;r1! 4.461 0.384 0.185 0.01 5.03 -0.01 tracted. Spectrographic exami4.069 0.560 0.448 40 0 01 5.08 3 33 30 +0.04 nation of some iridium residues Nitric 4 26 31 30 31 4 . 6 5 8 0.254 0.123 0.00 5.04 5.04 .. 1 after this treatment showed a 5 26 30 31 30 4.502 0.420 0.072 0.01 4.99 -0.05 4.780 0.214 0.00 6 25 31 31 30 0.00 5.02 -0.02 small amount of contamination Nitric , 7 32 29 24 24 4.479 0.251 0.012 0.01 4.74 5.04 -0.30 due to silica and this was present 3 8 30 26 15 22 4.482 0.179 0.007 0 . 0 1 4 . 6 7 -0.37 in the blank. Colorimetric tests 0.00 0.01 4.75 -0.29 9 34 30 27 20 4.549 0.201 for the presence of iridium in the Perchloric 10 29 28 25 27 4.614 0 . 1 7 0 0.097 0.00 4.88 5.04 -0.16 2 combined leachings of a single 11 29 23 25 28 4.278 0 , 4 1 0 0.201 0.00 4.89 -0.15 12 27 28 22 26 4.462 0.225 0.165 0.00 4.84 -0.20 sample were made; losses were Perchloric 13 38 31 27 26 4.117 0.762 0.067 0.03 4.95 5.04 -0.09 2 generally less than 0.01 mg. of 14 38 29 26 28 4.672 0.215 0.00 0.02 4.89 -0.15 15 40 30 26 27 4.531 0 416 0.009 0 . 0 1 4.96 -0.08 iridium. When dealing with samples of total weight 0.5 mg. Leacha 31 .. . . 0.451 0.016 0.005 0.47 4 Nitric 16 30 0.50 -0.03 of iridium or less, the amount 0.48 0.00 17 34 30 . . . . 0.479 0.0056 -0.02 0.45 . . . . 0.432 0 . 0 0 0.015 of iridium in the leachings -0.05 18 34 30 0.20 0.25 0.01 . . 0.186 0.0056 35 19 33 -0.05 (found by comparison with 0.01 .. 0.169 34 0.18 30 20 -0.07 n 167 0.18 21 .. 0.005 35 29 -0.07 standards) was added to the 0.14 0.15 .. 0 126 0.01 34 22 30 -0.01 0.005 35 0.13 0.123 23 30 .. -0.02 gravimetric result in order to 0.134 0.00 0.01 0.14 35 24 34 -0.01 produce a more accurate value Colorimetric estimation, not included in gravimetric value of the total collection of Colorimetric value, added t o gravimetric value. iridium. ~~~
~~
~
~
~~
V O L U M E 2 4 , NO. 5, M A Y 1 9 5 2
a43
Iron Ore. h svnthetic iron ore containing 46% iron(II1) oxide and 6Oy0 silica was used. (5.04 mg. of iridium taken) A4bisilicate flux (Table I, No. B u t t o n Weight, Iridium Recovered, 10) yielded the results shown in Flux Slag Grams Mg. Error. Table IV, Nos. 13 and 14. No. Composition No. 1st 2nd 3rd 1st 2nd 3rd" Total Mg. Discussion. Both copper 6 Subsilicate 1 39 31 ... 2.742 0.439 ... 3.18 -1.86 2 35 30 25 3.720 0.378 0.03 4.10 -0.94 and nickel oxides require basic 3 39 30 26 2.958 0.594 0.01 3.55 -1.49 4 42 28 25 3.394 0.369 0.07 3.76 -1.28 fluxes with excess litharge in 8 Subsilicate 5 35 23 20 3.636 0.047 0.01 3.68 -1.36 order to be slagged. High 6 35 20 24 3.228 0.075 0.04 3.30 -1 74 7 Monosilicate 7 30 15 . . , 1.849 0.382 .. . 2.23 -2.81 losses of iridium occurred in 8 30 13 ... 2.843 0.243 .. . 3.09 -1.95 these assays. Moreover, cop9 Subsilicate 9 26 32 22 1.797 0.054 0.01 1.86 -3.18 per and nickel were alwaye 10 28 32 21 2.130 0.035 0.00 2.17 -2.87 11 54 20 . . . 1.508 0.460 . . . 1.97 -3.07 found in the lead buttons and 12 57 23 ... 2.234 0.223 , .. 2.46 -2 58 their presence would have in10 Bisilicate 13 29 31 ... 5.129 0.00 .. .. .. 5 .. 1130 + 00 .. 0069 terfered in the cupellation if 14 28 30 . .. 5.054 0.045 5 + this process had been used ( 19). Colorimetric estimation, not included in gravimetric value. There was no significant difference in the losses of iridium in niter assays and assays of the DISCUSSION preroasted ore, and the period of fusion appears to play no part From a studv of Table 111. it is noted that the collection of , in the completeness of the collection of iridium. Assays 3, 4, 11, and 12 in Table 1V were placed in the furnace a t 1200" C. for 40 iridium in a single assay is not complete, except when the quanminutee nhile the othei assays were carried out as described tity of iridium to be collected is small. Two reassays of neutralearlier The use of a monosilicate slag in the assay of chalcopyi.e., monosilicate-or acid slags are sufficient to collect most of rite ore did not improve the collection of iridium, while theamount the iridium which is not collected in the first assay. When of contamination of the lead button was increased. Good recoveries basic-Le., subsilicate-slag compositions are used, there is a of iridium were made in the assay of the iron ore for which a bisilirelatively large loss of iridium. Nitric and perchloric acids are cate slag could be used. both suitable for parting lead-iridium buttons. The two proceHence, the fire assay for iridium is not suitable for copper or dures outlined for the analysis of the iridium residue yield satisfacnickel ores which require slags rich in litharge. A combination tory results. method of analysis of the ore in which a bi- or sesquisilicate slag The large loss of iridium in basic slags, but nct in neutral and composition could be employed would probably yield better reacid slags, is probably due to the formation of a compound of sults for iridium alone, although complex platinum ores are iridium n i t h one or more of the constituents of the molten flux. difficult to treat by this procedure. Spectrographic examination of slags from Nos. 7 t o 9 failed to show any lines of iridium. However, slag compositions which contained up t o 0.06 mg. of iridium per gram of slag-i.e., equivaTable V, Cupellation of Lead-Iridium Buttons lent to the loss of the whole 5-mg. sample to the slag-also failed (30-gram lead buttons) to show any iridium lines in an arc spectrum. It was also found Iridium Iridium Silver that after iridium had been slagged, a subsequent recovery of the Taken, Recovered, Loss, Taken, Nlg. Ratio hlg. Alg. NO. Mg. metal by a lead collection method is not efficient even after a num4.19 25: 1 3.94 0.25 1 100 ber of reassays. Some experiments were carried out in which 0.22 4.24 30: 1 4.02 2 125 0.20 4.52 40: 1 4.32 3 150 four samples, comprising two each of fluxes 4 and 5, were salted 4 100 1.50 70: 1 1.23 0.27 with 5 mg. of iridium, fused a t 1150" C. in the usual manner but 0.13 0.88 100:1 0.75 5 100 0.79 0.07 0.86 150: 1 6 150 without lead collection, reground, and reassayed. After two reassays the total weights of iridium recovered amounted to only 3.89 and 3.98 mg. from No. 4, and 3.80 and 3.78 mg. from No. 5, CUPELLATION OF LEAD-IRIDIUM BUTTONS or a recovery of only 80% in each trial. Reference has been made above to the fact that large losses of The fact that iridium does not alloy readily with lead may iridium occur when lead buttons containing iridium and silver account for the need of reassaying the slag one or more times. are cupelled to a silver bead. However, the cupellation process is When the melt is poured, some of the finely divided particles of of such importance in the rapid analysis of oies that a further iridium are redistributed throughout the slag. This explanation examination was undertaken. is in agreement Tvith observations on the assay of other platinum metals. Ruthenium does not alloy with lead and large losses of Preparation of Lead Buttons. Synthetic buttons were preruthenium to the slag occurred ($1). The opposite is true for pared by compressing reagent grade granulated lead together with accurately weighed quantities of iridium sponge and silver rhodium, where one reassay was sufficient for a good recovery ( 1 ) . powder in a steel mold. A layer of lead was placed at the bottom of the mold, a small amount of iridium and silver was added, then RECOVERY OF IRIDIUM FROM ORES BY FIRE ASSAY another layer of lead and more silver and iridium, etc. It was hoped that in this manner a partial distribution of the metals Sulfide Ore. A chalcopyrite ore, free of platinum metals and throughout the lead could be effected. gold, and which had an approximate composition of 237, copper, hlonosilicate fluxes were salted with 0.5 mg. of iridium and 100 20y0 iron, and 15% silica and a reducing power of 6.9 was mg. of silver; the lead buttons obtained from the assays were chosen. Two methods of treatment were used: (1) an oxidizing cupelled. fusion with potassium nitrate in which the sulfide is converted to Procedure. The lead buttons were placed in a furnace a t a sulfate and (2) preroasting in which the metal sulfides are changed temperature of 930' C. on preheated bone ash cupels. This temto oxides. A quantity of ore was mixed with the appropriate flux perature was maintained throughout the cupellation. The silver (shown in Table I, Sos. 6 to 8 ) , salted with iridium, and assayed. beads and the cupels were examined under a microscope. When The results of the analyses of the lead buttons are recorded in the beads were removed from the cupels, no attempt was made t o Table IV, Xos. 1 t o 8. clean the cupel material which adhered to the undersides of the Nickel Ore. rl synthetic nickel ore, containing 13y0 nickel and beads because the black iridium scales were easily dislodged and made up of nickel oxide and silica in the ratio of 1 to 5, was used. lost. Some of the beads (Table V, Nos. 1 t o 3) were parted by The flux mas S o . 9 in Table I. The results of the analyses are heating on a steam bath for 30 minutes with 10 ml. of 1 t o 4 shown in Table IY, Nos. 9 to 12 Table IV.
Recovery of Iridium from Ores
~~
~
~~
~
844
ANALYTICAL CHEMISTRY
nitric acid. The residues were filtered out on a Whatman KO.42, 7-cm. paper, washed well with 100 ml of hot water, ignited in air at 650” C., reduced in hydrogen, and weighed. A cupellation blank was subtracted. Spectrographic examination of the iridium residues and the blank showed them t o be badly contaminated. No iridium was detected in the parting acid. Another set of beads (Nos. 4 to 6) were arted with nitric acid and the residue was filtered, washed by fecantation, and then treated with hot aqua regia for 20 minutes. The residues were filtered and washed as outlined above. Spectrographic examination of the recovered indium and the blank showed that some silicon, silver, aluminum, and copper were present as impurities. This was due, no doubt. t o thevaryingamounts of insoluble substances from the cupel itself,
Discussion. When examined under a microscope, all the silveriridium beads had black scales at various places on the surface of the bead; this is a characteristic surface effect ( 5 ) . As the ratio of silver to iridium was increased, the scales appeared only near the base of the bead, just above the cupel. Furthermore, black particles of iridium oxide were scattered about the surface of the cupel, close to the silver bead. Both of these phenomena were observed even when the ratio of silver to iridium was as high as 200 to 1. The mechanically lost iridium was not included in the analyses of the silver bead. CONCLUSIONS
The distribution of iridium in fire assays has been examined. Reabsaying of slags is necessary for a complete collection of iridium and significant losses of iridium to basic slags were observed. The cupellation of lead-iridium buttons leads to serious mechanical losses of iridium, even when the ratio of silver to iridium is high. ACKNOWLEDGMENT
This lvork was supported by search Council (Canada).
5
grant from the Sational Re-
LITERATURE CITED (1) Allen, IT.F., and Beamish, F. E., ANAL.CHEM.,22, 461 (1950). (2) Barefoot, R. R., McDonnell, W.J . , and Beamish, F. E., Zbid., 23, 514 (1951). (3) Bugbee, E. E., “Textbook of Fire Assaying,” 3rd ed., Sew Tork, John Wiley & Sons, 1940. (4) Davis, C. W., U. S. Bur. Mines, Tech. Paper 270 (1921). (5) Forbes, E. C., and Beamish, F. E., IND.ENG.CHEM., ANAL.ED., 9, 397 (1937). (6) Gilchriat, R., J . Am. Chenz. Soc., 45, 2820 (1923). (7) Gilchrist, R., J . Research N a t l . Bur. Standards, 9, 64i (1932). (8) Ibid., 12, 294 (1934). (9) Ibid., 20, 745 (1938). (10) Ibid., 30, 89 (1943). ( 1 1 ) Gilchrist, R., and Wichers, E., J . A7n. Chena. Soc., 57, 2666 (1935). (12) Hill, M. A . , and Beamish, F. E., . ~ K . A L . CHEM.,22, 590 (1950). (13) Holzer, H., and Zaussinger, E., Z . anal. Chem., 111, 321 (1938). (14) Milazzo, G., and Paoloni, I,., Rend. ist. super. sanitct ( R o m e ) , 12, 693 (1949). (15) Plaksin, N., and Marenkov, E. A , Izrest. A k a d . Natkk S.S.S.R., Otdel. Khim. S a u k , 1948, 209. * (16) Pollard, W. B., Bull. Inst. M i n i n g M e t . , No. 497, 9 (1948). (17) Schoeller, W. R., and Powell, A. R., “Analysis of Minerals and Ores of the Rarer Elements,” London, C. Griffin & CO., 1940. (18) “Scott’s Standard Methods of Chemical Analysis,” Vol. 1 , 5th ed., New York, D. Van Nostrand Co., 1939. ,. 12, (19) Seath, J., and Beamish, F. E., IND.ENG.CHEX,,a h . i ~ED., 169 (1940). (20) Shepard, D. C., and Dietrich, W.F., “Fire Assaying,” Kew Tork, McGraw-Hill Book Co., 1940. . 20, (21) Thiers. R., Graydon, W., and Beamish, F. E., - 4 s . 4 ~CHEM., 831 (1948). RECEIVED for review November 9, 1951.
Accepted January 26, 1952.
Instrument for Automatic Determination of Melting Point RALPH ri. M U L L E R ~m n SEYMOUR T. ZENCHELSKY~ Washington Square College of Arts and Science, New York University, New York 3, N . Y
M
ANY techniques exkt for the determination of the melting points of pure organic compounds (3, IS). In the practical organic chemistry laboratory the capillary method is most frequently employed. Nearly all the literature on the determination of melting points deals with the impmvrment and simplification of this procedure. This method, as well as all others thus far employed, depends upon the subjective decision of the operator for a criterion of melting. Two exceptions may be noted: Lawe employed an electrical indication of melting, and Kardos employed an optical one (3, 5 , 9 ) . Keither of these methods, however, nor any other, has been employed in an instrument for the automatic determination of melting point. The instrument described in this report wae developed for the purposes of the practical organic chenlistry laboratory. It was designed to give rapid and precise melting point data without continuous attention by an operator during the determination, and it employs samples of micro size. INSTRUMENT REQUIREMENTS
Because the definition of melting point involves the existence of an equilibrium between liquid and solid, the ideal measurement would be made while the system is maintained in such a state of dynamic equilibrium. A closed-loop servosystem ~ - 0 u l d supply or remove heat in proportion t o the relative amounts of solid and liquid present, causing the temperature to oscillate 1 Present address, The Los Alamoe Scientific Laboratory, Los Alamos, N. M. 2 Present address, School of Chemistry, Rutgera University, New Brunswick. N. J.
within narrow limits. The mean temperature would be taken as the melting point, provided that the temperature excursions were small enough. Such a state of affairs is impossible to achieve, aa all liquid. supercool, and thus permit smooth servo control from one direction only. A practical instrument would have to approach the melting point in the conventional manner. It would require three basic components: (1) a means for automatically increasing the temperature of the sample a t a predetermined rate, the hot stage assembly; (2) a temperature-indicating device, the thermometer; and (3) a means for automatically determining when melting has occurred as well as for causing the temperature of the melting point to be indicated on the thermometer for as long as is desired after melting has occurred, the sensing and control circuit. The sensing and control circuit requires an objective criterion of melting. The one employed in this instrument makes use of the fact that polycrystalline materials scatter light while shiny metallic surfaces give specular reflection. Thus if a beam of light is incident upon a polished metal surface, it will be reflected in a beam a t an angle that is equal t o the angle of incidence. At that angle a photocell provided with a lens and diaphragm would register maximum light intensity. If a thin film of finely powdered crystalline material is placed on the metallic surface, the light intensity a t the photocell will diminish because the incident light is now scattered in many directions. Heating the metal surface will cause the powder t o melt, leaving a thin transparent film of liquid on the shiny metal surface, Thie system gives specular reflection once more, causing the photocell to register a maximum again.