Separation of Palladium from Lead and Colorimetric Determination of

Arthur L. Erikson , Robert L. Tromp , Robert A. Nielsen , Myra D. Anderson , Terry C. Chapman , and W. A. Emel. Analytical ... J. J. Morrow and J. J. ...
0 downloads 0 Views 574KB Size
Separation of Palladium from lead and the Colorimetric Determination of Palladium with Potassium Iodide J. G. FRASER, F. E. B E A M I S H , and W. A. E. M c B R Y D E University o f Toronto, Toronto, Ontario, Canada

No proved procedures have been published for the determination of palladium in lead assay buttons. Since the lead-palladium alloy dissolves completely in 1 to 2 nitric acid, the palladium must be separated from a nearly saturated lead nitrate solution. Conventional methods of precipitating lead as sulfate and leaving the precious metal in solution cause coprecipitation losses. Five-milligram quantities of palladium were separated successfully from the lead by a double precipitation with salicylaldoxime. Up to 1 mg. of palladium was completely extracted with chloroform after the palladium had been complexed with dimethylglyoxime; and subsequent to the destruction of the organic matter the metal was determined colorimetrically using potassium iodide as the color reagent. These two methods permit the investigation of fire assay losses of palladium and add to a general scheme of analysis of lead buttons containing the platinum metals.

T"?

4 investigation of the efficiency of the fire assay for palladium required a procedure for the determination of the precious metal in a lead button of 15 to 30 grams. Unlike the other platinum metals, palladium is not "parted" by dilute nitric acid, but is completely dissolved with the lead. The large amount of lead makes impracticable the separation of palladium by hydrolytic methods. The following report deals with methods for separating palladium from lead.

APPARATUS A Y D REAGENTS

Apparatus. The fire assays were made in a Williams and Wilson 15 KVA Glohar assay furnace. Absorption measurements were made with a Beckman Model DU spectrophotometer. Palladium Solutions. Powdered palladium sponge obtained from J. Bishop Co., Platinum Works, Malvern, Pa., was examined spectrographically and showed only traces of platinum and lead as impurities. A 2.000-gram portion of this palladium was dissolved in 70 ml. of aqua regia and the solution was evaporated to dryness on a steam bath. The residue was tahen up in 20 ml. of aqua regia and the solution evaporated again; this procedure was repeated twice with hydrochloric acid and twice with water. The residue was then dissolved in 20 ml. of hydrochloric acid and the solution diluted to 150 ml. This was filtered through a Whatman S o . 42 filter paper and diluted to 2 liters. The filter paper was ashed and the weight of residue proved insignificant Portions of the solution were diluted to give palladium solutions containing 0.2000, 0.050, and 0.010 mg. per ml. The first of these dilute solutions was standardized by precipitation with salicylaldoxime. The results of three parallel determinations were 5.000,5.003, and 5.003 mg., respectivelv. This agreed well with the theoretical amount of palladium taken, which ~vits1.991 mg. Salicylaldoxime Reagent. Salicylaldoxime, obtained from Eastman Kodak Co., Rochester, N. Y., was recrystallized from petroleum ether containing a small amount of benzene. A hot, saturated aqueous solution was prepared and filtered before use. Lead Nitrate Solution. A solution containing 240 grams of Baker and Adamson reagent grade lead nitrate and 1 ml. of nitric acid per liter was used. Potassium Iodide. A solution containing 500 grams of Baker and Adamson reagent grade potassium iodide and 5 ml. of ammonia per liter was used. Sodium Sulfite. A freshly prepared solution was used containing 1.5 grams of May and Baker anhydrous sodium sulfite in 250 ml. of water. Dimethylglyoxime. A ITo aqueous solution of the sodium

salt of diniethylglpoxime was used. The reagent was obtained from Tower Drug & Chemical Co., Rochester, N. Y. COLORIMETRIC DETER.IIINATION OF PALLADIUM IN LEAD BUTTONS

colorinietric procedure. for For various reasons none of th the determination of palladiuni proposed to date could bc applied Patisfactorily to lead assay buttons. The separation of the palladium by solvent extraction followed by its colorimetric determination with potassium iodide was used for buttons cont :tining from 1 to 10007 of palladium. I t is generally acknowledged that, excess potassium iodide must he avoided for the quantitative precipitation of palladium iodide. While some use of the pink color produced with excess potawium iodide has been made as a qualit,ative test for palladium, apparently this reaction has not, been used for a quantit,ative colorimet,ric method. The authors determined that, when a sufficiently large excess of potassium iodide was added to an acidified solution of palladium, the maximum color developed immediately with amounts of palladium up to 1 mg. Choice of Wave Length. An absorption spectrum was drterniined and is given in Figure 1. For the analytical determination absorbancy measurements were made a t 408 mp. A siniilar curve n.a* obtained using Photovolt narroiv-band filters in a Klett-Summerson colorimeter. This instrument was not satisfactory for two reasons: The hcst filter available was centered a t 420 mp, which lies on the slope of the absorption maximum peak, and the limited responw a t 420 mp of the barrier layer cells used in the Klett-Summerson instrument resulted in a great decre:jsc in sensit,ivit'y. Amount of Reagent Required. When a 500-7 sample of palladium was treated with 2 ml. of potassium iodide reagent, a precipitate formed which did not dissolve on standing. Five milliliters of reagent were sufficient to prevent formation of this precipitate, but gave an absorption approximately 7% below that obtained with 10 nil. of reagent. (These expcriments mere carried out with a reagent one half the concentration of that finally recommended.) With greater amount's of reagent there TI-BS little change in absorption. On a molar basis this is cquivalent to 3200 moles of potassium iodide for 1 mole of palladium. Assuming the colored complex to bc (PdId)-- this corresponds to an 800-fold excess. Effect of Acid. K i t h pure palladiurn solutions either sulfuric or hydrochloric acid can be used. The amount of acid made no difference wit,h final acid concentrations up to 2N. I n the recommended procedure sulfuric acid was used since it avoided the precipitation of lead iodide by permitting the removal of any small amounts of lead as the sulfate before the addition of t,he reagent. Time of Standing. Acid iodide solutions slowly liberate iodine on standing. One absorption maximum for iodine occurs a t approximately 350mw; however, the solution still absorb? strongly a t 408 mp. The difficulty caused by iodine formed in the palladium solution was avoided successfully in two ways. The first method was to zero the spectrophotometer on a water blank and to read the absorbance exactly 12 minutes after the addition of the reagent. This time was chosen as it enabled several samples to he done in a timed sequence. Standards were required for each set of unknowns and had to be treated exactly as the unknowns. The second and the preferred method was to add a 495

496

ANALYTICAL CHEMISTRY

small amount of sodium sulfite to the solution to reduce the iodine. Samples treated with sulfite were stable for a t least 18 hours. Effect of Reagent Impurities. Using the first of the procedures mentioned, interesting results were obtained with iodides from different manufacturers. Each batch of reagent gave a different absorbancy for the same amount of palladium. On increasing the amount of reagent two effects were observed: (1) With some reagents the absorbancy would increase with increasing amounts of reagent and (2) the absorbancy would decrease with increasing amounts of reagent. The first was due to traces of some oxidant in the reagent such as iodate, iron(III), or copper(I1). The second effect was caused by some impurity which sloivly formed a complex more stable than the iodide. These suppositions were supported by noting the time dependence of the various reagents. Whereas it mas normal to find the absorbancy slowly increasing because of iodine liberation, those reagents which showed the second effect also exhibited a decrease in absorbancy with time. It was noted that phosphate, fluoride, and bromide did not cause this color fading. However, slight traces of cyanide, when added t o the pink palladium-iodide solution, did cause fading. Two drops of 0.5% potassium cyanide were sufficient to cause the color developed by 500y of palladium to disappear slowly. The addition of sulfite eliminated the first trouble, but all batches of potassium iodide reagent should be checked for the presence of impurities causing the fading of color.

1.2

I.o t

z"a

O.%

m

a

2

0.6

m

0.4

a? 0 400

5 00 WAVE LENGTH MC)

bo0

Figure 1. Absorption Spectrum of Palladium-Iodide Complex 10 p.p.m. of palladium

Sensitivity and Range. The authors found that 10.5 p.p.m. of palladium corresponded to unit absorbancy in 1-cm. cells in the Beckman spectrophotometer a t 408 mg. The solutions obeyed Beer's law throughout the working range of the spectrophotometer. The optimum range of concentration was 1 to 10 p.p.m. of palladium metal when the solution was contained in 1cm. cells. As the circumstances required, the range was lowered to 0.2 to 2.0 p.p.m. by the use of 5-cm. cells. Solvent Extraction. Young ( 5 ) demonstrated that the palladium dimethylglyoxime complex could be extracted with chloroform. Traces of palladium \yere thus separated from silver beads for subsequent analysis by dithisone titrations. Ayres and Tuffly ( 1 j stated that for larger amounts (10 to 20 mg.) of palladium the procedure was unsatisfactory. By using large amounts of chloroform, the present authors were able to extract completely up to lOOOy of palladium from solutions containing 15 grams of lead. The chloroform was evaporated and the organic matter destroyed with nitric and sulfuric acids. To prevent possible interference from the nitric acid, the solution was evaporated to fumes of sulfur trioxide several times after the ad-

Figure 2. Absorbancy-Concentration Plots for Palladium Solutions Resulting from Different Treatments 1. 2. 3.

Palladium extracted from salted as5ay buttons Palladium extracted from lead nitrate solutions Unextracted palladium solutions

dition of water. The resulting solution was diluted with water filtered to remove traces of lead sulfate, and the palladium-iodide color developed. Three extractions with chloroform were found sufficient to remove all the palladium, but as a safety measurc four extractions were usually carried out. With amounts of palladium above 250 y, a precipitate of palladium dimethylglyoxime formed but this did not hinder the extraction. The colorimetric procedure was applied to extractions of known amounts of palladium from lead solutions. Straight-line plots of absorbancy versus concentration were obtained in ranges of 100 to lOOOy per 100 ml. and 10 to lOOy per 100 ml. I n one experiment amounts of palladium from 0 to 207 R ere added to the lead solution and extracted. The color was developed in a 25-ml. flask and absorbancy measurements were made in a 5-em. cell. The plot obtained was linear and the point obtained lyith 2 y of palladium lay on the line. This indicated that the extraction 1% as complete and that the procedure could be used for lead buttons with a palladium content from traces up to 1 0 0 0 ~ . Slight differences in the slope of the absorbancy-concentration plots were obtained between extracted and nonextracted samples. Differences were also obtained in the slope of samples extracted from lead nitrate solutions and those extracted from salted buttons as described in procedure for assay buttons (see Figure 2) This is not surprising since lead buttons will contain traces of various impurities. Whatever the cause of this change of slope, it did not vitiate results if the recommended procedure was followed. Linear absorbancy plots resulted from the salted buttons obtained from the five different types of flux thus far investigated. During these experiments palladium n as also extracted from silver nitrate solutions and anal) xed successfully with potassium iodide. The only modification required in the procedure was to filter the final solution after the addition of potassium iodide in order to remove any silver iodide. I t is anticipated by the authors that this procedure ~ 1 1 1be of value for analyzing silver beads. Interfering Cations. The elements most likely to be encountered when the lead button, resulting from an assay of a palladium-containing ore, is dissolved in nitric acid are iron, copper, nickel, and possibly small amounts of platinum None of the other platinum metals \%illdissolve in nitric acid. Iron was the only one of these elements which adversely affected the extraction of the palladium. As little as 3 mg. of iron prevented the extraction of 60y0 of the palladium in a 250-y sample. The disodium salt of ethylenediaminetetraacetic acid (Versene) was used successfully t o complex the iron and prevent interference

497

V O L U M E 26, NO. 3, M A R C H 1 9 5 4 with t,he palladium extraction. (Phosphate, fluoride, tartrate, etc., could not be w e d in the presence of lead.) One half gram of this reagent was needed t o complex 50 mg. of iron. Platinum, nickel, and copper did not interfere x i t h extractions. This n.as proved as follon-s: Three lead nitrate 'solutions containing 2507 of palladium were taken and to the first were added 5 mg. of platinum, to the second 0.6'7 gram of nickel, and to the third 0.92 gram of copper. When these solutions were extracted and analyzed colorimetrically, they showed 255, 243, and 250-1 of palladium, respectively. I n the case of lead buttons containing copper, slightly more nitric acid than the 3 or 4 drops recommended in the procedure following was required to dissolve any basic copper nitratr that had formed on evaporation of the original nitric acid solution of the button. While any basic copper nitrate remained undissolved the pH was above 4. .It this p H t'he copper formed a soluble complex with dimethj-lglyoxime and caused losses in the palladium extraction. Procedure for Assay Buttons. The lead button, freed from slag, was dissolved in 90 ml. of 1 to 2 nitric acid. The solution was evaporated to dryness and the residue dissolved in 100 ml. of water containing 3 or 4 drops of nitric acid. The solut,ion was transferred t o a 250-ml. pear-shaped separator)- funnel. It, did not mat,ter if hits of slag which adhered to the button were added to the funnel as they floated on the chloroform la>-er. If the butt,on contained iron, 0.5 gram of Versene was added. Three millilitrre of dimethylgl>-oxime were added and t,he solut,ions were miscd b y shaking and allo\\ed to st'and for 1 hour. The solution was extracted Lvith four portions of chloroform (50, 25, 15, and 10 ml., resnectively). The chloroform was caught in a 125-ml. coriiral beaker and evaporated on a steam bath. Twelve milliliters of a 1 to 1 solution of concentrated nitric and sulfuric arids n-ere added and the solution was evaporated on a hot plate to fumes of sulfur trioxide. Water was added and the evaporation repeated threr times. Twenty-five milliliters of water were added arid after cooling the solution was filtered into a 100-ml. volumetric flask. The filter paper was washed \yell with water. Potassium iodide reagent was added followed by 5 ml. of sodium sulfite solution. The amount of potassium iodide reagent was fixed by the sample with the largest palladium content. One milliliter was required for each lOOy of palladium. T h e sample was then diluted to volume. A standard curve was prepared as follows for each type of flux used. Several blank lead buttons were obtained which were dissolved in beakers containing known amounts of evaporated palladium solution. These solutions were t'hen treated as outlined in the previous paragraph. The zero setting on the spectrophotometer was made with a solution obtained from a button that had not been salted. A number of samples of a flux calculated to yield a basic slag were salted with standard palladium solutions. These were placed in the furnace at' 1500' F. and the temperature was raised to 2200 a F. The samples were then removed and poured into a conical steel mold The lead buttons were separated from the slag and analyzed in the manner already described. The slags were ground and reassayed after the addition of more litharge and flour. The results are shown in Table I, Nos. 1, 2, and 3.

cylaldoxime. Flagg and Furman ( 2 ) showed that lead begins to precipitate with salicylaldoxime a t p H 5 . The use of this reagent was investigated by the authors as a means of determining t'he palladium content in lead buttons. Results obtained with a single precipitation of 4 mg. of palladium from a solution containing 24 grams of lead nitrate were high. The precipitates were ignited and dissolved in aqua regia. The resulting solutions were checked for lead content by Sandell's dithizone procedure ( 4 ) . The average amount of lead present in four precipit,ates examined was 1.14 mg. To determine the purity obtained by reprecipitation the first precipitates were ignited and the residues dissolved in aqua regia. After several evaporations to dryness with hydrochloric acid, the residues xere taken up in water containing 4 drops of nitric acid. Amounts of Versene ranging from 0 to 1 gram were added and the palladium was reprecipitated. When these precipitates were analyzed for lead they showed 8, 10, 16, and 8 y . This indicated that separation from the lead was virtually complete with a double precipitation, even without the complexing agent for the lead. The results obtained for the palladium on thc second precipitation were erratic. This was due to losses incurred during the ignition of the precipitate. This difficulty was overcome by filtering the first precipitate through a fine paper, folding thie paper carefully, and wrapping it in a second paper. This was placed in a crucible supported on a silica triangle and was immediately given the full heat of a AIeker burner. After treatment with aqua regia as described above, the palladium was then reprecipitat,ed. The results from three samples t,hus trrated were 4.975, 4.990, and 4.962 mg. These result,s agree satisfactorily with the 5.002 mg. obtained from single precipitations with the same aliquot of standard palladium solution. Filtrate Losses. Four samples of standard palladium solution were precipitated by salicylaldoxime. The first sample x a s n-ashed into the filter crucible with hot water only. The second sample was washed with 20 ml. of 30% alcohol solution after the hot water wash. The third was washed with 40 ml. of alcohol and the fourth with 60 nil. The filtrates and washes were analyzed colorimetrically and showed l , 0, 0, a n d 3 7 of palladium, respectively. These results show that filtrate losses were negligible and that a hot n-ater n-ash is just as effective as an alcohol wash. Filtrates from t'he first precipitation in the presence of lead were checked as follows. Dimethylglyoxime was added to the filtrates from the first precipitation of the palladium contained in several assay buttons. These filtrates were extracted and analyzed colorimetrically. The colorimetric standards were prepared by adding known amounts of palladium to a solution containing 24 grams of lead nitrate. These were treated with salicylaldoxime reagent and after 1 hour on the steam bath, 3 ml. of dimethylglyoxime solution were added. The palladium was extracted with chloroform and analyzed in the prescribed manner. The palladium remaining in three filtrates was found t o be 7, 5 , and 87.

Procedure for Assay Buttons. The button Tyas dissolved-in 90 ml. of 1 to 2 nitric acid, diluted with 50 ml. of water and filtered while hot. After washing the paper well with hot lyater, the paper Holzer ( 3 )first demonshated t h a t palladium could be quantiand the gangue contained on it lvere retained for inclusion in the tatively precipitated from a slightly acidic solution with saliassay of the slag. The filtrate and washings were evaporated to drvness on a steam bath. The resultinn residue f a s dissolved in 100 ml. of water contzining 4 drops of nitric acid. This solution was filtered while hot to remove any silica which resulted from Table I. Analysis of Lead Buttons from Fusions the evaporation. (The amount of silica obtained Palladium seemed to depend on the nature of the slag uped Palladium Found, Mg. Sample Taken, Difference, and how well it separated from the button.) This 1st button 2nd button 3rd button NO. ME. Total 1Ig. paper was saved and included in the reassay. 1 0.100 0.100 0,005 ... 0.105 +0.005 2 0.250 0.247 0.003 ... 0.250 Ten milliliters of salicylaldoxime reagent were 3 0,500 0.505 0.013 0.518 +'O :01 added to the hot filtrate and the precipitate 4,474 4 5.00 0.298 0: 002 4.77 -0.23 was allowed to coagulate on the steam bath for 5 5,oo 4.638 0.245 0.009 4.89 -0 11 4,564 6 5.00 0 360 0 000 4.92 -0.08 1 hour. After cooling for '/z hour, the precipi0 T h e probability t h a t most of this palladium was lost t o the slag will be supported in a tate was filtered off on a Whatman No. 42 7-cm. subsequent report on assay losses. paper. The beaker was freed from any adhering precipitate by means of a '/4 circle of filter GRAVIMETRIC DETERMINATION OF PALLADIUM LEAD BUTTONS

IK

498

ANALYTICAL CHEMISTRY

paper. The paper and residue were partially dried under a heat lamp and tightly folded. This was wrapped in a 5.5cm. paper and tamped into a Coors 00000 porcelain crucible. This crucible was placed inside a larger crucible, ignited over a Meker burner a t full heat without first allowing the paper to char, as is the usual practice. The contents of the crucible were reduced in hydrogen over a Meker burner, flushed with nitrogen, and allowed to cool in a stream of nitrogen. The crucible and contents were then placed in a 125-ml. conical beaker and 50 ml. of aqua regia were added. When the palladium was dissolved the crucible was removed with platinum-tipped tongs and rinsed well with water. The solution was evaporated to dryness, fresh aqua regia was added, and the evaporation was repeated twice. After three evaporations with hydrochloric acid, the residue was dissolved in 8 ml. of hot 1 to 32 hydrochloric acid. This was diluted with 25 ml. of water and filtered while hot. The filtrate and washings were diluted to 100 ml., the palladium n a i reprecipitated and coagulated, and the mixture was cooled in the same way as the first precipitation. The precipitate was caught on a Royal Berlin porcelain filter crucible of A-2 porosity, dried for 3 hours a t 115” C., and weighed. The results of several assays using a basic slag are shown i n Table I, Nos. 4, 5, and 6. Only the palladium contained in the first button was determined by the gravimetric procedure. The colorimetric method was used for the second and third button..

require the colorimetric procedure; in this case one must use for the standard curve blank buttons (or lead nitrate solutions) of composition comparable to those obtained from the assay. The gravimetric procedure is recommended only for buttons obtained from materials rich in palladium. In the double precipitation of palladium with salicylaldoxime filtrate losses were shown to be negligible. The amount of lead contaminating the second precipitate was shown to be insignificant Both methods were applied successfully to regular assay buttons from synthetic ores, although only results from a basic ore are given in this paper. A critical examination of the distribution of palladium with the baqic and other slags is being made.

DISCUSSION

(1) Ayies, G. H., and Tuffly, B. L., AXAL.CHEW,24, 949 (1952). (2) Flagg, J. F., and Furman, N. H., IND.EXG.CHEW,AN