- Email: [email protected]
Determination of palladium, platinum and rhodium in geologic materials by fire assay and emission spectrography
~mta.
1968.Vol. 1s. pp.111IO 117. pusunOn Rar.
Rimed h N-
-
DETERMINATION OF PALLADIUM, PLATINUM AND RHODIUM IN GEOLOGIC MATERIALS BY FIRE ASSAY AND EMISSION SPECTROGRAPHY* JOSEPH HAFFTY and LEONARDB. RILEY U.S. Geological Survey, Federal Center, Denver, Colorado 80225, U.S.A. (Received 14 July 1967. Accepted 5 August 1967)
S--A method is described for the determination of palladium down to 4ppb (parts per billion, lo*), platinum down to IOppb and rhodium down to 5 ppb in 15 g of sample. Fire-assay techniques are used to preconcentrate the platinum metals into a gold bead, then the bead is dissolved in eyua regia and diluted to volume with 1M hydrochloric acid. The solution is analysed by optical emission spectrography of the residue from 200 ,ul of it evaporated on a pair of flat-top graphite electrodes. This method requires much less sample handling than most published methods for these elements. Data are presented for G-l, W-l, and six new standard rock8 of the U.S. Geological Survey. The values for palladium in W-l are in reasonable agreement with previously published data. PROCEDURES for the determination of palladium, platinum and rhodium are available in the literature, but many are only suitable for much higher concentrations than that of most geologic materials which the U.S. Geological Survey is called upon to analyse. Some procedures require considerable time and analytical shill and are generally cumbersome and not too reliable in routine use. Still other procedures are designed for one of the three elements, giving little attention to the other two. The method described here is relatively free from extensive sample manipulations, adequately sensitive, and can determine all three elements in essentially a single procedure.
EXPERIMENTAL Fire assay
Gold (weighing between 0.80 and 1.20 mg), which is used as a collector for the platinum metals, is added in the form of O-007in. dia. wire to yield approximately the same concentration of gold in the fu& solution as in the spectrographic standards. A piece of lead foil 4 in. square and @016 in. thick is sharply folded at the center with pliers. The fold is then opened, the gold wire placed in the centre of the crease, the fold closed on the wire and again folded at the centre to give a thickness of four layers. All edges are then crimped with pliers to ensure enclosure of the wire. This lead envelope is added as the last step in the preparation of the fusion mixture described below. The fusion and cupellation step8 in general follow conventional fire-assay practices as described by Bugbee’ and are only summarized here to point out variation8 found helpful. A flux is formed by combining sodium carbonate, lead@) oxide, silica, borax glass and calcium fluoride together with flour as a reducing agent or potassium nitrate as an oxidizing agent. These am combined in a “30-g” tireclay crucible with a maximum capacity of about 260 ml. The amount8 of the flux component8 are dependent on the composition of the sample to be analysed. The object is to form at least a two-phase melt: a liquid slag approximating to a complex borosilicate, and a metallic lead phase, controlled in sire, to collect the noble metals present. The addition to the flux of 1-3 g of calcium fluoride, not * Publication authorized by the Director, U.S. Geological Survey. 8
111
112
JOSEPHIIAFFIY and LEONARD B. RILEY
routinely used in gold fire-assaying, has been found beneficial for many types of samples analysed for the platinum metals. After the flux components are in the crucible, 15 g (or 30 g) of sample are added and the contents thoroughly mixed. Then the lead foil containing the gold wire is added. Part of the borax glass for the charge may be added as a cover for a 30-g sample, or for a sample liable to form dust, or for a sample relatively rich in iron(II1) oxide. The crucible is placed in the muffle of an assay furnace preheated to a temperature of 850-900”. After the door is closed the temperature is raised to 1050-1120” over a period of 30-45 min. The crucible is then removed and the molten charge is poured into an iron mould and allowed to cool. The slag is carefully chipped and worked away from the lead with a hammer and brush, and the lead is tapped into a cube, producing the lead button. The lead of the lead button is separated from its noble metal content by cupellation. The cupels, made of compressed bone ash, are placed in the furnace, preheated to 95&1000”, for about 10 mm. Then the lead buttons are placed in the cupels and the furnace door closed until the lead melts and a bright surface is formed on the lead. The door is opened and the temperature first lowered to about 830”, then gradually raised to a fmishing temperature of about 900” as the lead is oxidized and the oxide absorbed by the cupel. The cupellation is finished when only a noble metal bead remains on the cupel. Since silver is not to be determined, higher temperatures may be used than in the conventional silver assay, thus reducing the dangers of “freexing” (lead oxide covering the molten lead and preventing further oxidation). Gnce the bead is obtained, the cupel is removed from the furnace and allowed to cool. No precautions are necessary to prevent “sprouting” as in the silver assay, since there is about 1 mg of gold present, The bead is examined under a stereoscopic microscope (about 45 x magnification) while still on the cupel and the colour, surface, and shape are noted. The presence of certain elements is indicated when the colour has drastically changed from that of gold. No correlation, as yet, has been made with the shape or surface of the bead and the presence or concentration of specific elements. This is still under investigation. The bead is then carefully transferred and weighed. Relatively pure gold beads are nearly spherical and need more care in handling than hemispherical silver beads. No attempt is made to clean the bead, e.g., by brushing, thus avoiding any losses of metals. The small amount of lead-saturated bone ash retamed on the bead adds little to the weight and is easily dissolved in the subsequent treatment. The weight of the gold wire added to the flux initially is compared with the weight of the bead obtained. Any sizable gain (or slight loss) is noted and aids in calculating the dilution factor if the weight has substantially increased. The bead is transferred to a l-ml volumetric IIask to which 0.1 ml of uq~ regiu (3 : 1 hydrochloric acid:&& acid)is added, and allowed to stand overnight at room temperature. To ensure complete dissolution of the bead, the solution is heated for 15-20 min on a steam-bath. After the bead has diiolved. an aliquot of a 4 mg/l solution of molybdenum (the internal standard) in 2M hydrochloric acid is added so that the concentration is 0.2 mg of molybdenum per 1. after dilution to volume with 1M hydrochloric acid. All beads analysed to date have been soluble in aqua regiu. However, if any metals remain undiilved, the presence of other platinum metals should be suspected. When a sample is known from a previous semi-quantitative spectrographic analysis to be high in platinum metals, the amount of gold is increased so that a 2: 1 ratio of gold to total platinum metals is obtained. This ratio has been found satisfactory for dissolution of the platinum alloys. Larger volumetric equipment and more acid may then be necessary as stated by Murt and Cot-son.* Dilutions are made from-the solution to obtain a desirable concentration for-measurement of the platinum metals. Gold in solution (10 mglml) is added to the final dilution in order that the gold concentration be about 1000 mg/l and approximately equivalent to the concentration of gold in the standards. S’ctropphic adysis Two &in. dia. flat-top, graphite electrodes are first waterproofed with 4 drops each of a petroleum
ether solution of Apiexon “N” grease (5 g/l). Then 100 ,ul of the unknown sample or the standard, both containing the internal standard, are added to each electrode in SO-p1increments and evaporated by means of a heat lamp. The electrodes are kept under the heat lamp until placed in the arc-spark stand. Each exposure consists of exciting the residue of 200 ~1 of solution on the two electrodesS. The sample-bearing electrodes are then placed in the arc-spark stand and the residue is excited and the spectrum recorded, with the a paratus and operating conditions listed in Table I. The emulsion is calibrated ! y means of an iron bead arced at 5 A d.c. and exposed for 15 set at a transmission of 48 %. The formation of the iron bead is described by Haffty.’ The iron lines and relative intensities used for plate calibration were selected from a list of homologous lines.S*a They are reproduced in Table II for the convenience of the reader. The calibration curve is established by plotting transmission (ordinate) us. intensity (abscissa) on log-log paper.
Determination of palladium, platinum and rhodium TABLEI.-APPARATUS
AND SPECI~~OORAPHIC OPERAT?NQ CONDITIONS
Excitation source-intermittent d.c. arc. Primary power source-full-wave rectified 280-V. 6O-cycle power supply-initiated by a high-voltage condensed spark synchronized to initiate each half cycle. 2-OA Radio-frequency current, Initiating circuit parameters 0.0025 ,eF Capacitance, 15 PH (residual) Inductance, 16 R (dial setting 5) Primary resistance, 0.5 R (residual) Secondary resistance, 15,000 v Discharge voltage, 11 Number of discharges per half cycle, 7.0 A Current, radio-frequency, Spectrograph-folded. 3-m Rowland circle-mounted grating with 21,ooO lines/in. giving a reciprocal linear dispersion of 4 A/mm in the first order. 240&36OOA Wavelength region, Slitwidth, 4OP arc image focused on grating Illumination, 3mm Analytical gap, Transmission, 100% 20 set Exposure+ Kodak SA-1 Emulsion, ASTM type C-3 Electrodes,
TABLEII.-IRON LINESUSED FOR PLATE CALIBRAnON Fe line,
Fe line,
A 3157.89 3175.45 ~178.02 3196.93 320540 321594
Intensity
A
Intensity
73 125
3217.38 3222.07 3225.99 3239.44 3251.24 326824
165 625 770 285 62 29
4: 210 22!!
TABLEIII.-ANALYTICAL LINES USED IN DETERMWATION OFPALLADIUM,PL.4~ANDRHODfUhf
Element Pd Pt Rh
A”alyYl Pd Pd Pt Rh Rh Rh
line,
3242.70 340458 2659.45 2490.77 3396.85 3434.89
Concentration range based on analytical curve, mgll @05&O 0.025-05 0.125-5-O 0*125-5-o 0*05-5~0 0~05-125
The MO 2816.15 A line was used as internal standard line throughout.
113
JOSEPHHAFFW and LEONARDB. RILEY
114
Each plate contains the exposures of 10 or 11 samples in duplicate, plus 9 standards and one iron bead to calibrate the emulsion. After the plate has been processed, transmittance measurements of selected analytical and internal standard lines (Table III) are obtained by means of a microphotometer. From the calibration curve, intensity values are obtained for the standards which are used in constructing the analytical curves. The analytical curves are established by plotting the ratio of the intensities of the analytical and internal standard lines us. concentration on logarithmic coordinates. The concentrations of the elements in the unknown samples are read from the analytical curves. The absolute weight of metal is calculated from the volume of solution in which the bead was dissolved. This weight is then divided by the weight of sample taken to give ppm or ppb [parts per billion (log)] in the sample. Preparation of stpndardr
Stock solutions, calculated to contain 1000 mg of the element per l., are made by dissolving the ammonium salts of the appropriate chloro-complexes in 2M hydrochloric acid. These stock solutions are then serially diluted with 2M hydrochloric acid to give concentrations of 100, 50, 25, 10, . . . 0.5 mg/l. The standards for analysis are made by transferring O-2ml of a gold stock solution (10 mg/ml) to each of nine 2-ml volumetric flasks. To these, 0-l ml each of the palladium, platinum and rhodium solutions are added. In addition to this, O-1ml of a solution of molybdenum (4 mg/l) in 2M hydrochloric acid is added as the internal standard and the mixtures are diluted to volume with 1M hydrochloric acid. The resulting standards contain palladium, platinum and rhodium in concentrations of 50,5,2*5,1*25,0*5 . . .0*025 mg/l in solutions containing gold (1000 mg/l) and molybdenum (0.2 mgll). RESULTS
The precision of the method is illustrated by the results in Table IV. The standard deviations were obtained by running several 15-g portions of the same sample on various plates at different dates.2*7 TABLEIV.-DATA ON PRECISION OF METHOD Average concentration, Element
PP”
Palladium Platinum Rhodium
0.426 0.395 o-0143
d n X S.D. C.V.
= = = = =
S.D. O-023 OX@77 O-0008
C.V.
n
5.4 1.9 5.6
5 5 2
difference between duplicate values. number of duplicate determinations. average of the individual results. 2/(X da/h) = standard deviation. lOO(S.D.)/x = coefficient of variation.
In order to obtain an indication of the accuracy of the method at concentrations present in geological materials, aliquots of known solutions of palladium, platinum and rhodium were added to 15-g portions of quartz sand, evaporated under heat lamps, and carried through the method described. The solutions were added in O-l-ml increments and distributed over the surface of the sand, which had previously been heated for about 20 min in an evaporating dish. The total amount of solution added to any one dish was no more than O-6 ml and with proper care the solution did not seep through the sand to the evaporating dish. After the sand was added to the fire clay crucible, the additional precaution was taken of dry-rinsing the evaporating dish with the soda needed for the charge and then transferring this to the crucible. However,
Determination of palladium, platinum and rhodium
115
the platinum
in samples 10 and 11 was accurately weighed and added as the metal. The results are presented in Table V. For low-grade platinum ores, 30 g (approx 1 assay ton) have been used. When the weights of the elements given in Table V are considered on the basis of this weight of sample, they represent a concentration range of 17 ppb to 4.23 ppm. The analytical results for G-l, W-l, and six new standard rocks are presented in Table VI. Each run is the average of duplicate determinations. More detailed data on the localities from which the six new standard rocks were collected, and the results for many elements, were presented by Flanagan.8 TABLEV.-RECOVERY OF PALLADIUM,PLATINUM AND RHODIUM ADDED TO QUARTZ SAND
Sample No.
:
3 4 5 6 P
1: 11
Platinum
Palladium Added Recovered /g
rug
%
0.50 1.0 0.50 1.0 5.0 5.0 20
0.54 1.15 0.57 1.13 198
108 115 114 113 96 98
20.8 93 94
Added M 0.50 ;:; 10 5:;
104 93 94
1: 100 113 127
Rhodium Added
Recovered ,g % 0.55 1.12 2.02 440 4.95 9.65 20.9 106 102 103 126
110 112 101 88 99 19d 106 102 91 99
Recovered
pg
pg
%
0.50 1.0
0.57 1.01
114 101
5.0
5.4
108
DISCUSSION
Fire ussuy To obtain maximum recovery of the platinum metals from a sample, help to know its approximate mineral or chemical composition so that combination and proportions of the constituents of the assay flux can The tabular material below shows flux compositions used for a granite dotite; all quantities are in grams. Granite (G-l) Ore NasCOI PbO SiOI Borax glass CaP: Flour (for 28-g lead button)
15 20 50 : 1 3
30 30 60 0 3 1 3
it is a great the proper be selected. and a peri-
Peridotite @CC-l) :: 32 6 35 1 5 (excess to reduce all Pbo)
Relatively high concentrations of certain minerals or elements may require special treatment. Beamish stresses the problem of analysing for platinum metals in the presence of large amounts of nickel. Large amounts of organic matter such as coals require careful oxidation to avoid losses. Placer concentrates create separate problems, and slags from certain finely ground chromite ores may show considerable amounts of
116
JOSEPHm
and LEONARD
B. RILEY
unattacked chromite when examined under a microscope. Hence, fire assaying for platinum metals is more involved than routine fire assaying for gold. Results The results for palladium in W-l shown in Table VI are in reasonable agreement with previously published data. lo The platinum content in W-l, although varying slightly for different weights of sample taken (15 and 30 g), is ascribed a weighted average of 18 ppb. Two different lots of the dunite (DTS-1) were used for the two runs and this may account for the apparent slight variation in the platinum content. TAIILIX VI.-PALUDIUM, U.S.G.S. sample
PL.AllNUM, AND RHODIUM
Rock type
DO.
CONTENT IN STANDARD
Weight of
Pd, ppb
Pt, ppb
sample,
1
2
1
7 <4
ROCKS
Rh* PPb
2
1
2
15
t5 t5 <5 <5 <5 <5 <5 <5 <5 <5
<5 <5
g G-l W-l W-l G-2 GSP-1 AGV-1 BCR-1 BCR-1 PCC-I DTS-1
Granite Diabase Granite Granodiorite Andesite Basalt Peridotite Dunite
30 30 IS 30 30 30 30 15 15 15
t4 14 14 <4 <4 <4 <4 t4 7 <4
The elements listed in the table have been assigned conservative “limits of determination.” It will be noted that these limits are the same for both 15- and 30-g samples. Lower limits for the 30-g samples are possible and in favourable instances have been obtained. However, until more data are available the conservative figure given seems advisable. The term “limit of determination” is used here to mean that if the element in question were present in the sample in a concentration at or above this limit, not only would its presence be detected, but also its concentration could be stated, each with a reasonable degree of certainty. Below this limit the element, in a given sample, may be detected (and under favourable conditions may be stated) but, even with equal care, there is a reasonable chance it may be missed. Zusammenfassang-Ein Verfahren zur Bestimmung von Palladium bis herunter zu 4 ppb (Teile pro Milliarde lOa), Platin bis 10 ppb und Rhodium bis 5 ppb in 15 g Probe wird beschrieben. Dutch Schmelzverfahren werden die Platinmetalle in einer Goldperle vorangereichert; dann wird die Perle in Konigswasser gel&t und mit 1 M Salztiure zur Marke aufgefiillt. Die Losung wird analysiert durch opt&he Emissionsspeltrographie des Riickstandes, der beim Verdampfen von 200 ~1 auf ein Paar flachen Graphitelektroden zurilckbleibt. Diese Methode erfordert vie1 weniger Arbeits@nge mit der Probe als die meisten bekannten Vorschriften fiir diese Elemente. Ergebnisse fir G-l, W-l und sechs neue Standardgesteine des US Geological Survey werden mitgeteilt. Die Palladiumwerte in W-l stimmen befriedigend mit friiher publizierten Daten iiberein.
Determination
of palladium,
platinum and rhodium
117
R&nn&Gn decrit une methode pour le dosage du palladium jusqu’a des quantites aussi petites que 4 p.p.b. (parties par billion, IV), du platine jusqu’a 10 6p.b. et’du &dium*jusqu’b 5 p.p.b. da& I5 g d%chantillon. On utilise des techniaues de erillaee nour n&concentrer les metaux du platine dans uie perle a’or,-p& la pkrle est dissoute dans l’eau regale et dilute au volume a l’acide chlorhydrique 1M. On analyse la solution par spectrographic d’bmission optique du residu obtenu en en Cvaporant 200~1 sur une paire d’electrodes de graphite a t&te plate. Cette methode necessite beaucoup moins de manipulation d’tchantillon que la majeure partie des m&odes publiQs pour ces elements. On presente les don&es pour G-l, W-l et six nouvelfes roches &talons du U.S. Geological Survey. Les valeurs pour le palladium dans W-l sont en accord raisonnable avec les don&es anterieurement publiees. REFERENCES 1. E. E. Bugbee, A Textbook of Fire Assaying, 3rd Ed., Wiley, New York, 1948. 2. American Society for Testing and Materials, Methods for Emission Spectrochemical Analysis, 4th Ed., pp. 218 and 397. Philadelphia, 1964. 3. J. Haffty and D. M. Pinckney, U.S. Geol. Survey Prof. Paper No. 575-B, 1967, B-178. 4. J. Haffty, U.S. Geol. Survey Water-Supply Paper No. 1540-A, 1960. 5. H. M. Crosswhite, Spectrochim. Acta, 1950,4, 122. 6. H. Bastron, P. R. Bamett and K. J. Murata, U.S. Geol. Survey Bull. No. 1084-G, 1960. 7. W. J. Youden, Statistical Methods for Chemists, p. 16. Wiley, New York, 1951. 8. F. J. Flanagan, Geochim. Cosmachim. Acta, 1967,31,289. 9. F. E. Beamish, Talunta, 1960, 5, 7. 10. J. H. Croclcet and G. B. Skippen, Geochim. Cosmochim. Actu, 1966,30, 129.
1968.Vol. 1s. pp.111IO 117. pusunOn Rar.
Rimed h N-
-
DETERMINATION OF PALLADIUM, PLATINUM AND RHODIUM IN GEOLOGIC MATERIALS BY FIRE ASSAY AND EMISSION SPECTROGRAPHY* JOSEPH HAFFTY and LEONARDB. RILEY U.S. Geological Survey, Federal Center, Denver, Colorado 80225, U.S.A. (Received 14 July 1967. Accepted 5 August 1967)
S--A method is described for the determination of palladium down to 4ppb (parts per billion, lo*), platinum down to IOppb and rhodium down to 5 ppb in 15 g of sample. Fire-assay techniques are used to preconcentrate the platinum metals into a gold bead, then the bead is dissolved in eyua regia and diluted to volume with 1M hydrochloric acid. The solution is analysed by optical emission spectrography of the residue from 200 ,ul of it evaporated on a pair of flat-top graphite electrodes. This method requires much less sample handling than most published methods for these elements. Data are presented for G-l, W-l, and six new standard rock8 of the U.S. Geological Survey. The values for palladium in W-l are in reasonable agreement with previously published data. PROCEDURES for the determination of palladium, platinum and rhodium are available in the literature, but many are only suitable for much higher concentrations than that of most geologic materials which the U.S. Geological Survey is called upon to analyse. Some procedures require considerable time and analytical shill and are generally cumbersome and not too reliable in routine use. Still other procedures are designed for one of the three elements, giving little attention to the other two. The method described here is relatively free from extensive sample manipulations, adequately sensitive, and can determine all three elements in essentially a single procedure.
EXPERIMENTAL Fire assay
Gold (weighing between 0.80 and 1.20 mg), which is used as a collector for the platinum metals, is added in the form of O-007in. dia. wire to yield approximately the same concentration of gold in the fu& solution as in the spectrographic standards. A piece of lead foil 4 in. square and @016 in. thick is sharply folded at the center with pliers. The fold is then opened, the gold wire placed in the centre of the crease, the fold closed on the wire and again folded at the centre to give a thickness of four layers. All edges are then crimped with pliers to ensure enclosure of the wire. This lead envelope is added as the last step in the preparation of the fusion mixture described below. The fusion and cupellation step8 in general follow conventional fire-assay practices as described by Bugbee’ and are only summarized here to point out variation8 found helpful. A flux is formed by combining sodium carbonate, lead@) oxide, silica, borax glass and calcium fluoride together with flour as a reducing agent or potassium nitrate as an oxidizing agent. These am combined in a “30-g” tireclay crucible with a maximum capacity of about 260 ml. The amount8 of the flux component8 are dependent on the composition of the sample to be analysed. The object is to form at least a two-phase melt: a liquid slag approximating to a complex borosilicate, and a metallic lead phase, controlled in sire, to collect the noble metals present. The addition to the flux of 1-3 g of calcium fluoride, not * Publication authorized by the Director, U.S. Geological Survey. 8
111
112
JOSEPHIIAFFIY and LEONARD B. RILEY
routinely used in gold fire-assaying, has been found beneficial for many types of samples analysed for the platinum metals. After the flux components are in the crucible, 15 g (or 30 g) of sample are added and the contents thoroughly mixed. Then the lead foil containing the gold wire is added. Part of the borax glass for the charge may be added as a cover for a 30-g sample, or for a sample liable to form dust, or for a sample relatively rich in iron(II1) oxide. The crucible is placed in the muffle of an assay furnace preheated to a temperature of 850-900”. After the door is closed the temperature is raised to 1050-1120” over a period of 30-45 min. The crucible is then removed and the molten charge is poured into an iron mould and allowed to cool. The slag is carefully chipped and worked away from the lead with a hammer and brush, and the lead is tapped into a cube, producing the lead button. The lead of the lead button is separated from its noble metal content by cupellation. The cupels, made of compressed bone ash, are placed in the furnace, preheated to 95&1000”, for about 10 mm. Then the lead buttons are placed in the cupels and the furnace door closed until the lead melts and a bright surface is formed on the lead. The door is opened and the temperature first lowered to about 830”, then gradually raised to a fmishing temperature of about 900” as the lead is oxidized and the oxide absorbed by the cupel. The cupellation is finished when only a noble metal bead remains on the cupel. Since silver is not to be determined, higher temperatures may be used than in the conventional silver assay, thus reducing the dangers of “freexing” (lead oxide covering the molten lead and preventing further oxidation). Gnce the bead is obtained, the cupel is removed from the furnace and allowed to cool. No precautions are necessary to prevent “sprouting” as in the silver assay, since there is about 1 mg of gold present, The bead is examined under a stereoscopic microscope (about 45 x magnification) while still on the cupel and the colour, surface, and shape are noted. The presence of certain elements is indicated when the colour has drastically changed from that of gold. No correlation, as yet, has been made with the shape or surface of the bead and the presence or concentration of specific elements. This is still under investigation. The bead is then carefully transferred and weighed. Relatively pure gold beads are nearly spherical and need more care in handling than hemispherical silver beads. No attempt is made to clean the bead, e.g., by brushing, thus avoiding any losses of metals. The small amount of lead-saturated bone ash retamed on the bead adds little to the weight and is easily dissolved in the subsequent treatment. The weight of the gold wire added to the flux initially is compared with the weight of the bead obtained. Any sizable gain (or slight loss) is noted and aids in calculating the dilution factor if the weight has substantially increased. The bead is transferred to a l-ml volumetric IIask to which 0.1 ml of uq~ regiu (3 : 1 hydrochloric acid:&& acid)is added, and allowed to stand overnight at room temperature. To ensure complete dissolution of the bead, the solution is heated for 15-20 min on a steam-bath. After the bead has diiolved. an aliquot of a 4 mg/l solution of molybdenum (the internal standard) in 2M hydrochloric acid is added so that the concentration is 0.2 mg of molybdenum per 1. after dilution to volume with 1M hydrochloric acid. All beads analysed to date have been soluble in aqua regiu. However, if any metals remain undiilved, the presence of other platinum metals should be suspected. When a sample is known from a previous semi-quantitative spectrographic analysis to be high in platinum metals, the amount of gold is increased so that a 2: 1 ratio of gold to total platinum metals is obtained. This ratio has been found satisfactory for dissolution of the platinum alloys. Larger volumetric equipment and more acid may then be necessary as stated by Murt and Cot-son.* Dilutions are made from-the solution to obtain a desirable concentration for-measurement of the platinum metals. Gold in solution (10 mglml) is added to the final dilution in order that the gold concentration be about 1000 mg/l and approximately equivalent to the concentration of gold in the standards. S’ctropphic adysis Two &in. dia. flat-top, graphite electrodes are first waterproofed with 4 drops each of a petroleum
ether solution of Apiexon “N” grease (5 g/l). Then 100 ,ul of the unknown sample or the standard, both containing the internal standard, are added to each electrode in SO-p1increments and evaporated by means of a heat lamp. The electrodes are kept under the heat lamp until placed in the arc-spark stand. Each exposure consists of exciting the residue of 200 ~1 of solution on the two electrodesS. The sample-bearing electrodes are then placed in the arc-spark stand and the residue is excited and the spectrum recorded, with the a paratus and operating conditions listed in Table I. The emulsion is calibrated ! y means of an iron bead arced at 5 A d.c. and exposed for 15 set at a transmission of 48 %. The formation of the iron bead is described by Haffty.’ The iron lines and relative intensities used for plate calibration were selected from a list of homologous lines.S*a They are reproduced in Table II for the convenience of the reader. The calibration curve is established by plotting transmission (ordinate) us. intensity (abscissa) on log-log paper.
Determination of palladium, platinum and rhodium TABLEI.-APPARATUS
AND SPECI~~OORAPHIC OPERAT?NQ CONDITIONS
Excitation source-intermittent d.c. arc. Primary power source-full-wave rectified 280-V. 6O-cycle power supply-initiated by a high-voltage condensed spark synchronized to initiate each half cycle. 2-OA Radio-frequency current, Initiating circuit parameters 0.0025 ,eF Capacitance, 15 PH (residual) Inductance, 16 R (dial setting 5) Primary resistance, 0.5 R (residual) Secondary resistance, 15,000 v Discharge voltage, 11 Number of discharges per half cycle, 7.0 A Current, radio-frequency, Spectrograph-folded. 3-m Rowland circle-mounted grating with 21,ooO lines/in. giving a reciprocal linear dispersion of 4 A/mm in the first order. 240&36OOA Wavelength region, Slitwidth, 4OP arc image focused on grating Illumination, 3mm Analytical gap, Transmission, 100% 20 set Exposure+ Kodak SA-1 Emulsion, ASTM type C-3 Electrodes,
TABLEII.-IRON LINESUSED FOR PLATE CALIBRAnON Fe line,
Fe line,
A 3157.89 3175.45 ~178.02 3196.93 320540 321594
Intensity
A
Intensity
73 125
3217.38 3222.07 3225.99 3239.44 3251.24 326824
165 625 770 285 62 29
4: 210 22!!
TABLEIII.-ANALYTICAL LINES USED IN DETERMWATION OFPALLADIUM,PL.4~ANDRHODfUhf
Element Pd Pt Rh
A”alyYl Pd Pd Pt Rh Rh Rh
line,
3242.70 340458 2659.45 2490.77 3396.85 3434.89
Concentration range based on analytical curve, mgll @05&O 0.025-05 0.125-5-O 0*125-5-o 0*05-5~0 0~05-125
The MO 2816.15 A line was used as internal standard line throughout.
113
JOSEPHHAFFW and LEONARDB. RILEY
114
Each plate contains the exposures of 10 or 11 samples in duplicate, plus 9 standards and one iron bead to calibrate the emulsion. After the plate has been processed, transmittance measurements of selected analytical and internal standard lines (Table III) are obtained by means of a microphotometer. From the calibration curve, intensity values are obtained for the standards which are used in constructing the analytical curves. The analytical curves are established by plotting the ratio of the intensities of the analytical and internal standard lines us. concentration on logarithmic coordinates. The concentrations of the elements in the unknown samples are read from the analytical curves. The absolute weight of metal is calculated from the volume of solution in which the bead was dissolved. This weight is then divided by the weight of sample taken to give ppm or ppb [parts per billion (log)] in the sample. Preparation of stpndardr
Stock solutions, calculated to contain 1000 mg of the element per l., are made by dissolving the ammonium salts of the appropriate chloro-complexes in 2M hydrochloric acid. These stock solutions are then serially diluted with 2M hydrochloric acid to give concentrations of 100, 50, 25, 10, . . . 0.5 mg/l. The standards for analysis are made by transferring O-2ml of a gold stock solution (10 mg/ml) to each of nine 2-ml volumetric flasks. To these, 0-l ml each of the palladium, platinum and rhodium solutions are added. In addition to this, O-1ml of a solution of molybdenum (4 mg/l) in 2M hydrochloric acid is added as the internal standard and the mixtures are diluted to volume with 1M hydrochloric acid. The resulting standards contain palladium, platinum and rhodium in concentrations of 50,5,2*5,1*25,0*5 . . .0*025 mg/l in solutions containing gold (1000 mg/l) and molybdenum (0.2 mgll). RESULTS
The precision of the method is illustrated by the results in Table IV. The standard deviations were obtained by running several 15-g portions of the same sample on various plates at different dates.2*7 TABLEIV.-DATA ON PRECISION OF METHOD Average concentration, Element
PP”
Palladium Platinum Rhodium
0.426 0.395 o-0143
d n X S.D. C.V.
= = = = =
S.D. O-023 OX@77 O-0008
C.V.
n
5.4 1.9 5.6
5 5 2
difference between duplicate values. number of duplicate determinations. average of the individual results. 2/(X da/h) = standard deviation. lOO(S.D.)/x = coefficient of variation.
In order to obtain an indication of the accuracy of the method at concentrations present in geological materials, aliquots of known solutions of palladium, platinum and rhodium were added to 15-g portions of quartz sand, evaporated under heat lamps, and carried through the method described. The solutions were added in O-l-ml increments and distributed over the surface of the sand, which had previously been heated for about 20 min in an evaporating dish. The total amount of solution added to any one dish was no more than O-6 ml and with proper care the solution did not seep through the sand to the evaporating dish. After the sand was added to the fire clay crucible, the additional precaution was taken of dry-rinsing the evaporating dish with the soda needed for the charge and then transferring this to the crucible. However,
Determination of palladium, platinum and rhodium
115
the platinum
in samples 10 and 11 was accurately weighed and added as the metal. The results are presented in Table V. For low-grade platinum ores, 30 g (approx 1 assay ton) have been used. When the weights of the elements given in Table V are considered on the basis of this weight of sample, they represent a concentration range of 17 ppb to 4.23 ppm. The analytical results for G-l, W-l, and six new standard rocks are presented in Table VI. Each run is the average of duplicate determinations. More detailed data on the localities from which the six new standard rocks were collected, and the results for many elements, were presented by Flanagan.8 TABLEV.-RECOVERY OF PALLADIUM,PLATINUM AND RHODIUM ADDED TO QUARTZ SAND
Sample No.
:
3 4 5 6 P
1: 11
Platinum
Palladium Added Recovered /g
rug
%
0.50 1.0 0.50 1.0 5.0 5.0 20
0.54 1.15 0.57 1.13 198
108 115 114 113 96 98
20.8 93 94
Added M 0.50 ;:; 10 5:;
104 93 94
1: 100 113 127
Rhodium Added
Recovered ,g % 0.55 1.12 2.02 440 4.95 9.65 20.9 106 102 103 126
110 112 101 88 99 19d 106 102 91 99
Recovered
pg
pg
%
0.50 1.0
0.57 1.01
114 101
5.0
5.4
108
DISCUSSION
Fire ussuy To obtain maximum recovery of the platinum metals from a sample, help to know its approximate mineral or chemical composition so that combination and proportions of the constituents of the assay flux can The tabular material below shows flux compositions used for a granite dotite; all quantities are in grams. Granite (G-l) Ore NasCOI PbO SiOI Borax glass CaP: Flour (for 28-g lead button)
15 20 50 : 1 3
30 30 60 0 3 1 3
it is a great the proper be selected. and a peri-
Peridotite @CC-l) :: 32 6 35 1 5 (excess to reduce all Pbo)
Relatively high concentrations of certain minerals or elements may require special treatment. Beamish stresses the problem of analysing for platinum metals in the presence of large amounts of nickel. Large amounts of organic matter such as coals require careful oxidation to avoid losses. Placer concentrates create separate problems, and slags from certain finely ground chromite ores may show considerable amounts of
116
JOSEPHm
and LEONARD
B. RILEY
unattacked chromite when examined under a microscope. Hence, fire assaying for platinum metals is more involved than routine fire assaying for gold. Results The results for palladium in W-l shown in Table VI are in reasonable agreement with previously published data. lo The platinum content in W-l, although varying slightly for different weights of sample taken (15 and 30 g), is ascribed a weighted average of 18 ppb. Two different lots of the dunite (DTS-1) were used for the two runs and this may account for the apparent slight variation in the platinum content. TAIILIX VI.-PALUDIUM, U.S.G.S. sample
PL.AllNUM, AND RHODIUM
Rock type
DO.
CONTENT IN STANDARD
Weight of
Pd, ppb
Pt, ppb
sample,
1
2
1
7 <4
ROCKS
Rh* PPb
2
1
2
15
t5 t5 <5 <5 <5 <5 <5 <5 <5 <5
<5 <5
g G-l W-l W-l G-2 GSP-1 AGV-1 BCR-1 BCR-1 PCC-I DTS-1
Granite Diabase Granite Granodiorite Andesite Basalt Peridotite Dunite
30 30 IS 30 30 30 30 15 15 15
t4 14 14 <4 <4 <4 <4 t4 7 <4
The elements listed in the table have been assigned conservative “limits of determination.” It will be noted that these limits are the same for both 15- and 30-g samples. Lower limits for the 30-g samples are possible and in favourable instances have been obtained. However, until more data are available the conservative figure given seems advisable. The term “limit of determination” is used here to mean that if the element in question were present in the sample in a concentration at or above this limit, not only would its presence be detected, but also its concentration could be stated, each with a reasonable degree of certainty. Below this limit the element, in a given sample, may be detected (and under favourable conditions may be stated) but, even with equal care, there is a reasonable chance it may be missed. Zusammenfassang-Ein Verfahren zur Bestimmung von Palladium bis herunter zu 4 ppb (Teile pro Milliarde lOa), Platin bis 10 ppb und Rhodium bis 5 ppb in 15 g Probe wird beschrieben. Dutch Schmelzverfahren werden die Platinmetalle in einer Goldperle vorangereichert; dann wird die Perle in Konigswasser gel&t und mit 1 M Salztiure zur Marke aufgefiillt. Die Losung wird analysiert durch opt&he Emissionsspeltrographie des Riickstandes, der beim Verdampfen von 200 ~1 auf ein Paar flachen Graphitelektroden zurilckbleibt. Diese Methode erfordert vie1 weniger Arbeits@nge mit der Probe als die meisten bekannten Vorschriften fiir diese Elemente. Ergebnisse fir G-l, W-l und sechs neue Standardgesteine des US Geological Survey werden mitgeteilt. Die Palladiumwerte in W-l stimmen befriedigend mit friiher publizierten Daten iiberein.
Determination
of palladium,
platinum and rhodium
117
R&nn&Gn decrit une methode pour le dosage du palladium jusqu’a des quantites aussi petites que 4 p.p.b. (parties par billion, IV), du platine jusqu’a 10 6p.b. et’du &dium*jusqu’b 5 p.p.b. da& I5 g d%chantillon. On utilise des techniaues de erillaee nour n&concentrer les metaux du platine dans uie perle a’or,-p& la pkrle est dissoute dans l’eau regale et dilute au volume a l’acide chlorhydrique 1M. On analyse la solution par spectrographic d’bmission optique du residu obtenu en en Cvaporant 200~1 sur une paire d’electrodes de graphite a t&te plate. Cette methode necessite beaucoup moins de manipulation d’tchantillon que la majeure partie des m&odes publiQs pour ces elements. On presente les don&es pour G-l, W-l et six nouvelfes roches &talons du U.S. Geological Survey. Les valeurs pour le palladium dans W-l sont en accord raisonnable avec les don&es anterieurement publiees. REFERENCES 1. E. E. Bugbee, A Textbook of Fire Assaying, 3rd Ed., Wiley, New York, 1948. 2. American Society for Testing and Materials, Methods for Emission Spectrochemical Analysis, 4th Ed., pp. 218 and 397. Philadelphia, 1964. 3. J. Haffty and D. M. Pinckney, U.S. Geol. Survey Prof. Paper No. 575-B, 1967, B-178. 4. J. Haffty, U.S. Geol. Survey Water-Supply Paper No. 1540-A, 1960. 5. H. M. Crosswhite, Spectrochim. Acta, 1950,4, 122. 6. H. Bastron, P. R. Bamett and K. J. Murata, U.S. Geol. Survey Bull. No. 1084-G, 1960. 7. W. J. Youden, Statistical Methods for Chemists, p. 16. Wiley, New York, 1951. 8. F. J. Flanagan, Geochim. Cosmachim. Acta, 1967,31,289. 9. F. E. Beamish, Talunta, 1960, 5, 7. 10. J. H. Croclcet and G. B. Skippen, Geochim. Cosmochim. Actu, 1966,30, 129.