Sample Preparation and Bulk Analytical Methods for PGE ERIC L. HOFFMAN ‘Activation Laboratories Led, Ancaster, Ontario, Canada, L9G 4V5 and BERNIE DUNN ‘Actlabs-Skyline “Tucson, Arizona 85745, US.A. Introduction The geologic secting of primary platinum-group element (PGE) deposits vaties from low-grade - large tonnage (Lac des Inks Complex in Canada, Watkinson et al., his volume) co high-grade, natrow “tect” cype deposits (Merensky Reef in South Africa, Barnes and Maicr; Cawthorne etal, this volume), Stillwater Complex in United States (Zientek ct a., this vol- tums), Great Dyke in Zimbabwe (Oberthiir, this volume). ‘Nickel-Cu ores (Norilk-Talnatch in Russia, Kozyrev et a, this ‘volume, Sudbury in Canada, Farrow and Lightfoot, this vol- time, and Kambalda in West Australi) are also major contribu- tors of PGEs as by-products. The diverse nature of the ‘mineralization results in grades, which vary considerably. New geochemical exploration techniques for PGE. have been made viable as a result of the development of new analytical instru- ‘ments, principally fourth generation ICP/MS technology. ‘Whether it is ore grade assessment, ot geochemical surveys of lake-botiom sediments or tree tops, these exploration tech- niques require accurate and precise analy data ranging from the sub-ppb to multiple ppm levels. Field sampling procedures are beyond the scope of this papers however, proper sampling and sample preparation proce- dures are of paramount importance prior to considering the ‘method of analysis to be used for the PGE determination. The PGE can occur as discrete mineral phases or in solid solution in ‘mineral phases (Cabri, 1981, 2002; Mitchell and Keays, 1987; ‘Talkington and Watkinson, 1986; Scoates and Eckstrand, 1986. The dbwibaion of PGF can be ease within samples parte larly if they are present as discrete platinum-group minerals, (PGM), Ths “mugger” ter for PCE i smart that known ‘ occur commonly for gold, but perhaps not quite as extreme ‘a the PGMs usually are nor composed solely of the PGE, ‘whereas, native gold particles would be almost entirely gold ot gold-ilver alloys. The required sub-sample size that should be ‘hosen for analysis will depend on PGM particle size and the rioealogy of the ample ‘There are many methods of analysis available for determin- ing Pr and Pd only. The mose common method is the standard lead fre assay using silver asa collector for the Pe and Pd fol- lowed by inductively coupled plasma/optia! (also commonly known as acomic) emission spectrometry (ICP/OES or ICP/AES), or direct eutrent plasma spectrometry (DCP for the final analytical finish. A. variation on the fead fire assay using ‘gold asa collector is required co determine Rh quantitatively as Rh forms insoluble alloys in the Ag colleceor matrix that cannot be dissolved quantitatively. These techniques are rapid, accurate and low cost. For the ultimate in sensitivity, inductively coupled plasma mass spectrometry (ICP/MS) is used for the analytical finish, The ICP/MS method has great application for large-scale lake-bottom sediment surveys (Ontario Geological Survey, 2000), and lithogeochemical surveys looking for depletion and enrichment in magmatic stratigraphy (Hulbere, 2001). The lease used analytical finishes for PGE in modern laboratories are atomic absorption spectromeuy (AAS) and graphite furnace ‘atomic absorption spectrometry (GFAAS). ICP methods are faster, cheaper, and have fewer elemental interferences. The nickel sulphide five assay technique isthe most effective for the determination ofall PGE plus Au. After collection in the nickel sulphide matcix, the button produced is dissolved in com- centrited hydrochloric acid and che residue analyzed by inseru- ‘mental neutton activation analysis (INA), ot it can be further digested in qua regia and analyzed by ICP/MS. Telluciumn co- precipitation may be requiced to eliminate some interferences, and to ensure chat che PGE are not partially dissolved. These pro- cedures ate usually slower and more cosy chan the standard lead. ire assay procedure, bue chey are the only practical methods for quantitative determination ofall the PGE plus Av. Certain matrices, such as water, can be analyzed direcly by TCPIMS; howeves, at natural levels, (cl nanograrlltr), the PGFs are unlikely o be detected. In addition, adsoxption on to the polyethylene containers i a distinct possibile, despite acid- ification co pH 2. This effect has been described ly for ‘Au by Hall eral. (1986) and probably also occurs with the PGE. Pre-concentration by 2 factor of 10 would probably be required 1 establish background levels for PGE in water. Plant ash can also be analyzed direcly by ICP/MS (Dunn ‘etal, 2001} for Pr and Pd providing appropriate analytical cor- vierections are applied. An advancage ofthis technique is that only 0.25 g t0 0.5 g of ash is required for digestion. This tepresencs & very much larger volume of dried vegetation, For example, a 196 ash yield (ash yield is defined as ash weight/dry weighe x 1008), would require 50 g of dry vegetation to give 0.5 g af ash, Typi- «ally itis only convenient co collect and ash 50°g c0 100 g of vegetation as this represents a relatively large sample volume. In addition, geologises and geochemists may request 2 60 plus smult-element package of elements (including Cu, Ni, Cx, Co, As, Sb, Se, Te, Bi) which can be determined at the same time as the Prand Pa by ICP/MS. Vegetation ash hat also been analyzed by Pb fire assay collection (Dunn, 1986; Duna et al, 1989). This technique has potential blank problems, ic requires 1 g of ‘mote of ash, and provides data only for Pe, Pd, and Au, making this a much less attractive option. Selective extractions are growing rapidly in theic use for min- eral exploration, These analytical procedures attack specific min- manganese (Enzyme Leach*), amorphous Mn and Fe oxides, and weakdy bound ions (PGETer. sas01%), and weakly bound ions (MMI™). These can be ane- Iyzed forall of the PGE Leach and PGE Tearasols™) (Fill and Claris 2001) 0 for Pe and Pd (MMI™). Generally, these techniques extract only small amoums of intefering dle ‘ments, but ineerelement corrections mus sl be applied, For certain rock matrices, is posible to analyze rocks and soils for PGE by sagua regia digestion of a large sample with sub- sequent analysis by ICP/MS. However, ome PGE phases eur as minerals and alloys, which are notclssolved with acids. Detoo. tion limits are ceatvely high (Faction of ppm, i, 0.2 ppm for Pa) for some PGE due to corrections which must be applied for intefeing elements, Extractions into MIBK or DIBK may be used to reduce this type of interference, but this technigue is not tecommended for the mast reliable PGE results. Analysis of PGE by gravimetry, DC arc/spark emission, and spectrophotometry have been made zedundane by modern tech nology and are not discussed further, ‘There are a number of laboratories which proclaim to have “special proprietary processes for PGE.” Most of thete laboraro- ries use cither agua regia digestion and analysis by AAS, ICP/OES or DCP with or without some type of “special fire ‘say" (ison, nickel, copper, and tin have been noted). Analyses Petfonmed in this manner are unceliable a best and will usually report multi-ounce/ten results due to measuring interferences ‘caused by elements such as Fe, Ni, Ca, Sn, and possibly others. ‘Sampling Pitard (1989) notes that “preparation isa sequence of non- Selective observations such as transfer, crushing, grinding, pul- veriaing, drying, mixing, et., carried out on a batch of matter in order to bring it under a convenient form for the next pro- cessing stage, which can be another sampling stage or the ulti amate analysis, Preparation shall not be cofised oth sampling, which is the selection process itelE” Any of the preparation steps listed can lead to substantial ercor if not propery. Hellman (1999), AusIMMC-AIG (1999), Bloom and ‘Titaro ~The Geology, Geochemistry, Mineralogy and Mineral Beneficiaton of PGE. (1997), and Brooks (2001) have described a number of thes Preparation failures. Geological matctals collected for analysis should be repre- sentative of the material being sampled. Theoretically, che size of sample should depcad on the patticle size distribution of the PGE-bearing phases, and consideration should be given to the deposit geology. ‘There are many PGE-beating phases, chat include arsenides, bismuthides, selenide, telluride, sulphides, sative metas, and alloys, and the PGE may be present in solid folution in various sulphide minerals. Cabti (1993, 2002) describes more fully the mineralogical possibilities, which exoced 100 PGM and over 20 PGE bearing mineral species, sv eral of which are common sulphides. Determination of PGE mineralogy is complex and there are few skilled mineralogist in this field. Determination of the PGM patticle size, and hence required sample size (Frangois-Bongaegon, 1998), is nota crivial task and usually it is pecformed after a deposit of some conse quence has been oudined. ‘Commonly; the appropriate sie of sample is determined by tial ‘and error through an iterative process of sampling and. analysis until an appropriate sized sample is determined. Ini- tially, samples colleced depend on the proposed genetic model for the mineralization. For example, clivomite seams are sam- pled one way, whereas, sulphide mincralizaion, (or establishing background levels in rock) ate sarmpled differenly. Further dis: ‘cussion of field sampling is beyond the scope of this paper. Suf- fice it to say that a minimum requizement for determining sampling variance should be to take replicate samples in the field. This will give an estimate of how much variance is caused by field sampling, and permits adjustment of the sampling pro- ‘gram to an acceptable level. Sampling remains the largest source of error for PGE explo- sation programs, The use of chip sampling and composite sam_ pling may provide che best represeativity ofthe sample. “High reding” of chromice seams or Cu-Ni mineralization should be avoided in order to obvain reliable estimates of grade. Brooks (2001) provides many examples of how sampling errors have resulked in a number of exploration and mining development projects to be severely compromised, Consistency in the sclecced. ‘sampling procedure used for any given project is keynote. ‘Sample Preparation (Once rock samples ative in the laboratory, they muse be Prepared accordingly for analysis, A jaw crusher finely crushes the entire sample on the primary crush. Secondary cone erushy ‘rs ate no longer widely used. ATM Rhino™ crusher of rolls crusher is widely used in many commercial laboratoties for the Drirwaty crush and will crush to 80% -2 mm (-8 mesh) if prop- erly maintained. Typically, rotay or rifle splitters are used to produce a representative sample, which can then be pulverized (© 95%6 -75 y4m (-200 mesh) (some labs use 95% - 105 pm oF 150 mesh). Usually, « 100 g to 1 kg split can be pulverized depending on the mills available in the laboratory: For most samples itis bese to pulverize using sealed bowl pulveriers [ring and puck type (sometimes called swing mills), or “Aying saucer® 72Horr AND DUR oso essen ype mills such as those produced by Labtechnics]. Ring and puck mills usually provide the finest pulp, ulpe from ring mills, which pulverize 100 g to 200 g, ate dumped onto a “cloth” or sheet and the “pour to fil” process is used. This technique does not cause segregation problems. Brooks (2001), recommends caution when using “lying saucer” type pulverizers; he showed an example of up to 35% difference taken from samples prepared using this type of pul- vetizer from samples from the Bendigo gold deposit in Aus- tralia, He theorized that minerals with high specific gravity such as Au or the PGM segregate and setde to the bortom of the bowl. This can particularly be a problem if the sample is erized for too long a time period. Common practice when using bowls pulverizing 2.5 kg or mote is to take a scoop of pulverized material out ofthe bow, usually near the top of the pile, and discard the remaining materia. It is virtually impos- sible to dump the powder out of the bow] duc to its weight and Brooks’ warning must be carefully considered when dsing this sample preparation protocol. A solution co determine if there is problem may be to take several spits as pulp repli- cat fom the op, sides nd btm ofthe pale ample when ie is still in the original bovd and evaluate the variance. Brooks (2001) also notes: *— errors of the order of 30% 10 4096 of the true values of ‘gold, uranium. copper, and platinum to name a fas are readily ‘noraduced by such practices as unfethful halving of core, incorrect feeding of saraple crusher, incre: feeding of piers, sucha Jones ‘ifflesplisters — or even wing incorrectly designed equipment. Soft igh value minerals such as native gold, elecrum and molybdenite readily contaminate semple preparation equipment, while dense miners such as gold, elec, uraninite and the platinum group etal: —” end 10 separate fromm comminused host rocks because the Latter have specific grivties thas are of the onder of a quarter to an cighth ofthe heawy minenas of interest. He goes Further to recommend against using, contra-rotar- {ng disc pulverizers, which will segregate minerals based on den- sity. The situation may be farther aggravated by improper mat rolling with improper sub-sampling, The heavy minerals segre- gate on the mat surface, and have an overlying layer ofthe lower grade p EEE vata dot deanersand to can proces ui ‘ment between samples is suggested to avoid cross contamina- tion, Contaminasion from sample comminution equipment also can occur For an example, inthe late 1990s, ic was found that bowls of the “Aying saucer” type were made from steel coming from scrap iron originally used for bal mills from a gold mine. Gold contamination of more than 10 000 samples occurted before the problem was found. Clearly when, ‘new equipment or supplies ate used, appropriate process blanks ‘must be run to identify these problems and correct them before ‘using new process equipment or supplies Quality Controt ‘After the pulp has been prepared, randomly selected pulp duplicates or even replicates should be chosen for replicate analyses. Routinely, at leat 1 in 20 samples should be replicate. More Frequent replicates should be chosen if the analytical ‘method chosen requites this. An additional control is to take preparation duplicates which are splits of the -10 mesh reject ‘material which can be sub-sumpled by mechanical spliting and then separately pulverized. If suficient preparation and pulp duplicates are chosen for analyses, variance can be compused from each of these steps to determine if the sample preparation steps are adequate, If variance exceeds expecrations, moxlifca- tions to the procedure are advisable. In extreme “nugger effecc” cases, a metallic sereen analysis may be advised. Ina metallic screen analysis a large sample is pulverized and complerely screened at say -150 mesh. All the coarse mineralization (native ‘gold, metallic PGM), which may have been fattened on pulver- izing is screened out and assayed separately. This removes the ‘coarse nuggets, which can cause homogeneity problems and ensures all coarse particles are assayed from the whole sample. The -150 mesh (in duplicate) masetial is analyzed separately. The assays aze cecombined based on the otiginal distribution of the oversize and undersize material and a combined assay is arrived at. Te is important chat rock samples be finely ground to achieve, for example, 95% that will pass through a -200 mesh C74 ym), particulary when chromite is present. If rock samples are not finely ground chromice grains may not be complerely melted duting fire assay procedures. ‘Many geologists elecr to sieve their soils or sediments to - 80 mesh (-177 jm). This is the absolute coarsest particle size required for 2 fre assay. This course particle size, however, may result in poor PGE recovery and/or reproducibility. Use of a -150 mesh (-105 im) sieve size ensures more complete dissolu- sion of grains in dhe fre assay process. Depending on what infor- mation the geologise wanss to obtain (eg. chemical versus ‘mechanical dispersion) the appropriate sample preparation pro- cedure can be chosen. This may require a two-stage preparation program including sieving and pulverizing. Scrice sample preparation procedures should be followed since they probably represent (outside of field sampling), the largest component of the total analytical errot. Preparation blanks, preparation duplicates and sample replicates should be routinely reporced with analytical data. This is becoming mandatory with the International Standards Organization accreditation to ISO/IEC Guide 25 or its new replacement ISO. 17025 and accreditation to CAN-P-1579, the Suandand for Mineral Analysis Laboratories developed by the Standards Council of Canada (SCC). Reference Materials and Validation af Results ‘An analytical laboratory normally assumes that blanks and concrol standards are included with sample submissions so that further validation of laboratory analytical results ean be assessed. It appears, however, that this docs noe always happen in routine sample submissions to commercial abora- tories, partly because there is a scarcity of International Ces tified Reference Materials for PGE, Most, with the exception 73.+ The Geology, Geochemistry, Mineralogy and Mineral Benefciaton of PGE <0 :+-+----:0:++ ‘Table 2. Some of the more widely used analytical certified reference materials for PGE. ‘Standard ‘Souree Pt a Rh * 8 ma 7 (oob}____toob) (xb) oot) toptd loot) ob) SauT ANTEX erenshy Ret) 30 «1540 240 7% 6 430 310 Preis ‘CANWET (supe conceal) a0 aD 330 10" = 210" 1310 Pmea ‘CANE iNiCu mate) 73101001020 350" - 700" 3300 PTaL ‘CANNET ftinferous back sand) 3050 ~ - ~ = - - SUA CCANMET WCCO oe) 410 370 - - - - 200 st. CCANWAET (rama or tangs) 1a 106 95 38 80" 109 8 WoB CCANIET (gabbeo rock) 61 139 032 033 03 29 wel CCANMET tmineratzed gabbro) 73 382 2 46 aa 5 no ws ‘CATWETtmassive subhise) iat 188 28 235 ng 9 279 Wer CANMET ater peridot) 285 235 134 135, a 2 2 BL CANE (base) 58. ee or 9.15 ~ oa" 63 “ves are ecoreended, ony, oreo inrairal purposes. ‘of SARM-7, are sold in volumes of 100 g to 400 g (Table 1), and use of 30 g to 50 g aliquots for fire assay quickly con- sumes these standards. Custom standacds of lower cost and higher volume can also be made from a variety of source materials to match exploration needs. Typically, the selected selected can be prepared and homogenized then submieted for “round robin" testing to arrive at consensus values. Standatd prepatatian can be performed by a number of government facilities (eg, CANMET and Ontario Geo- science Laboratories) and/or by commercial laboratories, of sample preparation equipment manufacturers (e.g., Rock- labs}. Matcices of standards should be similar to the material being submitted for analysis otherwise che procedures will not be adequately tested: i.c., PGE chromite standards should be used for chromite PGE. prospects and sulphide standards for sulphide projects. Also, a range of standards covering a range of values should be used, tather than a si ale standacd. Some variability should be expected due to nate tral variation which is documented by the entity supplying the standard reference material, CANMET (Canada Centce for Mineral and Energy ‘Technology), as part of PTP-MAL (Proficiency Testing Pro- gram for Mineral Analysis Laboratories), provides profi- Giency testing of laboratories for Pe and Pd. These proficiency tests for Pt and Pd occur twice per year and are known to the participating laboratories. Tn the last PTP- MAL test of 2000, 45 laboratories worldwide participated, but only 25 participated for Pt and Pd. As general good prac- tice, a recent report by the Totonto Stock Exchange and Ontario Securities Commission (1999) recommends a repre- sentative selection, usually 5% to 10%, of pulps and rejects should be sent to another laboratory fo confirm analytical results and to ensure that a laboratory does not have bias in results. This cannot be ovedly stressed as many laboratories may have difficulty with PGE analyses. Lead Fire Assay Collection Since antiquity, the method of choice for gold analysis has been fire assay. The first recorded use of che fire assay tech- nique was that of Erker, asayer to the Holy Roman Empire and fire assay probably even predates that time. The classical lead fire assay, which uses silver as a collector, collects Pt and Pa as well as Au, if the sample is fused properly: The princi- pple behind this technique is chat the sample is mixed with a flux which lowers che melting point of the geological sample material, forming totally molten liquid at 100°C to 1100°C. Usually the fusion time is 45 to 60 minutes, The Mux is usually composed of litharge (PbO), soda ash (Na CO), sodium borate (borax, Na,B,O,), silica and houschold flour (added as a reductant) to which Ag has been added as a col- Tector (usually in solution for highest purity). Details on the lead fire assay procedure are available in Bugbce (1940), Steele etal, (1970), Moloughney and Faye (1976), Beamish and Van ‘Loon (1977), Hafiy et al. (1977), Moloughney (1986), and ‘Hoffman et al. (1998), The Au, Pt, Pé, and Ag are highly sol- uuble and compatible with lead which has been reduced from PbO by the carbon present and is not compatible with the boro-slicate lag. As the lead droplets in che molten mixture setile to the botom of the crucible, due to the high specific stavity of lead, the Au, Pt, Pd, and Ag are normally scavenged. ‘effectively from the melt by the lead. The flux must be of the appropriate viscosity to allow the PbO to be reduced to the Pb droplets and come into intimate contact with the sample material before it has a chance to settle under gravity to the bottom of the crucible. Ifa flux is too fluid, seeding occurs too rapidly; if che ux is too viscous it results in Pb shot (molten Pb droplets) to be present in the slag and this may cause some retention of PGE and Au in the slag. "The crucible melt is poured into a mold which allows the Jad wo sink co the bottom: the slag, which contains most of the silicates as well as most unwanted elements (e.., Fe, Cu, Ni, V) + solidify on top of the lead button. This straightforward pro- ‘cedure concentrates Pt, Pd, and An from a large sample and achieves separation from the interfering matrix for the subse- quent instrumental procedures. Variations in flux composition are required for different sample types. Ifthe sample has not been properly fuxed there are telleale signs (button size not correct, flux has etched the ‘crucible, fux boiled over, lead shot presene in slag, slag too vis- ‘cous, ete.) The experienced fire assayer can usualy identify these signs and adjust the Bux accordingly. Analytical recovery viay cauve arg to eras. slg may hid A Pt Ps. ‘ery foro, dct, may rest in low P,P Feeney. Contaminates Pb bution, resus in fow PY, PA recovery, ay form Cu, Matt or dope wich wil scaerge PSE, Cu Fie Asay odicnonRegueod (Se a eve bor ese Anal a by (tat tog sr ato acy bre hoge 00, sce Rar ao, es8H eeashsan ihe Ges. ‘cous slag. Nay cause fag to emus. stra bora, increase fsa rato. ‘uiphide Contamaton of Pb Moy case sipide droplets nag or Potassium nate (ater edad 38 an oxcier and ex Warge 2 suphce matte 10 frm wih wl scavenge PGE. 0 ease a, oat ean ae ‘moet sue, ee Wels cupel. Au and PP may be absorbed into cup. Look ei cor ngs cpl. Seren of ae seat Hy et a 0997), will suffer this is not done properly. Assembly line proce- lures, with up 10 seven fusion potirings at one time, may not allow the Bre assayer to properly view the viscosicy of each jpored fusion and should be avoided unless toal PGE recovery lenot a requirement or you are dealing with same sample rhate- Fal, Unknown mineral/macrix samples fiom prospects or dif- fering sources should be custom-fluxed to produce the best analytical resus. "The lead burton, weighing 25 g to 30 g, conains the Ax, [e:Pd, and Ag. For the subsequent proceduces, the lead muse be liminared in a process called cupellation by placing the lead button into a cupel made of bone ash or magnesia and heating £0.900°C to 960"C. The lead is primarily absorbed (95% «) ineo the cupel leaving a 3 mg to 5 mag Ag bead (prill or doré) con- taining the Au, Pt, and Pd as well as some, but not all, of the Rh and Le The expertise of the fre assayer is required to detezmine temperatures of fusion and cupellation and rates of heating based on the composition of the sample. PGE and gold may be associated with Se or Te which at levels of 10's of ppm can wet the surface of the cupel and cause the PGE and Au to be sub- stantially lost to the cupel by absorption. There are usually tell- tale signs that this has occurred with distinctive color sings showing on the cupel. Occasionally, additional seeps (<.g., scot fication} are required to creat impute Pb buttons containing Te otherwise, low recoveries will occu. ‘At chs poine, the Au, Pr, and Pd have been effectively sep- arated from the inteeferents. Van Wyk and Dixon (1983) dever- ‘mined that unless a ceducing atmosphere in the crucible is maintained, and the ratio of PbO to reducing agent is 9 «0 12, recoveries of Pt, Pd are low. Their recovery results on SARM-7 (Merensky Reef) and UG-2 (upper chromicite, Bushveld Com- plex) increased by 59% and 16% respectively by changing the ‘tio ro the 9 to 12 target. Diamantaros (1977, 1986), and Hafity et al. (197) also describe other potential ineerfering ele- ments such as Cz, Cu, Ni, and S. Although the separation tech- nique is fairly simple, more sl is required than for simple gold analysis. The mode of occurrence of the PGE is usually in association with high chromite, MgO, Cu, and Ni mineral, selenides, tellurides and/or sulphides. This requires adjustments 1 the flux as described in Table 2, because one fie assay flux does not fic all sypes of samples. The Ag doré bead method presents a major problem for Rh and Ir analysis. Rhodium and Ir may not be dissolved complezely in the bead as it may have formed insoluble alloys. ‘A.solution to this problem is to use Au as a collector instead of Ag, Generally, it is very useful to provide the laborarory with sample descripsions so thar the samples can be fluxed properly the first time, particularly if the amount of sample material available'is at premium, Although the lead fire assay fusion is reducing, it is more commonly regarded as an oxidizing fire assay as a result of the cupellation stage, where the lead burton is oxidized, This results in complete loss of Os as the volatile and potentially lethal (if in a high enough concentration) osraium tetraoxide. Table 3 shows 2 comparison of detection limits for Au, Pt, and Pd from the price lists of a number of commercial laboratories. These detection limits will vary with sample size used, dilution and type of instrument. Table 4 shows results for Aa, Pr, and Pd on chromite (18.5% Cr,0;) submixed co a commercial laboratory without indicating the natute of mate- tial. When improperly floxed the results shown a8 “orig” were obtained. The sample was then properly fuxed with the results labeled repeat obrained. Analytical Finishes for the Lead Fire Assay Collection ‘There are three main instrumental techniques used co ana~ Iyze the resulting agua regia solution produced from the disso- lution of the doré bead: atomic absorption spectromerry (AAS), ICPIOES, and ICP/MS. AAS has the potest sensitivity (unless a graphite faumace is used) for Pe and Pd, and ehe most inter- clement interferences; ICP/OES has invetmediate sensitivity; and ICP/MS has the best sensitivity. ICP/MS also has the high- est capital cose and sill requirements, and AAS the lowest. 'AAS and graphite furnace aromic absorption spectrometry (GFAAS) have been described by Senepfe and Grimaldi (1969), Fryer and Kertich (1978), Aruscavage et al. (1984) and Koneas et a. (1986). Their procedure involves (parting) dissolving the ‘Ag doré bead in HNO, acid and then dissolving the gold flake ‘containing Pr, Pd with aqua repia. The AAS technique does not Ihave the required detection limits for low level Pt, Pd determi- nations, GEAAS is bertes, but the technique is relatively slow and shows no advantages over ICP/OES or ICP/MS determina- tions, GFAAS also has 2 relatively small linear dynamic range (ive. the range of concenteations over which instrument response is linea) requiring more frequenc dilution.‘Table 3. Comparison of the detection limits for Pt, Pd, Rhy and Au ‘on Ag and Au doré beads from the lead fire assay fusion (all ppb) ++ The Geology, Geochemistry, Mineralogy and Mineral Beneficiation of PGE 22 .-....0-0+- secs eesssere ‘Table 4. Comparison of chromite in triplicate analyzed by inade- ‘quate fluxing of sample Pal Pr Rh ma FaaAS 10 50 = 3 FAGrAAS, 05 1 - 2 Face 4 5 5 2 FaDoP 1 1» 2 FAICE/ANS oa, OL ost 2 "determined A or ead “he dtecton its de tore assay reagent ak inttns and can vary consi ably depending onthe laboratn, couce ol eagets a crtarmation we Note: Ostecton its yxy wih ana instrument abcelay, mete The ‘ove comolaton was de using nue of camera sboraones pulses ree ts Other less common methods of analysis involving the Pb fire asay collection include radiochemical separations of Pt and. Pe after irradiating the sample in a nuclear reactor. The sample then undergoes the Pb fixe assay silver cupellaion and the resule- ant Ag bead is measured by INAA. This method has been described by Taskstra et al. (1970), and Rowe and Simon (1971), Caughlin (1989) has desctibed a method for the analy- sis ofthe doré bead by atomic fuorescence spectroscopy. ICPIOES or DCP/OES (DCP is mostly obsolece) are much faster than AAS or GFAAS and have 2 much better sensitivity for Pr than che AAS. Wemyss and Scott (1978), Watson et al (1983) and Date ct a. (1987), have described the analysis of the PGE by ICP/OES analysis. With this technique (rapid and vir tually interference fice) the doté bead, usually from a 30 g fice assay charge, is dissolved in agua regia and diluted for analysis. Dilution amounts vary with the laboratory and will ease detec- tion limits co vary with the laboratory as well. The solution is then aspirated into an argon plasma at 8000° Kelvin. The emis- sion spectra are measured and quantified. Analytical finishes also include ICP/MS. Gregoire (1988) cscribed this technique using a third generation ICP/MS coupled to a graphice furnace. Currenc ICP/MS technology is an order of magnitude more sensitive, negating the need for a sgeaphite Furnace. With this technique, the sample solution is also aspirated into an argon plasma at about 8000° Kelvin. The elements in solution are ionized and ate introduced into a mass spectrometer kept under high vacuum. The massce of the ele- ‘ents ate measured and quantified by che detector associated with the mass spectrometer. Typically, ic is possible to obtain and Dabvch deposits in gabbroic rocks fom North America. bv The Geology, Geochemistry, Mineralogy and (Mine) Benefiiaion of Pioum-Group Fletens. Edd LJ. Cabri, Canadian Institute of Mining, Metlhungy and Petroleum, this volume, p- 299.319. WATSON, AE,, RUSSELL, GM., MIDDLETON, H.R. and DAVEN- PORT, RE, 1983, The application of the inductively coupled plasms sytem co the simulancous determination of precious met als, Mink, M88, 10 p. WEMYSS, RB, and SCOTT, REL, 1978, Simultaneous decerminacion of platinum group meals and god, in ores and related plane maria by inducivelycaupled plesma-opialemision pecromet:y. Ana pica Chemise. 50, p. 1694-1697. ZIENTEK, M.L, COOPER, RW, CORSON, SR. and GERAGHTY, E.R, 2002, Platinum group element mineralization in the Silwarer Complex, Montana. in The Geology, Geochemistry, Min: and Minceal Beefication of Platin-Group Elements. Eited LJ. Cabei. Canadian Intute of Mining, Meslay and Pero: team, this volume, . 459-481. oie