0039-9140/88 $3.00 + 0.00 Pergamon Press plc
Tdanta,Vol. 35,No. 4,pp. 297-300, 1988 Printed in Great Britain
DETERMINATION OF ARSENIC IN ORES, CONCENTRATES AND RELATED MATERIALS BY CONTINUOUS HYDRIDE-GENERATION ATOMIC-ABSO~TION SPECTROMETRY AFTER SEPARATION BY XANTHATE EXTRACTION ELSE+M. DONALDSON and MAUREEN E. LEAVER Mineral sciences Laboratories, Canada Centre for Mineral and Energy Technology, Department of Energy, Mines and Resources, Ottawa, Canada (Received 23 September
1987. Accepted
29 October
1987)
Summary-A recent graphite-furnace atomic-absorption method for determining -0.2 pg/g or more of arsenic in ores, concentrates, rocks, soils and sediments, after separation from matrix elements by cyclohexane extraction of arsenic(II1) xanthate from u 8-1OM hydrochloric acid, has been modified to include an alternative hydride-generation atomic-absorption finish. After the extract has been washed with IOM hydr~hloric acid-2% thiourea solution to remove co-extracted copper and residual iron, arsenic(III) in the extract is oxidized to arsenic(V) with bromine solution in carbon tetrachloride and stripped into water. Following the removal of bromine by evaporation of the solution, arsenic is reduced to arsenic(II1) with potassium iodide in -4M hydrochloric acid and ultimately determined by hydride-generation atomicabsorption spectrometry at 193.7 nm, with sodium borohydride as reductant. Interference from gold, platinum and palladium, which are partly co-extracted as xanthates under the proposed conditions, is eliminated by complexing them with thio~mi~rba~de before the iodide reduction step. The detection limit for ores and related materials is _ 0.1 peg of arsenic per g. Results obtained by this method are compared with those obtained previously by the graphite-furnace method.
The accurate
determination
of arsenic
at trace and
pg/g-levels in ores, concentrates, rocks, soils, processing products and related materials is important in many CANMET projects, including the Canadian Certified Reference Materials Project (CCRMP), and recently in this laboratory a relatively rapid graphitefurnace atomic-absorption spectrometric (GFAAS) method was developed for the dete~ination of -0.2 rig/g or more of arsenic in these materials.’ This method involves the preliminary separation of arsenic, after its reduction with titanium(III), from iron, lead, zinc, copper and other matrix elements by cyclohexane extraction of arsenic(II1) xanthate from -8-10M hydrochloric acid. After several washing steps to remove co-extracted copper, residual iron and chloride, arsenic is stripped from the extract with 16M nitric acid and ultimately determined by GFAAS in a 2% nitric acid-0.1% thiourea medium with palladium as matrix modifier. It was considered that, with an alternative stripping step involving bromine oxidation of arsenic(II1) in the extract followed by back-extraction into water, as described in an earlier method for arsenic based on xanthate extraction and a spectrophotometric molybdenum blue finish,2 this method could easily be modified to include a hydride-generation atomic-absorption (HGAAS) finish which would be very useful for
routine work. Results obtained for various reference materials by this modified method are compared with those obtained previously by the GFAAS method. An advantage of the proposed method for arsenic over many other HGAAS methods is that arsenic is separated from other interfering hydride-forming elements, oiz. lead, tin, bismuth, antimony, selenium and tellu~um. EXPERIMENTAL Apparatus
A Varian-Techtron model AA6 spectrometer, equipped with a IO-cm laminar-flow air-acetylene burner and an arsenic hollow-cathode lamp, was used in this work. It was interfaced with a Varian model VGA-76 automated vapourgeneration accessory, which features a continuous-flow design incorporating a peristaltic pump and is coupled with a flame-heated silica absorption cell. With this system, the sample solution, hydrochloric acid and sodium borohydride solution are pumped through a reaction or mixing coil and the arsine and other gaseous reaction products formed are then swept by means of a stream of nitrogen into a gasliquid separator to remove liquid and water vapour. From there they are swept into the silica cell-heated by an airacetylene flame-where the arsine is thermally decomposed and where, after a suitable delay period to allow the production of a steady-state atomic-absorption signal, the absorbance is integrated for a suitable time. A schematic diagram and a detailed description of the development of this accessory can be found elsewhere.3 The spectrometer and vapour generator were used under the following conditions. Wavelength: 193.7 nm Lamp current: 8 mA Spectral band-width: 0.5 nm
Crown Copyright reserved. 297
298
ELSIE M. DONALDKIN
Air flowmeter reading: 6.5 (_ I1 l./min) Acetylene flowmeter reading: 3.0 (-2 l./min) Integration period for absorbance measurement: 3 set after -40 set delay time Hydrochloric acid (10M) uptake rate: -. 1 ml/min Sodium borohydride solution uptake rate: _ 1 ml/min Sample solution uptake rate: -8 ml/min Nitrogen flow-rate: -90 ml/min Reagents Sodium borohydride, 0.4% solution containing - 0.5% sodium hydroxide. Store in a plastic bottle. Prepare a fresh solution every two days. Thiosemicarbazide solution, 2.5%. Dissolve 2.5 g of the reagent by heating with 100 ml of water containing 2 ml of concentrated hydrochloric acid. Cool and filter the solution into a plastic bottle. Prepare a fresh solution every three days. Potassium iodide (IO%bscorbic acid (1%) solution. Store in a plastic bottle. Hydrochloric acid, 1OM. Store in a plastic bottle. Doubly demineralized water was used throughout and lOO- and I-fig/ml arsenic solutions and all other solutions used in the extraction step were prepared as described previously.’ Calibration solutions Prepare 0.005, O.Ol-, 0.015 and 0.02~pg/ml arsenic solutions by adding 0.5, 1, 1.5 and 2 ml, respectively, of I-pg/ml standard arsenic solution to lOO-ml standard flasks. Dilute each solution to _ I5 ml with water, then add 10 ml of concentrated hydrochloric acid and 4 ml of 2.5% thiosemicarbazide solution and mix thoroughly. Add 1 ml of 10% potassium iodide-l % ascorbic acid solution, mix and allow the solutions to stand for -45 min to ensure the complete reduction of arsenic (Note 1). Add 10 ml more of concentrated hydrochloric acid to each solution, then dilute to volume with water. Prepare a zero calibration solution in a similar manner (Note 2). Procedure After sample decomposition (including a blank) and the extraction of up to N 1 pg of arsenic, contained in a suitable aliquot of the initial 10M hydrochloric acid sample solution, as described previously,’ wash the extract once (or twice if necessary) with 5 ml of 1OM hydrochloric acid-2% thiourea solution, as described, to remove co-extracted copper. Drain off the acid phase, then add 3 ml of 20% bromine solution in carbon tetrachloride to the extract, stopper the funnel and mix thoroughly. Allow the solution to stand for -5 min to ensure the complete oxidation of arsenic(III) to arsenic(V), then add 5 ml of water and shake the funnel for -30 sec. Allow the layers to separate, then drain the aqueous phase into a loo-ml Teflon beaker. Wash the stem of the funnel with water and collect the washings in the beaker. Repeat the stripping step twice more, washing the stem of the funnel with water each time. Evaporate the resulting blank and sample solutions to _ 10 ml (Notes 3 and 4), then dilute them to _ 15 ml with water, add 10 ml of concentrated hydrochloric acid and proceed with the reduction of arsenic as described above, using half the volumes of thiosemicarbazide and potassium iodide solutions recommended for the calibration solutions. After the 45-min reduction period, transfer the resulting solutions, without further addition of concentrated hydrochloric acid, to 50-ml standard flasks, then dilute them to volume with water and mix. Using the zero calibration solution to zero the spectrometer, measure the absorbance generated by the calibration solutions, followed by the blank and sample solutions (Note 5) under the conditions described under “Apparatus”. Determine the arsenic concentration of both solutions either by reference to a graph plotted from the absorbance values obtained for the calibration solutions or calculate the
and
MAUREEN E. LEAVER
concentrations from the values obtained for the calibration solutions that bracket the sample and blank concentrations (Note 6). Calculate the arsenic content of both solutions in
ng and correct the result obtained for the sample solution by subtracting that obtained for the blank solution. Notes 1. A standard arsenic solution prepared by dissolving arsenic(II1) oxide in sodium hydroxide solution as recommended previously’ may contain some arsenic(V) if it has stood for a considerable length of time, because arsenite is slowly oxidized by air to arsenate in an alkaline medium.4 However, this will not cause problems in tests involving arsenic(III) as long as the test solution is taken through the reduction step. 2. Although the calibration solutions become deep yellow on standing they are still stable for at least five days. 3. At this point the solution should be colourless, indicating that all the bromine has been removed. If some remains it will interfere with the reduction of arsenic(V) to arsenic(II1) by iodide. This will produce a low result for arsenic. 4. For samples of low arsenic content, a 25-ml final sample solution volume can be used, if desired, instead of 50 ml as recommended in the procedure. However, in this case the solution should be evaporated to -7 or 8 ml, followed by the addition of 5 ml of concentrated hydrochloric acid, 1 ml of thiosemicarbazide solution and 0.25 ml of potassium iodide-ascorbic acid solution for the reduction step. For samples of high arsenic content, a RIO-ml final volume can be used under the conditions described for the calibration solutions. 5. If dilution is necessary, dilute suitable aliquots of both the blank and sample solutions with freshly prepared zero calibration solution and correct the result (ng of arsenic) obtained for the diluted sample solution by subtracting that obtained for the diluted blank solution. 6. Although the calibration graph is not linear it was found that more accurate results, with minimal positive error, were obtained for arsenic if they were calculated on the basis of linearity between two calibration solutions that closely bracket the sample concentration. Much greater errors were obtained when calibration curves were used. This problem should not occur with the newer microprocessorcontrolled atomic-absorption spectrometers which feature direct concentration read-out with curve linearization. RESULTS AND DISCUSSION
Stripping and HGAAS
determination
of arsenic
Previously’ arsenic(II1) xanthate in the cyclohexane extract was oxidized with, and stripped into, 16M nitric acid before the ultimate determination of arsenic by GFAAS. However, treatment of the extract with bromine solution in carbon tetrachloride, followed by stripping of the arsenic into water as described in earlier work,* was considered advantageous for hydride-generation work because the residual bromine in the strip solution can be more quickly and efficiently removed by heating and evaporation of the solution than can nitric acid, which requires fuming of the solution with sulphuric or perchloric acid. Reduction of the resultant arsenic(V) to the tervalent state required for the hydride-generation step is readily accomplished by making the solution -4M in hydrochloric acid, adding potassium iodide and allowing -45 min for complete reduction.5s6 A potassium iodide concentration of -0.4% during the
~te~~nation reduction step was found to be adequate. As recommended by other investigators,3*6 using the same hydride-generation accessory, 10M hydrochloric acid and 0.6% sodium borohydride solution, which was confirmed to be the optimum concent~tion, were used in the acid and borohydride channels of the hydride-generator. Under these conditions, the final hydrochloric acid concentration of the sample solution, after the iodide reduction and dilution of the solution to volume with water, can vary from -0.5 to at least 1OM without affecting the sensitivity for arsenic. At the pumping rates specified, this corresponds to hydrochloric acid concentrations in the range N 1.4-9M during the arsine generation step. Other workers7-9 have also found that hydrochloric acid concentrations > 1M have no significant effect on the reaction. For convenience a final hydrochloric acid con~ntration of 2M was chosen for the sample and calibration solutions in this work. Effect of diverse ions As found previously,’ platinum is slightly coextracted and palladium and gold are largely coextracted as xanthate complexes under the conditions, -g-lOM hydrochloric acid, used for the extraction of arsenic. Copper, which is also partly co-extracted, is largely removed from the extract by washing it with 10M hydrochloric acid containing thiourea. Although easily reducible elements such as these and
Table I. Determination
of arsenic
299
the other platinum-group elements are known to interfere severely in the determination of arsenic by hydride-generation,” interference from these elements can be eliminated or minimized by complexing them with ~ios~i~r~de.5,9*” Tests showed that, when 2ml of 2.5% ~io~rni~rba~de solution were added before the iodide reduction step, up to at least 0.2 mg of gold, platinum and palladium and 2 mg of copper can be present individually in the final solution (50 ml), or half these amounts of these elements can be present wllectively, without causing significant error in the result. However, as mentioned previously,’ these amounts of gold, platinum and palladium are considerably more than would be expected in the final sample solution on the basis of the largest sample taken and the largest aliquot used for extraction. Similarly, considerably less than 1 mg of copper wouid be present after the extract is washed as described above. Low results were obtained for arsenic in the presence of palladium and copper when the thiosemicarbaxide solution was added after the potassium iodide solution. Under these conditions palladium and copper form iodide complexes.
Table 1 shows that the mean results obtained for arsenic in various CCRMP reference materials and in several National Research Council Canada and National Bureau of Standards marine and river
of arsenic in CCRMP and other reference ores, concentrates and related materials
Sample* CCU-1 Copper concentrate CCU-la Copper concentrate RL-I Uranium ore UTS-4 Uranium tailings SY-2 Syenite rock SY-3 Syenite rock MRG-1 Gabbro SO-l Regosolic soil SO-2 Podzolic soil SO-3 Calcareous C Horizon soil SO-4 Chemozemic A Horizon soil FER-I Iron formation rock FER-2 Iron formation rock FER3 Iron formation rock FER4 Iron formation rock NRCC MESS-l Marine sediment NRCC BCSS-1 Marine sediment NBS 1645’River sediment NBS 1646 Estuarine sediment
Certified value and 95% confidence limits, As, Mglg 41.0 + 4.0 53 + 5# 19.6 & 1.1 38.0 f 2.0 177 19ll 0.q 1.9 + 0.3” 1.2 f 0.2” 2.6 f O.l* 7.1 & 0.7” 5.9s l.8b l.lb 3.4b 10.6 f 1.2 11.1 + 1.4 66c 11.6+ 1.3
As found, fig/g GFAAS method? 44.1 f 0.8 48.7 f 0.7 # 19.8 f 2.2 39.6 f 2.6 17.1 f 1.3 17.9 * 1.1 0.5, f 0.0, 2.1 f0.3 1.1 *to.1 2.5 j: 0.2 6.3 + 0.2 5.9 & 0.4 1.6 f 0.0 0.8, f 0.0,
2.9 + 10.1 h 10.5 f 64.2 f 10.4 +
0.1 0.8 0.4 5.9 0.7
Proposed methodS 44.1 + 3.0 (5) 48.6 + 1.4 (5) # 21.5f0.4(4) 40.5 + 1.8 (5) 16.1 f 0.9 17.9 f 1.7 0.7, & 0.0, 2.1 kO.2 1.2 f 0.2 2.4 + 0.3 6.7 + 0.6 6.0 + 0.5 (4) 1.9 + 0.2 1.1 f 0.1 3.4 + 0.4 (4) 9.8 i 0.7 10.2 & 0.5 69.9 + 3.4 (6) 10.2 + 0.4
*CCRMP reference materials except where indicated otherwise. Except for CCU-la nominal compositions are given in the previous work.’ The approximate percentage chemical composition of CCU-la is -27 Cu, -27 Fe, _ 3 Zn and -35 s. ?Mean values and standard deviations obtained previously.’ $Mean and standard deviation for 3 values except where indicated otherwise in parentheses. ~Consensus mean value (excluding gross outliers) obtained during the interla~rato~ certitication programme. # Mean of 5 results obtained by the authors during the interlaboratory certification programme. l/Most recent usable value.i2 “CCRMP value given for information only (not certified). bMean value obtained during the interlaboratory programme.” Tentative value.
300
ELSIEM. DONALDWNand MAUREENE. LEAVER
sediments are, in most cases, in good agreement with the certified values, with values given for information only or with the consensus mean or tentative values obtained during interlaboratory certification programmes. Except for NBS 1645, they are also in excellent agreement with the results obtained previously by the GFAAS method and with other reported values.’ Although the mean value obtained for NBS 1645 is slightly higher than the GFAAS value, it is still within the wide range of values (61-81 pg/g) reported by other workers.’ The mean result obtained for CCU-la, and which is at present undergoing certification by the CCRMP, which was not included in the previous work,’ is in excellent agreement with the mean GFAAS value obtained in this laboratory for certification purposes, but slightly lower than the current consensus mean value. In the present work, each of the individual results obtained for the reference materials was the mean of 3 or 4 HGAAS runs involving duplicate measurements each time. The precision for arsenic by the proposed HGAAS method (Table 1) is comparable with that obtained by the GFAAS method. However, the practical detection limit, calculated as three times the amount of arsenic equivalent to the standard deviation of the reagent blank, based on a 50-ml aliquot (0.5 g) of sample solution taken through the extraction step, is -0.1 pg of arsenic per g. This is about half the limit
found for the GFAAS method.’ The sensitivity or characteristic concentration is N 0.16 ng of arsenic per ml for 0.0044 absorbance. The reagent blank varied from -5 to 90 ng, depending on the size of the aliquot taken for extraction. This is comparable with the blanks obtained with the GFAAS method.’ REFERENCES 1. E. M. Donaldson,
Tulunru, 1988, 35, 47.
2. Idem, ibid., 1971, 24, 105. 3. B. T. Sturman, Appl. Spectrosc.,
1985, 39, 48.
4. I. M. Kolthoff and E. B. Sandell, Textbook of Quuntitative Inorganic Analysis, 2nd Ed., Q. 591. Macmillan, New York, 1948. 5. S. P. Kellerman, MINTEK Report NO. M39, Council for Mineral Technology,
Randburg, South Africa,
1982.
6. K. Brodie, B. Frary, B. Stunnan and L. Voth. Variun Instruments
I. L. Ebdon, 8. 9. 10. 11. 12. 13.
at Work, 1983, No.
J. R. Wilkinson
AA-38.
and K. W. Jackson, Anal. Chim. Actu, 1982, 136, 191. E. de Oliveira, J. W. McLaren and S. S. Berman, Anal. Chem.. 1983. 55. 2047. R. K. Anderson,’ MI Thompson and E. Culbard, Analyst, 1986, 111, 1143. L. M. Voth-Beach and D. E. Shrader, Spectroscopy, 1985, 1, 60. G. F. Kirkbright and M. Taddia, Anal. Chim. Acra, 1978, 100, 145. S. Abbey and E. S. Gladnev.- Geostds. Newsl.. 1986., 10, 3. S. Abbey, C. R. McLeod and W. Liang-Guo, Geol. Surv. Canada Paper 83-19, 1983.