ANALYST, NOVEMBER 1986, VOL. 111 1255 Inductively Coupled Plasma Emission Spectrometric Determination of Boron and Other 0x0-anion Forming Elements in Geological - Materials* Gwendy E. M. Hall and Jean-Claude Pelchat Geological Survey of Canada, 60 1 Booth Street, Ottawa, KIA OE8, Canada A simple, rapid method is described for the determination of boron in different types of geological samples. Measurement is made by inductively coupled plasma emission spectrometry on an aqueous leach of a sodium carbonate - nitrate melt, which effectively separates the analyte from potentially interfering major elements such as iron, calcium and magnesium. A determination limit of 1 pg g-1 of B is achieved with a precision of about 3% (relative standard deviation) at levels greater than 5 pg 9-1 of B.Application of this method to include the determination of vanadium down to 4 pg g-l, molybdenum down to 2 pg g-1 and tungsten down to 5 pg g-1 is demonstrated by the use of 18 international reference standards. Extension of this method to include the determination of chromium and phosphorus is cautioned, as success is dependent on the sample type. Keywords: Boron determination; 0x0-anion forming elements; geological materials; inductive1 y coupled plasma emission spectrometry Boron is an important trace constituent of the hydrothermal fluids from which several types of mineral deposit are precipitated and it is frequently found in anomalous amounts in the host rocks of these deposits. However, the use of this element as an indicator of mineralisation has been handicapped by the absence of effective analytical methods for geological materials.The boron content of geological materials has been deter- mined almost exclusively in the past by one of two techniques: emission spectrography using a d.c. arc1 and spectropho- tometry based on the formation of the boron - curcumin complex.2 Recently, Baucells et al.3 have improved on the emission spectrographic method to achieve a minimum detectable concentration of 11 pg g-1 of B and Troll and Sauerer4 have combined extraction of boron by ethylhexane- 1,3-diol (after carbonate fusion) with spectrophotometry of the carminic acid complex to determine boron in the range 1-1000 pg g-1. The determination of boron in rhyolites and tektites in the range 10-200 pg g-1 has been described by Kluger and KoeberF using the tetrafluoroborate-selective electrode at 1-2 "C (to prevent hydrolysis) after dissolution with hydrofluoric acid and buffering with hexamethylene- tetramine.However, this method, together with those based on spectrophotometry, has not been applied routinely to the analysis of many types of geological samples, largely owing to their lack of simplicity and universality. Higgins6 determined boron in 18 international reference standards by prompt- gamma neutron-activation analysis and obtained good agree- ment with literature values at high levels, but the correlation was poor at low abundances, probably owing to uncertainty in the sodium correction factor. In 1982, Owens et al.7 published the first paper to describe a method to determine boron specifically in geological refer- ence materials by inductively coupled plasma emission spec- trometry (ICP-ES).A carbonate fusion was carried out and the solution was neutralised and measured against similarly prepared synthetic standards using a Perkin-Elmer ICP/5000 spectrometer. A detection limit of 5 pg g-1 was obtained. Borsier and Garcia8 later described an automated ICP method to determine major and trace elements, including boron, in geological samples after a sodium peroxide fusion. The detection limit and precision were poor. In 1984, Din9 * Geological Survey of Canada Contribution No. 20786. described an elaborate procedure to separate boron from potentially interfering iron by successive fusions with potas- sium dihydrogen phosphate and potassium hydroxide prior to ICP emission measurement at 249.68 nm.Thompson and WalshlO commented on the lack of exploitation of ICP emission gectrometry for boron by analysts in geochemistry. Subsequently, Walshll described a method to determine boron in rocks by ICP-ES on a solution obtained by aqueous leaching of a potassium carbonate fused melt from which the bulk of the potassium had been precipitated using perchloric acid. I t is this approach that has been investigated here, modified and extended to determine other elements in addition to boron. A strong attack is required for boron-bearing minerals, such as tourmaline, so fusions with alkali metal carbonates, alone and with oxidants, were examined. Potassium rather than sodium carbonate was thought to be desirable because the potassium could be precipitated later as its perchlorate salt, whereas sodium carbonate has the advantage of a lower melting point (853 vs.903 "C) and is not hygroscopic. The temperature and time of fusion were both optimised once the preferred decomposition was chosen. As the major portion of analytical time is often spent on the decomposition stage, other elements of interest that would also be brought into solution were determined, viz., molybdenum, vanadium, chromium, tungsten and phosphorus, which form oxy-anions after fusion. A thorough study was made of all emission lines of suitable sensitivity for each element, checking for spectral interferences by scanning a 0.08 nm window centred on each line while aspirating high concentrations of likely interferents. The spectral line atlas compiled by Brenner and Eldad12 was found to be useful in planning this interference study.Experimental Reagents De-ionised, distilled water was used throughout. Carbonates and nitrates were of Baker analysed reagent grade. Individual standard stock solutions were obtained from Spex Industries, Metuchen, NJ, USA. All linear polyethylene laboratory ware was washed in 30% nitric acid and repeatedly rinsed with de-ionised, distilled water.1256 ANALYST, NOVEMBER 1986, VOL. 111 Instrumentation A Jobin-Yvon (ISA, Metuchen, NJ, USA) Model 38 high- resolution sequential ICP emission spectrometer was used. Details of the instrumentation and operating conditions used are given in Tables 1 and 2, respectively.The path from the viewing zone of the argon plasma to the entrance of the spectrometer was flushed with nitrogen, as was the spec- trometer itself, allowing emission measurement in the low ultraviolet region of the spectrum. Both Meinhard and GMK nebulisers were tested. The sample gas flow was regulated with a mass flow controller for better precision, and a humidifier was used with the nebuliser to improve high salt tolerance. Procedure Preliminary investigation of sample decomposition methods Prior to deciding the final flux mixture, various fusions and sinters were investigated. International reference materials were used to test the accuracy and precision of these decompositions. First, potassium carbonate alone and a 5 : 1 mixture of K2C03 and KN03 were tested in a muffle furnace at 950-1050 "C.Potassium, rather than sodium, as the cation initially appeared beneficial because perchloric acid could be added after the dissolution, resulting in precipitation of insoluble potassium perchlorate. It was thought that this would drastically decrease the dissolved salt content of the analyte solution at the ICP and hence reduce the problem of nebuliser clogging and drift. However, a salt build-up was seen to occur at the top of the torch and clogging was still evident with both the Meinhard and GMK nebulisers. It was found that platinum crucibles were seriously attacked by the flux, especially in the presence of a sample containing a high concentration of Fe". Nickel crucibles were used instead, but occasional creeping up and over the walls occurred.Decom- position by sintering over a flame (heating below the melting-point of the flux) was attempted, followed by addition of HC104. Low results were obtained for Mo, V, Cr and W; these were attributed to incomplete attack. Table 1. Instrumentation R.f. generator . . . . 2.5 kW, frequency 27.12 MHz Monochromator . , . . 1-m Czerny-Turner, nitrogen-purged for Grating . . . . . . Holographic3600groovesmm-~, Torch . . . . . . . . Jobin-Yvon demountable quartz with Nebuliser . . . , . , Meinhard concentric, Type C high salt low UV wavelengths resolution 0.01 nm first order argon sheathing gas (TR-30-C2) with Scott-type spray chamber Pump . . . . . . . . Gilson, peristaltic, Model Minipuls I1 Computer . . . . . . Columbia Commander Micro, 96 kbyte memory; dual floppy disc; software by Jobin-Yvon Table 2.Operating conditions R.f. forward power . . Plasma gas flow-rate . . Auxiliary gas flow-rate . . Sample gas flow-rate . . Nebuliser pressure . . Sample uptake rate . . Slit widths: Entrance . . . , . . Exit . . . . . . Height . , . . . . Photomultiplier tube . . Observationzone . . 1.05 kW (<lo W reflected) 13 1 min-' 0.15lmin-I 0.6 1 min-1 38 Ib in-2 1.35 ml min-I 40 pm 40 pm 7 mm 1200 v 12 mm above initial radiation zone An alternative flux consisting of Na2C03 and NaN03 (oxidant) was tested, and, as reduction of the Na content of the analyte solution was not possible via precipitation, the salt tolerance level of the ICP for alkaline sodium solutions was established. It was found that a 4% Na2C03 solution could be nebulised for hours without clogging.Fig. 1 depicts the variation in the signal intensity of each analyte standard with time; each point represents the mean of three 0.5s integration readings. The relative standard deviation (RSD) ranges from 1.6% for B to 2.9% for Mo, indicating no deleterious effect of the high salt concentration. Recommended sample decomposition A 200-mg sample, ground to at least - 100 mesh, was placed in a 25-ml nickel crucible and thoroughly mixed with 1 g of the flux, a 5 : 1 mixture of Na2C03 and NaN03. The crucible was placed in a muffle furnace and the temperature raised to 900 "C and maintained there for 15 min following the melting stage. After cooling, the contents of the crucible were leached with water and warmed on a hot-plate, stirring to ease dissolution.The contents (ca. 20 ml) were transferred into a 50-ml polyethylene test-tube and warmed for several hours. After cooling, the solution was centrifuged and the clear supernatant liquid transferred into a 25-ml polyethylene calibrated flask and diluted to volume with water. The pH of this solution was 12. Calibration Blanks and calibration standards were taken through a procedure similar to the decomposition using Eppendorf micropipettes to deliver accurate amounts of standard stock solutions directly on to the flux itself. The results were compared with those for calibration standards made from the background matrix of "fused blank" solution. Analyte spectral wavelengths were chosen to be the most sensitive available, with the least amount of interferences likely to be encountered with typical sample material.The positions for background correction measurement were deter- mined by studying spectral line atlases and by scanning the 202.03 nrn 2.93% 2.14% ;I---- 1 pg ml-1 E 207.91 nrn > c .- w I I .- .- 249.77 nrn 1.59% 1 pg ml-l .- Y 2.00% 1 pg ml-l 2 Cr II 267.71 nrn 31 1.83 nrn I .awO 15 30 45 60 75 Time/mi n Fig. 1. Variation of emission signal with timeANALYST, NOVEMBER 1986, VOL. 111 1257 analyte lines while nebulising samples with high contents of potentially interfering elements. These wavelengths are shown in Table 3 together with the standard concentrations used for calibration. There is an aluminium recombination interference on the Mo 202.03-nm line, necessitating correc- tion on both sides of the line.Correction coefficients were found to be unnecessary with the chosen wavelengths. An integration time of 500 ms was selected as the optimum for each measurement. Recalibration was carried out after ten samples. Analysis of international standard samples Eighteen reference materials from the US Geological Survey (USGS) , the Canadian Certified Reference Materials Project (CCRMP), Group International de Travail (GIT-IWG) and the Institute of Geophysical and Geochemical Prospecting, China (IGGE) were analysed by the proposed method. The analysis was carried out in two steps: it was found to be more efficient to determine P, Mo and W for the entire suite of samples and then to determine B, Cr and V in a second step because the scanning rate of the JY-38 is slow.With the knowledge that the mineral chromite was unlikely to be attacked efficiently by a single fusion with the carbonate - nitrate mixture, the standard IGS 30 (Institute of Geological Sciences, UK) containing 23.95% of Cr was included to test the recovery at high levels. Analysis of “in-house-” control samples The method described was also applied to a suite of “in-house” control rock standards collected from sediment- hosted Pb - Zn deposits in the Yukon Territory, which had been well characterised by various analytical techniques. 13 These samples contain up to 5% of sulphur, contained in the minerals sphalerite, galena, pyrite (sulphide) and barite (sulphate). As the sulphide was thought to have a deleterious effect on the oxidative, alkaline fusion, the samples were initially ignited at 900 “C for 30 min and the results compared.The international reference standard GSD-6 was included in this test for comparison purposes. Results and Discussion The alkaline carbonate fusion allows separation of oxyanions, such as borate, from major element cations in geological samples, such as calcium and magnesium, and hence aids in minimising possible interferences in ICP-ES measurements. Iron did not pose a problem as a potential spectral interferent as most of the element remains in the carbonate - oxide residue; this is in agreement with Bock,l4 who found less than 0.1% of solubilised iron after a carbonate fusion. GXR-1, which contains about 25% of Fe, yielded a solution for analysis containing only 5 pg ml-1 of Fe.Thus, the lengthy sample preparation procedure described by Din9 proved unnecessary. The recovery of synthetic solutions carried through the fusion procedure was >%YO for all elements and thus it was concluded that there was no significant loss in the sample preparation steps. The solution detection limit for each analyte, defined as that concentration generating a signal equal to three times the standard deviation of the blank signal (ten readings), is shown in Table 4. The corresponding real detection limits in the samples themselves (factor of 125) are degraded by the presence of the analytes, especially boron, in the flux reagents. Here, the determination limit is defined as the instrumental detection limit (30) times the dilution factor multiplied by a constant that differs between elements and accounts for the variability introduced during sample prepara- tion prior to measurement.This constant was not calculated statistically, but was conservatively chosen based on the data. The sensitivity for molybdenum and tungsten is not adequate for many geological samples, as indicated by a comparison of the determination limits and concentration ranges found in rocks15 (Table 4); however, it was thought useful to include these two elements in this “package” for geochemical explora- tion purposes. The determination limit established for molyb- denum is double that obtained by atomic absorption spec- trometry (AAS) on a routine basis in our laboratories and that for tungsten is double that obtained by the spectrophotometric dithiol method.16 The results obtained for four separate determinations (over a period of 3 months) of boron and the other refractory elements in 18 geological reference materials are given in Table 5 together with the consensus literature values.1lJ7-20 The “consensus” values,l7J8 with their statistically based uncertainty estimate, are shown in order to give the reader an appreciation of the quality of data used by compilers.The absence of a question mark in the other citations implies a “recommended” or “usable” value as opposed to a “pro- posed” value. The comparison of results obtained on pre-ignited vs. unignited samples, which were analysed in duplicate, is given in Table 6. The values shown in parentheses are the authors’ “accepted” values13 arrived at by different analytical tech- niques performed at various laboratories.With reference to Tables 5 and 6, the data obtained for each element will be discussed individually. Boron The accuracy obtained by the proposed method (Table 5) appears reasonable, considering the paucity of previous data for establishing recommended values. The results obtained by this method lie within 6% of the recommended values given for GXR-1, GXR-4, SY-3 and GSD-6, the only samples of the entire group for which reliable values are assigned by the compilers. There are too few data for MRG-1 to comment on the accuracy of the mean boron value (1.9 pg g-1) obtained. The literature values cited by Gladney et al. 17 for BCR-1 range from 0.7 to 12 pg g-1 of B, another indication of the difficulties encountered in the determination of boron at these levels.Neither Abbey20 nor Govindarajulg reported boron values for SDC-1. Our method gives values for BCR-1, SDC-1 and SGR-1 that are considerably lower than those given by Walshll by a similar method. The precision ranges from 1.7% RSD for MA-N to 47% RSD near the determination limit for MRG-1 and averages 3.4% above 5 pg 8-1 of B. The method appears to be effective with samples that contain significant Table 3. Spectral lines and calibration standards Background positionlnm Analytical Element wavelengthlnm Low High P . . . . . . 178.225 I - 0.020 Mo . . . . 202.03011 0.016 0.019 W . . . . . . 207.911 I1 - 0.024 B . . . . . . 249.773 I - 0.020 Cr .. . . . . 267.716 I1 - 0.019 V . . . . . . 311.838 I1 - 0.016 Calibration standardslyg ml-I Low Medium High Reagent blank 10 50 Reagent blank 1.0 3.0 Reagent blank 1.0 3.0 Reagent blank 1 .O 3.0 Reagent blank 1.0 3.0 Reagent blank 1.0 3.01258 ANALYST, NOVEMBER 1986, VOL. 111 Table 4. Detection and determination limits and precision data obtained by the proposed method of analysis Detection limit in Element solution*lpg ml-1 B . . . . . . . . 0.002 Mo . . . . . . 0.006 v . . . . . . . . 0.005 Cr . . . . . . 0.004 w . . . . . . 0.015 P . . . . . . . . 0.05 Determination limit in sample?/ 1 .o 2.0 4.0 2.0 5 50 g-l Typical RSD, YO ( n = 4) 2.5 at 32 pg g-1 (QLO-1) 14 at 8 pg g -1 (GSD-6) 2.4 at 90 pg g-l (GXR-4) 6 at 13 pg g-l (GXR-1) 7 at 28 pg g-l (GSD-6) 2.5 at 780 pg g-l (MAG-1) Concentration range in igneous and sedimentary rocks$/ g-l 3-100 0.2-2.6 20-250 4-2980 170-1100 0.1-1.8 * Defined as the concentration of an element in 4% Na2C03 - NaN03 solution that gives an emission signal equal to three times the 1- Based on a 0.2-g sample with 1.0 g of flux, diluted to a final volume of 25 ml.$ Reference 15. standard deviation of the blank. Table 5. Analysis of international reference standards by the proposed method. All values in yg g-l unless noted otherwise Mean f SD (n = 4) Sample B AGV-1 (andesite) . . 6.1 f 0.4 BCR-1 (basalt) . . . . 2.2 f 0.4 BHVO-1 (basalt) . . ~ 1 . 0 MAG-1 (marine mud) . . QLO-1 (quartz latite) . . 32.2 f 0.8 (6.9 f 3.8)* (6 f 4),* (4.8)$ (3)$ 132 f 3 * 130?),§ (130)$ ~- (37?)7§ (32)$ RGM-1 (rhyolite) .. 23.8 f 0.9 SCo-1 (shale) . . . . SDC-1 (mica schist) . . SGR-1 (shale) . . . . STM- 1 (nepheline syenite) . . GXR-1 (jasperoid) . . GXR-4 (copper mill-head) . . GXR-5(soil) . . . . SY-3(syenite) . . . . MRG-l(gabbro) . . FeR-2 (iron formation) MA-N(granite) . . . . GSD-6 (stream sediment) . . * Reference 17. t Reference 19. $ Reference 11. § Reference 20. 7 Reference 18. (31?),§ (23)$ 71.5 f 1.9 (66?),§ (73)$ 10.1 ? 0.7 (14)$ 50.3 f 1.4 (50?),§ (68)$ 4.0 f 0.4 (5)$ 15.1 f 0.6 (15.3)§ 4.0 f 1.1 (4.3)§ 19.8 f 0.4 (20 f 8)7J 104 f 2.8 1.9 f 0.9 (13?)§ 56.8 k 1.1 (61?)1- 17.2 f 0.3 (17?)§ 48.2 f 2.1 (1 lo)§ (5W Mo 2.3 f 0.3 (2.8 f l.O)* <2.0 (2.8 f 1.7)* <2.0 2.3 f 0.3 2.3 f 0.4 (2.6?) § 2.5 f 0.3 (2.3?)§ 2.4 f 0.4 (1.4?)§ <2.0 34.5 k 1.2 (36?)§ 5.0 f 0.3 (5.2)§ 17.8 f 2.2 (15 f 6)fi 311 f 4 (310 f 40)7 30.2 f 1.4 (28 f 6)7 <2.0 (2.5?)§ <2.0 3.5 f 0.5 <2.0 (l?)§ (1?)§ (3711- 8.1 f 1.1 (7.8)§ V 118 k 3 (123 a 12)* 413 t- 6 (404 f 40) * 299 f 5 (320?) § 138 4 3 (140)§ 50.3 f 2.1 11.6 k 0.8 (14?)§ 129 f 3 (135?)§ 94.8 k 3.2 (105?)§ 124 f 2 (125)§ <4.0 81.2 k 2.3 (61) 8 (79 f 9)7 90.1 +_ 2.2 (90 _t 5)fi 57 f 3.9 (66 f 11)7 47.1 ? 1.9 (51)s 511 f 10 (520)§ 30.4 k 2.3 (37?) t <4.0 (4.6?) § 140 f 3 (140)§ Cr 12.4 f 1.1 11.9 5 0.8 (16 f 4)* 328 f 4 (300) § 107 f 3 (105)§ <2.0 (4.2?)§ <2.0 (4?)§ 73.2 5 2.2 (71?)§ 70.5 f 2.3 (66?)0 32.7 f 2.9 (12 f 3)* (33?)§ (4?)§ <2.0 13.2 f 0.8 (13 f 3)7 68.8 f 3.1 (64 f 6)7 108 f 3 (106 f 9)fi 6.2 f 1.2 489 f 6 (450)§ 44.9 f 3.0 <2.0 (lo)§ (47P (3?)§ 192 f 4 (190)§ P 2125 f 14 1547 f 29 (1580 f 150)* (2100 f 200)* 0.118% k 0.004% (0.122%)§ 0.078% 2 0.002% (O.O78% ?) § 0.115% 4 O.OOI% (O.113%)§ 0.015% f O.OOI% 0.100% f 0.003% (0 .O96% ?)O O.O7O% f 0.002% (0.078%) § 0.121% f 0.002% (0.126 Yo ?) § (0.002% ?) § 0.072% f 0.002% (O.O69%)§ 615 f 7 (630 & 80)7 1280 +_ 27 311 f 6 (320 f 40)7 ( 130O)l-l 0.226% f 0.004% (0.23 Yo ) § 0.021% f 0.001% (0.026%) 0 0.116% f 0.002% (O.I16%)’f 0.601% f 0.004% (O.606%)§ O.1O3% 0.002% (0.100%)§ W <5 <5 (0.4?)§ <5 (0.06?)1- <5 <5 <5 (1.6?)§ <5 <5 (0.8?)§ <5 (0.53)t <5 (3.8?)§ 203 _t 6 (210 f 40)7 3 1 f 4 (32 ? 3)fi <5 (1.1 f 0S)Y <5 <5 <5 66 -C 2 (70?)§ 26 f 2 (2515 amounts of sulphur (Table.6) and prior ashing has no effect; the comparison data were obtained from various commercial laboratories using hydroxide fusion and ICP-ES and spectro- photometry.Molybdenum The accuracy, indicated by the results and comparative data in Table 5 , is good. Although the mean value obtained some- times appears significantly different from that cited by Gladney et a1.173 (BCR-1, GXR-1, GXR-5), it always falls within the range. The values obtained by this method for BCR-1, GXR-1 and GXR-5 are close to those given by Abbey.20 The average precision at concentrations greater than 5 pg 8-1 of Mo is 7% RSD. The data presented in Table 6 for the “in-house” control samples agree well with the “accepted” values obtained by AAS after an HF - HC104 - HN03 attack under pressure. The determination limit, with a\o 00 a\ c 0 B Mo V Cr P W r Table 6. Analysis of “in-house” control samples, with and without ashing at 900 “C.Values in yg g-l unless noted otherwise + A* Bt F A* B t Sample A* Bt XY29 (sulphide mineralisation in carbonaceous, cherty mudstone, 3.5% S) . . 31,31 32,31 (3W (167) (102) (64) (64) (293) (74) (50) XY03 (carbonaceous, cherty mudstone, 0.9% S) . . . . . . . . . . . . . . 166,168 170,169 NSH-1 (carbonaceous, siliceous shale, 0.7% S) . . . . . . . . . . . . . . 102,99 101,103 CHRT (carbonaceous chert, 0.7% S) . . , . 65,65 66,64 SLBA (weakly baritic, massive sulphide + cherty mudstone, 2.3% S) . . 63,65 61,63 . BASL (baritic massive sulphide + cherty mudstone, 4.9% S) . . . . . . . . . . 291,288 289,293 XYPC (phosphatic chert, 3.1% S) . . . . 72,71 75,72 GSD-6 .. . . . . . . . . . . . . 51,49 50,48 * A, Ignited at 900 “C. t B, Unignited. $ Values in parentheses are accepted values. A* B1- A* B t A* B t 449 , 460 445,458 (461) 393,395 400,408 (395) 173,162 49,56 (172) O.35%, 0.32% 0.30% , 0.29% (0.43 ‘/o ) 6.7,6.1 7.0,6.4 (6) 229,235 237,234 (235) 73,75 68,72 (76) 394,404 413,401 (430) 521,512 528,525 270,260 265,262 (517) (260) 82,86 83,88 68,68 61,67 (85) (71) O.I5%, 0.15% O.I5%, 0.15% (0.15 Yo) 0.11%,0.12% 0.12%,0.12% (0.12%) (560) 343,351 318,317 <5 <5 171,180 169,183 (4) ( 182) (147) 150,144 136,145 782,787 782,792 (770) 197,189 190,197 (196) 86,90 78,54 (89) 60,43 70,68 (300) 0.10%,0.10% 0.10%,0.10% (0.1OYo) O.84%, 0.76% OM%, 0.94% (2.96% ) 5.9,6.9 7.4,6.8 11,lO (7) 12,11 (10) (25) 27,25 24,28 24,21 22,20 27,29 32,29 7.9,7.8 7.5,g.l (23) (31) (7.8) 82,85 86,82 465,468 458,469 138,142 141,139 (86) (460) (140) 34,36 21,24 149,156 100,56 191,193 189,190 (35) (149) (190)1260 ANALYST, NOVEMBER 1986, VOL.111 dilution factor of 125, is adequate only for samples enriched above crustal abundance. Vanadium The results for vanadium compare well with literature values, with the possible exceptions of those for QLO-1, SY-3 and MRG-1, which appear slightly low. The value of 57 pg 8-1 for GXR-5 also seems low; however, Abbey20 recommends a value of 60 pg g-1 of V. This small bias to the low side is not borne out in the analysis of “in-house” controls where the data agree well with those obtained by AAS. The determination limit of 4 pg 8-1 is sufficient and the RSD averages 3.5%.Chromium There are three samples in Table 5 for which the results indicate significant deviations (positive and negative) from the recommended values, viz., BHVO-1, SY-3 and MRG-1. No apparent reason is evident and the repeatability of these values is excellent (e.g., 1.2% RSD for BHVO-1). Chromite is a difficult mineral to solubilise and a fusion of this type must be repeated several times for complete dissolution to take place unless a stream of oxygen is introduced above the melt. This fact was verified when the proposed method was applied to the international reference samples DTS-1 (dunite) and PCC-1 (peridotite), for which valuesofonly0.10% ofCr (cf.., 0.42%20)and0.11% ofCr(cf., O.28yo2O), respectively, were obtained. Analysis of the IGS reference sample IGS 30 recovered only 9.9% of the total chromium.It was also observed that the recovery was poor for samples containing significant amounts of sulphur, and this led to the analysis of in-house controls with and without prior ashing. It can be seen from Table 6 that the Cr values are low and erratic in samples containing >2% of sulphur, unless they are ignited prior to fusion, whereupon they agree well with established values obtained by AAS or ICP-ES measurement following fusion with Na202 or LiB02. Phosphorus Most of the results for phosphorus obtained by this method compare very favourably with the recommended values in Table 5 , with the possible exceptions of SDC-1 and MRG-1, which are slightly low. The precision averages 2.3% RSD. However, perusal of Table 6 will quickly indicate that this method fails drastically when applied to the in-house controls containing >2% of sulphur.The values in parentheses were obtained by ICP-ES following fusion with LiB02 and showed good agreement between laboratories using this now common method. The recovery is particularly poor for the samples BASL and SLBA, containing barite, and for XYPC, the phosphatic chert with 17% of CaO. It is likely that the phosphate anion has remained in the insoluble residue with barium and calcium carbonates and the solubility would be improved with a greater flux to sample ratio. Tungsten Most of the international reference samples analysed contain tungsten at levels below this determination limit of 5 pg 8-1; agreement for the remaining four samples is good.The results are also very comparable for the in-house controls where the established data are derived from application of the alkaline fusion - dithiol spectrophotometric method.16 A study is currently in progress to lower the determination limit to a more practical level of 0.1 pg g-l by ICP mass spectrometry. Conclusion It has been demonstrated that the ICP-ES determination of boron in a wide range of geological materials decomposed by sodium carbonate - sodium nitrate fusion is accurate with good precision of about 3% RSD. The method also provides excellent data for vanadium and is reliable for the determina- tion of molybdenum and tungsten at levels above crustal abundances, that is, greater than 2 pg g-1 of Mo and 5 pg g-1 of W. When applying this method to the analysis of geological materials for chromium, care must be taken to ignite the samples prior to fusion if more than 2% of sulphur is present.The low results obtained on applying this method to the determination of phosphorus in baritic or sulphidic samples limit its usefulness and caution is advised, although good agreement was shown in the data on 18 international reference standards. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20 * References Dale, L. S., Appl. Spectrosc., 1979, 33, 404. Dible, W. T., Truog, E., and Berger, K. C., Anal. Chem., 1954, 26, 418. Baucells, M., Lacort, G., Roura, M., and Rauret, G., Appl. Spectrosc., 1984, 38, 572. Troll, G., and Sauerer, A., Analyst, 1985, 110, 283. Kluger, F., and Koeberl, C., Anal. Chim. Acta, 1985,107,127. Higgins, M. D., Geostand. Newsl., 1984, 7, 31. Owens, J . W., Gladney, E. S . , and Knab, D., Anal. Chim. Acta, 1982, 135, 169. Borsier, M., and Garcia, M., Spectrochim. Acta, Part B, 1983, 38, 123. Din, V. K., Anal. Chim. Acta, 1984, 159, 387. Thompson, M., and Walsh, J. N., “A Handbook of Inductively Coupled Plasma Spectrometry,” Blackie, Glasgow, 1983, p. 99. Walsh, J. N., Analyst, 1985, 110, 959. Brenner, I. B., and Eldad, H., ZCPZnf. Newsl., 1984, 10,451. Hall, G. E. M., and Vaive, J. E., in “Current Research, Part A,” Geological Survey of Canada, Ottawa, 1986, Paper 86-IA, p. 71. Bock, R., “A Handbook of Decomposition Methods in Analytical Chemistry,” Blackie, Glasgow, 1979, p. 111. Rose, A. W., Hawkes, H. E., and Webb, J. S., “Geochemistry in Mineral Exploration,” Academic Press, London, 1979, p. 553. Stanton, R. E., “Rapid Methods of Trace Analysis for Geochemical Application,” Arnold, London, 1966, p. 86. Gladney, E. S., Burns, C. E., and Roelandts, I., Geostand. Newsl., 1983, 7, 3. Gladney, E. S., Burns, C. E., and Roelandts, I., Geostand, Newsl., 1984, 8, 119. Govindaraju, K., Geostand. Newsl., 1984, 8, 3. Abbey, S . , “Studies in Standard Samples of Silicate Rocks and Minerals 1969-1982,” Geological Survey of Canada, Ottawa, 1983, Paper 83-15, 114pp. Paper A6189 Received March 17th, 1986 Accepted June 16th, 1986