1802
Anal. Chem. 1981, 53, 1802-1806
Simultaneous Determination of Major, Minor, and Trace Elements in Marine Sediments by Inductively Coupled Plasma Atomic Emission Spectrometry J. W. McLaren," S. S. Berman, V. J. Boyko, and D. S. Russell Analytical Chemistry Section, Chemistry Division, National Research Council of Canada, Oiia wa, Ontario, Canada, K 1A OR9
A procedure which permits the slmultaneous determlnatlon of six major and minor elements (AI, Fe, Ca, Mg, Na, P) and elght trace elements (Be, Co, Cu, Mn, Ni, Pb, V, Zn) In near-shore marlne sedlments by inductively coupled plasma atomlc emlsslon spectrometry Is descrlbed. Dlsolutlon of the samples Is achleved wlth a mlxture of nltrlc, perchlorlc, and hydrofluorlc aclds In sealed Teflon vessels. Accurate callbratlon for all elements can be achleved with slmple aqueous standards, provlded that proper correctlon for varlous spectroscoplc interferences Is made. The accuracy of the method has been conflrmed by analysis of two marlne sedlment reference materials, MESS-1 and BCSS-1, recently produced In these laboratories. The method was not sultable for determlnatlon of As, Cd, and Mo in these materlals because of Inadequate sensitivity nor for Cr and ll because of Incomplete dlssolutlon.
The precision, speed, and sensitivity of inductively coupled plasma atomic emission spectrometry (ICP-AES), as well as its remarkably low susceptibility to concomitant interferences, have led to rapid acceptance of this relatively new technique by geochemists and environmental scientists. At least at present, accurate ICP-AES analyses of geological materials require dissolution of the samples; whether or not complete dissolution is necessary depends on the elements to be determined. The choice of an appropriate dissolution procedure is also complicated by the temptation, on the one hand, to dilute the original sample as little as possible so that the maximum number of trace elements can be determined, and, on the other hand, the desire to avoid dissolved solids concentrations greater than 1 to 2% in the solution presented to the ICP. Fortunately, the sensitivity of ICP-AES permits the direct determination of many trace elements in geological materials even when the dissolution procedure involves a dilution of 100-fold or more. Accurate results are obtained, however, only if proper attention is paid to a variety of spectroscopic interferences which can arise when trace elements are determined in the presence of much higher concentrations of iron, aluminum, calcium, magnesium, and titanium. Recently, as part of the National Research Council of Canada Marine Analytical Chemistry Standards Program, this laboratory undertook production of two near-shore marine sediment reference materials, with the acronyms MESS-1 and BCSS-1, with reliable results for about 25 major, minor, and trace elements. ICP-AES played an important role in this project. This report is a detailed description of a procedure which permitted accurate direct determination of six major and minor constituents and eight trace metals in these sediments. It will be seen that the method of sample dissolution has a major impact on the scope of the subsequent multielement determinations by ICP-AES. The dissolution of geological materials can be accomplished by a variety of techniques which fall into two general cate-
gories: acid digestion and fusion (1).Previous studies (2-5) involving trace element determinations in sediments by atomic spectroscopy have favored acid digestion procedures although some problems of incomplete recovery have been encountered (5). Since near-shore marine sediments are usually composed primarily of clay minerals derived from continental weathering, along with minor amounts of other minerals, procedures developed for dissolution of silicate rocks are often appropriate also, although consideration must be taken of the sometimes appreciable organic content of near-shore sediments. There have been several reports (6-9) of multielement analysis of silicate rocks by ICP-AES, involving dissolution by a fusion procedure if major constituents were determined. A recently reported dissolution procedure (10) involving fusion with sodium hydroxide in graphite crucibles (at a flux to sample ratio of 101) was found to be suitable for preparation of a wide variety of geological and environmental samples for determinations of up to 50 elements at major, minor, and trace levels by ICP-AES. The mixed acid digestion procedure developed in this work results in dissolution of a 0.5 g sediment sample in 50 mL of 1 M hydrochloric acid with a total dissolved solids concentration less than 1%. Reagent blanks for all elements of interest are negligible. Once the dissolution procedure had been developed, the only obstacles to be overcome were spectroscopic interferences. Interferences due to line overlaps and to background shifts were observed, despite a careful choice of wavelengths appropriate for geological matrices. Line overlaps are, of course, a classical problem of atomic emission spectrometry (12) for which the use of.a medium-to-high resolution spectrometer is only a partial solution. The stability of the ICP makes the determination of accurate and reproducible correction factors (13, 14) an effective means of dealing with the remaining overlaps. Background shifts in ICP-AES have been the subject of a series of papers (15-1 7). Corrections for these types of interferences were made in this work by subtraction of background measurements made by means of a rotatable quartz refractor plate which shifted the entire emission spectrum in the exit focal plane (18). EXPERIMENTAL SECTION Reagents. Acids used in this work were purified by subboiling distillation (11). High-purity water was produced by passing distilled water through a deionizing system (Cole-Parmer Instrument Co., Chicago, IL). Sampling and Processing of the Sediments. The near-shore marine sediments analyzed in this work were taken from two sites in the Gulf of St. Lawrence. The wet sediments were freeze-dried and then put through a mechanical roller to break up the lumps formed in the freeze-drying process. Next, the sediments were passed through a 120-mesh stainless steel screen to remove particles larger than 125 pm. The screened material was blended in a large mixer. Prior to bottling of the sediments, subsamples were removed for homogeneity testing by ICP-AES and X-ray fluorescence spectrometry. When homogeneity was achieved, the sediments were bottled. The bottled sediments were y-irradiated to minimize possible effects of bacterial action.
0003-2700/81/0353-1802$01.25/00 1981 American Chemical Soclety
ANALYTICAL CHEMISTRY, VOL. 53,
Table I. Wavelengths and Background Measurement Positions for Marine Sediment Analysis
Table 11. Correction Factors for Spectroscopic Interferences
background measurement positions, nm Wavelength, nm B 1249.773 M o I1 202.030 As I 198.696 Cd 1228.802 Zn I 213.856 P 1214.914 Pb I1 220.353 Co I1 237.862 N i I1 23’1.604 Mn I1 269.373 A1 I 237.335 Fe I1 259.940 v’ I1 292.402 Cr I1 267.716 Mg I1 239.079 Cu 1324,754 Na 1330,237 Ti I1 336.121 Be I1 313.042 Ca I1 317.933 a
2
1 -a
-
202.003 193.678 228.778
202.057 193.714 228.826
-
-
214.895 220.329
214.931 220.376
-
-
231.585
-
-
231.625
-
267.688
wavelength, nm A1 I 237.335 Zn I 213.856 Co 11 237.862
interf‘erent
correction factor? (me L-l)a/ (mg L-’)i 2.0 x 10-3 1.35 x 10-4
1803
-
source of interference
Fe Fe? Fe Fe 1213.869 A1 2.57 X A1 I 237.841 Fe 1.11 x 10-3 Fe I1 237.853 Fe Mn I1 259.373 1.11 x 10-3 Fe I1 259.373 V I1 292.402 4.8 x 10-5 A1 A1 292.452 Fe Fe 292.435? 2.3 x 10-5 Mg 5.4 x Mg? Cu 1324.754 Fe 1.2 X Fe I1 324.739 a The subscripts “a” and ‘7’’ denote analyte and interferent, respectively.
-
267.745
-
-
324.719
-
324.789
-
312.988
-
-
NO. 12, OCTOBER 1981
312.988
Dash indicates background correction not performed.
Dissolution of the Sediments. Sediment samples (0.5 g) were placed in “LORRAN Teflon pressure decomposition vessels (3) (H. K. Morrison and Sons, Ltd., Mount Uniacke, N. S., Canada) and to each were added 3 mL of concentrated nitric acid, 1 mL of concentrated perchloric acid, and 3 mL of concentrated hydrofluoric acid. The sealed vessels were completely immersed in a boiling water bath for 1-2 h. After being cooled, the contents of the vessels were transferred to Teflon beakers q d evaporated to dryness on a low-temperature hot plate. The residues were dissolved and diluted to 50 mL with 1 M HC1. A small amount (1-2 mg) of undissolved material always remained, so the samples were either centrifuged before analysis or allowed to stand overnight to let this material settle. Reference Solutions. The reference solutions were prepared by dilution of stock solutions of high-purity materials with 1 M HC1. The 20 elements listed in Table I were divided into three groups. Group 1included all the major elements: Al, Fe, Ca, Mg, Ti, and N a Isolating this group from the trace elements eliminates possible contamination problems and some complications due to spectroscopic line overlap interferences. Group 2 contained the elements B, P, As, Mn, Zn, and Pb, while group 3 contained Cd, Co, Cr, Cu, Be, Mo, Ni, and V. The minor and trace elements were separated into these two groups to avoid some chemical incompatabilities (e.g., Pb2+ with Cr042-) and also some line overlap interferences (As on Cd I 228.802 nm). The calibration concentration range for each element was chosen on the basis of preliminary analyses of the sediments. From five to seven reference solutions were run for each element. It is noteworthy that none of the reference solutions for the minor and trace elements contained any of the niajor elements; no attempt was made to “matrix match” them to the sediment solutions. ICP-Eohelle Spectrometer. The instrument used in this work has been described in previous publications (18-21). The operating conditions were the same as previously described (20,21). Choice of Wavelengths for Sediment Analysis, The wavelengths listed in Table I were chosen for sediment analysis after consideration of a number of factors. These included the expected concentration of each element in the sample solutions, relative severity of spectroscopic interferences at various wavelengths considered, and physical constraints arising from the necessity to incorporate 20 acceptable lines into a single multielement cassette. For many of the major elements, it was desirable to use a less sensitive line to avoid nonlinearity of the calibration curve. On the other hand, it was necessary to use one of the most sensitive lines for many of the trace elements. Potentially suitable lines chosen from various tables (22-24) were evaluated by performing short (-0.1 nm) wavelength scans (18)
to check for possible spectroscopic interferences. Initially, a sediment solution, the 1 M HC1 blank and a reference solution containing the anal@ were run. These three scans immediately revealed whether background correction would be necessary and also whether the analyte line was overlapped by one or more interfering lines (unless the interfering lines were exactly coincident with the analyte line). Pure solutions of the suspected interferents were then run to positively identify all interferences. In most cases, the line overlap interferences encountered could be handled by determining a correction factor, as described below, to be entered in the data processing software. If the interference was judged to be too severe, other lines were evaluated by the same process. Determination of Correction Factors for Line Overllap Interferences. In casea of full or partial overlap of an interfering line with an analyte line, it is necessary to determine a correction factor which can be multiplied by the determined concentration of the interfering element to provide a correction to be subtracted from the raw concentration data for the analyte. The correction factors used in this work are listed in Table 11. Not surprisinl:ly, many of the line overlap interferences are due to iron, a major constituent of these sediments which also has a very rich ICP emission spectrum. Correction factors were determined by rimning several very pure solutions of each of the interferents at the same time as the instrument was calibrated for the analytes. A least-squares linear regression of the intensity at the analyte wavelength as a function of the interferent concentration was performed (Le., all interferences were assumed to be linear with Concentration). The ratio of the slope thus obtained to the slope of the analyte calibration (Le., the sensitivity) is the correction factor (CF) which is entered in the data processing software. The correction applied, AC,, is thus defined by AC, = CF X C, where the subscripts ‘‘2’ and “a” denote interferent and analyte, respectively, and both the concentrations have units of mg L-l. In determining these correction factors, it is essential to ensure by an independent method that the solutions of the interfering element do not contain significant amounts of the analyte as an impurity. Also, it is necessary to operate the instrument under exactly the same conditions as will be used in subsequent analyses as the correction factors may be quite sensitive to changes in certain of the key ICP operating parameters, notably the power and the carrier gas flow rate (14). Data Acquisition and Processing. Two FORTRAN programs, SEDANL and SEDCAL, control data acquisition and processing, respectively. These two functions have been separated largely for convenience in manipulation of the correspondiing programs. SEDANL contains FORTRAN-callable assembler language subroutines, provided by Digital Equipment Corp. in the computer software package, which allow the user to write instrument control programs entirely in FORTRAN. Details of these programs are available from the authors on request. To start a series of analyses, the operator loads SEDANL. Any number of the 20 available channels (Table I) can be activated for a particular analysis. The channels for which background
1804
ANALYTICAL CHEMISTRY, VOL. 53, NO. 12, OCTOBER I981
! WAVELENGTH Figure 1. Wavelength scans in the vicinity of the Co 11 237.862 nm line: - - -, 1 M HCI; -, 200 pg L-' CO; ..*,600 mg L-' Ai; - * -, 400 mg L-' Fe; -, sediment solution.
correction is desired and the background measurement positions for each channel are then read from a data file. The sequential nature of the data acquisition process in this instrument has been previously described (18,20,21). As each sample or reference solution is run, the raw emission intensity data are transmitted to a file on the data disk as well as to the terminal. Reference solutions are interspersed with samples thoughout the analyses and can be be rerun at any time, as no results are calculated until all standards and samples have been run. At that point, the operator closes the data file. Background correction is performed only on those channels where wavelength scanning has shown that it is necessary, since unnecessary background measurements would increase the analysis time for each sample. Despite the complexity of ICP emission spectra of sediment solutions, accurate background correction by a simple one- or two-point background measurement was usually possible. The wavelengths at which background correction was found to be necessary and the corresponding background measurement positions are indicated in Table I. In general, background measurements were made 0.02-0.03 nm above and below the analyte wavelength by means of the rotatable qqartz refractor plate (18). For Be, however, background measurement was made only at the lower wavelength position because of a Ti I1 line at 313.080 nm. To calculate the results of the analyses, the operator loads the second program, SEDCAL. SEDCALfirst searches the data file and assembles the calibration data, pairing the intensity data for the reference solutions with the corresponding elemental concentrations, which are also stored on the data disk. Linear regressions of emission intensity on concentration are performed for all channels and the calibration parameters for each, including slope, intercept, and computed detection limit, are printed out. Our solution detection limit is defiied as that concentration of analyte producing a response equivalent to 3 times the standard deviation of the ordinate intercept (i.e., the blank, or background) calculated from the linear regression analysis (20). The results for the samples are then calculated, with application of corrections for line overlap interferences where necessary. RESULTS AND DISCUSSION Wavelength Scanning Experiments. For characterization of spectroscopic interferences and for determination of appropriate wavelengths for background correction, the usefulness of short wavelength scans in the vicinity of analyte wavelengths cannot be overstated. In some cases, the scan for a sediment solution showed only a single symmetrical peak superimposed on a relatively flat background (B, Ca, Fe, Mg, Na, Ti), but in others, a background shift (As, Cr, Cu, Mo,
WAVELENGTH Figure 2. Wavelength scans in the vicinity of the Be I1 313.042 nm ilne: - -, 1 M HCi; -, 100 mg L-' Ti; - -, 1 mg L-' V; -, sediment solution.
-
-
Table 111. Major and Minor Element Analysis of Two Marine Sediments
MESS-la constitaccepted uent ICP-AES value A1,0, 11.05i: 11.03 i: Fe,O, Na,O MgO
CaO P,o,
0.71 4.42i 0.27 2.68 i: 0.20 1.42 i 0.04 0.674 i: 0.022 0.153 * 0.004
0.38 4.36 i 0.25 2.50 i 0.15 1.44 i: 0.09 0.674 i 0.064 0.146 i 0.014
BCSS-1a accepted ICP-AES value 11.89 i: 0.78 4.77 i 0.20 2.80 i: 0.18 2.35 i 0.16 0.763 i 0.013 0.168 i 0.011
11.83 i 0.41 4.70 i. 0.14 2.72 i: 0.21 2.44 i 0.23 0.760 i 0.074 0.154 i 0.016
a Results in %; precision expressed as the 95% confidence interval for a single result.
Ni, P, Pb), a line overlap (Al, Mn, V, Zn), or both (Be, Cd, Co) were revealed. Figures 1 and 2 are overlays of several scans performed for two of the more difficult trace elements, cobalt and beryllium. The Co I1 237.862 nm line is subject to two line overlap interferences, that of the intense A1 I line a t 237.841 nm and of the Fe I1 line at 237.853 nm. Figure 1, which includes scans for pure solutions of iron, aluminum, and cobalt, as well as for a sediment solution, indicates the severity of these interferences. In this case, there was little choice but to determine correction factors for the iron and aluminum interferences, as indicated in Table 11. Despite the size of these corrections (totaling about 80% of the observed intensity at the peak wavelength) reliable results were obtained for cobalt by ICP-AES according to the data of Tables I11 and IV. A possibly superior wavelength for cobalt determinations in geological materials may be the ion line at 228.616 nm. Although there are many Co I1 lines of comparable sensitivity around 230 nm, the 228.616 nm line is one of the few which is not overlapped by an Fe line. Many of the lines suffer worse interference from iron than does the 237.862 nm line. Our preliminary wavelength scanning experiments had indicated
ANALYTICAL CHEMISTRY, VOL. 53, NO. 12, OCTOBER 1981
Table V. Estimated Detection Limits for Trace Element,s in Sediments by This Procedure
Table IV. Trace Element Analysis of Two Marine Sediments MESS-la eleaccepted ment ICP-AES value -.
As Be
2.0 0.2
i
Cd
Go
Cr Cu Mn Ni
Pb V Zn
1805
-
-.
10.5 i 2.2
BCSS-1 accepted ICP-AES value
10.9 i 1.7 1.9 2 0.2 0.59 i 0.10
1.4 i 0.2
10.8 t
10.4 i 3.6
1.9 '71
i
11
24.0 i 25.1 i 2.5 3.8 5 2 9 i 34 5:13i 25 30.2 i 29.5 i 2.2 2.7 36.7 r 34.0 i 4.9 6.1 74.2 i 72.4 i 2. 7 5.3 1 8 3 t 10 191 i 17
17.0 1.3
i
238t 4 55.9 r 3.6 24.8 r 5.4 92.0 i 4.2 112i 7
11.2 I 1.2 1.3 I 0.3 0.25 i 0.04 11.42 2.1 123 i 1 4 18.5 I 2. 7 229 i 1 5 55.3 r 3.6 22.7 f 3.4 93.4 I 4.9 1 1 9 i 12
Results in p g / g ; precision expressed as the 95% confidence interval for a single result. a
the suitability of the 228.616 nm line for sediment analyses but it unfortunately could not be included in the multielement cassette because of a geometrical conflict with another line. The sensitive Be I1 313.042 nm line is convenient for sediment analysis provided a spectrometer with adequate power to resolve it from neighboring Ti and V peaks is used. Figure 2 illustrates the potential problems through overlay of scans for four solutions: the 1 M HC1 blank, a 100 mg L-l Ti solution, a 1 mg L-' V solution, and a sediment solution. The Be line lies within the OH band region, as evidenced by the two OH peaks within the short scan (-0.1 nm) for the 1M HC1 solution. The wing of an off-scale Ti I1 line at 313.080 nm can be seen in the scans for a pure titanium solution (100 mg L-l) and a sediment solution containing approximately 40 mg L-* Ti. A further complication is a V I1 line a t 313.027 nm which is coincident with one of the pair of OH peaks. This line is clearly resolved from the Be peak despite a wavelength difference of only 0.015 nm. A slight background shift is also observed. The only suitable location for background measurement is at the low wavelength extreme of the scanning region. Comparison of the scans for the 1 M HC1 blank and the sediment solution indicate that background measurement at any other location within the scanning range would result in a large error. Marine Sediment Analysis. The ICP-AES results reported here were generated during a large program of analysis undertaken in this laboratory in the preparation of two reference near-shore marine sediments, named MESS-1 and BCSS-1. These materials have been exhaustively analyzed by a variety of techniques, including wavelength dispersive X-ray fluorescence spectrometry (XRF), isotope dilution spark source mass spectrometry (IDSSMS), neutron activation analysis (NAA), instrumental photonuclear activation analysis (IPAA), flame atomic absorption spectrometry (FAAS), and graphite furnace atomic absorption spectrometry (GFAAS), as well as ICP-AES. Statistical analysis of the pooled results has permitted the determination of reliable concentrations for 12 major and minor elements and 13 trace elements. Good agreement between the results of at least three independent methods in the usual criterion for accepting a value for a particular element. It should be noted that sample preparation techniques for the various methods were quite different. Direct analysis of the sediments by IPAA was done after they
element As Be Cd
co
cu Mn
detection limit, Mg/g 5 0.05 0.2 0.5 0.1 0.05
element Mo Ni
Pb V Zn
detection limit, p g / g 1.5 0.6 3 0.3 0.1
-
were pressed into disks. The XRF method involved fusion of the sediment with a mixture of lithium tetraborate, sodium tetraborate, lanthanum oxide, and sodium carbonate. Tlne dissolution procedures used for ICP-AES and GFAAS were similar, both involving digestion with nitric, perchloric, arid hydrofluoric acids, although the samples for GFAAS were dissolved in open beakers. The mixed acid digestion procedure and ICP-AES determination described here gave excellent results for 14 of the 20 elements listed in Table I. The good agreement of ICIPAES results with the accepted values obtained from determinations by several independent techniques is illustrated in Tables I11 and IV. All 14 major, minor, and trace elements were determined simultaneously without the need for intermediate dilutions. Accurate calibration was accomplished with easily prepared solutions of the analytes in 1M HC1, without any attempt at matrix matching to counter chemical interferences of the sample matrix. Six of the 20 elements listed in Table I could not be determined by the procedure described here. The evaporation of the residues from the mixed acid digestion to near dryness in order to eliminate excess hydrofluoric acid was preferred to the addition of boric acid (2, 4 ) because it resulted in solutions of much lower dissolved solids concentration. This was desirable not only to minimize problems of nebulizer drift but also to take maximum advantage of the remarkable freedom of the ICP from nonspectroscopic matrix effects (25-27). A consequence of this procedure, however, is th,at silicon is almost completely lost, and boron is partially loEit, from the solutions as volatile SiF4and BFB. As was mentioned in the Experimental Section, the mixed acid digestion procedure always left a small amount (1-2 mg) of undissolved material. At first, it was assumed that this residue did not contain significant amounts of any of the elements of interest, but comparison of ICP-AES results for chromium with those from IDSSMS, NAA, IPAA, and XRB left no doubt that the mixed acid digestion procedure yielded results which were approximately 80% of the accepted values. In addition, ICP-AEX results for titanium were low compared to XRF and IPAA results, indicating 75-80% recovery. It is well-known that certain chromium-containing minerals (e.g:., chromite) are not dissolved by nitric/perchloric/hydrofluoric acid mixtures. Examination of the combined solid residues from several BCSS-1 sample dissolutions by spark source ma3s spectrography indicated that incomplete dissolution was rl?sponsible for the low chromium and titanium results. Three elements, arsenic, molybdenum, and cadmium, could not be determined by this procedure because of inadequate sensitivity. The accepted values for arsenic and cadmium in Table IV are all less than 3 times the estimated ICP-AES detection limits of Table V, which were calculated by multiplying solution detection limits obtained from calibration statistics by 100, the dilution factor of the dissolution procedure. The molybdenum concentration in these sedimenik was determined by spark source mass spectrometry to be about 2 Fg/g, again very close to the estimated ICP-AES detection limit. The detection limits estimated for arsenic
1808
Anal. Chem. W l , 53, 1806-1808
and molybdenum in sediments may in fact be slightly optimistic in that they do not take into account the very large background shifts below about 210 nm encountered with samples containing high aluminum concentrations (17). The detection limits for elements such as arsenic and molybdenum with lines in this region are almost certain to be poorer in the presence of a high concentration of aluminum than in the 1 M HC1 solutions used for calibration in this work. An increase in the background a t the As I 193.696 nm line can also result from the broadened wing of the C I 193.091 nm line. Carbon may be present as an impurity in the argon or as residual soluble organic materials in the samples. The slightly less intense As I line at 197.197 nm would perhaps be less susceptible to carbon interference, but the sensitivity of both of these lines is inadequate for accurate arsenic determination in these samples. It appears that the sensitivity of ICP-AES is nearly sufficient for accurate cadmium determinations in marine sediments. The estimated detection limit for the Cd I 228.802 nm line used in this work is 0.2 pg/g. This line was chosen because of its high sensitivity and its freedom from iron interferences. This latter attribute is not shared by alternative cadmium lines of comparable sensitivity at 214.438 nm and 226.502 nm. However, the 228.802 nm line is subject to possible arsenic interference because of the As 1228.812 nm line. Although the interference is not so severe in our highresolution instrument that it could not 228.812 nm handled by a correction factor, this approach would require an accurate arsenic determination, which for these sediments was precluded by inadequate sensitivity. Perhaps the best line for cadmium determination in marine sediments would be the ion line a t 226.502 nm. Although this line is overlapped by an iron line a t 226.505 nm, the correction factor is relatively small (2.3 X and here a t least the concentration of the interferent can be very accurately determined. This approach was applied successfully to determine cadmium at about the 2 pg/g level in a fresh water sediment but was unsuccessful for a marine sediment in which the cadmium concentration was about 0.2 pg/g (18). This is hardly surprising when it is noted that multiplication of the correction factor by a typical solution iron concentration of 300 mg L-l yields a bogus cadmium concentration of 0.07 mg L-l, equivalent to about
7 pg/g of cadmium in the sediment! Thus, it appears that
accurate determinations of Cd in marine sediments by ICPAES will usually require a preconcentration of the Cd or at least a separation from the matrix. LITERATURE CITED (1) . . Maxwell, J. A. “Rock and Mineral Analysis”, Wlley-Intersclence: New York, 1968, Chapter 5. (2) Rantala, R. T. T.; Lorlng, D. H. At. Abs. News/. 1975, 74, 117-120. (3) Rantala, R. T. T.; Loring, D. H. At. Abs. Newsl. 1973, 72, 97-99. (4) Agemian, H.; Chau, A. S. Y. Anal. Chim. Acta 1975, 80, 61-66. (5) Sinex, S. A.; Cantillo, A. Y.; Helz, G. R. Anal. Chem. 1980, 52, 2342-2346. (6) Waish,J. N. Spectrochim. Acta, Part 8 1980, 358, 107-111. (7) Uchida, H.; Uchlda, T.; Iida, C. Anal. Chlm. Acta 1980, 776, 433-437. (8) Burman, J.-0.; Ponter, C.; Bostrom, K. Anal. Chem. 1978, 50, 679-680. (9) Burman, J.-0.; Bostrom, K. Anal. Chem. 1979, 57, 516-520.
(IO) Floyd, M. A.; Fassel, V. A.; D’Sllva, A. P. Anal. Chem. 1980, 52, 2168-21 73.
(11) Dabeka, R. W.; Mykytluk, A.; Berman, S. S.; Russell, D. S. Anal. Chem. 1978, 48, 1203-1207. (12) Boumans, P. W. J. M. Spectrochim. Acta, Part 8 1980, 358, 57-71. (13) Boumans, P. W. J. M. Spectrochlm. Acta, Part 8 1976, 376, 147- 152. (14) Botto, R. I. In “Developments In Atomlc Plasma Spectrochemical Analysis”; Barnes, R. M., Ed.; Heyden and Son: Philadelphia, PA, In press. (15) Larson, 0. F.; Fassel. V. A.; Winge, R. K.; Kniseley, R. N. Appl. Spectrosc. 1978, 30, 384-391. (16) Fassel, V. A.; Katzenberger, J. M.; Winge, R. K. Appl. Spectrosc. 1979. 33. 1-5. (17) Larson, G sF.; Fassel, V. A. Appl. Spectrosc. 1979, 33, 592-599. (18) McLaren, J. W.; Berman, S. S. Appl. Spectrosc., in press. (19) Berman, S. S.; McLaren, J. W. Appl. Spectrosc. 1978, 32, 372-377. (20) Berman, S. S.; McLaren, J. W.; Willie, S. N. Anal. Chem. 1980, 52, 488-492. (21) Berman, S. S.; McLaren, J. W.; Russell, D. S. In “Developments In (22) (23) (24) (25) (26)
Atomic Plasma Spectrochemical Analysis”; Barnes, R. M., Ed.; Heyden and Son: Phlladelphia. PA, In press. Winge, R. K.; Peterson, V. J.; Fassel, V. A. Appl. Spectrosc. 1979, 33, 206-219. Boumans, P. W. J. M.; Bosveld, M. Spectrochim. Acta, Part 8 1979, 348, 59-72. Parsons, M. L.; Forster, A,; Anderson, D. “An Atlas of Spectral Interferences In ICP Spectroscopy”; Plenum Press: New York, 1980. Larson, G. F.; Fassel, V. A.; Scott, R. H.; Knlseley, R. N. Anal. Chem. 1975, 47, 238-243. Kalnlcky, D. J.; Fassel, V. A.; Kniseley, R. N. Appl. Spectrosc. 1977,
31. 137-150. (27) Boumins, P:W. J. M.; DeBoer, F. J. Spectrochim. Acta, Part 8 1977, 328, 365-395.
‘. RECEIVED for review March 12,1981. Accepted June 1,1981.
Determination of Arsenic Compounds in Biological Samples by Liquid Chromatography with Inductively Coupled Argon Plasma- Atomic Emission Spectrometric Detection Masatoshl Morita, * Takashi Uehiro, and Keiichlro Fuwa National Institute for Environmental Studies, 16-2 Onoga wa, Yatabe, Tsukuba, Ibaragi, 305 Japan
Speclation and quantitative analysis of arsenlc compounds in biological samples are performed by using high-performance liquid chromatography (HPLC) with inductively coupled argon plasma-atomlc emission spectrometrlc detection (ICP). HPLC Is used to separate mixtures of arsenlc compounds on anion and cation exchange columns using phosphate buffer. The ICP is used as a selective detector by observlng As emlssions at 193.6 nm. Detection ilmit is 2.6 ng Asis.
It has been shown that arsenic is incorporated into both marine and freshwater organisms in the form of water-soluble
and lipid-soluble arsenic compounds (1-3). Recent studies to identify the chemical forms of these arsenic compounds have shown the presence of arsenite (As(III)),arsenate (A@)), methylarsonic acid (MAA), dimethylarsinic acid (DMAA), and arsenobetaine (AB) (4-6). Methylated arsenicals also appear in the urine and plasma of mammals, including man, by biotransformation of inorganic arsenic compounds (7-11). Several methods have been devised to characterize these arsenicals. Gas-phase methods have been described for the separation and identification of inorganic species as well as for MAA and DMAA (6,1240).Braman and Foreback succeeded in sep-
0003-2700/81/0353-1806$01.25/00 1981 American Chemical Society