(4) W. F. Carey, J. Assoc. Anal. Chem.. 51, 1300 (1968). (5) W. R. Penrose, Crit. Rev.Environ. Control, 5, 465-482 (1974). (6) Y. Talmi and C. Feldman, "The Determination of Traces of Arsenic: A Review", in "Arsenical Pesticides", ACS Symposium Series No. 7, E. Woolson. Ed., American Chemical Society, Washington, D.C., 1975, pp 13-34.
(7) (8) (9) (10) (1 1)
C. E. Gleit and W. D. Holland, Anal. Chem., 34, 1454 (1962). H. A. Schroeder and J. J. Balassa, J. Chronic Dis., 19, 85 (1966). C. E. Mulford, At. Absorp. Newsl., 5, 135 (1966). P. R. Walsh, J. L. Fasching, and R . A. Duce, Anal. Chem., 48,1014 (1976). P. R. Bevington, "Data Reduction and Error Analysis for the Physical Sci-
ences", McGraw-Hill. New York, 1969. (12) J. M. Wood, Science, 183, 1049 (1974). (13) G. R. Sandberg and I. K. Allen, "A Proposed Arsenic Cycle in an Agronomic Ecosystem", in "Arsenical Pesticides", E. A. Wooison. Ed.. ACS Symposium Series No. 7, American Chemical Society, Washington, D.C., 1975, pp 124-147. (14) G. 0. Doak and L. D. Freedman, "Organometallic Compounds of Arsenic, Antimony, and Bismuth", Interscience, New York, 1970.
for review October 2of
Accepted February
20, 1976.
Matrix Effects and Their Control during the Flameless Atomic Absorption Determination of Arsenic Paul R. Walsh" and James L. Fasching Department of Chemistry, University of Rhode Island, Kingston, R.I. 0288 1
Robert A. Duce Graduate School of Oceanography, University of Rhode lsland, Kingston, R.1. 0288 1
Various matrices have been found to affect the flameless atomic absorption analysis of arsenic with the Heated Graphite Atomizer. Sodium in the presence of sulfate presentsthe most serious interference. The use of magnesium in these solutions is described to enhance the sensitivity and control the interference effects. Environmental matrices have been analyzed for arsenic by this method. Results are consistent with those obtained by standard addition and neutron activation techniques.
In flameless atomic absorption determinations of arsenic, certain matrices may cause chemical interferences in which reactions of concomitant materials with arsenic may affect the production rate and population of gaseous elemental arsenic from which absorbance measurements are obtained. Ion exchange ( I ) , tantalum lined graphite tubes (2), and solvent extraction ( 3 )have been used to eliminate chemical, as well as the spectral ( 4 ) interferences related to the flameless determination of arsenic. Recent studies on the flameless atomic absorption determination of arsenic in steel and alloys ( 5 , 6 )has shown, however, that arsenic absorbances are highly dependent on certain solution components and that matrix modification techniques, in which a specific reagent is added to.remove or control an interference, should be applied (7). Ammonium nitrate, for example, has been used as an agent for the removal of well documented sodium chloride interferences (7,8).The addition of a protective agent enhances the absorbance signal by forming compounds that atomize more efficiently than the unprotected analyte-interferent matrix (9). For arsenic sample matrices, a protective agent may be used to reduce volatility during drying and ashing cycles or to promote more efficient volatization of arsenic in the atomization cycle. This report presents a study of arsenic absorbances in various mineral acids and metal salt matrices. In particular, the matrix effects of certain environmental constituents which many analysts encounter is discussed in conjunction with a protective agent that may be used in their presence. Neutron activation analysis is used to verify arsenic determinations in natural aerosol matrices by flameless atomic absorption. 1014
ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
EXPERIMENTAL Apparatus.Atomic Absorption. A Perkin-Elmer Model 503 atomic absorption spectrophotometer equipped with the Model 2100 Heated Graphite Atomizer (HGA), deuterium background corrector, and chart recorder was used with an arsenic electrodeless discharge line source. The 193.7-nm arsenic resonance line was used a t all times. Instrumental settings were in accordance with manufacturer's operation procedures. The optimum time-temperature program of the Model 2100 HGA was experimentally determined to be: dry, 125 "C for times ranging from 10 to 40 s for sample volumes of 5 to 20 fil; ash, 600 "C for 20 s; atomize, 2300 "C for 8 s. The normal nitrogen purge cycle (at 30 units) was used for all measurements. Eppendorf and Oxford pipets were used for sample injections as well as for sample preparation. Sample injection volumes were typically 20 11. Calibration curves for arsenic were linear over wide concentration ranges. For 20-fil volumes, a detection limit of 5 &g/l. As a t a S/N ratio of 2 was obtained. Neutron Actiuation. Samples were irradiated in the Rhode Island Nuclear Science Center 2 Megawatt swimming pool reactor a t a thermal flux of 4 X 1OI2 neutrons/cm2/s. After a 24-h decay period, the 559-keV gamma ray of the '6As ( t l l z = 26.4 h) isotope was monitored by counting on an Ortec 40 cm3 Ge(Li) coaxial detector (resolution of 2.3 keV for the 1332-keV gamma ray of 6oCo) coupled to a Nuclear Data (Model 2200) 4096 channel analyzer. Data were stored on magnetic tape for processing by computer programs. Reagents.All standard solutions were made from 1000 pg/ml stock solutions obtained from Fisher Scientific Co., Pittsburgh, Pa. Suprapur acids from E. M. Laboratories, Inc., Elmsford, N.Y. were used where acids are specified. Standard solutions and test solutions were made on a daily basis, although no variation in arsenic solutions was noticed over a three-week period a t concentrations ranging from 0.02 to 10 fig/ml. RESULTS AND DISCUSSION
The absorbance of 0.5 to 1Fglml arsenic concentrations in various mineral acids indicated that maximum values are obtained from solutions of dilute (0.01-1N) nitric acid. As found previously by other workers (5, IO),hydrochloric and hydrofluoric acids gave very reduced signals and these anions should be avoided for this application. Arsenic absorbances in 0.01 to 0.1 N sulfuric acid are slightly reduced relative to nitric acid solutions. Nitric acid is, therefore, used as the solvent for subsequent sample preparation unless otherwise noted. Acid concentrations are kept below 2 N because of erratic absorbance readings and enhanced graphite tube dete-
Table I. Arsenic Absorbances in Various Metal Salt Matrices Absorbance of 0.47 pg/ml As in 500 pg/ml of the metal and solution listed Metal None Ca"(CO3) Cu11(N03):! Ga111(N03)3 Fe"'C13 @(Nod Mg"(N03h Mn"(NO& Na'(NO3) N i 11S04 a
0.1 N "03 0.20 0.35 0.46 0.32" 0.41 0.27 0.50 0.39 0.23 0.55"
1.75 mg/ml C1 480 pg/ml SO4 (0.05 N HCl) (0.01 N H2SO4) 0.07 0.22 0.28 O.2gn
...
0.24 0.30 0.28 0.20 0.44
0.14 0.35 0.41 0.33" 0.40 0.14O 0.46 0.46 0.10
Table 11. Absorbance of 0.47 bg/ml As in the Presence of Sulfate, Trace Metals, and Sodium Sodium, Sulfate, pg/ml Trace metals pg/ml Absorbance 0 0 0 0.24 2 pg/ml Fe, Cu, Mn,
480
2 pg/ml Fe, Cu, Mn,
0 0 0
0.22 0.21 0.28
0
0.27
0
0.27
io0 47
0.24 0.17
235
0.07
470
0.02
Ni 960
0 100 500
1000
Double absorbance peak
rioration at higher levels. Arsenic absorbances in numerous metal salt matrices and various acid media are presented in Table I. The mean absorbance of the data in Table I is 0.34 with a relative standard deviation of 33%. Absorbance data have a precision of 0.01 absorbance unit. The range of the absorbance values emphasizes the effect of metal salt components on the efficiency of arsenic atomization. In general, absorbances are greatest in 0.1 N " 0 3 and least in 0.05 N HC1, reflecting the normal trend discussed earlier. The presence of a metal did, in most cases cause a significant enhancement of the arsenic absorbance relative to the corresponding metal free solution. Inspection of Table I shows that sodium in the sulfate matrix gave the lowest arsenic absorbance while nickel and magnesium metal solutions consistently gave the highest. Nickel, gallium, and potassium solutions sometimes gave double arsenic absorbance peaks most likely attributable to heterogeneous arsenic speciation and slow atomization. The arsenic absorbance in these metal solutions was, therefore, less precise. The double absorbance peaks could often be eliminated by atomizing at 2500 "C. This temperature enhanced the deterioration of the graphite tube and sensitivity. The large depression of the arsenic absorbance in the sodium and sulfate matrix is very significant since both sodium and sulfate are major constituents in many sample matrices including natural waters, aerosols, and biological materials. In addition, the use of sulfuric acid in a sample pretreatment could seriously affect a subsequent arsenic determination by flameless atomic absorption. These factors suggested that a more datailed study of this specific interference was warranted. The absorbance of arsenic in the presence of sulfate, trace metals, and sodium is presented in Table 11. The data in Table I1 indicate that a signal depression by sulfate is not apparent in the presence of trace quantities of iron, copper, manganese, and nickel. The effect of sodium, however, in the sulfate matrix is very pronounced and proportional to the increase in sodium content. In Table 111, a comparison of the arsenic absorbances of various sodium and sulfate solutions with and without background correction is presented. Clearly, the effect of the sodium sulfate is twofold. First, the background corrected data imply that the arsenic atomization efficiency is suppressed by the sodium sulfate. Second, the non-background corrected data show that there ,is a molecular background absorption associated with the highest concentrations of sodium and sulfate. The reduced arsenic atomization may
0 0
480 960 0
Ni 2 wg/ml Fe, Cu, Mn, Ni 0 2 pg/ml Fe, Cu, Mn, Ni 2 pg/ml Fe, Cu, Mn, Ni 2 pg/ml Fe, Cu, Mn, Ni
Table 111. Absorbance of 0.47 pg/ml Arsenic in Sodium and Sulfate Matrices with and without Background Correction Absorbance Na, pglml
Sulfate, pg/ml
0 47 235 470
0 100 500 1000
With *H arc Without *H arc 0.28 0.17 0.07 0.02
0.28 0.17 0.12 0.32
be due to a slow rate of decomposition of the sodium sulfate matrix. The data in Table I, however, suggest that most divalent metal sulfates readily decompose during the atomization cycle, with nickel and magnesium being the most efficient. This information led to the hypothesis that a divalent metal, nickel and magnesium in particular, could be used in the role of matrix modification to increase the volatility of arsenic in the sodium sulfate matrix. Figure lA, B, and C show the variability of the arsenic absorbance as a function of sodium and sulfate a t different magnesium concentrations in 0.1 N "03. Each graph represents data for a specific magnesium concentration. Absorbance is plotted vs. sodium concentration with lines connecting those points which have the sulfate concentration indicated. In Figure lA, the effect of sodium and sulfate on the arsenic absorbance is clearly seen. It is interesting to note that the sodium alone has little effect on the arsenic absorbance. Only when sulfate is added does the signal depression and variability become apparent. Figures 1B and C indicate that the effect of sodium and sulfate is reduced with both 100 Wg/ml and 500 pg/ml magnesium. The arsenic absorbances are greater and have a smaller range for the higher Mg levels, as illustrated by the mean absorbances and relative standard deviations for the various Mg concentrations; no magnesium, 0.15 f 30%;100 pg/ml magnesium, 0.21 f 14%; and 500 pg/ml magnesium, 0.26 f 9%. There is a definite enhancement and stabilization of the arsenic absorbance in the presence of magnesium. Similar results are obtained when nickel is used as the protective agent and atomization is done a t 2500 "C. This matrix modification technique, using magnesium as a protective agent, was used to determine arsenic in marine particulate aerosol samples that contain large quantities of sodium and sulfate as sea salt constituents. Some urban area particulate aerosol samples, which also contained large quantities of anthropogenic, as well as marine, sulfate, and some remote continental particulate aerosol samples were also ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
1015
0'3 [NO
Table IV. Arsenic ( f i g ) Found in Ashing Residues of Atmospheric Particulate Samples Method
MAGNESIUM
~~
Sample"
I
200pg/mlS04 W V
z U m
4
A Id0
2b0
pg/ml
0.3 r
do
460
SODIUM
G1 G2 G3 G4 G5 G6 G7 Y1 Y2
Y3
-
0
0 g g / m L SO4
a
m
100 k g / r n i SQSL~
U
2 0 0 p g / m i SO4
z
c1 c2
c3
Mg add AAS 2.07 f 0.13 0.87 f 0.07 0.39 f 0.03 0.69 f 0.04 0.27 f 0.03 0.81 f 0.07 0.60 f 0.05 9.5 f 0.9 7.0 i 0.6 7.0 f 0.6 21.0 i 2.0 18.5 f 1.7 16.9 f 1.9
Std addition AAS 2.06 f 0.13 0.91 f 0.07 0.32 f 0.04 0.74 f 0.04 0.22 f 0.03 0.80 f 0.07 0.53 f 0.05 9.8 f 0.9 6.7 f 0.5 7.2 f 0.5
...
... ...
NAA ...
... ...
... ... ...
... 11 f l 8.0 f 0.5 7.9 f 0.6 18.6 f 1.3 20.1 f 1.8 18.6 f 1.5
a G series, marine location; Y series, urban location; C series, continental location.
B
W
2AO 3iO p g / m l SODIUM
Id0
O! CL
0.1
4AQ
1 500 p g / r n l MAGNESIUM
0
0
100
200
300
400
p g / m l SOD1 U M
Figure 1. Absorbance of 0.47 pglml arsenic in 0.1 N H N 0 3 as a function sodium and sulfate at various levels of magnesium
of
without using standard addition is a suitable alternative. Clearly, the matrix modification, or protective agent technique, is most applicable to samples of variable composition. This study has shown that arsenic absorbances can be enhanced by many metal salts and depressed especially by sodium and sulfate, and also by halides. Awareness of these factors allows the analyst to choose, and apply, a major solution constituent such as magnesium that will control the atomization efficiency of the analyte in a flameless atomic absorption determination. ACKNOWLEDGMENT We are grateful to the nuclear reactor staff a t the Rhode Island Nuclear Science Center for providing space and facilities for this work. We also acknowledge the helpful criticisms and technical advice of Gerald L. Hoffman of the University of Rhode Island, Kingston, R.I. LITERATURE CITED
analyzed for arsenic. The continental aerosol samples were collected at a site where a smelter is believed to be introducing large quantities of arsenic into the atmosphere. All samples were collected on Whatman 41 filters. Sample destruction was by low temperature ashing (Tracerlab Model 505, LFE Corporation, Waltham, Mass.), a procedure that causes significant losses of arsenic (11) and almost total loss of chloride through volatilization. The ashing residues, however, could be used for comparative arsenic analyses by the matrix modification technique (using 500 +g/ml Mg as the protective agent) and standard addition using flameless atomic absorption and also by standard neutron activation. Arsenic determinations in the ashing residues of the various particulate aerosol samples are presented in Table IV. From these data, reasonable agreement among these methods is apparent. Although neutron activation is considered a more absolute technique, the relative inaccessibility of the method precludes its use by many workers. As seen here, arsenic determinations a t trace levels by flameless atomic absorption
1016
ANALYTICAL CHEMISTRY, VOL. 48, NO. 7, JUNE 1976
(1) R. b i r d , S. Pourian. and S. Gabrielian, 166th National Meeting, ACS, August 1973, Division of EnvironmentalChemistry, Reprints, Vol. 13, No. 2, Paper 15. (2) R. Baird and S. Bagrielian, Appl. Spectrosc., 28, 273 (1974). (3) K. C. Tam, fnviron. Sci. Techno/., 8, 734 (1974). (4) M. J. Adams, G. F. Kirkbright. and P. Rienvatana, At. Absorp. Newsl., 14, 105 (1975). (5) D. B. Ratcliffe, C. S. Byford, and P. B. Osman, Anal. Cbim. Acta, 75, 457 ( 1975). (6) W. B. Barnett and E. S. McLaughlin, Jr., Anal. Chim. Acta, 80, 285 (1975). (7) R. D. Ediger, At. Absorp. Newsl., 14, 127 (1975). (8) R. D. Ediger, G. E. Peterson, and J. D. Kerber, At. Absorp. Newsl., 13, 61 (1974). (9) G. F. Kirkbright and M. Sargent, "Atomic Absorption and Fluorescence Spectroscopy", Academic Press, New York, 1974, p 523. (10) F. Shaw and J. M. Ottaway, Analyst, (London), 99, 184 (1974). (11) P.R.Walsh, J. L. Fasching,andR. A Duce, Anal. Cbem., 48, 1012(1976).
RECEIVEDfor review October 20,1975. Accepted March 12, 1976. Presented in part at the 26th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 3-7, 1975, Cleveland, Ohio. This research was supported by the Office of the International Decade of Ocean Exploration, National Science Foundation, under NSF Grant GX-33777.
