Anal. Chem. 1984, 56,2724-2726
2724
Determination of Barium in Seawater by Direct Injection Graphite Furnace Atomic Absorption S,pect romet ry Kevin K. Roe and Philip N. Froelich* Department of Oceanography, Florida State University, Tallahassee, Florida 32306
Barlum Is determined In aqueous solutlon by dlrect Injection grhphlte furnace atomic absorptlon spectrometry. ",NO, modlfler ellmlnates seawater NaCl in a slow 1400 OC char step. Atomlratlon Is at 2700 OC. Detectlon Is at 553.6 nm, and callbratlon Is by peak area. Sensltlvlty is 0.2 abs s/ng (abs = absorbance unit) and Is Independent of matrix sallnlty. Detectlon llmits are 30 pg of Ba (600 ng L-') for 50-pL sample Injections of seawater and 10 pg of Ba (200 ng L-') for freshwaters. Preclslon Is about f13%. There are no spectral Interferences due to calclum or Ba halldes. Results of the analyses of seawater samples agree well with values determined by Isotope dllutlon.
We report here an accurate and precise method for simple and rapid determinations of barium in seawater and other aqueous solution. Knowledge of the distribution of barium in the ocean is based entirely on determinations by isotope dilution mass spectrometry (1-7). In our work on the biogeochemistry of the congeners of silicon (germanium, barium, and radium) in rivers, seawater, and estuaries, we required a less sophisticated Ba technique and thus investigated the possibility of modifying previous graphite furnace atomic absorption techniques to the available instrumentation. In the application of graphite furnace atomic absorption spectrometry to Ba analyses in a seawater matrix by direct injection, several problems must be overcome: furnace emission in the visible a t the analytical wavelength of Ba (8, 9),matrix interferences due to NaCl smoking ( I O ) , chloride vapor-phase interferences ( I I ) , Ba ionization (12), barium carbide formation (13, 14), and spectral interferences from Ca (15-17). Partial solutions to some of these problems exist in the literature, but no single approach has documented a verifiable methodology: no standard GFAA analytical technique is available for Ba analyses in seawater or in estuaries. The technique we report here is not as sensitive nor as precise as the isotope dilution method but is more widely applicable to the analyses of natural waters and has been adapted to the determination of Ba in dissolutions of solid phases, such as biogenic opal and whole plankton (Roe and Froelich, unpublished data).
EXPERIMENTAL SECTION Apparatus. We use a Perkin-Elmer 5000 atomic absorption spectrophotometer equipped with an HGA-400 graphite furnace and a tungsten iodide background corrector. Argon is used as the purge gas. We found Ringsdorff graphite tubes (Perkin-Elmer part no. B-0091504)provide consistent results and last longer than regular pyrolitically coated graphite tubes. Peak response is monitored with a Soltec 1241 fast-response recorder and a Hewlett-Packard 3390A integrator. The monochrometer is peaked at 553.6 nm, slit width at 0.4 nm, with background corrector on. Careful alignment of the furnace and lamps (primary source and background lamp) is critical to ensure absence of base-line shifts caused by furnace emission during atomization. Standards and Reagents. The primary standard is prepared from a 1000 ppm Ba atomic absorption standard (BaCl,, Fisher Scientific Co.) by dilution of 1mL of the standard plus 4 mL of 16 N HNO, (G. F. Smith Co.) to 1L with Ba-free deionized water. 0003-2700/84/0358-2724$01.50/0
Secondary standards are prepared by appropriate dilution of the primary standard with 0.4% HNO, (v/v) and stored in conventional polyethylene bottles. We have detected no difference between freshly diluted standards and standards up to 6 months old. Our routine secondary working standards are 0, 10, 20,30, 40,60, and 80 pg L-' which are adequate for the range of standard additions needed for most natural waters. Two matrix modifiers can be used for seawater, ",NO3 or HN03. ",NO3 ("Baker Analyzed" reagent grade) is dissolved in deionized water to make a 1.4 M solution. Nitric acid is diluted to 10% (v/v) with deionized water. Method. Determination of barium concentration in seawater is performed by the method of multiple standard additions. The solutions to be analyzed are prepared by pipetting 500 pL of the seawater sample, 250 wL of aqueous working standard, and 250 ML of 1.4 M NHlN03 (or 10% HNO,) into polyethylene vials. The aqueous standards are chosen to include the 0 wg L-* Ba standard and two other standards such that the Ba added is approximately 1and 2 times the Ba concentration in the seawater sample. Best results are obtained when the total Ba injected is between 100 and 2000 pg. The total injection volume can be varied up to a maximum of 100 wL. Due to the high char and atomization temperatures necessary for determinations of seawater and the corrosive nature of the injection solution, the Ringsdorff tubes survive only about 25 injections at 50 pL each. Thus, replicate injections of each solution should be kept to a minimum of about four. A three-point standard curve of Ba in deionized water is then run for comparison of sensitivitiesbetween matrices. These should agree within 25%. The furnace program is dependent upon the sample matrix and the volume of injection. For seawater standard addition curves, the dry temperature is 130 "C, 1-5 ramp, and 90-5 hold for 50-pL injections (150 s for 100-pLinjections), char at 1400 "C, 60-5 ramp, and 30-5 hold, and the atomization temperature is 2700 "C at maximum power heating (0-5 ramp) and held for 7 s. Freshwaters can be analyzed directly vs. a Ba standard curve in deionized water with a shorter ramp time on the char step (10 s). During atomization, the argon flow rate is reduced to 100 mL min-' by using miniflow. Both peak height and peak area recorders are activated approximately 5 s before atomization.
RESULTS AND DISCUSSION Matrix Modification. Matrix modification is necessary for seawater samples to reduce smoking caused by incomplete removal of NaCl during the char step. In addition, an atomization interference can occur due to formation of vapor-phase Ba halide, suppressing atomic absorption. This vapor-phase interference is general to many metals (11)but is a particular problem for Ba in seawater since the Ba-C1 bond has a higher dissociation energy than matrix chlorides (18). Without matrix modification, Ba atopic absorption is greatly reduced even a t the char temperature of 1600 "C normally sufficient to remove NaC1. Intercalation of matrix chlorides by graphite causes retention of chlorides even at temperatures above 1600 "C and becomes a serious problem after the pyrolitic coating of the graphite tube is breached, limiting the number of useful seawater injections. In addition, loss of the coating enhances barium carbide formation, causing peak broadening and loss of sensitivity. Our procedure overcomes these problems by carefully controlled charring with a modifier. We tested three matrix modifiers for Ba determinations in NaCl solutions and in seawater: ",NO3 (9),HN03 (19),and 0 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
t4
"\o .. O!
' 0
500
0
1500
1000
2000
Char TernDerature ('C)
0.6
1
2725
0 0
0
B 0
Salinity (per mil)
E
Figure 2. Sensitivity of standard additions curves as a function of salinity in the Charlotte Harbor estuary (FL) with a 1-s ramp before the char step. The river sample at 0.3 per mil was done without modifier. Increasing the length of the char step eliminates the pronounced dip in sensitivity at midsalinities (see text).
0 0
B 0
0
W 500
1000
,--
1500
0
0.4
2000
Char Temperature ('Cl
0
Flgure 1. (A) Influence of char temperature on loss of Ba from 0.4% HNO, matrix. Total mass of Ba injected is 1000 pg (50 pL of 20 pg L-I Ba). There is no background absorbance. Atomization Is at 2700 "C. The error bars are f l u (n = 4). (B) Background absorbance as
0
-
0.3
0,
a function of char temperature. Absorbance is plotted as peak height to avoid integration of noise at high scale expansion. The Solutions injected were 20 pL of Atlantic deep water pius 20 pL of 0.7 M
0
\
% b!l
0.2
",NO,.
-9
ascorbic acid (20). Each modifier removes chloride at different temperatures. HN03 removes C1 as HC1 at drying temperatures. ",NO3 removes C1 as NH4C1which sublimates at 340 "C. Ascorbic acid acts as a surfactant, allowing efficient volatilization of small well-dispersed NaCl crystals at 1600 "C. It has been used to reduce matrix effects in the determination of cobalt, copper, and manganese in seawater (20) and in the determination of Ba in dissolutions of coral skeleton (Shen, personal communication). We tested both 1% and 2% ascorbic acid additions to a Ba solution in 3.5% NaCl solution. The sehsitivity approached that of Ba in 0.4% HN03 only for the 2% ascorbic acid injections, but reproducibility was poor. The boiling points of NaC1, BaC12,and Ba metal fall in such a narrow range that complete C1 volatilization cannot be accomplished without some Ba loss, so ascorbic acid is not an appropriate modifier for chloride matrices. Blanks. T o eliminate contamination, we clean all polyethylene vials and bottles by first rinsing and filling them with 2% HN03, sonication overnight, followed by several rinses with deionized water. There is no detectable blank in our deionized water. Blanks for 1.4 M ",NO3 and 1.6 M HN03 were determined by volume addition of each neat modifier (0-100 pL) in the furnace. Absorbance/microliter (abs/pL) (slope of the regression by both peak height and peak area) quantified by ",NO3 and HN03 standard additions yielded blanks of 1.0 f 0.2 pg L-I Ba for 1.4 M ",NO3 and 0.6 f 0.2 pg L-' for 1.6 M "OB. The zero standard (0 pg L-' Ba in DDW) contains no detectable Ba ( ..go.1
0
L
cn 0 \o
0 2 00
I
2400
2600
2800
Atomization Temperature
3000
("C)
Flgure 3. Sensitivity vs. atomization temperature. The solutions injected were 20 pL of Atlantic deep water containing 12 pg L-' Ba pius 20 /.LL
Of
0.7 M NHdNO,.
of Ba up to 1400 "C. Background absorbance in seawater samples falls dramatically with increasing char temperature up to 1200 "C, above which the background is low, stable, and amenable to correction (Figure 1B). Background may still occur at 1200 OC, appearing sporadically as a leading peak which is not present at 1400 "C. The char temperature is chosen as 1400 "C with a long char time to ensure that the matrix is fully removed before the onset of atomization. Estuarine waters of midsalinity present unusual problems. Without careful char optimization, sensitivity can decrease by 50% compared to freshwater or seawater samples even though standard additions and modifier concentrations remain constant (Figure 2). Tramontano (personal communication) found similar results when analyzing Ba in estuaries with NH4N03modifier, and McArthur (21) had similar problems with Mn and ",NO3. McArthur (21) eliminated the sensitivity shift by using a slow ramp on the char step (29 "C &). We find that increasing the ramp time from 10 s to 60 s increases Ba sensitivity in midsalinity waters to that of seawater and freshwaters. Atomization Temperature. Maximum power heating is necessary to volatilize Ba quickly from the walls of the tube. Sensitivity is a function of atomization temperature. It
2726
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
1.51
Table I. Barium in Natural Waters sample location
Ba
GFAA,pg L-l
B a IDMS, pg
L-I
seawatera
11 m 642 m 3076 m river waterb
5.4 f 0.9 7.9 f 1.0 21.3 f 2.8 70 f 8
4.5
8.5 19.5 64-70 (23)
GEOSECS 227: 25'00" 175'05'E (Central Northwest Pacific Ocean). IDMS data are from Chan (personal communication). Mississippi River: IDMS values represent range in samples collected between Venice, LA, and H e a d of Passes. GFAA sample collected a t Belle Chasse. LA.
continues to increase above 2700 "C (Figure 3), but tube life is diminished. Furnace emission and tube vibration are serious problems at 3000 "C where the pyrolitic coating fails after 10-15 injections. An atomization temperature of 2700 "C with 100 mL min-l argon flow produces sharp reproducible peaks, reduces memory effects, and permits two full standard addition curves of 35% salinity seawater before graphite tube failure. Interferences. Flame atomic absorption and emission measurements of Ba are reportedly plagued by emission of CaOH bands (543.0-622.0 nm) which peak near the Ba lines (15,16)and by Ba ionization (12). In order to investigate the possibility of a calcium spectral interference with GFAA methods, we prepared a Ba-free Ca solution. Ca in HC1 was purified of Ba by cation exchange (22). The separation was monitored by 133Batracer to produce a Ba-free Ca solution. The eluted Ca fraction was 42.5 mM Ca and contained undetectable 133Ba. Standard additions to the Ca solution yielded a result insignificantly different f r o p the NH4N03 matrix modifier blank. Thus, Ca interference is not a problem. Sturgeon and Berman (12) studied simultaneous atomic absorption, atomic emission, ionic absorption, and ionic emission of Ba in a graphite furnace at both 2300 and 2700 "C. At 2700 "C, a 12% enhancement of the Ba atomic absorption signal occurs if a 40-fold excess of Cs over Ba is added to suppress ionization. Our standard addition method will correct for small ionization effects in environments such as estuaries where K+, an ionization suppressor, increases with increasing salinity. Analytical Features of Merit. Sensitivity and Detection Limits. The average sensitivity determined from the slopes of 62 standard additions curves of seawater samples is 0.21 abs s/ng, with a standard deviation of 0.045 abs s/ng. This large variability reflects inherent differences in individual graphite tubes and is thus a measure of the range of sensitivities to expect in routine work, but not an indication of the technique's imprecision since each sample is internally calibrated. The standard deviation of base-line noise during atomization is 0.001 abs, corresponding to noise-limited level of detection by peak height of 10 pg of Ba or a concentration detection limit of 200 ng L-' for 50 pL of freshwater sample injections in the absence of blanks. Our modifiers contain detectable Ba blanks, so that 100-pL injections (25 pL of either NH4N03 or "OB modifier) contain 25 and 15 pg of Ba, respectively. Practical detection limits (3 times the standard deviation of the blank) are 30 pg of Ba or 600 ng L-l for 50 pL of seawater sample injections. Precision and Accuracy. We estimate precision to be f 1 2 % , based on the average uncertainty in the intercepts of 62 standard additions curves for seawater samples. Replicate determinations (n = 7) of one seawater sample containing 21.3
0
0
2000
4000
6000
8000
Bo ( p g )
Flgure 4. Range of response for Ba in 0.4% HN03 (50-pL injections, n = 4). The useful range is 0-2000 pg of Ba, denoted by the re-
gression line.
pg L-l Ba (GEOSECS 227,3076 m, Table I) yielded a precision of f13%. Multiple injections of river water and aqueous standards are reproducible to *5%. Comparisons of determinations by our technique with those by isotope dilution mass spectrometry are shown in Table I. The precision of IDMS determinations are usually 0.5-1 % (6). Our technique agrees well with IDMS values. Linearity. The calibration curve by peak area is linear from the detection limit to at least 2000 pg (Figure 4). Registry No. NaC1, 7647-14-5; ",NO3, 6484-52-2; "OB, 7697-37-2; Ba, 7440-39-3; water, 7732-18-5.
LITERATURE CITED (1) Wolgemuth, K. J Geophys. Res. 1970, 75,7686-7687. (2) Wolgemuth, K.; Broecker, W. S. Earth Planet. Sci. Lett. 1970, 8 , 372-378. (3) Bernat, M.; Church, T.; Allegre, C. J. Earth Planet. Sci. Lett. 1972, 16,75-80. (4) Bender, M.; Snead, T.; Chan, L. H.; Bacon, M. P.; Edmond, J. M. Earth Planet. Sci. Lett. 1972, 16, 61-03. (5) Li, Y.-H.; Ku, T. L.; Mathieu, G. G.; Woldemuth, K. Earth Planet. Sci. Lett. 1973,19,352-358. (6) Chan, L. H.; Edmond, J. M.; Stallard, R. F.; Broecker, W. S.;Chung, Y. C.; Weiss, R. F.; Ku, T. L. Earth Planet. Sci. Lett. 1978,32,258-267. (7) Chan, L. H.; Drummond, D.; Edmond, J. M.; Grant, B. Deep-Sea Res. 1977,2 4 , 613-649. (8) Beaty, R. D.; Cooksey, M. M. A t . Absorpt. News/. 1978, 17,53-58. (9) Conky, M. K.; Sotera, J. J.; Kahn, H. L. Instrumental Laboratory Inc., Report 11, 1979,pp 1-5. (10) Epstein, M. S.;Zander, A. T. Anal. Chem. 1979,51,915-918. (11) Slavin, W.; Carnrick, G. R.; Manning, D. C. Anal. Chem. 1984, 5 6 , 163-166. (12) Sturgeon, E. R.; Berman, S. S.Anal. Chem. 1981,53,632-639. (13) Lagas, P. Anal. Chlm. Acta 1978,98,261-264. (14) Styris, D. L. Anal. Chem. 1984,5 6 , 1070-1076. (15) Kolrtyohann, S. R.; Pickett, E. E. Anal. Chem. 1988, 38, 585-587. (16) Snellemen, W.; Rains, T. C.; Yee, K. W.; Cook, H.D.;Menis, 0. Anal. Chem. 1970,42, 394-398. (17) Ebdon, L.; Hunon, R. C.; Onaway, J. M. Anal. Chlm. Acta 1978, 96, 63-67. (18) Slavln, W.; Manning, D. C. frog. Anal. At. Spectrosc. 1982, 5 , 243-340. (19) Jasim, F.; Barbooti, M. M. Talanta 1981,28, 353-358. (20) Hydes, D. J. Anal. Chem. 1980,52,959-963. (21) McArthur, J. M. Anal. Chim. Acta 1977,93,77-83. (22) Bacon, M. P.; Edmond, J. M. Earth Planet. Sci. Lett. 1972, 16, 66-74. (23) Hanor, J. S.;Chan, L. H. Earth Planet. Sci. Lett. 1977,37, 242-250.
RECEIVE~ for review May 21,1984. Accepted August 16,1984. We thank L. H. Chan for providing barium data from GEOSECS 227. Funding was provided by National Science Foundation Grant DPP-82-14213 to K. A. Fanning, G. A. Vargo, and P.N.F.