Anal. Chem. 1997, 69, 21-24
On-Line Determination of Dissolved Silica in Seawater by Ion Exclusion Chromatography in Combination with Inductively Coupled Plasma Mass Spectrometry Akiharu Hioki,† Joseph W. H. Lam, and James W. McLaren*
Institute for National Measurement Standards, National Research Council of Canada, Ottawa, Ontario K1A 0R6, Canada
A method for the direct determination of dissolved silica in seawater by ion exclusion chromatography in combination with inductively coupled plasma mass spectrometry is described. This method was developed as a second, independent method to complement the usual colorimetric procedure in the determination of a certified concentration of dissolved silica in a planned seawater reference material. Ion exclusion affords a separation of the dissolved silica not only from the major seawater cations but also from potentially interfering anions. The detection limit, conservatively estimated at 2.3 ng g-1 as Si (0.08 µM), is superior to that achievable by direct analysis by inductively coupled plasma atomic emission spectrometry. Determinations of the so-called “micronutrients” (e.g., nitrate, phosphate, and dissolved silica) in seawater are among the most commonly performed analyses in oceanographic research and survey work. Surprisingly, there exists no certified reference material (CRM) that can be used to check the accuracy of such analyses, although intercomparison exercises have been conducted on a regular basis for the past 10 years.1 The lack of a seawater CRM for micronutrients can be attributed both to the difficulty of preparing a suitable material with an adequate shelf life2,3 and to the dearth of independent methods available for the determinations. The National Research Council of Canada (NRCC) has undertaken a project, in collaboration with the Bedford Institute of Oceanography (BIO) of the Canadian Department of Fisheries and Oceans, to address the need for a seawater CRM for micronutrients, with the initial objective being the preparation of a material with certified values for nitrate, phosphate, and dissolved silica. One of the requirements to achieve our objective is the development of alternative methods to complement the colorimetric methods that are almost universally used by oceanographers for these determinations. Good agreement between at least two independent methods of analysis is considered an essential prerequisite for certification. Refinement of a method for the direct determination of nitrate in seawater by ion chromatography with UV absorbance detection will be the subject of a separate report. In the present paper, we describe a method † On leave from National Institute of Materials and Chemical Research, 1-1, Higashi, Tsukuba, Ibaraki 305, Japan. (1) Kirkwood, D. S.; Aminot, A.; Perttila¨, M. Fourth Intercomparison Exercise for Nutrients in Seawater; ICES: Copenhagen, 1991; Cooperative Research Report 174. (2) Aminot, A.; Ke´rouel, R. Anal. Chim. Acta 1991, 248, 277-283. (3) Aminot, A.; Ke´rouel, R. Mar. Chem. 1995, 49, 221-232.
S0003-2700(96)00588-4 CCC: $14.00
© 1996 American Chemical Society
for the direct determination of dissolved silica in seawater by a combination of ion exclusion chromatography (sometimes referred to as ion chromatography-exclusion mode, abbreviated as ICE) with inductively coupled plasma mass spectrometry (ICPMS). Dissolved silica is believed to be present in seawater in the form of silicic acid, Si(OH)4, a weakly dissociated inorganic acid. Two related colorimetric methods are most often used for its determination.4 Both are based on the formation of a heteropoly acid by treatment of the sample with an acidic molybdate solution. In one of these procedures, the absorbance of the yellow silicomolybdic acid itself is measured; in the other, which is considerably more sensitive, spectrophotometry is carried out on the molybdenum blue complex which is formed upon reduction of the heteropoly acid. There have also been two reports of the determination of dissolved silica in seawater by inductively coupled plasma atomic emission spectrometry.5,6 The first dissociation constant of silicic acid in 0.5 M sodium chloride solution at 25 °C has been reported as 3.9 × 10-10;7 consequently, it seems likely that the major silicon-containing species in seawater is undissociated silicic acid.8 Silicic acid can be separated from the bulk chloride and sulfate ions by ion exclusion chromatography (ICE). There are a few reports on the determination of silicate by ICE, with detection by postcolumn reactions.9-11 In the present work, our aim was to take advantage of the very high sensitivity and selectivity of ICPMS by using it for on-line detection of silicon separated from the major seawater ions by ICE. This is the first reported use of either ICPMS or ICE for the direct determination of dissolved silica in seawater, and it is also the first reported application involving a combination of ICE and ICPMS. EXPERIMENTAL SECTION Reagents. Water purified by distillation was used. Unless otherwise stated, all reagents used were of analytical reagent grade. A commercial sodium silicate standard solution containing 1000 µg g-1 Si in 0.2% sodium hydroxide (Anachemia Science, Montreal, Canada) was used to prepare a 50 µg g-1 silicate (4) Riley, J. P. In Chemical Oceanography, 2nd ed.; Riley, J. P., Skirrow, G., Eds.; Academic Press: New York, 1975; Vol. 3, Chapter 19, pp 427-431. (5) Isshiki, K.; Sohrin, Y.; Nakayama, E. Mar. Chem. 1991, 32, 1-8. (6) Abe, K.; Watanabe, Y. J. Oceanogr. 1992, 48, 283-292. (7) Ingri, N. Acta Chem. Scand. 1959, 13, 758-775. (8) Spencer, C. P. In Chemical Oceanography, 2nd ed.; Riley, J. P., Skirrow, G., Eds.; Academic Press: New York, 1975; Vol. 2, Chapter 11, pp 250-251. (9) Sakai, H.; Fujiwara, T.; Kumamaru, T. Bull. Chem. Soc. Jpn. 1993, 66, 34013406. (10) Konno, S.; Goto, R. Kogyo Yosui 1994, 433, 50-54. (11) Sakai, H.; Fujiwara, T.; Kumamaru, T. Anal. Chim. Acta 1995, 302, 173177.
Analytical Chemistry, Vol. 69, No. 1, January 1, 1997 21
Table 1. ICPMS Operating Conditions rf power (kW) plasma gas flow rate (L/min) auxiliary gas flow rate (L/min) nebulizer gas flow rate (L/min) scanning mode dwell time (ms) m/z 28 and 29 other masses measurements/peak no. of sweeps/reading resolution (amu)
1.0 15 1.0 0.9 peak hop transient 800 400 1 1 0.8
working standard by dilution with water. This working standard was stable for many months if kept at pH g 5 in a polyethylene bottle.4 Isotopically enriched sodium bicarbonate (99.4 atom % 13C) was purchased from Isotec Inc., (Miamisburg, OH). The seawater sample used was collected in the North Atlantic Ocean at a depth of 200 m by means of multiple casts of a Rosette sampler loaded with 12-L Niskin bottles. Water was transferred at a rate of ∼1 L/min, under a positive pressure of nitrogen, from the Niskin bottles to precleaned 50-L polyethylene carboys provided by NRCC. During this transfer, the water was passed through Nuclepore Type QR-AM 5-in. cartridge polycarbonate membrane filters with a nominal pore size of 0.05 µm. These filters were mounted in Ametek No. 5 polypropylene 5-in. cartridge housings. Subsequently, the water was subjected to γ-irradiation to eliminate any remaining bacterial activity. Apparatus. A Model AGP-1 advanced gradient pump (Dionex, Sunnyvale, CA) was used for ICE. Samples were injected by means of the microinjection valve of a Dionex Model LCM liquid chromatography module, equipped with an injection loop of 100-µL nominal volume. The column used was a Dionex IonPac ICE-AS1 ion exclusion column (9 mm i.d. × 250 mm), which has a capacity of 27 mequiv. The eluent was water unless otherwise specified, and the flow rate was always 0.8 cm3 min-1. The effluent from the column could be directed either to waste or to the ICPMS instrument by means of a four-way, six-port Teflon rotary switching valve (Type 50, Rheodyne, Cotati, CA). The column was connected to this valve with a 0.3-m length of PEEK tubing (0.254 mm i.d., 1.59 mm o.d.). The switching valve was connected to the nebulizer of the ICPMS instrument with a 0.75-m length of Teflon PFA tubing (0.5 mm i.d., 1.6 mm o.d.). Short (7 cm) lengths of Teflon PFA tubing (0.5 mm i.d., 1.6 mm o.d.) were used immediately upstream and downstream of the valve to make the connections to the longer lengths of PEEK and Teflon PFA tubing. The inductively coupled plasma mass spectrometer was a Perkin-Elmer SCIEX (Concord, ON, Canada) Elan 5000 equipped with the standard cross-flow nebulizer, operated as recommended by the manufacturer. Typical operating parameters are summarized in Table 1. The alumina central tube of the torch was replaced with a quartz tube; surprisingly, this reduced the Si background signal.12 Data acquisition was not commenced until ∼6 min after sample injection. Prior to this, and between chromatographic runs, a peristaltic pump was used to supply water to the nebulizer via the switching valve. Seawater Analysis. An aliquot of ∼0.5 cm3 of the seawater sample was used to flush and fill the 100-µL injection loop. For (12) Takaku, Y.; Masuda, K.; Takahashi, T.; Shimamura, T. J. Anal. At. Spectrom. 1994, 9, 1385-1387.
22 Analytical Chemistry, Vol. 69, No. 1, January 1, 1997
Figure 1. ICE-ICPMS chromatograms at m/z 28 and 29 for a 100µL injection of an 18.4 µM Si standard solution. Nebulizer gas flow rate, 0.95 L/min.
Figure 2. ICE-ICPMS chromatograms at m/z 23, 28, and 29 for a 100-µL injection of a seawater sample with a dissolved silica concentration of about 20 µM. Nebulizer gas flow rate, 0.95 L/min.
∼6 min after sample injection, the column effluent was directed to waste, after which the valve was switched to direct this flow to the nebulizer, and data acquisition commenced. Data acquisition continued for 11 min or longer. RESULTS AND DISCUSSION ICE-ICPMS chromatograms for an 18.4 µM silicate standard in water are shown in Figure 1. Data were acquired at m/z 28 and 29, corresponding to the two most abundant isotopes of silicon (with abundances of 92.2% and 4.7%, respectively); the third isotope, 30Si, is completely obscured by the very large 14N16O peak. The chromatograms contain a single peak at ∼450 s. The relatively high background at m/z 28 comprises a large contribution from 14N2 and a small contribution from 28Si. The N2 component arises primarily from air entrainment in the ICP, while the Si component arises from ablation and/or vaporization of silica from the quartz torch. The relative magnitudes of these two components depend on a number of factors, including the ICP operating conditions. Prior to commencing chromatographic analysis, the nebulizer gas flow rate was adjusted to optimize the sensitivity for Si. The ratio of background-subtracted peak heights for m/z 28 and 29 agrees with the value predicted (19.6) from the abundances of the two Si isotopes. Chromatograms at m/z 23, 28, and 29 for a seawater sample with a dissolved silica concentration of ∼20 µM run under the same conditions are shown in Figure 2. The response at m/z 23 is featureless during
Figure 4. Chromatograms at m/z 13, 28, and 29 for a 100-µL injection of a 2.4 mM solution of 13C-enriched NaHCO3. Nebulizer gas flow rate, 0.95 L/min.
Figure 3. Dependence on the nebulizer gas flow rate of the two major peaks observed in the ICE-ICPMS chromatograms for a 100 µL injection of a seawater sample. 0, m/z 28, first peak; 9, m/z 28, second peak; 4, m/z 29, first peak; 2, m/z 29, second peak.
the period of data acquisition, indicating that a good separation from sodium (and, by inference, other seawater ions) has been achieved. (Sensitivity at m/z 23 was, however, reduced by applying an offset of ∼5.5 V by means of the OmniRange feature of the ELAN 5000.) The chromatogram at m/z 28 contains three peaks: the first a small peak at 400s, the second at 450 s (as for the silicate standard solution), and the third at ∼930 s. For the chromatogram at m/z 29, peaks corresponding to only the second and third of these peaks are visible. For the second peak (at 450 s), the ratio of peak heights for m/z 28 and 29 agrees with the value predicted from the abundances of the two Si isotopes. In addition, the dependences of the heights of these two peaks on nebulizer gas flow rate were found to be identical, as shown in Figure 3. For the third peak (at 930 s), however, the ratio of peak heights was much smaller than would be expected on the basis of natural abundances of the Si isotopes. Furthermore, it was found that the value of this ratio depended on the nebulizer gas flow rate. It can be seen from Figure 3 that the dependences of the heights of these two peaks on nebulizer gas flow rate were not identical. Both of these observations suggest strongly that the peaks observed at both m/z 28 and 29 at about 930 s arise from species other than silicon. Potentially interfering polyatomic species at m/z 28 that could be expected to arise from direct injection of seawater include 12C16O and 11B16OH; at m/z 29, a potentially interfering polyatomic species is 12C16OH. Since both carbonic and boric acids are weak inorganic acids, both might be expected to exhibit some retention on an ICE column. It was therefore important to establish that neither of these was coeluting with the silicate. The hypothesis that the peaks appearing at about 930 s for m/z 28 and 29
Figure 5. Chromatograms at m/z 10, 11, 28, and 29 for a 100-µL injection of the same seawater sample as in Figure 2. Nebulizer gas flow rate, 1.30 L/min.
corresponded to 12C16O and 12C16OH, respectively, was initially tested by obtaining a chromatogram for a solution of sodium bicarbonate. The m/z 28 and 29 chromatograms contained peaks at 930 s, the heights of which were in the same proportion as those in Figure 2. Further confirmation was obtained by obtaining the chromatogram for a solution of 2.4 mM 13C-enriched sodium bicarbonate. (This concentration is approximately the same as the sum of carbonate and bicarbonate species in seawater.) The results are shown in Figure 4. Sensitivity at m/z 13 was reduced by applying an OmniRange offset of 4.0 V. Only a very small peak at 930 s is visible for m/z 28; this is attributed to 12C16O. The much larger peak at m/z 29 is attributed to 13C16O; the peak for 13C16OH is, of course, obscured by 14N16O. The chromatogram for m/z 13 contains a peak at the same location, lending further support to these assignments. Boron is a conservative minor element in seawater, with a concentration of 0.416 mmol kg-1 (at 35‰ salinity). It exists as boric acid, mainly undissociated H3BO3.13 It was possible to establish that borate was not coeluting with silicate by monitoring m/z 10 and 11 in addition to m/z 28 and 29, as shown in Figure 5. Peaks at ∼550 s were observed in both the m/z 11 and 10 chromatograms. The ratio of peak heights (∼4) is in agreement with the value expected on the basis of natural abundances of (13) Bruland, K. W. In Chemical Oceanography, 2nd ed.; Riley, J. P., Chester, R., Eds.; Academic Press: New York, 1983; Vol. 8, Chapter 45, p 177.
Analytical Chemistry, Vol. 69, No. 1, January 1, 1997
23
10B
and 11B. The small peak at the same location in the m/z 28 chromatogram, assigned to 11B16OH, is well separated from the earlier Si peak at 450 s. The chromatograms shown in Figure 5 were obtained after increasing the nebulizer gas flow rate from 0.95 to 1.30 L/min. This change greatly reduced the N2 background at m/z 28, so that the relatively small 11B16OH peak was visible. Under these ICP operating conditions, the sensitivity for silicon was much reduced. These results indicated that the silicate peaks at 450 s observed at m/z 28 and 29 for a seawater sample were not subject to interference by species arising from carbonate or borate, because of the longer retention times of the latter. In fact, a much longer retention time for silicate, near the total permeation limit, was expected because undissociated silicic acid is assumed to be the major silicon-containing species in seawater. A comparison of the first dissociation constants of silicic, boric, and carbonic acids would suggest that all three should have similar retention times near the total permeation limit.14 The retention time for silicate was not appreciably altered by the use of 10 or 160 mM hydrochloric acid instead of water as the eluent. The much shorter retention time for silicic acid, while surprising, is clearly desirable not only because it results in higher sensitivity (i.e., less chromatographic peak broadening) and shortens the analysis time but also because it avoids interference from polyatomic species arising from carbonate. One feature of the m/z 28 chromatogram of Figure 2 remained to be investigated, the small peak at about 400 s that precedes the peak assigned to dissolved silica. This feature is not observed in the chromatogram for a silicate standard solution. It was necessary to establish whether this small peak arose from a silicon-containing species. Chromatograms of solutions of sodium sulfate and sodium chloride, with concentrations roughly equivalent to seawater concentrations, obtained under conditions similar to those used to acquire the data shown in Figure 2, contained a small peak at the same retention time. The chromatogram for the NaCl solution showed that the small peak at m/z 28 coincided with the elution of chloride, i.e., that its appearance coincided with the elution of unretained species. While the origin of this small peak remains unclear, it seems highly unlikely that it arises from a silicon-containing species in seawater. Seawater Analysis. The seawater sample used for this study was a prototype seawater CRM for nutrients with a dissolved silica concentration of ∼20 µM. An initial analysis of one bottle of this sample (No. 84) was carried out by the method of standard additions. Additions of 1 and 2 times the approximate dissolved silica concentration in the sample were made. The height of the m/z 28 peak at 450 s was used for quantitation. Peak height rather than peak area measurement was used because of the slight overlap of the Si peak with the unidentified peak at 400 s. The value determined was 20.3 ( 1.6 µM. Subsequent analyses of unspiked subsamples taken from five other bottles of the seawater were performed by making use of this standard additions calibration. A comparison of results of six pairs of determinations by the ICE-ICPMS method and the standard colorimetric method is shown in Table 2. Results for the former method are about 5% higher than those for the latter. While a standard t-test for comparison of results of pairs of determinations by two methods indicates that this discrepancy is significant at the 99% level of confidence (5 degrees of freedom), it should be remembered that (14) Tanaka, K.; Ishizuka, T.; Sunahara, H. J. Chromatogr. 1979, 174, 153-157.
24 Analytical Chemistry, Vol. 69, No. 1, January 1, 1997
Table 2. Comparative Analyses of a Seawater Sample for Dissolved Silica (Results in µM) bottle no.
ICE-ICPMS
colorimetry
3B 21 42 66 84 103A
20.7 21.0 20.7 20.7 20.3 20.3
19.5 19.7 19.8 19.7 19.5 19.7
mean SD
20.6 0.3
19.7 0.1
five of the six ICE-ICPMS results were derived from the slope of the standard additions calibration for one of these samples. Further analyses are necessary to resolve this apparent discrepancy. It was subsequently determined that the slope of the standard additions calibration did not differ significantly from the slope of an external calibration. Thus, there appears to be no significant matrix effect; external calibration rather than standard additions could be used for the determination. One problem that was observed over the period of several hours was a gradual decrease in sensitivity. It was not determined whether this change was a result of a change in chromatographic conditions or a change in ICPMS instrumental sensitivity. A change in the former could be caused by the gradual displacement of protons from the ion exclusion column by seawater cations, although even after 40 100µL injections only about 10% displacement should have occurred. It was observed that the height of the peak at 930 s in the m/z 28 chromatogram also decreased with time and that, for repeated injections of the same sample, the ratio of the heights of the peaks at 450 and 930 s was constant. Detection Limit. The detection limit of the method was estimated by determining the concentration of dissolved silica required to yield a peak height equivalent to 3 times the standard deviation of the peak observed for a commercially available “nutrient-free” seawater (Ocean Scientific Ltd.) While the concentrations of nitrate and phosphate in this material are very low, the dissolved silica concentration is about 1.0 µM. The calculated detection limit, 0.08 µM (2.3 ng g-1 as Si), is therefore conservative. CONCLUSION The ICE-ICPMS method fulfills our requirement to have at least one alternative to the standard colorimetric method for the determination of dissolved silica in seawater for the purpose of production of a certified reference material. While not as simple as the previously reported ICPAES method,5,6 it offers a number of advantages. These include a lower detection limit, the requirement for only a very small sample volume, the possibility of highly accurate and precise isotope dilution ICPMS analyses, and the possible detection of other silicon-containing species, if indeed these exist in seawater.
Received for review June 13, 1996. Accepted October 12, 1996.X AC960588L X
Abstract published in Advance ACS Abstracts, November 15, 1996.
