703
Anal. Chem. 1987, 59. 703-708
centration. However, the preservation effect was replicated many times, under both laboratory and field conditions, and the results were robust. The results are supported by a similar preservation effect noted in the literature for the insecticide fenitrothian sorbed on XAD-2 resin (11, 12).
CONCLUSIONS Both XAD-2 resin and C18-silica gel, when used to concentrate hydrocarbons from water, preserved extracted hydrocarbons from bacterial attack. The preservation of extracted samples was found to be unaffected by differences in dissolved oxygen, nutrients, or size of bacterial population. Safe storage times of columns in the field were a t least 2 months, and in the laboratory at least 3 months. These results contrasted sharply with equivalent stored water samples which showed degradation beginning after a few days and complete degradation after 10-15 days. The preservation effect is encouraging for the future use of solid-phase columns for the extraction of trace organics from water either in situ or in the laboratory. Column samples can be expected to be stable under reasonable conditions of handling and storage for lengths of time up to several months. Samples concentrated on columns can therefore be handled with fewer precautions than ordinary water samples; for instance, they could be mailed to laboratories for analysis without degradation occurring en route. Registry No. XAD-2, 9060-05-3; H20, 7732-18-5.
LITERATURE CITED Green, D. R. Sampling Sea Water for Trace Hydrocarbon Determina tion; National Research Council of Canada publication No. 16472; 1976. Junk, G. A.; Richard, J. J.; Grieser, M. D.; Witiak, D. et ai. J. ChromatOgr. 1974, 99, 745-762. Gallant, R. F., King, J. W.; Levins, P. L,; Piecewicz, J. F. Characterization of Sorbent Resins for Use in Environmental Sampling; - - EPA-600/778-054; 1978. Schatzberg, P.; Adema, C. M.; Thomas, W. M.; Mangum, S. R. Oceans '86 Conference Record; Marine Technology Society: Washington, DC, 1986; pp 1155-1159. Jadamec, J. R.; Kleineberg, G.; Adrick. M. S.; Su, C. H.; Hiltabrand, R. R.; Cutler, J. L., Jr., I n Proceedings of Marine Data Systems International Symposium; New Orleans, LA, April 30-May 2, 1986; Marine Technology Society: 1986; p 127-140. Green, D. R.; Stull, J. K.; Heesen, T. C. Mar. Pollut. Buii. 1986, 17, 324-329. Sander, L. C.; Wise, S.A. Anal. Chem. 1984, 5 6 , 504-510. Jobson, A.; Cook, F. D.; Westlake, D. W. S . Appi. Microbiol. 1972, 2 3 , 1062-1089. National Research Council. I n Oil in the Sea. Inputs, Fates and Effects; National Academy Press: Washington, DC, 1965; pp 32-36. Walker, J. D.; Colwell, R. R. Appi. Environ. Microbioi. 1976, 3 1 , 189- 197. Berkane, K; Caisse, G. E.; Mallet, V. N. J. Chromatogr. 1977, 139, 386-390. Mallet, V. N.; Brun, G. L.; MacDonald, A. N.;Berkane, K. J , Chroma tOQr. 1970, 160, 81-88.
RECEIVED for review July 29,1986. Accepted October 27,1986. This research was supported by the Marine Analytical Chemistry Standards Program of the National Research Council of Canada, and by the Unsolicited Proposal Program of the Canadian Department of Supply and Services.
Determination of Cobalt, Copper, Mercury, and Nickel as Bis(2-hydroxyethy1)dithiocarbamate Complexes by High-Performance Liquid Chromatography Jeffrey N. King and James S . Fritz*
Ames Laboratory-DOE
and Department of Chemistry, Iowa State University, Ames, Iowa 50011
Ammonium bls(2-hydroxyethy1)dithiocarbamate (HEDC) is used as a precoiumn derivatiring reagent for the reversedphase HPLC determination of Co( I I), Cu( I I), Hg( I I), and Ni( 11). The metal-HEDC complexes are soluble in water, which eliminates the need to extract them Into an organic solvent prior to analysis. Co( 11), Cu( I I), and Ni( I I)can be determined by direct aqueous InJectiononto a C18 column in the range of 0.005-10.0 mg/L, with a precision of 1.5-3.2%. Detection Is at 405 nm. The Hg( II)-HEDC complex is preconcentrated on an on-line adsorption column prior to HPLC analysis. After preconcentratlon, Hg( I I ) can be determined in the range of 0.02-25 pg/L with a precision of less than 2 % The analysis of spiked electroplating wastewaters showed good agreement with expected values.
.
The separation and determination of metal ions by both normal- and reversed-phase HPLC have received a growing amount of attention in recent years. Typically, these methods are based upon the precolumn formation, separation, and subsequent detection of the metal chelates. Several reviews have summarized the work being done in this field (1-3). The disubstituted dithiocarbamates have found extensive use for the separation of metal ions by HPLC. Their ability to form stable chelates with a number of metal ions makes 0003-2700/87/0359-0703$01.50/0
them ideal for this type of application. Separations have been reported by use of both normal-phase (4-10) and reversedphase (11-14) HPLC. Applications have included the analysis of electroplating solutions with a Waters Radial Compression column (15); the automated analysis of wastewater with electrochemical detection (16-19); the determination of trace metals in rice flour and citrus leaves (20);and the determination of low levels of precious metals (21). The dithiocarbamates used thus far for the HPLC separation of metal ions all form water-insoluble metal complexes. When these are used as precolumn derivatization reagents, the metal complexes precipitate as colloidal particles. Thus, the complexes have typically been extracted into an organic solvent such as chloroform prior to the chromatographic separation. The organic extraction can be directly injected onto the separating column when using normal-phase HPLC (9, 10). For reversed-phase HPLC, the organic extract is usually evaporated to dryness and then redissolved in the mobile phase prior to injection (15, 22). Both techniques lengthen the analysis time and increase the possibility of contamination. The direct injection of aqueous sample solutions has been accomplished by incorporating the dithiocarbamate reagent in the mobile phase (16-19, 22). The metal complexes are formed by on-column derivatization prior to the separation. However, detection of the metal complexes is complicated by @ 1987 American Chemical Society
704
ANALYTICAL CHEMISTRY, VOL. 59, NO.
5, MARCH
1, 1987
the large excess of reagent in the mobile phase. For electrochemical detection, a postseparation anionic stripper column has been used to remove excess reagent from the mobile phase (19). When UV detection is used, the monitoring wavelength must be chosen such that the background absorbance from the excess reagent is minimized (22). The purpose of this work is to examine the use of ammonium bis(2-hydroxyethy1)dithiocarbamate(HEDC) as a precolumn derivatizing reagent for the reversed-phase separation of trace metal ions. The hydroxy groups present in HEDC cause its metal complexes to be soluble in water at low concentrations. In previous work, we showed that metal ions complexed with HEDC can be concentrated from very dilute solution by adsorption onto Amberlite XAD-4 resin, with subsequent elution with an organic solvent and measurement by atomic absorption spectrometry (AAS) or absorption spectrophotometry (23). The solubility and sorption properties of the metal HEDC complexes make them ideally suited for reversed-phase HPLC. Because the metal complexes are water soluble, they do not need to be extracted into an organic solvent prior to injection. Since there is no excess dithiocarbamate reagent in the mobile phase, there is less potential for interference in the detection of the metal complexes by spectrophotometric methods. The use of HEDC as a precolumn derivatizing reagent provides the basis for the reversed-phase HPLC determination of Co(II), Cu(II), Hg(II), and Ni(I1) by direct aqueous injection. Trace levels of Hg(I1) can be determined after enrichment on an on-line preconcentrator column.
EXPERIMENTAL SECTION Apparatus. Initial investigations on the separation of metal dithiocarbamates were performed by use of a home-built liquid chromatograph. Components of the system consist of a Milton Roy minipump (Laboratory Data Control, Riviera Beach, FL), two 0-6000 PSI pressure gauges, a Valco six-port injection valve (Valco Instruments, Houston, TX), and a Tracor Model 970 variable-wavelength detector (Tracor Instruments, Austin, TX). Pump pulsations are dampened by placing a 10 cm X 4.6 mm i.d. column, packed with 37-44 pm silica, in-line between the two pressure gauges. Final separations and all quantitative work were performed with a Gilson binary gradient HPLC (Gilson International, Middleton, WI). This system consists of two Model 302 pumps, one Model 602 Manometric module, one Model 811 dynamic mixer module, and an Apple IIE based gradient controller. A Kratos Model 783 variable-wavelength detector (Kratos Analytical Instruments, Ramsey, NJ) was used for peak detection. Peak areas and retention times were measured with a HewlettPackard Model 3392 computing integrator (Hewlett-Packard,Palo Alto, CA). A Supelcosil (Supelco, Bellefonte, PA) C-18 reversed-phase column (25 cm X 4.6 mm i.d., 5-mm particle size) was used for the separation of Co(II),Cu(II), Hg(II),and Ni(II). For the specific separation and determination of Hg(II), a Supelcosil C-8 reversed-phase column (7.5 cm X 4.6 cm id., 3 pm particle size) was used. Both columns were protected by a Supelcosil C-18 guard column (2 cm X 4.6 cm id., 5-pm particle size) placed between the injector and the analytical column. The on-line preconcentration of mercury dithiocarbamate was accomplished by using the system shown in Figure 1. A Supelcosil C-18 guard column (2 cm X 4.6 cm id., 5-pm particle size) is placed in the injection loop and is used as the on-line preconcentrator column. In the load position (Figure la), aqueous sample is pumped through the preconcentrator column to waste by an auxiliary pump. When the valve is switched to the inject position (Figure lb), flow through the preconcentrator column is reversed and the adsorbed mercury(I1)dithiocarbamate is backflushed onto the analytical column. Absorption spectra of the metal dithiocarbamate complexes were obtained with a Beckman Model DB-GT grating spectrophotometer (Beckman Instruments, Irvine, CA). Reagents. Ammonium bis(2-hydroxyethy1)dithiocarbamate was prepared by a modification of the procedure described by
Flgure 1. Valving sequence for the on-line preconcentration of Hg(HEDC),:
adsorbed column.
(a) load sample onto preconcentrator column; (b) back flush Hg(HEDC), from preconcentrator column onto analytical
Fritz and Sutton (24). Dissolve 10.5 (0.10 mol) of diethanolamine in 50 mL of methanol. Add 150 mL of tetrahydrofuran and 10 mL (0.15 mol) of concentrated ammonium hydroxide. Cool the solution in an ice bath to below 10 "C and add dropwise, with stirring, 5.0 mL (0.11 mol) of carbon disulfide. Allow the solution to stand for several hours in the ice bath to crystallize the product. Filter, wash with tetrahydrofuran, and dry the crystals under vacuum at ambient temperature. Yield of ammonium bis(2hydroxyethy1)dithiocarbamate is 16.5 g (84%): mp 106 "C. Anal. Calcd for C5H14NzOzS2:C, 30.28; H, 7.12; N, 14.13. Found: C, 30.38; H, 7.14; N, 14.07. Bis(2-hydroxyethy1)dithiocabamate complexes of Ag(I),Cd(II), Co(II), Cu(II), Hg(II), Ni(II), Pb(II), and Zn(I1) were prepared by mixing aqueous metal nitrate and aqueous dithiocarbamate solutions. The resultant precipitates were filtered, washed, dried under vacuum, and recrystallized from ethanol/water. The pure metal dithiocarbamate complexes were dissolved in methanol and the absorption spectrum of each complex was recorded. Stock 1000 mg/L metal ion standards were purchased from Fisher Scientific (Fair Lawn, NJ). Trace metal ion solutions of known concentration were prepared by dilution of metal ion quality control standards supplied by the U.S. Environmental Protection Agency (Quality Assurance Branch, E.M.S.L., Cincinnati, OH). All solutions were prepared using Milli-Q purified water (Millipore, Bedford, MA). Triethylammonium acetate buffer (0.5 M) was prepared by dissolving 25.3 g of triethylamine in 300 mL of water and neutralizing to pH 6.5 with acetic acid. The solution was passed through a 1-cm bed of 74-105-pm Amberlite XAD-4 resin to remove impurities and then diluted to 500 mL with water. Chromatographic eluents were prepared from HPLC grade methanol and acetonitrile. All eluents were made-up to contain 0.025 M triethylammonium acetate buffer and were degassed by helium sparging. Metal ion standards and samples were buffered to pH 6.5 with 0.025 M triethylammonium acetate. A stock solution of 0.10 M HEDC was prepared fresh daily. Generally, 0.25 mL of this solution was added to 25 mL of standard or sample to form the metal dithiocarbamate complexes.
RESULTS AND DISCUSSION The analytical method is based upon the precolumn formation, separation, and subsequent detection of the metal bis( 2-hydroxyethy1)dithiocarbamatecomplexes. The ammonium salt of HEDC was used in this study for the formation of the metal dithiocarbamate complexes because it is easier to prepare and contains less contaminants than the sodium salt, On the basis of qualitative observations of the color of each salt after several months of storage, the ammonium salt also appeared to be more stable. Detection of Metal-HEDC Complexes. The results of our previous investigations on the adsorption of metal-HEDC complexes on XAD-4 resin suggested that the following metal ions may be amenable to HPLC analysis: Ag(I), Cd(II), Co(II), Cu(II), Hg(II), Ni(II), Pb(II), and Zn(I1). The spectrophotometric detection of these metal complexes is based upon their UV-visible absorption spectra, which are shown in Figures 2-5. The reagent has maxima at 290 nm (t 14 000) and 260 nm (t 12000). The simultaneous determination of Co(HEDC),, CU(HEDC)~, and Ni(HEDC), can be achieved hy monitoring the absorbance at 405 nm. All of the metal-
ANALYTICAL CHEMISTRY, VOL. 59, NO. 5, MARCH 1, 1987
50
705
~
I
1
I
I
I
I
8
WAVELENGTH (nm)
Flgure 2. Absorption spectra of bis(2-hydroxyethy1)dithiocarbamate complexes of Ag(1) and Cd(I1).
I
I
I
I
I
I
I
I
(
Flgure 5. Absorption spectra of bis(2-hydroxyethyl)dithiocarbamate complexes of Cu(I1) and Ni(I1).
Table I. Dependence" of log k'on Eluent Methanol Concentration element
correlation
slope
Y intercept
Co(I1) Cu(I1) Hg(W Ni(I1)
0.9998 0.9999 0.9996 0.9997
-6.70
2.62 2.63 2.95 2.38
-5.28 -5.70
-5.31
Separation conditions: column, 25 cm X 4.6 cm id., 5-pm, C18; eluents, methanol/water containing 0.025 M TEAA. Parameters are for the straight line equation: log 12' = m(methano1 fraction) + b.
WAVELENGTH ( n m )
Figure 3. Absorption spectra of bis(2-hydroxyethyl)dithiocarbamate complexes of Pb(I1) and Zn(I1).
"c 40
PO0
240
280
320
360
400
440
480
520
560
WAVELENGTH ( n m l
Figure 4. Absorption spectra of bis(2-hydroxyethyl)dithiocarbamate complexes of Co(I1) and Hg(I1).
HEDC complexes can be detected by monitoring the absorbance at 255 nm. Separation of Metal-HEDC Complexes. Initial investigations on the separation of CO(HEDC)~, Cu(HEDC)*,and Ni(HEDC)z showed that these complexes could be well-resolved, with good peak shape, using binary mixtures of methanol and water. However, it was noted that after several days of operation both peak shape and resolution began to deteriorate significantly. This was due to ligand-exchange reactions with trace amounts of metal oxides present in the chromatographic system. This problem has also been pointed out by several other workers ( I I , 1 5 , 2 5 - 2 7 ) . The problem was overcome by a two-step process. First, with the column off-line, stainless steel components of the HPLC system were passivated by flushing the system with a 5% (v/v) nitric acid solution. Second, with the column on-line, a 0.01 M solution
of disodium ethylenediaminetetraacetate was pumped through the system. It was also noted that the peak areas of the separated metal complexes varied considerably (>20%) with repeated injections of the same sample. It is believed that these variations are from interaction of the metal complexes with residual silanol groups on the C18 column packing. This has been observed for the reversed-phase separation of other polar compounds (28-30). The addition of 0.025 M triethylammonium acetate (TEAA) buffer effectively blocks the effects of the residual silanols, as shown by eq 1-3. The use of chromatographic eluents containing TEAA buffer gave reproducible quantitation of the metal complexes.
+ Et3"+ + -Si-O-+HNEt3 -Si-OH + Et3N * -Si-O-+HNEt3 -Si-OH + AcO- + -Si-O--.H+---OAc -Si-0-
(1)
(2) (3)
Studies were performed by use of preformed aqueous solutions of the eight metal-HEDC complexes with both methanol/water and acetonitrile/water eluents. The complexes of Cd(I1) and Pb(I1) gave ill-defined, irreproducible shoulders on the reagent peak. No peak was observed for the Zn(I1) complex. The Ag(1) complex could not be resolved from the excess reagent peak with either acetonitrile or methanol based eluents. The complexes of Co(III), Cu(II), Hg(II), and Ni(I1) were suitable for chromatographic analysis using methanol/water eluents. The dependence of the capacity factor, k', on the mobile phase composition for the four metal complexes is shown in Table I. A typical separation is shown in Figure 6. When acetonitrile-based eluents were used for the separation, the peak areas of the four metal complexes were reduced by as much as 40% compared to methanol-based eluents. The capacity factors were able reduced considerably more than would be predicted from published solvent isoeluotropic series for reversed-phase separations (31). Thus, a 15% acetonitrile
706
ANALYTICAL CHEMISTRY, VOL. 59, NO. 5, MARCH 1, 1987
i
0.010
Table 11. Detection Limits" for Metal-HEDC Complexes detection limit mass, ng concn, mg/L
element Co(I1) Cu(I1) Ni(I1)
W
0.005 0.007 0.009
0.05 0.07 0.09
V
5
"Based upon a 1O-fiL injection. Conditions are as described in Figure 7.
0006-
m
U 0
2
0004-
a
Table 111. Precision Data" for Determination of Co(II), Cu(II), and Ni(I1)
k
u w -Y 0002-
0 000
P-dJU I l l
I
I
2
6
8 1012
I
Co(I1)
area counts Cu(I1)
Ni(I1)
1
199 650 209 100 203 170 206 880 208 110 202 560 208 050 206 670 206 230 ,209 270
174 110 176 870 168 750 172 600 165 740 167 540 176 240 173 690 172 220 165 650
115 760 128 700 123030 118 170 125 010 122 620 120 170 122 810 117600 118620
205 969 3 169 1.54
171341 4 150 2.42
121 240 3 932 3.24
2 3 4 5 6 7 8 9 10
I
I
0
4
injection
TIME (minutes)
Figure 6. Separation of metal-HEDC complexes: 0.40 ng/pL each of Co(II), Ni(II), Cu(I1) and 1.6 ng/pL Hg(I1); 10-pL sample; eluent, 40% methanol, 0.025 M TEAA; flow rate, 1.0 mLlmin; detection at 255 nm.
O'Oo5
t
1
I
SD
cv, 9c
"Injection of 10 pL; standard contains 0.25 mg/L each of Co(II), Cu(II), and Ni(I1). Chromatographic conditions are as for Figure 7.
Table IV. Analysisn of Electroplating Wastewater for Co(II), Cu(II), and Ni(I1)
E 0002 a
0 000
mean
U ~.--U u 0
2
4
6
8
1012
TIME (minutes)
Figure 7. Separation of metal-HEDC complexes: 0.50 ng/pL each of Co(III), Ni(II), and Cu(I1); 10-pL sample; eluent, 45% methanol, 0.025 M TEAA; flow rate, 1.0 mL/min; detection at 405 nm. eluent is required to give the same k'values as a 45% methanol eluent. This is significantly less than the 36% acetonitrile predicted by the solvent isoeluotropic series. This effect has not been reported previously with other dithiocarbamate reagents. The apparent interference of acetonitrile with the separation process requires that only methanol-based eluents be used for quantitative work. The poor peak shapes for Cd(I1) and Pb(II), and the lack of a peak for Zn(II), are consistent with the results of other studies which have used other dithiocarbamate reagents (25). These metals form relatively unstable dithiocarbamate complexes and are apparently broken-up during the separation process. It should be noted, however, that other workers have successfully separated these metal ions by using other dithiocarbamates (29, 20, 22). Determination of Co(II), Cu(II), and Ni(I1). Co(II), Cu(II), and Ni(I1) can be quantitatively determined by direct aqueous injection after addition of HEDC to the sample. The solubility of the metal complexes eliminates the need for a preliminary extraction step into an organic solvent. The metal complexes are detected at 405 nm, which minimizes the possibility of interferences from UV absorbing organic com-
element
initial mg/Lb
mg/L added
Co(II1) Cu(I1) Ni(I1)
BDL' 0.13 f 0.01 0.57 f 0.02
0.26 0.34 0.21
mg/L found recovery, 70 0.26 f 0.01 0.46 k 0.01 0.79 f 0.03
100 & 4 97 3 105 1.2
* *
Chromatographic conditions are as described in Figure 7 . Average and standard deviation for five determinations. Below detection limit.
pounds when using a lower wavelength. A typical separation is shown in Figure 7 . Simultaneous calibration plots were prepared for the three metals over the concentration range of 0.02-10.0 mg/L. Excellent linearity is observed over approximately 3 orders of magnitude. Detection limits, expressed as both absolute mass units and concentration units, are shown in Table 11. These were determined by using a signal-to-noise ratio of 3:l. These are below previously reported detection limits for these metals using other dithiocarbamates (12,15,19). Precision data for ten replicate injections of a standard containing 0.25 mg/L each of Co(II), Cu(II), and Ni(I1) are shown in Table 111. The coefficients of variance are within expected limits of error for this type of analysis. To show the applicability of the method, a sample of electroplating wastewater from the waste treatment plant of a major home appliance manufacturer was analyzed. The sample was also spiked with the USEPA standard and recoveries of Co(II), Cu(II), and Ni(I1) were calculated. Results are shown in Table IV. These results were compared to those obtained from analysis of the samples of AAS and were found to be in good agreement. Preconcentration and Determination of Hg(I1). Mercury is one of the most toxic metals known and is considered a cumulative poison in the environment (32, 33). For
ANALYTICAL CHEMISTRY, VOL. 59, NO. 5, MARCH 1, 7
W
0007
-
0006
-
0005
-
0 024
1987 707
r--J
U
z U m a
0004-
0 v)
m
a
0003
-
0002
c
0001
- gz-
0
2
4
6
8
1012
14
TIME (minutes)
0 000
u
Flgure 9. Elution of Hg(HEDCl2 standard after on-line preconcentration:
TIME (minutes)
M TEAA, 0.001 M EDTA; detection at 275 nm; preconcentration and separation flow rate, 1.0 mL/rnin.
0
2
4
6
8
Flgure 8. Elution of Hg(HEDC)? standard: direct aqueous injection of 10 pL; 1.00 mg/L Hg(I1); eluent, 37% methanol, 0.025 M TEAA, 0.001 M EDTA; flow rate, 1.0 mLlmin; detection at 275 nm.
these reasons there is continuing interest in the determination of trace levels of mercury in environmental samples. Of the four metal-HEDC complexes studied, the Hg(HED02 complex has the strongest retention on a reversed-phase column. Thus, chromatographic and detection conditions can be adjusted to provide a quick and specific method for the determination of Hg(I1) in the presence of other metals. Chromatographic conditions are adjusted such that the Co(II), Cu(II), and Ni(I1) complexes are eluted at or near the void volume, while the Hg(HEDCI2 complex is quickly eluted as a sharp peak. The use of a short, 3-pm particle size C8 analytical column provides adequate resolution and a relatively short analysis time. A typical separation using direct aqueous injection is shown in Figure 8. The detection limit for Hg(I1) using direct injection (10-pL sample) is approximately 10 pg/L, which is considerably higher than the levels of interest in most environmental samples. The Hg(I1) detection limit can be lowered considerably by an on-line preconcentration step prior to the separation. Connection of a preconcentrator adsorption column with the analytical column by a switching valve is the most common arrangement (34). The aqueous sample is pumped through the preconcentrator column, where the Hg(HEDC)2 complex is enriched a t the top of the column. EDTA is added to the eluent to mask high concentrations of other metal ions which would otherwise react with HEDC. The switching valve is next rotated and mobile phase flows through the preconcentrator column in the reverse direction. The Hg(HEDC)2 complex is back flushed onto the analytical column for separation and detection. Figure 9 shows the preconcentration and analysis of 5.00 mL of a 10.0 pg/L Hg(I1) standard. There is virtually no loss in analytical column performance as a consequence of the essentially pluglike injection of the Hg(HEDC)2from the back flushed preconcentrator column. A plot of sample volume vs. Hg(HEDC), peak area for a 10.0 ccg/L Hg(II) standard is linear over the range of 2-10 mL. Thus, sample size can be adjusted to either increase or decrease the effective concentration range of the method. A calibration plot of Hg(I1) with a 5.00-mL sample size is linear over the concentration range of 0.05-25 pg/L, which is the concentration range of interest in many environmental samples. By use of a 10.0-mL sample size, a detection limit
10.0 pg/L Hg(I1); 5.00-mL sample size; eluent, 35% methanol, 0.025
Table V. Analysis of Electroplating Wastewater for Hg(II)o a m t found, initial
gg/L
BDLb
a m t added,
0.67
pg/L
rg/L
recovery, %
0.66 f 0.01
98 f 2
aChromatographic conditions as in Figure 9 Average a n d standard deviation for five determinations. * B e l o w detection lim-
it.
of 0.02 pg/L was obtained. This is a 1000-fold improvement over using direct aqueous injection. Precision data were obtained for ten replicate determinations of a 10.0 pg/L standard after preconcentration of a 5.00-mL sample size. The coefficient of variance for these determinations is 1.8%. A sample of electroplating wastewater was spiked with USEPA standard to check the method on a real sample. Results are shown in Table V. Quantitative recovery, with good precision, was obtained for the determination of Hg(I1). Registry NO.HEDC, 75074-70-3;CO,7440-48-4;CU,7440-50-8 Hg, 7439-97-6; Ni, 7440-02-0; H20, 7732-18-5.
LITERATURE CITED (1) (2) (3) (4) (5) (6) (7) (8)
(9) (10) (11) (12) (13) (14) (15)
(16) (17) (18) (19) (20) (21) (22) (23)
Willeford, B. R.; Veening, H. J. Chromatogr. 1982, 2 5 1 , 61-88. O'Laughlin, J. W. J . Llq. Chromatogr. 1984, 7 , 127-204. Nickless, G. J . Chromatogr. 1985, 313, 129-159. Uden, P. C.; Bigley, I. E. Anal. Chlm. Acta 1977, 94, 29-34. Heizmann, P.; Balischmlter, K. J. Chromatogr. 1977, 137, 153-163. Liska, 0.; Guiochon, G.; Colin, H. J. Chromatogr. 1979, 171, 145-151. Liska, 0.; Lehotay, J.; Bradsteterova, E.; Guiochon, G. J. Chromatogr. 1979, 171, 153-159. Lehotay, J.; Liska, 0.; Brandsteterova, E.; Guiochon, G. J . Chromatogr. 1979, 172, 379-383. Liska, 0.; Lehotay, J.; Brandsteterova, E.; Guiochon, G.; Colin, H. J. Chromatogr. 1979, 172, 384-387. O'Laughlin, J. W.; O'Brian, T. P. Anal. Lett. 1978, All(10), 829-844. Haring, N.; Baiischmiter, K. Talanta 1980, 2 7 , 873-879. Shih, Y.; Carr, P. W. Anal. Chlm. Acta 1982, 142, 55-62. Schwedt, G. Chromatographla 1979, 12, 289-293. Borch, R. F.: Markovitz. J. H.: Pieasants, M. E. Anal. Lett. 1979, 12(BE), 917-926. Hutchins, S. R.; Haddad, P. R.; Dilli, S. J. Chromatogr. 1982, 2 5 2 , 185-192. Bond, A. M.; Wallace, G. G. Anal. Chem. 1981, 5 3 , 1209-1213. Bond, A. M.; Wallace, G. G. Anal. Chem. 1982, 5 4 , 1706-1712. Bond, A. M.; Wallace, G. G.Anal. Chem. 1983, 5 5 , 718-723. Bond, A. M.; Wallace, G. G. Anal. Chem. 1984, 5 6 , 2085-2090. Ichinokl, S.; Yamazaki, M. Anal. Chem. 1985, 5 7 , 2219-2222. Mueiier, 8. J.; Lovett, R. J. Anal. Chem. 1985, 5 7 , 2693-2699. Smith, R. M.; Butt, A. M.; Thakur, A. Analyst (London) 1985, 110, 35-37. King, J. N.; Fritz, J. S. Anal. Chem. 1985, 5 7 , 1016-1020.
700
Anal. Chem. 7907, 59,708-712
(24) Fritz, J. S . ; Sutton, S. A. Anal. Chem. 1956, 28, 1300-1303. (25) ShR, Y. T.; Carr, P. W. Talanta 1981, 28, 411-414. (26) Vespelec, R.; Neca, J. J . Chromtogr. 1983, 287, 35-47. (27) Verzele, M. LC Mag. 1983, 1(4),217-216. (28) Rivler, J. E. J . Liq. Chromatogr. 1978, 7 , 343,366. (29) Hearn, M. T.: Greco. B. J . Chromatoor. 1983. 255. 125-136. (30) Hearn, M. T.; Grego, B. J . Chromato&. 1983; 266; 75-87. (31) Schoenmakers, P. J.; Billlet, J. A.; DeGalan, L . J . Chromatogr. 1981, 218, 261-284. (32) Browning, E. Toxicity of Industrial Metals ; Butterworth: London, 1969.
(33) Luckey, T. D.; Venugopal, B. Metal Toxicity in Mammals; Plenum Press: New York, 1977. (34) Karger, 6. L.; Giese, R. W.; Snyder, L. R. Trends Anal. Chem. 1983, 2(5). 106-109.
RECEIVED for review June 2, 1986. Accepted November 1, 1986. The financial support provided by the Maytag Company, Newton, IA, is gratefully acknowledged.
Determination of Bicarbonate by Ion-Exclusion Chromatography with Ion-Exchange Enhancement of Conductivity Detection Kazuhiko Tanaka’ and James S. Fritz*
Ames Laboratory-DOE
and Department of Chemistry, Iowa State University, Ames, Iowa 50011
Carbon dloxlde and blcarbonate are determlned In aqueous samples by lon-excluslon chromatography uslng water as the eluent and a conductlvky detector. The sensltlvtty of detectlon Is Improved approxlmately 10-fold by the use of two ion-exchange “enhancement” cdumns Inserted In S e r b between the separatlng column and the detector. The flrst enhancement column converts carbonic acld to potasdum blcarbonate and the second enhancement column converts the potasslum bicarbonate to potasslum hydroxlde. The method provides a fast, selectlve, and sensttlve way to determlne carbon dioxlde or bicarbonate. Appllcatlon to several types of aqueous samples Is demonstrated.
The quantitative determination of carbon dioxide and bicarbonate is a very important analytical problem, especially when low concentrations are to be measured. Recently, a simple method for the determination of CO2/HCO3- based on a conductometric sensor with a gas-permeable membrane has been developed (1, 2). Although the method is very sensitive to COz/HC03-, the detector response to concentration is not linear. An acidic or basic gas, such as SO2 or NH3, interferes in the C 0 2 determination. Ion-exclusion chromatography provides a convenient way to separate molecular acids from highly ionized substances. The separation column is packed with a cation exchange resin in the H+ form so that salts are converted to the corresponding acid. Ionized acids pass rapidly through the column while molecular acids are held up to varying degrees. A conductivity detector is commonly used. Carboxylic acids have been separated by using water, a dilute mineral acid, or a dilute benzoic or succinic acid as the eluent (3-6). Carbon dioxide or bicarbonate has been determined by ion-exclusion chromatography with water as the eluent and a coulometric detection of H+ from H2C03(7). The method is reasonably selective for C02/bicarbonate and has a linear calibration curve but is somewhat lacking in sensitivity. In the present work, methods for determination of the sum of carbon dioxide and sodium bicarbonate that use ion-exclusion chromatography and a conductivity detector are Present address: Government Industrial Research Institute,
Nagoya, 1-1,Hirate-cho, Kita-ku, Nagoya-shi, Aichi, 462, Japan.
presented. The simplest procedure uses a cation-exchange separation column with purified distilled water as the eluent. However, the use of ion-exchange “enhancement” columns connected in series with the separation column greatly improves the sensitivity of detection and gives a linear calibration curve over a large concentration range.
EXPERIMENTAL SECTION Apparatus. A rather conventional chromatographic system was used, which consisted of the following components: eluent reservoir, pump, pulse dampener, precolumn for removing COz gas from the eluent, a 100-pL sample loop, separating column, first and second enhancement ion-exchange columns, a Wescan 213 conductivity detector, and a strip-chart recorder. A flow rate of 1.0 mL/min was used for all of the work reported. The plastic separating column was 7.5 x 100 mm and was packed with a cation-exchange resin in the H+ form TSK SCX, 5 pm (TSK, Tokyo, Japan). The first enhancement column was constructed of plastic (4.6X 50 mm) and packed with a cationexchange resin in the K+ form (TSK SCX, 5 pm;TSK IC-Cation for cation chromatography use, 10 pm; TSK SP-5 PW for HPLC use, 10 pm; or TSK IC-Cation SW for cation chromatography use, 5 pm). The second enhancement column was constructed of plastic (4.6 X 50 mm) and packed with an anion-exchange resin in the OH- form, TSK SAX (5 pm). The precolumn was constructed of plastic (7.5 x 100 mm) and packed with an anion-exchange resin in the OH- form (TSK SAX, 5 pm). Reagent and Chemicals. Standard solutions of bicarbonate were prepared by using NaHC03 of reagent grade. Other standard solutions used were of reagent grade. All standard solutions were prepared by using distilled, deionized water from a Mi&-Q reagent grade water system (Millipore, Bedford, MA). Sample Preparation. All actual samples were filtered with a 0.45-pm poly(tetrafluoroethy1ene) (PTFE)membrane filter before injection into the column. RESULTS AND DISCUSSION Determination of Carbon Dioxide or Bicarbonate by Ion Exclusion. A 100-pL sample containing 1.0 mM NaHC03and 1.0 mh4 KCl was injected to test the separation ability of the system used. Distilled water as the eluent and conductivity detection were employed. The separation column contains a cation exchanager of high capacity and thus converts the sodium bicarbonate to carbonic acid and the potassium chloride to hydrochloric acid. Good resolution of the H+-C1- and H2C03was obtained, the retention times being approximately 2 min and 5 min, respectively. However, HzC03
0003-2700/87/0359-0708$01.50/00 1987 American Chemical Society
