Determination of Chloride, Nitrate, Sulfate, Nitrite, Fluoride, and

Oct 15, 1994 - National Water Research Institute, Environment Canada, 867 Lakeshore Road, ... Capillary zone electrophoresis (CZE) on-line coupled wit...
0 downloads 0 Views 725KB Size
Download PDF
Anal. Chem. 1994, 66, 4258-4264

Determination of Chloride, Nitrate, Sulfate, Nitrite, Fluoride, and Phosphate by On-Line Coupled Capillary Isotachophoresis-Capillary Zone Electrophoresis with Conductivity Detection Dugan Kaniansky,*st lmrich Zelensky,t Anna Hybenov8,t and Francis 1. Onuska: Department of Analytical Chemistry, Faculty of Natural Sciences, Comenius University, Mlynska Dolina CH-2, 842 15 Bratislava, Slovak Republic, and Aquatic Ecosystem Protection Branch, National Water Research Institute, Environment Canada, 867 Lakeshore Road, P. 0. Box 5050, Burlington, Ontario L7R 4A6, Canada

Capillaryzone electrophoresis (CZE)on-linecoupled with capillary isotachophoresis (lTP) in the column-coupling configuration of the separation unit was studied for determination of nitrite, fluoride, and phosphate persent in a large excess of chloride, nitrate, and sulfate. Such distributions of these anions are typical for many environmental matrices, and it is shown that the lTP-CZE tandem enables the lTP determination of the macroconstituents, while ITP-preconcentrated microconstituents cleaned up from the macroconstituentscan be determined by CZE with a conductivity detector. This approach was effectivewhen the concentration ratio of macro-microconstituents was less than (2-3) x lo4. The limits of detection achieved for the microconstituents (200,300, and 600 pptr for nitrite, fluoride, and phosphate, respectively) enabled their determinationsat low parts per billion concentrations in instances when the concentrations of sulfate or chloride in the samples were higher than 1000 ppm. Although in such situations the load capacity of the ITP column was not suEcient to provide the quantitations of the macroconstituents,the ITP sample cleanup enabled the CZE determinations of the microconstituents present in the samples at lo5- 106-fold lower concentrations. Experiments with samples of such extreme compositions showed that anionic impurities from the lTP electrolyte solutions transferred together with the analytes into the CZE column are probably the main hindrances in extending this dynamic concentration range by decreasing the l i i t s of the detection to a level associated with the noise characteristics of the conductivity detector. At present, ion chromatography (IC) has a dominant position among the separation methods used for the analysis of inorganic anions (see, e.g., refs 1-4). However, when the separation +

Comenius University.

* National Water Research Institute, Environment Canada. (1) MacCarthy, P.; Klusman, R W.; Rice, J. A Anal. Chem. 1989,61,269R (2) Frankenberger, W. T., Jr.; Mehra, H. C.; Gjerde, D. T.J. Chromatogr. 1990, 504,211. (3) MacCarthy, P.;Klusman, R W.; Cowling, S. W.; Rice, J. A Anal. Chem. 1991,63,301R (4) Tarter, J. G., Ed. Ion Chromatography; M. Dekker; New York and Basel, 1987.

4258

Analytical Chemistty, Vol. 66, No. 23, December 1, 1994

efficiencies typically achieved by IC are compared to those characterizing capillary zone electrophoresis (CZE) for the same group of analytes? it is apparent that the latter technique has advantages. As the effective mobilities of inorganic anions differ sufficiently due to differences in their ionic mobilities5j6and/or can be influenced in a desired way via appropriately chosen pH (differences in pKavalues) and/or via the use of suitable nonionic additives,7s8the selectivity factorsgare also favorable in achieving rapid resolutions of the analytes by CZE. Although in some instances it is possible to determine very low concentrations of inorganic anions by CZE,'0J1 problems occur when the concentrations of the sample constituents differ considerably. This is due to the fact that the determination of microconstituents may require the sample load to have a negative impact on both the migration velocities and the resolutions of the analytes.12 The use of indirect detection, as preferred in CZE of inorganic anions (see, e.g., refs 5, 10, l l ) , is also less favorable in achieving adequate load capacities since low detection limits are met through a low concentration of the carrier constituent.13 In this context, it is to be noted that the load capacity limits may hinder the CZE determination of inorganic anions also in instances when the analytes are detected by the detection techniques having no prior restrictions with respect to the concentration of the carrier electr01yte.l~ Of the inorganic anions which are currently monitored in water, chloride, sulfate, and in many instances also nitrate can be considered anionic macroconstituents. These are typically accompanied by nitrite, fluoride, and phosphate present in water samples at 102-105-fold lower concentrations. Such a wide concentration span probably cannot be covered by CZE, especially when restrictions of the detection in the indirect mode are taken (5) Jones, W. R ; Jandik, P. J. Chromatogr. 1991,546,445.

(6) Hirokawa, T.; Nishmo, M.; Aoki, N.; Kiso, Y.; Sawamoto, Y.; Yagi, T.; Akiyama, J. I. J. Chromatogr. 1983,271,D1. (7)Fukushi, K; Hiiro, K J. Chmmatogr. 1990,518, 189. (8) Madajovl, V.; Turcelovl, E.; Kaniansky, D. J. Chromotogr. 1992,589,329. (9) Foret, F.; BoEek, P. Adu. Electrophor. 1989,3,271. (10) Bandoux, G.; Jandik, P.; Jones, W. R J. Chromotogr. 1992,602,79. (11) Jackson, P. E.; Haddad, P. R J. Chromatogr. 1993,640,481. (12) Beckers, J. L.; Everaerts, F. M. J. Chromatogr. 1990,508,3. (13) Yeung, E. S. Acc. Chem. Res. 1989,22,125. (14) Amran, M. B.; hkkis, M. D.; Lagarde, F.; Leroy, M. J. F.; Lopez-Sanchez. J. F.; b u r e t , G. Fresenius J. Anal. Chem. 1993,345,420. 0003-2700/94/0366-4258$04.50/0 0 1994 American Chemical Society

nilrile fluoride

phosphate rulphale

i

Figure 1. Schematic illustration of the ITP determination of chloride, nitrate, and sulfate and the CZE determination of nitrite, fluoride, and phosphate in column-coupling CE instrument. 1, ITP separation of the anions in the ITP column (ITP stage); 2, removal of the macroconstituents (chloride, nitrate, sulfate) from the separation compartment; 3, transfer of the sample microconstituents (nitrite, fluoride, phosphate) into the CZE column (after the microconstituents,focused in the boundary layer between the sulfate and terminating zones, reach the position as shown in step 2, a high-voltage relay is switching the columns so that the driving current flows through the column); 4, CZE separation of the microconstituents (CZE stage). L, T, C refer to leading, terminating, and carrier electrolytes, respectively; D-ITP, D-CZE are detectors for the ITP and CZE columns, respectively; ITP is the ITP column; CZE is the CZE column; BF is the bifurcation block; E-TE is the direction to the terminating electrode; E-ITP is the direction to the counter electrode of the ITP column; E-CZE is the direction to the counter electrode of the CZE column; i-ITP is the direction of the driving current in the ITP stage; and i-CZE is the direction of the driving current in the CZE stage.

into account.15 It appears that at present CZE offers for this analytical problem only a moderate improvement over the early days of this technique.I6 Capillary isotachophoresis OTP) in the column-coupling configuration of the separation unit provides a two-step procedure (integrated into one electrophoretic run) for a simultaneous determination of the anionic macro- and microconstituents in water.17 However, its use is limited when the concentrations of the macroconstituents are ca. lO3-fold higher than those of the microconstituents. Two independent ITP runs provide a way to perform these measurements when the concentration ratio of macro-/microconstituents is in the range 103-104.1s For higher values of this ratio, the determination of the microconstituents is possible only when a suitable sample preparation procedure is employed. For fluoride and phosphate, such procedures which are compatible with the ITP determinations are described in the literat~re.l~,~O However, their use is both time and labor intensive. It was shown recently that an on-line combination of ITPwith CZE in the column-coupling capillary electrophoresis instrument is a convenient capillary electrophoresis (CE) alternative to the analysis of trace ionogenic constituents present in a large excess (15) Foret, F.; Fanali, S.; Ossicini, L;BoEek, P.J. Chromatogr. 1989,470,299. (16)Gebauer, P.;Deml, M.; BoEek, P.; Jan& J.J. Chromatogr. 1983,267,455. (17) Zelenskjr, I.; Zelenska, V.; Kaniansky, D.; HavaSi, P.; Lednarovd, V. J. Chromatogr. 1984,294,317. (18)TkPEova, J.; Zelenskjr, I.; Kaniansky, D. In Hydrochemia 85; BilikovA, A, Hucko, P., Lehockjr, J., Stankovif, V., Eds.; Water Research Institute: Bratislava, Slovakia, 1985;p 421. (19) Fukushi, K; Hiiro, K Bunseki Kagaku 1985,34,21. (20) Fukushi, K; Hiiro, K Bunseki Kagaku 1985,34,205.

of matrix ions.21-29 Analytical capabilities of such a combination are apparent from the following facts. (i) In the ITP column (capable of accommodating sample volumes up to several hundred microliters), the sample constituents are separated into a stack of the zones with trace analytes focused into a narrow band($. (ii) As matrix constituents are forced to migrate out of the separation compartment at the end of the I"column, the sample is cleaned up before its injection onto the CZE column. This removal of the matrix is well defined and very reproducible when it is based on the signal from the detector used to monitor the 1°F' separation (see, e.g., ref 28). (iii) The trace analytes present in the injected sample, concentrated and cleaned up from the matrix in the ITP stage (column), are completely transferred onto the CZE column as a narrow sample pulse. (iv) An overall selectivity of the analysis can be enhanced when (21)Kaniansky, D.;Mar&, J. J. Chromutogr. 1990,498, 191. (22) Stegehuis, D.S.; Irth, H.; Tjaden, U. R; van der Greef, J. J. Chromatogr. 1991,538,393. (23)Kfivinkova, L.;Foret, F.; BoEek, P.J. Chromatogr. 1991,545,307. (24)Stegehuis, D.S.; 'lladen, U. R; van der Greef, J.J. Chromatogr. 1992,591, 341. (25)Foret, F.; Szoko, E.; Karger, B. L. J. Chromatogr. 1992,608,3. (26)Reinhoud, N.J.; Tinke, A P.; 'Ijaden, U. R; Niesen, W. M. A; van der Greef, J. J. Chromatogr. 1992,627, 263. (27)Hirokawa, T.; Okushi, A; Kiso, Y. J. Chromatogr. 1993,634,101. (28) Kaniansky, D.; Mar&, J.; Madajova, V.; &munitova, E.J. Chromatogr. 1993, 638,137. (29)Kaniansky, D.;Ivhyi, F.; Onuska, F. I. Anal. Chem. 1994,66,1817.

Analytical Chemistry, Vol. 66, No. 23, December 1, 1994

4259

Table 1. Electrolyte Systems.

Table 3. Maximum Tolerable Concentrations (ppm) of Chloride and Sulfate for Varying Concentrations of Nitrate (ppm) in the ITP Determination

electrolyte ITP-1

parameter solvent anion concentration (mM) counterion cc-counterion concentration (mM) PH additive concentration (%,w/v)

ITP-2

L

T

L

T

C

Hz0 C1-

H2O Asp

HzO

HzO

HzO Asp

8 BALA BTP 3 3.4

C14 8 BALA BALA 4

Asp

3,4

4

3,4

MHEC

MHEC

0.2

02

02

Abbreviationsused: MHEC, methylhydroxyethylcellulose;B N P-alanine; Asp, aspartic acid; BTP, bis-tris propane, 1,5bis[tris(hydrolfymethy1)methylaminolpropane; L, leading; T, terminating; C, camer. Table 2. Parameters of the Regression Equations for the Calibration Lines, Limits of Detection, and Reproducibilities of the ITP Determinations of Chloride, Nitrate, and Sulfate.

anion chloride nitrate sulfate

a

0.56 -0.11 0.49

0.4279 0.2635 0.319

0.9984 0.9998 0.9999

LOD (ppm)

RSD (%)

5 2 1

4.2 1 0.5

Abbreviations used: a, intercept; b, slope; CC, correlation coefficient; LOD, limit of detection; RSD, median of the relative standard deviations for five samples (each sample analyzed five times). The calibration data were obtained from 60 runs for calibration mixtures containing chloride in the range 20-90 ppm, nitrate in the range 5-40 ppm, and sulfate in the range 25-125 ppm.

the separations in the ITPand CZE columns are based on different separation mechanis"* The aim of this work was to study analytical capabilities of the JTP-CZE tandem to determine chloride, nitrate, sulfate, nitrite, fluoride, and phosphate in the concentration range within which these anions are found in water samples. In previous work, the ITP stage was used merely for the cleanup of the samples and/or for the preconcentration of the trace analytes. Here, we extend the use of this stage to the determination of sample macroconstituents (chloride, nitrate, sulfate) while maintaining the cleanup and preconcentration capabilities. The CZE stage is intended to perform final separation of the trace anions (nitrite, fluoride, phosphate) with their detection by a conductivity detector. EXPERIMENTAL SECTION

Instrumentation. A CS isotachophoretic analyzer (VillaLabeco, SpiSskP Nova Ves, Slovak Republic) that enables CE separations in the columncoupling mode was used. This analyzer is provided with two conductivity detectors (to monitor the separations in both columns) and with a W absorbance detector. The separation unit of the analyzer was assembled from components provided by the manufacturer and also from those developed in this laboratory. The ITP column (Villa-Labeco) was provided with a 0.85 mm i.d. capillary tube (0.d. 1.15 mm) made of fluorinated ethylenepropylene (FEP) copolymer. The length of the capillary tube to

-

4260

5 10 20 30 40

4 10 BAL4 BALA

MHEC

regression equationb (s) 6 (s/ppm) CC

nitrate concn

Analytical Chemistry, Vol. 66, No. 23, December 1, 1994

maximum coned chloride sulfate

a

100 75 75 75 50

125 125 125 100 125

When both anions are present in the sample.

the conductivity cell was 150 mm. The CZE columns (VillaLabeco) used a 0.30 mm i.d. (0.d. 0.65 mm) capillary tube made of FEP (the length from the inlet to the conductivity cell was 150 mm) and a 0.35 mm i.d. (0.d. 0.70 mm) capillary tube made of the same material (the length from the inlet to the detection cell was 150 mm). The capillary tubes were placed in compartments made of Plexiglas that allowed efficient dissipation of heat produced on the passage of current. The samples were injected with the aid of a sampling valve with a 30 p L internal sampling loop. The data from the detectors were acquired and processed by an ITPSOFT. program (KasComp, Bratislava, Slovak Republic) which runs on a 386 PC computer. Chemicals. Chemicals used for the preparation of the electrolyte solutions were obtained from Serva (Heidelberg, Germany), Sigma (St. Louis, MO), Reanal (Budapest, Hungary), and Lachema (Bmo, Czech Republic). Some of them were purified by conventional methods, and analytical grade chemicals were used as delivered. Methylhydroxyethylcellulose 30 000 (MHEC) served as an anticonvective additive in the electrolyte solutions. A stock aqueous solution containing 1%(w/v) of the derivative was purified on a mixed-bed ion exchanger (Amberlite MB-1; Serva). Water from an Aqualabo two-stage demineralization unit (Aqualabo, Bmo, Czech Republic) was further purified by circulation in a laboratory-made demineralization system made of poly(tetrafhoroethylene) and packed with Amberlite MB-1 mixed-bed ion exchanger (Serva). At the outlet, water of ca. 0.06 pS-cm-' specfic conductivity was filtered through a 0.45 pm filter (Gelman, Ann Arbor, MI).

-

-

RESULTS AND DISCUSSION

SeparatingConditions. As mentioned above, in our approach the ITP determination of chloride, nitrate, and sulfate was combined with the CZE determination of nitrite, fluoride, and phosphate using the ITP-CZE tandem in the column-coupling system. This tandem offers several working modes,2lSz8and the one preferred in this work consists of the following steps (see also Figure 1). (1)Chloride, nitrate, and sulfate are separated by ITPinto their zones, while nitrite, fluoride, and phosphate are focused as a group into a narrow band migrating in the JTP stack behind the zones of macroconstituents. This is a favorable migration configuration as far as an efficient use of the ITP column is concerned.31 The (30) Kfivankovi, L.; Gebauer, P.; Thorman, W.; Mosher, R A; BoEek, P. 1. Chromatogr. 1993,638, 119.

Table 5. Regression Equations of Calibration Lines and Reproduclbilitiesof CZE Determinations of Nitrite, Fluorlde, and Phosphate.

anion

a

regression equation (mVs) b (mV*s/ppb)

nitrite fluoride phosphate

1.38 2.71 0.86

0.153 0.597 0.0987

RSD (%) B

CC

A

0.9994 0.9999 0.9999

0.87 0.7 2.21

0.95 1.11 2.36

Abbreviations used: a, intercept; b, slope; CC, correlation coefficient; RSD, relative standard deviation; A, model sam le containing 15 ppb of nitrite, 15 ppb of fluoride, and 30 ppb of phospgate; B, model sample containing 20 ppb of nitrite, 20 ppb of fluoride, and 40 ppb of phosphate. The calibration solutions contained 5-375 ppb of nitrite, 5-375 ppb of fluoride, ane 10-750 ppb of phosphate, and the regression equations were calculated from 14 data points.

50

150

t(s)

Figure 2. Electropherograms from the separations of a model mixture of inorganic anions (nitrate, 6 ppm; sulfate, 2.5 ppm; nitrite, 5 ppb; fluoride, 5 ppb; phosphate, 10 ppb; chloride present in the sample coincided with the leading electrolyte zone (L)). The ITP separation (lasting 8 min) was carried out in the electrolyte system ITP-1 (Table l ) , and the carrier electrolyte CZE (Table 1) was used in the separation of microconstituents. The migration positions of the microconstituents as obtained in a blank run (b) are marked with asterisks (compare the run with the sample (a)). The driving currents were 250 and 75 pA in the ITP and CZE stages, respectively. R, G, t refer to the increasing resistance, conductance, and time, respectively; L, T are the leading and terminating zones, respectively. An isotachopherogram from the blank run (omitted) showed only the boundary between the leading and terminating anions (impurities were not detectable). Table 4. Separation Efficienciesfor Nitrite, Fluoride, and Phosphate at Various Concentrations.

concn (ppb) 5 10 15 20 30 40 50 100

separation efficiency (N/m) * nitrite fluoride phosphate 224 000 150 000 120 000 88 000

84 000 56 000 58 000 55 000

68 000 68 000

27 000 15 000

174 000

160 000 121 000 84 000 79 000

a Sample volumes were 30 pL. Number of theoretical plates per meter.

signal from the conductivity detector employed in this stage provides data used in the quantitations of the macroconstituents. (2) Chloride, nitrate, and sulfate are removed from the separation compartment by migrating in the direction to the counter electrode of the ITP column. Timing of this removal is based on the signal from the conductivity detector as obtained in step 1. This cleanup step is stopped after the band of the microconstituents reaches a position close to the start of the CZE column in the bifurcation block (Figure 1). (3) The sample fraction containing the microconstituents and the rear part of the zone of the macroconstituent with the lowest effective mobility is transferred onto the CZE column. (31) Mikkers, F. E. P.; Everaers, F. M.; Peek, J. A. F.J. Chromatogr. 1979,168, 293.

nitrite fluoride

Phosphate,, sulphate

t

i

I,:

Figure 3. ITP and CZE determinations of macro- and microconstituents in a river water sample (Yamaska River, Quebec, Canada). The sample directly injected into the analyzer was found to contain 15.0 ppm, (RSD = 2.6%) of chloride, 7.3 ppm (RSD = 0.5%) of nitrate, 5.9 ppm (RSD = 0.9%) of sulfate, 13.1 ppb (RSD = 0.9%) of fluoride, and 60.9 ppb (RSD = 3.4%) of phosphate. Nitrite was below the limit of detection (200 pptr). The quantitations were based on five parallel runs. The time for the ITP separation was 8 min. An isotachopherogram from the blank run (omitted) showed only the boundary between the leading the terminating anions (impurities were not detectable). The separation conditions, the driving currents, and the meanings of the symbols are the same as in Figure 1.

(4)The microconstituents are resolved under the separating conditions employed in the CZE column, and their detection is performed by the conductivity detector. There are three basic alternatives in selecting the electrolyte systems for the ITP and CZE stages in the ITP-CZE tandem.30 Of these, the one using identical terminating and carrier constituents was preferred in this work because it does not require replenishment of the electrolyte solution in the ITP column after the sample fraction of analytical interest is transferred onto the CZE column. An actual choice of the constituents of the electrolyte solutions followed the above separation and sample Analytical Chemistry, Vol. 66, No. 23, December 1, 1994

4261

Table 6. Influence of a Varying Excess of Macroconstituents on CZE Determinations. of Trace Concentrations of Nitrite, Fluoride, and Phosphate

sample no. 1 2 3 4 5 7 8 (I

nitrite (ppb) taken CZE 5 5 5 5 5 5 5

5.6 4.7 5.1 5.4 4.5 4.9 0

fluoride (ppb) taken CZE 4.8 5.3 6.1 4.6 4.7 5.1 2.8

5 5 5 5 5

5 5

phosphate (ppb) CZE

taken 10 10 10 10 10 10 10

11.3 11.3 12.3 10 10 10.4 12.2

macroconstituents sulfate (ppm) chloride (ppm) 250 500 750 1250 1000 2000 2500

Mean values from two parallel runs.

cleanup scheme. From this scheme it is apparent that the leading electrolyte has to meet the following requirements. (i) To provide the ITP separation of the macroconstituents and at the same time a group separation and concentration of the microconstituents into a narrow band. It is also required that the migration coniiguration as given in (1) Figure 1) is achieved. (i) To reduce the number of anionically migrating constituents, i.e., to reduce the number of potentially interfering anionic constituents, a low pH of the leading electrolyte is mandatory. There are several leading electrolytes proposed for the ITP separation of the anions of our i n t e r e ~ t . ' ~ The % ~ ~one - ~ful6lling ~ the requirements stated above for this type of samples is described in Table 1 under ITP-1. Besides the microconstituents, a small part of the macroconstituents is transferred onto the CZE column. A part of this material is transferred with the microconstituents to guarantee their high recoveries on the column transfer (see Figure l ) , and the rest is due to diffusion of the leading electrolyte constituents into the CZE column during the ITP separation (the solutions in the columns are in direct contact in the bifurcation block; see Figure 1). Therefore, the separating conditions in the CZE stage were chosen not only to provide the resolutions of nitrite, fluoride, and phosphate but also to speed up their defocusing from the macroconstituents.36 From the ionic mobilities of the anions involved and from pKa values of the corresponding acids,6 it is clear that a low pH of the carrier electrolyte is an optimum. Further, in the choice of the carrier constituent (serving also as a terminating anion in the ITP stage), it was necessary to take into consideration also the requirements related to the use of the conductivity detector in the CZE stage to get favorable molar responses of the detector to the a n a l y t e ~while ~ ~ the electromigration dispersion is kept to a minimum. The composition of the carrier electrolyte used in this work (system CZE in Table 1) was a reasonable compromise considering all the above requirements. Determination of Chloride Nitrate, and Sulfate. Working conditions (leading electrolyte, sample injection volume, i.d. of the capillary tube, driving current) employed in the ITP ~~~

(32) Kaniansky. D.; Zelenskjr, I.; HavaSi, P.; CerovsM, M.J. Chromatogr. 1986, 367, 274. (33) Vacik, J.; Muselasova, I. J. Chromatogr. 1985,320, 199. (34) Zelenskj., I.; SimuniEova, E.; Zelenska, V.; Kaniansky,D.; HavaSi, P.; Chalhi, P J. Chromatogr. 1985,325, 161. (35) Kfivankova, L.; BoEek, P. Chem. Prfmrvsl 1988,38, 97. (36) Gebauer, P.; Thomann, W.; BoEek, P. J. Chromatogr. 1992,608,47. (37) Mikkers, F. E. P.; Everaerts, F. M.; Verheggen, Th.P.E.M. J. Chromatogr. 1979,169, 1.

4262

Analytical Chemisfry, Vol. 66, No. 23, December 1, 1994

determination of chloride, nirtate, and sulfate were in fact identical to those used for the same purpose in the ITP-ITP tandem.l7J8 Therefore, analytical performance data describing the ITP determination in this work (Table 2) were close to those reported in the quoted references. A series of model mixtures containing chloride, nitrate, and sulfate at varying concentrations were analyzed to find maximum concentrations of these anions that can be determined simultaneously with the microconstituents in one run. It is apparent (Table 3) that the concentration of nitrate in the sample (having the intermediate effective mobility between those of chloride and sulfate) limited an overall sample load. This is what was expected from a theoretical treatment of the ITP separation p r o c e ~ s . ~ ~ , ~ ~ From a practical point of view, it means that such a load capacity of the ITP stage enables the determination of the anions in a wide variety of water samples. CZE Determination of Nitrite, Fluoride, and Phosphate. As is clear from electropherograms in Figure 2 nitrite, fluoride, and phosphate were baseline resolved under the separating conditions employed in the CZE stage. It is also apparent that the transferred macroconstituents (a large peak migrating in front of nitrite in Figure 2) did not interfere with their determination. The reproducibilities of the migration times of the CZE analytes were 1%or less when their concentration in the samples were below 40-50 ppb. At higher concentrations, they were systematically biased by the electromigration d i ~ p e r s i o n This . ~ ~ dispersive phenomenon had a negative effect on the separation efficiencies of the analyte, as is clear from the data in Table 4. Nevertheless, parameters of the regression equations describing the calibration lines show that reliable quantitations of the microconstituents were possible under such conditions (Table 5 ) . Besides the separation and concentration of the analytes of interest, ITP also concentrates ionic impurities present in the leading and terminating electrolytes (originating from the components of which the solutions of the electrolytes were prepared) when they migrate isotachophoretically. For example, in the CZE separations shown in Figure 2, peak areas of the impurities in the migration positions of the analytes corresponded to ca. 2 ppb of nitrate and 1 ppb of fluoride. An impurity peak migrating in the neighborhood of phosphate (see the blank run in Figure 2b) was equivalent to ca. 3 ppb of this analyte. Fluctuations in the peak areas of these impurities (background noise) were in fact determinating the limits of detection (LODs) for our analytes. (38) BoEek, P.; Deml, M.; Kaplanova, B.; Janak, J.J. Chromatogr. 1978,160, 1.

@i

G

1 I

I

1

50

'50

t(s)

so

150

t(s)

Figure 4. Electropherogramsfrom the CZE determination of parts per billion concentrations of microconstituents present in a model mixture containing 1250 ppm of sulfate. (a) Blank run (the terminating electrolyte was injected);(b) sample containing microconstituentsand sulfate at a 10 ppm concentration; (c) sample containing 1250 ppm of sulfate; (d) the same as (c) only the sample also contained 5 ppb of each nitrite and fluoride and 10 ppb of phosphate. Asterisks in runs a and c mark the migration positions of the microconstituents. The ITP separation was carried out in the system ITP-2 (Table l), and the carrier electrolyte CZE (Table 1) was used in the CZE column. The driving currents were 250 and 75pA in the ITP and CZE stages, respectively. G refers to increasing conductance.

Estimates of the relative standard deviations of the peak areas of the impurities as obtained in repeated blank runs (six runs related to the experiments in Figure 2) served as a measure of these fluctuations. From the signal-to-noiseratio of three, we obtained the LOD values of 200, 300, and 600 pptr (parts per trillion) for nitrite, fluoride, and phosphate, respectively. These values are ca. 5-10 times greater than those which were expected from the noise characteristics of the conductivity detector. As no special attention was paid to the purity of the chemicals used in the preparation of the leading and terminating electrolyte solutions, it is reasonable to expect that the appropriate measures taken in this respect could reduce the LOD values, making them comparable to those associated with the performance of the conductivity detector. Analytical Potentialities of the l"-czE Tandem. Electropherograms given in Figure 3 illustrate the use of the ITPCZE tandem with the conductivity detectors for the analysis of a practical sample. In this instance the CZE column of a larger i.d. (0.35 mm) was employed to test a column of a higher load capacity. Relevant data (see figure legend) show that, also under such conditions, very reproducible CZE determinations were possible. In some instances (see, e.g., refs 18-20 and 39) it is necessary to analyze samples containing chloride, sulfate, and nitrate in a very large excess relative to nitrite, fluoride, and phosphate. The (39) Shotyk, W. J. Chromatogr. 1993,640, 309.

potential of lTP-CZE tandem for the analysis of such sample types was evaluated with a series of model mixtures containing constant concentrationsof the microconstituents, while the concentrations of chloride and sulfate varied (See Table 6). As is apparent from the data in Table 3, the concentrations of chloride and sulfate were in some of the model samples several times above the load capacity of the l"column for these anions. The data in Table 6 show that microconstituents were tidy recovered by the ITP pretreatment when the concentration of chloride was 2000 ppm or less or the concentration of sulfate was 1250 ppm or less. As low parts per billion concentrations of the microconstituents could also be determined at such concentrations of sulfate and/or chloride, we can state that the ITP-CZE tandem enables the CZE determination of nitrite, fluoride, and phosphate in a 1@-106-fold excess of the macroconstituents. The main problems in these experiments were associated with the purities of the chemicals from which the model mixtures were prepared. For example, from the electropherograms in Figure 4, we can see that the injection of sulfate at a 1250 ppm concentration was responsible for an increase of the peak area of phosphate equivalent to ca. 10 ppb concentration of this anion in the sample (see the electropherograms in Figure 4a and 4c). The origin of this peak was confirmed in a run with a chloride sample (containing chloride at a concentration leading to the same prolongation of the 1°F' separation,while the background signal in the migration position of phosphate was identical to that in the blank run; Figure 4a). Therefore, on spiking the sample with a 10 ppb concentration of phosphate, it was necessary to make a correction for a signjficantly higher blank value and thus introduce a higher random error into the determination. Similarly, the nitrate spikes, even considerably higher nitrite concentrations were detected. A prolongation of the ITP pretreatment by injecting the samples containing 1250 ppm of sulfate (12 min longer ITP separation to comparison to the blank run in Figure 4a) led to an increase of the peak areas of some of the unidentified constituents (see, e.g., a peak migrating immediately behind phosphate in Figure 4). In this instance, the origin of these peaks was associated with impurities in the lTP electrolyte solutions. CONCLUSIONS From the results presented in this work, it seems appropriate to state that the on-line coupling of ITP with CZE in the columncoupling configuration of the separation unit offers a very promising alternative to the determination of the studied anions in situations when their concentrations differ considerably. It is also apparent that the load capacity of the ITP column sets a limit for a simultaneous determination of both macro- (sulfate, nitrate, and chloride) and microconstituents (nitrite, fluoride, and phosphate). Thus, when the concentration ratio of macro-/microconstituents was higher than (2-3) x 104, the analysis had to be divided into two runs (separate determinations of both groups of anions). In this way, e.g., the ITP cleanup provided a way to determine the microconstituents present in samples with a 105-106-fold excess of macroconstituents. The anionic impurities from the electrolyte solutions accumulated in the ITP stage between the leading and terminating zones were main disturbances in the CZE analysis of microconstituents. In this respect, only the use of high purity chemicals in the preparation of the electrolyte solutions for the ITP stage can define limits of the KF-CZE tandem. The use of conductivity Analytical Chemistry, Vol. 66, No. 23, December 1, 1994

4263

detector in CZE has some limitation^.^^ Nevertheless, this work showed that its performance was not limiting in achieving sub parts per billion detection limits for the studied analytes. In this context, it can also be concluded that only advances related to the purities of the electrolyte solutions in the ITP stage will define its limitations and, eventually, initiate further research related to the detection.

4264 Analytical Chemistry, Vol. 66, No. 23, December 1, 1994

ACKNOWLEDGMENT This work was supported by a grant provided by Slovak Grant Agency for Science. Received for review June 29, 1994. Accepted September 8, 1994.@ Abstract published in Advance ACS Abstracts, October 15, 1994.