Anal. Chem. 2001, 73, 267-271
Single-Run Capillary Electrophoretic Determination of Inorganic Nitrogen Species in Rainwater Audrius Padarauskas,* Vaida Paliulionyte, and Birute Pranaityte
Department of Analytical and Environmental Chemistry, UNESCO Trace Element Institute Satellite Center, Vilnius University, Naugarduko 24, LT-2006 Vilnius, Lithuania
A capillary electrophoretic (CE) method for the simultaneous determination of nitrate, nitrite, and ammonium ions has been developed. Direct (NO3-, NO2-) and indirect (NH4+) UV detection at 214 nm in conjunction with electromigration sampling from both ends of the capillary was used. The optimized separations were carried out in 10 mmol/L imidazole sulfate, 2 mmol/L 18crown-6, and 0.02 mmol/L tetradecyltrimethylammonium hydroxide electrolyte (pH 4.0). The method permits excellent separation of three nitrogen species in only 4 min. A 1 × 10-4 mol/L KBr solution was used as an internal standard to limit possible electrokinetic injection biases. Experimental results showed that the use of an anionic internal standard for cationic analytes and vice versa gives only slightly better precision than analysis with no internal standard. Using Br- internal standard for NO3- and NO2- ions and K+ for NH4+, peak area RSD values decrease significantly. The proposed system was applied to the speciation of inorganic nitrogen ions in rainwater samples. The recovery tests established for external calibration and standard addition techniques using one or two internal standards were within the range 100 ( 10% The CE results agree with those obtained by spectrophotometric methods. The nitrogen cycle is of particular significance to a number of biological and nonbiological processes in the environment.1 Natural and anthropogenic effects can cause localized interrelated changes to the cycle. Anthropogenic externally supplied nitrogen loading is a key factor controlling primary production in nitrogenlimited estuarine, coastal, and oceanic waters.2 One of the most rapidly growing sources of anthropogenic nitrogen loading is atmospheric deposition. Atmospheric deposition includes all biologically available inorganic nitrogen species that are products of the atmospheric reactions of nitrogen oxides emmited from fossil fuel combustion and biomass burning and ammonia as volatilization product of agricultural waste and fertilizers.3 To assess the contribution of various sources to inorganic nitrogen in atmospheric deposition, it is important to develop a rapid and simple analytical technique for the simultaneous * Corresponding author: (fax) (+370-2) 330987; (e-mail) audrius.padarauskas@ chf.vu.lt. (1) Stumm, W.; Morgan, J. Aquatic Chemistry, 3rd ed.; Wiley: New York, 1995. (2) Paerl, H. W.; Whitall, D. R. Ambio 1999, 28, 307. (3) Paerl, H. W. Ophelia 1995, 41, 237. 10.1021/ac000674s CCC: $20.00 Published on Web 12/08/2000
© 2001 American Chemical Society
determination of inorganic nitrogen species. Many methods have been used for the determination of nitrate, nitrite, and ammonium in environmental samples. The most important are spectrophotometric and ion chromatographic (IC) methods. Spectrophotometric techniques often involve conversion of nitrite into an azo dye coupled with aromatic amines.4,5 Nitrate is previously reduced to nitrite, usually with Cd, and is then determined by difference. In the spectrophotometric determination of ammonium, the Nessler6 and indophenol7 methods appear to be dominant. All these methods, however, are plagued with many interferences such as dissolved organic matter and common metal ions. Moreover, the use of three separate analyses for each nitrogen species is labor-intensive as well as time-consuming. The most common analytical technique used for nitrogen species in the last two decades is IC.8 However, the determination of anions and cations using IC requires entirely different conditions (stationary and mobile phases). Mou et al.9 used a mixed anion-cation-exchange column for the simultaneous determination of all three nitrogen ions. Nitrate and nitrite were detected by UV spectrometry and ammonium ion by use of a chemically suppressed conductivity detector. The major disadvantage of this system is relatively low selectivity in comparison with conventional IC columns. In recent years, the use of capillary electrophoresis (CE) for the analysis of ionic analytes has grown significantly. Because of its higher resolution, shorter analysis time, lower consumption of reagents, and greater simplicity in operation compared to IC, CE has received a great deal of attention for the determination of inorganic ions. Numerous applications of CE have been reported for the determination of nitrate and nitrite anions10-16 or the (4) Eaton, M. A., Clesceri, L. S., Greenberg, A. E., Eds. Standard Methods for the Examination of Water and Wastewater, 18th ed.; American Public Health Association: Washington, DC, 1994. (5) Verma, K. K.; Verma, A. Anal. Lett. 1992, 25, 2083. (6) Thompson, J. F.; Morrison, G. R. Anal. Chem. 1952, 23, 1153. (7) Scheurer, P. G.; Smith, F. Anal. Chem. 1955, 27, 1616. (8) Haddad, P. R.; Jackson, P. E. Ion Chromatography-Principles and Applications; Elsevier: Amsterdam, 1990. (9) Mou, S. F.; Wang, H. T.; Sun, Q. J. Chromatogr. 1993, 640, 161. (10) Nielen, M. W. F. J. Chromatogr. 1991, 546, 173. (11) Romano, J.; Jandik, P.; Jones, W. R.; Jackson, P. E. J. Chromatogr. 1991, 546, 411. (12) Kelly, L.; Nelson, R. J. J. Liq. Chromatogr. 1993, 16, 2103. (13) Li, K.; Li, S. F. Y. J. Liq. Chromatogr. 1994, 17, 3889. (14) Shamsi, S.; Danielson, N. D. Anal. Chem. 1994, 66, 3757. (15) Jimidar, M.; Hartman, C.; Cousement, N.; Massart, D. L. J. Chromatogr., A 1995, 706, 479. (16) Okemgbo, A. A.; Hill, H. H.; Siems, W. F.; Metcalf, S. G. Anal. Chem. 1999, 71, 2725.
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001 267
ammonium cation17-19 in various aqueous and other samples. However, as the separation by CE is based on the difference in electrophoretic mobilities of the analytes, the determination of fast anions and cations in a single analysis under conventional CE conditions is not possible. In this paper, we report on the CE approach for a simultaneous separation of nitrate, nitrite, and ammonium ions based on electrokinetic sample injection from both ends of the capillary and both direct and indirect UV detection. The quantitative aspects of the internal standardization technique using one or two oppositely charged internal standards are discussed. Finally, determination of nitrogen species in rainwater samples using CE method was compared to the conventional spectrophotometric methods. EXPERIMENTAL SECTION Instrumentation. Separations were performed on a P/ACE 2100 apparatus (Beckman Instruments Inc., Fullerton, CA) equipped with a UV detector with wavelength filters (200, 214, 230, and 254 nm). Fused-silica capillary (Polymicro Technology, Phoenix, AZ) of 75 µm i.d.(375 µm o.d.) and 85 cm total length (50 cm to the detector) was used. Samples were introduced by electromigration injection (5 kV) from both ends of the capillary. System Gold software was used for data acquisition. UV detection was employed at 214 nm. All experiments were conducted at 25 °C. Reagents and Chemicals. Deionized water was obtained by passing distilled water through a Waters Milli-Q water purification system (Millipore, Eschborn, Germany). Imidazole was purchased from Sigma (St. Louis, MO). All other reagents were of analyticalreagent grade obtained from Merck (Darmstadt, Germany). Stock solutions (0.01 mol/L) of analyte ions and internal standard were prepared from appropriate salts. Tetradecyltrimethylammonium (Merck) hydroxide (TTAOH) was prepared from bromide salt by conversion using an OH--form anion-exchange material ARA-12P (Reachim). Procedures. The working electrolytes were prepared fresh daily. All electrolyte solutions were filtered through a 0.45-µm membrane filter and degassed by ultrasonication. The capillary was rinsed with 1.0 mol/L sodium hydroxide and water for 5 min and then equilibrated with carrier electrolyte for 30 min at the beginning of each day. Between all electrophoretic separations, the capillary was rinsed for 2 min with carrier electrolyte. RESULTS AND DISCUSSION Separation Optimization. In our previous work20,21 for the simultaneous CE separation of common inorganic anions and cations based on electromigrative sample introduction from both ends of the capillary and indirect UV detection, we used an electrolyte containing two UV chromophores: cationic imidazole or copper(II) chelate with ethylenediamine (Cu(En)22+) and anionic nitrate or chromate. However, most environmental and (17) Ba¨chmann, K.; Steeg, K. H.; Groh, T.; Roder, A.; Haumann, I.; Boden, J. Int. J. Environ. Anal. Chem. 1992, 49, 87. (18) Simunicova, E.; Kaniansky, D.; Loksikova, K. J. Chromatogr., A 1994, 665, 203. (19) Havel, J.; Janos, P.; Jandik, P. J. Chromatogr., A 1996, 745, 127. (20) Padarauskas, A.; Olsauskaite, V.; Schwedt, G. J. Chromatogr., A 1998, 800, 369. (21) Padarauskas, A.; Olsauskaite, V.; Paliulionyte, V. J. Chromatogr., A 1998, 829, 359.
268
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
biological samples contain large amounts of chloride, sulfate, and hydrogen carbonate anions which interfere in the determination of anionic nitrogen species, especially nitrite. Relatively high UV absorbance in the wavelengths range between 200 and 220 nm of both nitrate and nitrite anions enables the use of a direct detection technique for these analytes.16 This approach has the advantage of eliminating interferences from nonabsorbing anions such as chloride, sulfate, etc. Moreover, in this case, the carrier electrolyte must contain only one cationic UV chromophore, which significantly improves the baseline stability and detection sensitivity of both anionic and cationic analytes. The difference in effective electrophoretic mobilities of nitrate and nitrite ions is negligible.22 This fact often causes problems with simultaneous CE determination of these analytes, especially when their concentrations differ significantly. Changes in the electrophoretic mobility of an anion are most dramatic in the pH range at or near the pKa of the ion.23 An acidic electrolyte offers the possibility to improve the resolution between NO3- and NO2due to the reduction of the effective mobility of weakly acidic nitrite (pKa ) 3.4). Because ammonium has no UV absorbance, for the simultaneous UV detection of all three nitrogen species, the carrier electrolyte must contain a cationic UV chromophore. In our work, imidazole was selected for this purpose. This amine is well suited as an UV chromophore because its electrophoretic mobility is only slightly lower than the mobility of ammonium.24 In addition, imidazole exhibits spectral characteristics required for indirect UV detection at 214 nm. Preliminary experiments were performed in 10 mmol/L imidazole sulfate electrolyte. Investigation of electrolyte pH effect on the resolution of NO3- and NO2- anions demonstrated a good resolution in the pH range below 4.0. Although total resolution was highest at lower pH values, we selected pH 4.0 in order to reduce separation time and to obtain higher detection sensitivity for nitrite. All CE separations were performed in the 85-cm total length capillary in which the detector window was placed at 50 cm from the injection side (according to the nomenclature used in conventional CE). Common environmental samples contain relatively large amounts of calcium, magnesium, potassium, and sodium cations that are also detected under the conditions used. Consequently, in our system, more selective separation should be obtained when nitrogen anions are eluted before metal cations. For this reason, a positive power supply was used because in this case migration times for cationic analytes are slightly longer than those for anions. An addition of 0.02 mmol/L TTAOH to the carrier electrolyte additionally reduces the migration times for anions and increases those for cations. Thus, nitrogen anions and common cations are fully separated but the separation of NH4+ and K+ is difficult due to their identical electrophoretic mobilities at a neutral and slightly acidic pH.22 The use of a crown ether to solve this problem has been reported in several publications.25,26 The effect of 18-crown-6 concentration in the carrier electrolyte on the migration times of the NO3-, NO2-, NH4+, and K+ ions can be seen in Figure 1. The increase in concentration of (22) Marcus, Y. Ion properties; Marcel Dekker: New York, 1997. (23) Smith, S. C.; Khaledi, M. G. Anal. Chem. 1993, 65, 193. (24) Beck, W.; Engelhardt, H. Chromatographia 1992, 33, 313. (25) Ba¨chmann, K.; Boden, J.; Hauman, I. J. Chromatogr. 1992, 626, 259. (26) Riviello, J. M.; Harrold, M. P. J. Chromatogr., A 1993, 652, 385.
the amount of the analyte injected (na) can be expressed as follows:27
na )
Figure 1. Effect of 18-crown-6 concentration in the electrolyte on the migration times of ions. Electrolyte, 10 mmol/L imidazole sulfate, 0.02 mmol/L TTAOH, pH 4.0; applied voltage, +30 kV.
(µa ( µeo)r2π Uinjtinjca L
where µa is the electrophoretic mobility of the analyte, µeo is the mobility of the electroosmotic flow (EOF), r is the inner radius of the capillary, L is the total length of the capillary, Uinj is the injection voltage, tinj is the injection time, and ca is the molar concentration of the analyte. In the case when the mobility of the analyte is in the direction opposite to that of the EOF, only analytes with absolute values of µa exceeding the magnitude of the EOF can be injected. On the basis of eq 1, two different kinds of bias can be observed using EK injection mode for the quantitative analysis.28 One is brought about by the different mobilities of the species in the sample solution. This effect causes a distortion in the ratio of peak areas for ions having different mobilities. All the three inorganic nitrogen species have similar mobilities (7.40 × 10-4, 7.46 × 10-4, and 7.60 × 10-4 cm2/V s for NO3-, NO2-, and NH4+, respectively); therefore, the mobility bias should be negligible. The second bias (matrix bias) is related to the ionic strength of the medium in which the analytes are dissolved. Since the mobility of the ions is strongly influenced by the conductivity of the sample, even if the concentration of the analytes injected is held constant, the injected amount of the analytes will vary according to the composition of the sample matrix. This bias means that the peak areas obtained for the same ions in samples of different matrixes cannot be compared directly. Matrix bias can be minimized or entirely eliminated by using an internal standard for the quantitative analysis.29 In our system for the simultaneous determination of both anionic and cationic analytes, only anionic, only cationic, or both anionic and cationic internal standards can be used. For example, for anionic analyte and anionic internal standard, the ratio of the amounts injected is given by
na (µa - µeo)ca ) nst (µst - µeo)cst
Figure 2. Electropherogram of a standard solution under optimum conditions. Electrolyte, 10 mmol/L imidazole sulfate, 2 mmol/L 18crown-6, 0.02 mmol/L TTAOH, pH 4.0; applied voltage, +30 kV; UV detection at 214 nm; electrokinetic (5 kV; 5 s) injection from both ends of the capillary.
18-crown-6 reduces the mobility of potassium ion without any significant changes in the migration times for other analytes studied. At 2 mmol/L crown ether concentration, ammonium and potassium ions are well resolved. Figure 2 shows an example of the simultaneous separation of inorganic nitrogen species and potassium ion obtained under optimum conditions. All ions are fully separated in less than 5 min. Choice of the Internal Standard. The proposed CE system requires electrokinetic (EK) sample introduction. In EK injection,
(1)
(2)
For cationic analyte and anionic internal standard, eq 2 is expressed
(µa + µeo)ca na ) nst (µst - µeo)cst
(3)
Comparison of eqs 2 and 3 demonstrates that fluctuations of the EOF from run to run during the EK injection are more important if internal standard and analyte have opposite charges. For quantitative analysis, we have selected KBr as an internal standard. Both K+ and Br- ions have mobilities close to those of the analytes; therefore, mobility bias should be negligible. A 1 × 10-4 mol/L concentration of KBr was used in all measurements. (27) Rose, D. J.; Jorgenson, J. W. Anal. Chem. 1988, 60, 642. (28) Huang, X.; Gordon, M. J.; Zare, R. N. Anal. Chem. 1988, 60, 377. (29) Jackson, P. E.; Haddad, P. R. J. Chromatogr. 1993, 640, 481.
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
269
Table 1. Repeatability for EK Injection (5 kV, 5 s, Six Replicate Injections)
Table 3. Recovery of Nitrogen Species Added to Rainwater Sample (n ) 3)
internal standard
NO3-
peak area RSD, % NO2-
NH4+
method
analyte
found, mg/L
added, mg/L
total found, mg/L
recovery, %
no standard BrK+
6.1 2.3 5.4
7.4 2.7 5.0
6.2 5.4 2.4
external calibration
NO3-
2.21
NO2-
nfa
NH4+
0.73
NO3-
2.18
NO2-
nf
NH4+
0.68
0.50 2.00 0.50 2.00 0.50 2.00 0.50 2.00 0.50 2.00 0.50 2.00
2.63 4.17 0.46 2.12 1.36 2.95 2.55 4.22 0.45 1.88 1.24 2.63
97 99 92 106 111 108 95 101 90 94 105 98
Table 2. Calibration Data and Detection Limits for Ions Studied internal standard analyte BrK+
NO3NO2NH4+ NO3NO2NH4+
equation of regression lines y ) 0.216 + 3.62 × 104c y ) 0.124 + 2.81 × 104c y ) 0.073 + 1.64 × 104c y ) 0.260 + 1.63 × 104c y ) 0.214 + 1.25 × 104c y ) 0.102 + 7.69 × 103c
standard addition
correlation detection coefficient limit, (r2) × 107 mol/L 0.999 0.998 0.996 0.994 0.993 0.999
2 3 8 2 3 8
An injection of this pure solution did not show any presence of the analytes. In Table 1, the peak area RSD values obtained for 5 × 10-5 mol/L standard solutions (six replicate injections) without and with internal standard are compared. Using the internal standartization technique, the analyte peak areas were divided by appropriate internal standard peak areas and the ratios were employed as responses. As can be seen, the use of an anionic internal standard for cationic analyte and vice versa gives only slightly better repeatability than analysis with no internal standards. When we apply Br- as the internal standard for NO3- and NO2- ions and K+ for NH4+ ion, peak area RSD values decrease significantly. The explanation for these results is that the ratio of injected amounts for oppositely charged ions may change from run to run because of fluctuations in EOF rate (see eq 3). It is further possible that the conditions of injection (e.g., position of the electrode and the capillary, relative physical heights of the solutions, etc.) in the two sample vials could differ, causing higher RSD values when analyte and internal standard are injected from different capillary ends. It should be noted that each analyte’s RSD value increases with increasing difference between the analyte’s migration time and that of the internal standard used. Migration times RSD values were in the range 0.25-0.70% for all ions studied. Validation of the Method. The linearity of the calibration curves was measured by triplicate injections (5 kV, 5 s) of standards containing 1 × 10-4 mol/L KBr at seven different concentration levels (5 × 10-6-5 × 10-4 mol/L). For the evaluation of calibration curves, the peak areas of the analytes were divided by the peak area of the internal standard and plotted versus analyte concentration. As a criterion of linearity, deviation within 5% of the mean response factor was used. Table 2 summarizes the results of the calibration curves for all three analytes. Valid calibration is demonstrated over 2 orders of magnitude with both internal standards, but the correlation coefficients were higher in the case when analyte and standard had the same charge. The detection limits (LODs) were determined for EK injection of standard solutions containing 1 × 10-4 mol/L KBr at 5 kV for 270 Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
a
nf, not found.
Figure 3. Electropherograms of a rainwater sample spiked with 1 × 10-4 mol/L KBr (a) using hydrodynamic (3.43 × 103 Pa, 3 s) and (b) electrokinetic (5 kV, 5 s) injection. The upper electropherogram is obtained from two separate runs using positive (for cations) and negative (for anions) power supply. Other conditions as in Figure 2.
10 s based on 3 times the baseline noise and are also summarized in Table 2. It should be noted that LODs can be reduced at least by a factor 10 when the concentration of KBr in the standard solutions is of the same order as analyte concentration. The LODs achieved with the proposed method were suitable for rainwater samples, and further optimization was not performed. Analysis of Rainwater Samples. To evaluate the proposed system for real samples, it was applied to the speciation of inorganic nitrogen ions in rainwater samples. The pH of the
Table 4. Data Comparison (mg/L) of Rainwater Samples (n ) 5) sample no
analyte
1 (pH 5.85)
NO3NH4+ NO3NH4+ NO3NH4+
2 (pH 6.30) 3 (pH 6.14) a
CEa
spectrophotometry
4.93 (2.8) 1.42 (3.1) 2.55 (2.5) 0.61 (4.2) 1.84 (3.9) 0.79 (4.7)
4.80 1.35 2.47 0.56 1.72 0.85
Values in parentheses are RSD (%).
sample solution can effect the quantification of analytes using an electrokinetic injection technique. The analytes display changes in electrophoretic mobility if the pH is changed in the vinicity of their pKa. In most cases, the pH of the rainwater samples varies in the range 5-7. Although in this pH range any changes in the effective charges and the amounts for the nitrogen species should occur, several standard solutions containing KBr internal standard were investigated in the pH range 4.5-7.5. No significant changes in the peak areas (divided by internal standard peak areas) were observed. Consequently, the additional pH adjustment for rainwater samples was not performed. The suitability of external calibration and standard addition methods was evaluated for quantitative analysis. External calibration using K+ as an internal standard is impossible because K+ ions also present in rainwater. For this reason the calibration in this method was performed using Br- for all three analytes. In the standard addition method, both Br- (for NO3- and NO2-) and K+ (for NH4+) internal standards were used. The recovery tests for spiked rainwater sample were established for both calibration procedures, and the results are presented in Table 3. As can be observed, the recoveries in all the cases were within the range 100 ( 10%. Slightly better recoveries for ammonium were obtained using the standard addition technique with K+ as the internal standard. It should be noted that in comparison with the conventional CE system with hydrodynamic sample injection, lower concentration rainwater samples can be analyzed using a proposed method
because on-column sample preconcentration occurred by electrostacking. This effect is demonstrated in Figure 3 in which two electropherograms obtained for the same sample injected by hydrodynamic (a) and by electrokinetic (b) techniques are compared. According to theoretical calculations in both electropherograms, approximately the same absolute amounts of the analytes should be injected. The upper electropherogram is obtained from two separate analysis because use of hydrodynamic injection simultaneous separation of cations and anions is not possible. Several rainwater samples were analyzed by the proposed CE method using a standard addition procedure and by spectrophotometric methods.4 The results are compared in Table 4. No nitrite was detected by both methods in any of the samples in this study. As can be seen, the CE method showed good agreement with data obtained from spectrophotometric techniques. The analysis does not require any preliminary treatment of the samples except filtration and addition of an internal standard. CONCLUSIONS A new CE system based on the electromigrative sample introduction from both ends of the capillary was developed for single-run determination of nitrate, nitrite, and ammonium. The precision of EK injection is improved by using an internal standard. It was also shown that higher precision can be obtained if two oppositely charged internal standards are used. The developed CE method is fully suitable for the analysis of inorganic nitrogen species in the rainwater samples. The CE method is extremely fast because it requires one run for all three analytes, provides acceptable precision, and can be efficiently used in the routine analysis. ACKNOWLEDGMENT This work was supported by the Lithuanian State Science and Studies Foundation.
Received for review June 14, 2000. Accepted October 24, 2000. AC000674S
Analytical Chemistry, Vol. 73, No. 2, January 15, 2001
271