Anal. Chem. 1999, 71, 3046-3053
Thermospray Methods for Rapid, Sensitive, and Nonchromatographic Speciation of Chromium Oxidation States Xiaohua Zhang and John A. Koropchak*
Department of Chemistry & Biochemistry, Southern Illinois University at Carbondale, Carbondale, Illinois 62901
A sensitive technique for speciation and quantification of Cr(III) and Cr(VI) has been developed using thermospray (TSP) sample introduction with inductively coupled plasma atomic emission spectrometry (ICPAES). For unacidified solutions, the sensitivity for Cr(III) was found to be lower than that for Cr(VI). The sensitivity for Cr(III) was further depressed to a negligible level by adjusting sample and thermospray operating parameters. The low sensitivity for Cr(III) was thought to result from the precipitation of that species to form Cr(OH)3, which deposited within the vaporizer. For acidic solutions (1% v/v HNO3), the sensitivities for both species were essentially identical. On the basis of these results, methods for speciation of Cr(III) and Cr(VI) were developed. With samples buffered to pH 4.4, Cr(VI) could be selectively determined. With acidic sample aliquots (1% v/v HNO3), the total chromium concentration could also be determined, and the Cr(III) concentration could be calculated by difference. Parameters affecting Cr(III) sensitivity, such as control temperature, pH, and pump flow rate, were studied in addition to optimal TSP-ICPAES parameters. The limits of detection (LODs) for Cr(VI) and for total Cr were 0.47 and 0.61 µg/L with standard deviations of 1.5% and 2.0%, respectively. Good accuracy and precision of the method were demonstrated for analysis of spiked tap water and lake water samples. Mobile phase ion-pairing chromatography with ICPAES detection provided comparable results for moderately high concentration samples. Accuracy of measurements for Cr(VI) was within 1% of the certified value for NIST standard reference material 2109. In aqueous solutions, chromium exists predominantly in one of two primary forms: Cr(III) and Cr(VI). Of these two relatively stable oxidation states, Cr(III) is considered to be essential to mammals for the maintenance of glucose, lipid, and protein metabolism, whereas Cr(VI) is a toxic and carcinogenic form.1-3 Since chromium is a widely used metal, it can enter the environment through industrial waste, such as from steel works, elec(1) Nriagu, J. O.; Nieboer, E. Chromium in the Natural and Human Environment; Wiley: New York, 1988; p 185. (2) Committee on Animal Nutrition, Board on Agriculture, National Research Council. The Role of Chromium in Animal Nutrition; National Academy Press: Washington, DC, 1997; Chapter 3. (3) Singh, J.; Carlisle D. L.; Pritchard D. E.; Patierno, S. R. Oncol. Rep. 1998, 5, 1307-1318.
3046 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
troplating, tanning, or dying industries. Drinking water may be contaminated by such industrial waste or by the corrosion inhibitors used in water pipes. Precise knowledge of the concentrations of different chromium species present in a system is of great significance to an accurate assessment of the environmental and biological impact of chromium. As a result, the development of speciation techniques with sufficient selectivity and sensitivity for Cr is an important challenge for analytical chemists. Existing speciation methods for chromium include extraction,4 ion exchange,5 coprecipitation,6 and electrochemical methods,7 which are all laborious and time-consuming. Another traditionally used method is colorimetry,8,9 with which Cr(VI) can be determined after reaction with DPC (1,5-diphenyl carbohydrazide). As Cr(III) does not interfere in the Cr(VI)-DPC reaction, both Cr(VI) and total chromium (after the on-line oxidation of Cr(III) to Cr(VI) by an oxidant such as Ce(IV)8) are sequentially determined, and the Cr(III) content is obtained by difference. Generally, this method has problems in handling complex sample matrixes and must be combined with a conversion step which is well-known to be error-prone. The best limit of detection (LOD) reported with this method is 5 µg/L.9 High-performance separation techniques can achieve the selectivity required for speciation measurement, especially when coupled with an element-selective detector (an atomic or mass spectrometer).10 These so-called hyphenated techniques are currently preferred approaches for speciation study. The most attractive separation methods for this approach are flow injection (FI) on-line preconcentration11-15 and HPLC.16-20 The FI separation process is quite similar to batch filtration or solvent extraction. (4) Sugiyama, M.; Fujino, O.; Kihara, S.; Matsui, M. Anal. Chim. Acta 1986, 181, 159-168. (5) Johnson, C. A. Anal. Chim. Acta 1990, 238, 273-278. (6) Lan, C.-R.; Tseng, C.-L.; Yang, M.-H.; Alfassi, Z. B. Analyst 1991, 116, 3538. (7) Boussemart, M.; van den Berg, C. M. G.; Ghaddaf, M. Anal. Chim. Acta 1992, 262, 103-115. (8) De Andrade, J. C.; Rocha, J. C.; Baccan, N. Analyst 1985, 110, 197-199. (9) Girard, L.; Hubert, J. Talanta 1996, 43, 1965-1974. (10) Lobinski, R. Appl. Spectrosc. 1997, 51, 260A-275A. (11) Sperling, M.; Xu, S.-K.; Welz, B. Anal. Chem. 1992, 64, 3101-3108. (12) Posta, J.; Berndt, H.; Luo, S.-K.; Schaldach, G. Anal. Chem. 1993, 65, 25902595. (13) Gaspar, A.; Posta, J.; Toth, R. J. Anal. At. Spectrom. 1996, 11, 1067-1074. (14) Rao, T. P.; Karthikeyan, S.; Vijayalekshmy, B.; Iyer, C. S. P. Anal. Chim. Acta 1998, 369, 69-77. (15) Shah, A.; Devi, S. Anal. Chim. Acta 1990, 236, 469-473. (16) Roychowdhury, S. B.; Koropchak, J. A. Anal. Chem. 1990, 62, 484-489. (17) Tomlinson, M. J.; Wang, J.-S.; Caruso, J. J. Anal. At. Spectrom. 1994, 9, 957-964. 10.1021/ac9813585 CCC: $18.00
© 1999 American Chemical Society Published on Web 06/08/1999
Its purpose is to gain a higher sensitivity and better LOD for lowconcentration samples by preconcentration. In most proposed online methods, one sorbent is used to preconcentrate and determine either Cr(III) or Cr(VI) in a single run. Sometimes the other species, which is unretained, is determined by the on-line detector, but with a poorer LOD than the preconcentrated species.13 Alternatively, the concentration of the other species may be calculated by difference if the total chromium concentration is separately determined13 or one may use another FI separation step for selective preconcentration.14 Detailed reviews of techniques, sorbents, flow rates, RSDs, LODs, linear dynamic ranges, etc. for this method of chromium speciation report that LODs for Cr(III) are between 0.02 and 55 µg/L, with a majority above 0.50 µg/L, while LODs for Cr(VI) are from 0.02 to 20 µg/L with a majority above 1.0 µg/L.11,14 The lowest LODs are achieved at the expense of long preconcentration times.14 The preconcentration times for FI separations are typically at least several minutes, and as long as 50 min.15 Chromatographic separation invariably introduces a significant dilution factor (10-100 times), meaning that the LOD for the injected sample will be a comparable factor higher than that for continuous analysis. Barnowski et al.20 have summarized the LODs for HPLC methods for chromium speciation. Even with highly sensitive inductively coupled plasma mass spectrometry (ICPMS) detection, the best LOD reported is still above 0.1 µg/L for both chromium species. Further, the chromatographic methods often require more than 5 min to complete a separation. Therefore, development of faster, more sensitive methods is still of great interest. For element-selective detection, plasma spectrometric techniques are favored over atomic absorption spectrometry (AAS) because of their generally higher sensitivity.10 Inductively coupled plasma atomic emission spectrometry (ICPAES) and ICPMS are the common detectors used in speciation studies. One aspect of ICP methods that has long been considered to be a hindrance to detection is the sample introduction process.21 Conventional sample introduction systems involve the pneumatic generation of aerosols which are processed in a spray chamber prior to injection into the ICP. However, this type of sample introduction is highly inefficient; at solution uptake rates of 1-2 mL/min, the transport efficiency is less than 3%.22,23 Among attempts to improve sample introduction efficiency, thermospray techniques have been developed and the subject of reviews.24,25 Thermospray aerosols are generated by forcing a liquid sample through a capillary tube that is heated to partially vaporize the solvent, resulting in a blast of vapor that converts the remaining liquid to droplets. It has been shown that thermo(18) Powell, M. J.; Boomer, D. W.; Wiederin, D. R. Anal. Chem. 1995, 67, 24742478. (19) Tomlinson, M. J.; Caruso, J. A. Anal. Chim. Acta 1996, 322, 1-9. (20) Barnowski, C.; Jakubowski, N.; Stuewer, D.; Broekaert, J. A. C. J. Anal. At. Spectrom. 1997, 12, 1155-1161. (21) Browner, R. F.; Boorn, A. W. Anal. Chem. 1984, 56, 875A-888A. (22) Montaser, A.; Golightly, D. W. Inductively Coupled Plasmas in Analytical Atomic Spectrometry, 2nd ed.; Wiley-VCH: New York, 1992; p 235. (23) Montaser, A.; Minnich, M. G.; McLean, J. A.; Liu, H.; Caruso, J. A.; McLeod, C. W. Sample Introduction in ICPMS. In Inductively Coupled Plasma Mass Spectrometry; Montaser, A., Ed.; Wiley-VCH: New York, 1998; p 83-264. (24) Koropchak, J. A.; Veber, M. Crit. Rev. Anal. Chem. 1992, 23, 113-141. (25) Conver, T. S.; Yang, J.; Koropchak, J. A. Spectrochim. Acta, Part B 1997, 52, 1087-1104.
spray sample introduction provides analyte transport efficiencies on the order of 20-50% at 1-2 mL/min solution uptake rates and improved LODs compared to those for conventional pneumatic nebulization.22-28 Compared with pneumatic nebulization, thermospray is a hightemperature and high-pressure process. Under such conditions, the properties of a compound such as electrical conductance29 and chemical reactivity30 may be quite different from those at room conditions. In our previous work, valence discrimination effects for certain elements have been observed with thermospray sample introduction (TSP) for ICPAES.16,30,31 The sensitivities for As(III), Sb(III), and Se(IV) were found to be significantly lower than those for As(V), Sb(V), and Se(VI), respectively. The mechanism identified for such results was that, in the thermospray system, the lower valence elements are reduced to the zerovalent metal forms which are insoluble and trapped in the vaporizer. Minimizing the sensitivity of the lower valence to a negligible level compared to that of higher valence allows a nonchromatographic separation. On the basis of this idea, a method for Se speciation was developed involving the addition of a mild reducing agent to depress Se(IV) sensitivity to a negligible level such that Se(VI) can be selectively determined. A second sample containing an oxidant to prevent Se(IV) reduction allowed the total concentration to be determined.30 Recent work has predicted that the solubility of Cr(III) and other metal species may be lower under conditions such as the high temperature and pressures of hydrothermal fissures32,33 and other high-temperature systems34 than the solubilities predicted by conventional solubility calculations. Pourbaix diagrams suggest that the solubility for Cr(III) can be less than 1 ng/mL at high temperature.34 If thermospray conditions can be established to this efficiency and below the LOD for Cr by ICPAES or ICPMS, a rapid, sensitive method for speciation of Cr can be developed. This work will describe our efforts to develop such a method. The feasibility of this method for speciation of Cr(III) and Cr(VI) will be demonstrated by analysis of spiked tap water and lake water samples. The accuracy of the method was tested by using a standard reference material and comparison to mobile phase ion-pairing chromatography. EXPERIMENTAL SECTION Instruments and Operating Conditions. All spectrometric measurements were performed using a Liberty 220 ICP spectrometer (Varian, Victoria, Australia), consisting of a plasma source powered by a crystal-controlled high-frequency generator operating at 40.68 MHz and a vacuum path monochromator with a Czerny-Turner mount. The spectrometer has a resolution of 9 (26) Yang, J.; Conver, T. S.; Koropchak, J. A.; Leighty, D. Spectrochim. Acta, Part B 1996, 51, 1491-1503. (27) Koropchak, J. A.; Veber, M.; Herries, J. Spectrochim. Acta, Part B 1992, 47, 825-834. (28) Conver, T. S.; Yang, J.; Koropchak, J. A.; Shkolnik, G.; Flajnik-Rivera, C. Appl. Spectrosc. 1997, 51, 68-73. (29) Goemans, M. G. E.; Funk, T. J.; Sedillo, M. A.; Buelow, S. J.; Anderson, G. K. J. Supercrit. Fluids 1997, 11, 61-72;. (30) Yang, J.; Conver, T. S.; Koropchak, J. A. Anal. Chem. 1996, 68, 40644071. (31) Yang, J.; Koropchak, J. A. Appl. Spectrosc. 1997, 51, 1573-1578. (32) Baes, C. F.; Mesmer, R. E. The Hydrolysis of Cations; John Wiley & Sons: New York, 1976; p 212. (33) Helgeson, H. Geochim. Cosmochim. Acta 1992, 56, 3191-3207. (34) Beverskog, B.; Puigdomenech, I. Corros. Sci. 1997, 39, 43-57.
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
3047
Table 1. Initial Experimental Conditions pump flow rate (mL/min) ICP conditions forward power (kW) outer Ar flow (L/min) intermediate Ar flow (L/min) carrier Ar flow (L/min) viewing height above load coil (mm) thermospray control temperature (°C) spray chamber temperature (°C) condenser temperature (°C)
0.45 1.0 15.0 1.5 0.50 1 130 100 10
pm for the second order. The Cr(II) 267.716 nm line was used as the analysis line in the second order. All data were collected at an integration time of 1 s with three replicates for standards and five replicates for blanks. The sample introduction system is similar to that reported elsewhere.30 A fused-silica capillary thermospray nebulizer was employed for aerosol generation in this work. Briefly, the thermospray vaporizer consisted of a resistively heated stainless steel tube (1.6-mm o.d., 0.5-mm i.d., 50-cm length), into which a fusedsilica capillary (0.36-mm o.d., 50-µm i.d.) was inserted and placed such that the capillary tip was just 3-5 mm beyond the end of the stainless steel tube. In this case, the capillary was of constant internal diameter and not terminated with an exit aperture as used before.30 Liquid samples were pumped through the fused-silica capillary. Solvent vaporization and aerosol generation resulted from the energy transfer from the heated stainless steel tube through the fused-silica wall and the surrounding annular space to the flowing liquid stream. The particular thermospray system used herein was a prototype constructed by Leeman Labs, Inc. This system employs a fused silica thermospray nebulizer and glassware arrangement as described by Koropchak et al.30 but uses digital controllers to monitor and control the operating temperatures as opposed to the analog triac controllers used in the prior publications. Specifically, two CN 9000A digital temperature controllers from Omega Engineering, Inc. (Stamford, CT) were used to monitor and regulate the thermospray control and spray chamber temperatures. A Varian 2510 HPLC pump continuously delivered a carrier flow to the nebulizer. The sample solutions were introduced in a flow injection manner using a Rheodyne (Cotati, CA) model 7125 metal-free injector following a 1-mL PEEK injection loop. Finally, the Ar carrier gas flow rate was monitored and controlled using a mass flow controller (model FC-280, Tylan General, Torrance, CA). Thermospray aerosols were input into a heated cylindrical Pyrex glass chamber. Desolvation of the aerosols was accomplished with a Friedrich condenser cooled to 10 °C. The condenser temperature was regulated using a refrigerated recirculating bath (Neslab, Newington, NH). The dry aerosol from the condenser was introduced into the ICP. The starting experimental conditions are listed in Table 1. The ICP operating conditions (viewing height, carrier gas flow rate), the control temperature for thermospray, the condenser temperature, and the spray chamber temperature were optimized to provide the maximum signal-to-background ratio for Cr(VI) at 267.716 nm. These conditions were used throughout this work, unless stated otherwise. 3048 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
Chemicals and Reagents. Analytical grade chemicals and deionized water were used for the preparation of all the solutions used in this study. A chromium(III) stock solution (1000 µg/mL) was prepared from chromium nitrate [Cr(NO3)3‚9H2O] from MC/B (Norwood, OH). A chromium(VI) stock solution (1000 µg/ mL) was prepared from potassium dichromate (K2Cr2O7) obtained from GFS (Columbus, OH). Acetic acid/sodium acetate buffer was prepared by mixing acetic acid stock solution and sodium acetate stock solution in different volume ratios to achieve different pHs. The acetic acid stock solution (0.20 M) was prepared from acetic acid (glacial) obtained from J. T. Baker Chemical Co. (Phillipsburg, NJ). Sodium acetate solution (0.20 M) was prepared from anhydrous sodium acetate from E. K. Industries, Inc. (Addison, IL). The pH meter used was a model 955 Accumet mini pH meter from Fisher Scientific (Fair Lawn, NJ), which was calibrated with pH 4.0 and 10.0 buffer solutions that also came from Fisher Scientific. The working solutions were prepared daily by diluting the stock solutions. Mobile Phase Ion-Pairing HPLC Separation (MPIP) of Chromium(III) and Chromium(VI). The mobile phase was metered with a Varian 2510 HPLC pump. The system also included a Rheodyne model 7125 injector fitted with a 100 µL injection loop and an Adsorbosphere C18 column (Alltech Associates, Inc., Deerfield, IL) with 4.6-mm i.d., 15-cm length, and 3-µm particle size. The mobile phase was 4.5 mM pentanesulfonic acid (Na salt) from Aldrich (St. Louis, MO). The pH of the eluent was adjusted to 3.75 with dilute acetic acid. The flow rate of the mobile phase was 1 mL/min. The eluent from the column was nebulized by a type K pneumatic nebulizer (J. E. Meinhard Associates, Inc., Santa Ana, CA) and introduced into the ICPAES above for detection. RESULTS AND DISCUSSION Thermospray Behavior of Chromium(III) and Chromium(VI). For solutions of Cr species that were not acidified (unlike the usual case for elemental analysis), we noticed an unusual behavior with thermospray sample introduction. For 1 µg/mL solutions, the sensitivity for Cr(III) was less than 200 counts above the background level (∼1000 counts), while the sensitivity for Cr(VI) was in the expected, “normal” range which is more than 30 000 counts above background. We adjusted the sample solution pH by adding either nitric acid or ammonia. The effect of pH on the sensitivity of both chromium forms is shown in Figure 1, which indicates that the sensitivity for Cr(VI) is relatively constant as a function of pH but the sensitivity for Cr(III) is quite dependent on pH. In the middle pH range (pH 3-5), the sensitivity is close to the background level, while at lower or higher pH (pH 5), the sensitivity increases gradually to a level comparable to that for Cr(VI) (e.g., at pH 1.5-2.0). These results are different from those for nonmetals reported earlier,30,31 which were more sensitive to the presence of redox agents in the solution. Redox agents were found to have no significant effects on the responses for Cr(III) and Cr(VI). Cr(III) has a strong tendency to form complexes.35 With the addition 5 × 10-4 M EDTA to unacidified solutions of either chromium species, the sensitivities measured for both species (35) Katz, S. A.; Salem, H. The Biological and Environmental Chemistry of Chromium; Wiley-VCH Publishers: New York, 1994; Chapter 2.
Figure 1. pH effect on chromium sensitivity: O, Cr(VI); b, Cr(III).
were almost the same, meaning that the Cr(III)-EDTA complex could be transported to the ICP for detection. This result indicates that the CrIII(aq) ion may change its form during the thermospray process, with the new form of chromium retained in the thermospary capillary, leading to lost sensitivity for Cr(III), while chelated forms do not undergo this conversion. The pressure in the thermospray system was found to rise slowly after injection of unacidified Cr(III) solutions, which suggested that substance was depositing inside the fused-silica capillary since the thermospray system pressure primarily results from the small internal diameter of the fused-silica capillary. If no step was taken to remove the deposit, the capillary would be plugged eventually after multiple injections. Peeling off the polyimide coating of a plugged capillary by immersing it into concentrated sulfuric acid for several days, we observed under a microscope that the capillary was filled with a green deposit in the middle part of the capillary (i.e., 20 cm away from the tip). When 10% (v/v) nitric acid was injected after the injection of a Cr(III) solution, the pressure showed a rapid drop and returned to the original pressure prior to Cr(III) injection. The time scan for the acid injection verified that chromium was flushed out by the injection of acid. From the standpoint of capillary plugging, it is not necessary to flush the capillary after every Cr(III) aliquot injection, but only when the pressure is close to the pump maximum. In our case, with a maximum pump pressure set to 300 kg/cm2 (i.e., 4200 psi), we injected the acid solution only when the pressure increased to around 200 kg/cm2 (original pressure was about 100 kg/cm2). For 1 µg/mL Cr(III) solutions, flushing the capillary was required after about 20 injections. The flush injection brings the pump pressure back to the original value almost immediately. To determine if the precipitation process was unique to thermospray, we measured unacidified samples using pneumatic nebulization ICPAES. At room temperature, an aliquot of 1 µg/ mL Cr(III) solution was centrifuged, and the upper layer solution was collected for PN-ICPAES measurement. With PN-ICPAES, the sensitivities obtained were the same for the Cr(III) solution and a 1 µg/mL Cr(VI) solution, which indicated that Cr(III) in such solution would not precipitate at room conditions. After heating a 1000 µg/mL Cr(III) solution in a boiling water bath for 5 h, we did not observe any visible precipitate. The solution was later centrifuged, and the upper layer solution was taken to make
a 1000-fold dilution. Comparing the sensitivities of the Cr(III) solution prepared in this way with that of a 1 µg/mL Cr(VI) solution using PN-ICPAES, we found there was no significant difference between them. This result suggested that even with a 1000 times higher concentration and a higher temperature, at atmospheric pressure, the precipitation reaction of Cr(III) does not occur. On the basis of these observations, we hypothesize that, in the high-temperature, high-pressure environment of the thermospray system, Cr(III) forms chromic hydroxide [Cr(OH)3] by hydrolysis. Considering the Cr(OH)3 solubility product constant (KSP ) 10-30.2), the molar concentration of 1 µg/mL Cr(III), and the pH of the aqueous solution (pH 4.6), at room temperature and pressure, the conversion of Cr(III) to Cr(OH)3 is not expected to occur. A distribution plot showing the relationship between different inorganic chromium species and pH shown elsewhere provides the same conclusion.11 As a result, the thermospray process initiates this conversion. Cr(OH)3 is insoluble in aqueous solution and deposits inside the capillary. Thus, it is not transported to the plasma. Cr(OH)3 exists as hydrous chromic oxide, Cr2O3‚nH2O, and is green.36 This color is consistent with that of the deposit within the capillary as observed under a microscope. With the precipitation and deposition hypothesis, the sensitivity dependence on pH (Figure 1) is rationalized. At lower or higher pHs, Cr(III) remains in soluble forms, such as chromic or chromite ion, which are transported with aerosol to the plasma for detection, as is Cr(VI). At the mid-pH range, Cr(OH)3 forms and is trapped in the thermospray system, eliminating the signal for Cr(III). To study the mechanism of deposition, we investigated the possibility that the fused silica of the capillary played a role in the chromium deposition, e.g., Cr(OH)3 binding with the fusedsilica silanol groups. To do so, we changed the fused-silica capillary vaporizer to a 100 µm i.d. stainless steel capillary vaporizer. The same result was obtained, which suggested that the deposition process is a physical process of precipitation onto the surface of the capillary. The vaporizer temperature is a primary operating parameter in thermospray systems because it controls the degree of solvent vaporization, influences the pressure within the nebulizer, and hence modifies the physical and chemical environment that the analyte experiences. With the vaporizer used in this study, the control temperature is referred to as the vaporizer temperature measured with the aid of a thermocouple located about 40 cm distant from the tip of the stainless steel capillary. We studied control temperature effects on solutions of different pHs. The dependence of the sensitivity for Cr(III) and Cr(VI) on control temperature is shown in Figure 2. In this figure, three different pHs were chosen to represent low-pH, mid-pH, and high-pH ranges, respectively. In all pH ranges, control temperature does not have much effect on sensitivity for Cr(VI). For Cr(III) at the mid-pH (3-5) range, the sensitivity increases slowly with increasing temperature. At the lower pH (pH 5), the sensitivity of Cr(III) is the same as that for Cr(VI) at low control temperature but shows a lower signal than Cr(VI) at higher control temperature, and the sensitivity difference between them enlarges with increasing temperature. The results suggest (36) Udy, M. J. Chromium; Reinhold: New York, 1956; p 118.
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
3049
Table 2. Control Temperature (°C) Effect on Sensitivity (counts ppm-1 s-1) and Signal-to-Background Ratio (S/B) of Chromium in Unacidified Solutionsa sensitivity
S/B
control temp
blank
Cr(VI)
Cr(III)
Cr(VI)
Cr(III)
90 100 120 125 130 135
750 814 897 900 844 850
20 782 20 250 28 711 32 222 25 098 20 803
1375 768 378 281 412 543
27.7 24.8 32.0 35.8 29.7 24.5
1.83 0.94 0.42 0.31 0.49 0.64
a All figures are based on the average of three replicates for samples and five replicates for blanks. The RSDs for all the measurements are within 5%.
Figure 2. Effects of thermospray control temperature on chromium sensitivity: O, Cr(VI); b, Cr(III). Key: (a) pH 1.5; (b) pH 4.0; (c) pH 8.0.
that, for Cr(III), the control temperature effect interacts with the solution pH effect. 3050 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
The effects of spray chamber temperature on sensitivity are similar for both species of chromium in either acidified solution or nonacidified solution and close to the observations shown in Figure 2a,b, meaning that the spray chamber temperature effect is similar to the control temperature effect. We hypothesize that this effect occurs because the spray chamber temperature also affects the vaporizer temperature to some extent. Optimization of Speciation Conditions. On the basis of the different thermospray behaviors of Cr(III) and Cr(VI) in unacidified solutions, we developed a nonchromatographic method to separate these two chromium forms. Although the sensitivity for 1 µg/mL Cr(III) in unacidified solution was about 150 times lower than that for Cr(VI), the remaining low sensitivity indicated that Cr(III) was not completely deposited under the conditions used so far. To develop a method for chromium speciation, we investigated the establishment of conditions where Cr(III) could deposit even more efficiently. The mechanism discussed previously for lower sensitivity of Cr(III) suggested that the thermospray process made the formation and precipitation of Cr(III) possible inside the vaporizer. Thus, any factors that affect thermospray behavior would influence the Cr(III) deposition as well. Thermospray control temperature is one primary factor. From the control temperature effect on both species (Figure 2b), lower control temperatures in unacidified solution are favorable to eliminate the Cr(III) signal. Focusing on the low control temperature range from 90 to 135 °C, the sensitivity and signal-to-background ratio data for both species of chromium are shown in Table 2. At 125 °C, Cr(VI) has the highest signal-to-background ratio (S/B) and Cr(III) has the lowest one. So, 125 °C was chosen as the control temperature for the remaining experiments. The pump flow rate determines the amount of liquid put into the vaporizer per unit time, which will affect the heat transfer and solvent vaporization rate of the thermospray process. As a result, this parameter was expected to also have an effect on Cr(III) deposition. We changed the pump flow rate from 0.25 mL/min to 0.45 mL/min in 0.05 mL/min increments with the consideration of the maximum system pressure. It was found that Cr(VI) sensitivity was highest and the deposition of Cr(III) was most complete at a flow rate of 0.35 mL/min. The solution pH obviously has a strong effect on Cr(III) deposition. We finely adjusted pH using 2 × 10-4 M acetic acid/ sodium acetate buffer solutions. Table 3 shows that, at pH 4.4,
Table 3. pH Effect on Cr(III) Sensitivity (counts ppm-1 s-1) at 125 °C Control Temperature and 0.35 mL/min Pump Flow Rate pH 3.6
3.8
4.0
4.2
4.4
4.6
4.8
5.0
5.2
5.4
sensitivity 240 281 184 107 12 141 207 352 578 822 S/Ba 0.22 0.26 0.17 0.10 0.01 0.13 0.19 0.32 0.53 0.80 a
Signal-to-background ratio.
Figure 3. Calibration curves for chromium in acidic solution (1% v/v nitric acid) with 5 × 10-4 M EDTA: O, Cr(VI); b, Cr(III).
the sensitivity for Cr(III) is almost zero; in essence, Cr(III) is quantitatively deposited at this pH. Overall, the conditions identified where Cr(III) can deposit most completely and Cr(VI) can be selectively determined are as follows: 125 °C control temperature, 0.35 mL/min pump flow rate, and pH 4.4. These conditions can be chosen if it is only Cr(VI) that is to be determined. The sensitivity ratio of Cr(VI)/Cr(III) was above 2500 under these conditions. If total chromium and chromium(III) by difference are to be determined, a second step is to find a condition where both chromium forms have the same sensitivity and thus can be used to determine the total concentration. In acidic solutions (1% v/v HNO3), Cr(III) and Cr(VI) behaved similarly, but the sensitivity for 1 µg/mL Cr(VI) was about 20% lower than that for Cr(III). Adding 5 × 10-4 M EDTA to both solutions eliminated this disparity (Figure 3). It should be pointed out that since an acid solution would flush out Cr(OH)3 that deposited previously, the system needs to be cleaned by an injection of 10% (v/v) HNO3 solution before detecting the total amount of chromium. If only the deposition of Cr(III) or detection of Cr(VI) is of interest, no more care needs to be taken as long as the pressure is in a tolerable range. Under those two conditions developed so far, Cr(VI) and Cr total concentration can be determined separately. The Cr(III) concentration can be obtained by calculating the difference. The conditions established also allow the option of solely determining the Cr(VI) concentration, if preferred.
Precision and Limits of Detection (LODs). The reproducibility of the entire system is dependent on several variables. Previous work has shown that the precision of thermospray as a sample introduction device has proven to be poorer than that of pneumatic sample introduction but comparable to those of other sample introduction methods.16,19,25 The limitation on precision of the thermospray-ICPAES system as a whole is related to the reproducibility of sample injection and the stabilization of the thermospray behavior, which can introduce pulsation to the flow system. The use of a pulse dampener or membrane desolvation has been shown to alleviate this problem.26,37 To demonstrate the system short-term precision, five injections of a 1 µg/mL Cr(VI) solution at pH 4.4 or in 1% HNO3 (v/v) acidified solution were measured sequentially. The resultant RSDs were 1.5% and 2.0% for Cr(VI) and total Cr measurements, respectively. With the 1 mL sample loop, one injection analysis could be done within 3 min, although this time could likely be reduced substantially with a well-designed flow injection system. Limits of detection for Cr(VI) and total chromium were determined using 3 times the standard deviation of a blank signal. The LODs were 0.47 µg/L for the Cr(VI) detection and 0.61 µg/L for the total chromium detection. Method Application and Accuracy. The method developed was applied to the determination of Cr(III) and Cr(VI) in spiked tap water samples. The results are listed in Table 4. In general, both the accuracy and the precision are satisfactory with average recoveries of 100% for Cr(VI) and 102% for Cr(III), even at a concentration as low as 20 µg/L and with a 10-fold concentration difference between Cr(III) and Cr(VI). For the solutions with a concentration difference of 50-fold (e.g., 20 µg/L Cr(VI) and 1000 µg/L Cr(III), or vice versa), the recoveries for total concentration and higher spiked concentration species are good; however, those for lower spiked concentration species have some deviation, and recoveries as high as 150% were observed. Because the Cr(VI) determination is based on complete deposition of Cr(III), any undeposited Cr(III) will obviously affect Cr(VI) accuracy, especially when the Cr(VI) is trace and the Cr(III) is much more concentrated. In addition, the Cr(III) concentration was calculated by difference, so any error for the data from either Cr(VI) or total Cr will affect the result for Cr(III). As concentrations of chromium in unpolluted natural waters are usually
