Anal. Chem. 1984, 56, 1993,-1994 (5) Lowry, S.R.; Huppier, D. A. Anal. Chem. 1081, 53,889. (6) Rossiter, V. Am. Lab. (Fairfield, Conn.) 1982, 14 (June), 71-79. (7) Gariock, S. E.; Adams, G. E.; Smith, L. Am. Lab. (Falrfield Conn.) 1982, 14 (Dec), 49-55. (8) Kuehi, D.; Lemeny, G. J.; Griffiths, P. R. Anal. Spectrosc. 1980, 34, 222. (9) Shafer, K. H.; Cook, M.; DeRoos. R.; Jakobsen, R. J.; Rasario, J. D. Appl. Spectrosc. 1981, 35,459.
1993
(IO) Lowry, S. R.; Huppler, D. A. Anal. Chem. 1983, 55, 1288.
RECEIVED for review January 18, 1984. Accepted April 16, 1984. The financial assistance of the Air Force Wright AerLaboratories, Wright-Patterson AFB, OH! is gratefully appreciated. Onautical
Automated Determination of Sulfur( I V ) Using the Schiff Reactlon Gregory L. Kok,* Sonia N. Gitlin, Bruce W. Gandrud, and Allan L. Lazrus National Center for Atmospheric Research, Boulder, Colorado 80307 The recent modification of the Schiff reaction by Dasgupta e t al. (I) to determine S(1V) in aqueous solution is an improvement over the generally used West-Gaeke (2)technique. The improvements by Dasgupta et al. include the removal of the mercuric chloride fixing solution, improvement in sensitivity, and unambiguous analysis of S(1V) bound as the formaldehyde adduct, hydroxymethanesulfonic acid (HMSA). The latter is important since Kok et al. have found that the HMSA is resistant to hydrogen peroxide oxidation (3). In measurement of Southern California cloud water, micromolar concentrations of S(1V) and hydrogen peroxide (H202)have been found to coexist (4). The original studies by Dasgupta et al. oriented the analytical technique to the analysis of SO2 in the gas phase ( I ) . In this paper the technique is adapted for automated analysis using the Technicon segmented flow system. The reagent concentrations are optimized for the analysis of S(1V) in cloud and precipitation samples.
EXPERIMENTAL SECTION Reagents. Pararosaniline hydrochloride (PRA), J. T. Baker, was purified according to the technique of Scaringelli (5). A stock solution of 0.1% PRA was prepared in 1 N HC1. The analytical reagent was prepared by diluting 33 mL of the PRA solution and 40 mL of concentrated HCl in 250 mL of water. A stock buffered formaldehyde (HCHO) stabilizing solution was prepared from 8.5 mL of 37% HCHO, 36.4 g of 1,2-cyclohexylenedinitrilotetraacetic acid (CDTA) (Aldrich)and 8.0 g of NaOH diluted to 1L total volume. The final pH is approximately 4.8. The CDTA is neutralized to the disodium salt and serves as both the buffer and chelating agent for removal of metal ion interferences. To stabilize the S(1V) from oxidation and to provide for color development with the PRA, buffered HCHO stabilizing solution must be added to the samples. A ratio of 1 part of stabilizing solution to 15 parts of sample is used. The addition of the stabilizing solution to the sample should be done immediately after sample collection. The automated Technicon sampler contains a reservoir which provides solution to the analytical system between actual samples. This background solution is used to provide the base line signal of the test. This solution is prepared by diluting 1part of stabilizing solution to 15 parts of deionized water. It is important that the concentration of HCHO, as derived from the stabilizing solution, be identical in both the sample and background solutions. In the absence of S(1V) some reaction of the PRA and HCHO takes place to produce a small absorbance. If the HCHO concentrations are not matched in the sample and background solutions a bias will occur in the analytical results. Removal of the hydrogen peroxide (H202)interference is accomplished by destroying the H202with the enzyme catalse. A catalase working solution is prepared by diluting bovine catalase (Sigma Chemical Co. (2-100) 1:lOO. Standards of S(1V) were prepared from sodium metabisulfite or sodium formaldehyde
bisulfite. The use of Brij-35 surfactant solution (Technicon Instruments) is not recommended, as a high background absorbance is produced. Without the surfactant, minor base line drift is observed when working at high sensitivity. Instrumentation. The analytical system is designed around a Technicon Auto Analyzer I1 system. The flow cell in the colorimeter has a path length of 5.0 cm and absorbance measurements were made at 580 nm. Figure 1 shows the flow system designed to automate the Dasgupta et al. procedure for the determination of S(1V). Reagent flow rates were chosen to parallel the dilutions used in the original procedure (1). The autosampler is setup with a timing cam which provides for 20 samples per hour with a 1:2 sample-to-rinseratio. The flow manifold shown in Figure 1 incorporates four steps necessary for the analysis. The first step is the removal of the air bubble which enters the system with the change of the sampling arm from the rinse to the sample position. The second step in the manifold is the addition of the segmenting air bubble. If the additional air bubble introduced by the sampling arm is not removed it will disrupt the segment sequencing. The use of the air bar on the Technicon pump I1 is highly recommended for precise addition of the segmenting bubbles in the flow system. This provides higher reproducibility under the high liquid flow conditions used in the analysis. The third step in the manifold is the addition of 0.10 mL m i d of 5 N NaOH. The purpose of the NaOH is to breakup the HMSA to form S032and HCHO. The one-turn mixing coil used after the addition of the NaOH is fabricated by cutting down a standard multiturn mixing coil. The short mixing coil reduces the time for air oxidtaion of the S(1V). In the fourth step acidic PRA is added. The mixing in this step is crucial, since a competition for the S(1V) is setup between the acidic PRA to form the colored complex and the HCHO to re-form unreactive HMSA. The manifold is designed so that the sample is added to the acidic PRA, providing the highest PRA concentration possible. The fiial pH of the system must be between pH 0.9 and 1.1. Following the addition of reagents, a 10-min delay coil is used, allowing time for full color development.
RESULTS AND DISCUSSIONS The analytical technique is linear over the S(1V) concento 7.9 X lo4 M. The linear regression tration range 4.9 X equasion over this concentration range is: AbS = (4.31 x 104)[S(IV)- 0.016, r = 0.999 for n = 9. When extrapolated to zero concentration, the line gives a significant negative intercept. A series of calibration points over the concentration range 0-6.2 X lo-' M S(1V) indicate significant curvature and a zero intercept for the low concentration calibration. In detailed work on fluorimetric determination of S(1V) a t low concentrations, Dasgupta observed a similar negative intercept and curvature of the calibration line (6). The cause of this curvature is not presently understood.
0003-2700/S4/0356-1993$01.50/00 1984 American Chemical Society
1904
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
AUTO SAMPLER
*
ACIDFLEX PUMP T U B E
TECHNICON A P E COLORIMETER 58Onm5 c m CELL
-
-
Figure 1. Flow system diagram for the S(IV) analytical technique.
The reproducibility of the automated technique is better than 2% at the 2.7 X lo-’ M S(1V) concentration level. The detection limit for S(IV) is 1.5 X lo-’ M.
INTERFERENCE STUDIES In the original work Dasgupta et al. examined Fe(III), Cu(II), V(V), and Mn(I1) as interferences in the analytical technique (1). Only Mn(I1) interfered through an apparent redox reaction with the PRA. In the present work Mn(I1) was examined as an interference. In precipitation and cloud water the concentration of manganese will vary widely and is generally higher than the level examined by Dasgupta et al (I). Average manganese concentrations of 2.6 X lo4 M have been measured in Southern California cloud water with a maximum of 1.7 X 10” M (7). The stock stabilizing solution containing 0.1 M CDTA gives a final concentration of 6.3 X lom3M CDTA in the sample which is sufficient to prevent interference by 1.8 X M Mn(I1) in the presence of 0.05 to 1.0 X M S(1V). An interference in the analytical technique was also observed from the HzOzoxidation of S(1V). Previous studies have shown that Hz02and S(1V) can coexist when the S(1V) is bound as HMSA (3). In this analytical procedure the HMSA is broken up to form free S(1V) which can be rapidly oxidized by Hz02. Figure 2 shows the results when samples containing HzOzare analyzed. The first peak, A, is a standard of 1.2 X l0-e M S(IV). The second peak, B, shows the response when 3.0 X lo4 M HzOzhas been added. The H202 oxidizes the S(1V) after it is liberated from the HMSA and before it can react with the PRA. In peak C, 20 MLof the catalase working solution was added to a 3-mL sample of 3.0 X lo4 M H20zand 1.2 X lo4 M S(1V) prior to the analysis. The catalase quantatively destroys the HzOzwithout influencing the S(IV)analytical results. A parallel series of analyses were run using the manual procedure (1)and a similar Hz02interference was observed. Under the conditions used, the reaction of catalase with H20zis essentially instantaneous. Catalase can be used to remove organic peroxides, but a minimum reaction time of 15 min must be given before running the sample through the analytical procedure. Catalase can be mixed with the stabilizing solution directly; however,
0.0 0
6
12
IE
TIME (rnin)
Figure 2. Sulfur(1V) analytical system response and intcrference from
hydrogen peroxide. the catalase enzyme has a lifetime of approximately 24 h in the concentrated stabilizing solution and fresh solutions must be prepared daily.
CONCLUSION The modified Schiff reaction ( I ) which does not require mercuric chloride in the analytical procedure is a significant improvement over the originial West-Gaeke S(1V) analytical technique (2). Automation of the modified technique provides a detection limit of 1.5 X lo-’ M S(1V) and the capability to analyze 20 samples per hour. Registry No. H202,7722-84-1; S, 7704-34-9; Mn, 7439-96-5; water, 7732-18-5. LITERATURE CITED (1) Dasgupta, P. K.; DeCesare, K.; Ullrey, J. C. Anal. Chem. 1080, 52, 1912-1922. (2) West, P. W.; Gaeke, S. G. Anal. Chem. 1056, 28, 1816-1819. (3) Kok, G. L.; Gltlln, S. N.; Lazrus, A. L., unpublished results. 1984. (4) Rlchards, L. W.; Anderson, J. A.; Blumenthal, D. L.; McDonald, J. A.; Kok, G. L.; Lazrus, A. L. Afmos. Envlron. 1083, 17, 911-914. (5) Scaringelll, F. P; Saltzman, B. E.; Frey, S. A. Anal. Chem. 1067, 39, 1709-1719. (6) Dasgupta, P. K. Anal. Chem. 1081, 53, 2084-2087. (7) Rlchards, L. W.; Anderson, J. A.; Blumenthal, D. L.; Duckhorn, S. L.; McDonald, J. A. “Characterlzation of Reactants, Reaction Mechanisms, and Reaction Products Leading to Extreme Acid Rain and Acid Aerosol Conditions In Southern, California”; Final report to the California Alr Resources Board; CARB Agreement No. A0-140-32, June 1983.
RECEIVED March 12,1984. Accepted April 30,1984. This work was supported in part by the Electric Power Research Institute under research project 1630-12. The National Center for Atmospheric Research is sponsored by the National Science Foundation.
