Sensitive and Selective Flow Injection Analysis of Hydrogen Sulfite

---

Anal. Chem. 2001, 73, 3187-3192

Technical Notes

Sensitive and Selective Flow Injection Analysis of Hydrogen Sulfite/Sulfur Dioxide by Fluorescence Detection with and without Membrane Separation by Gas Diffusion Hasson Mana and Uwe Spohn*

Institute of Biotechnology, University of Halle-Wittenberg, D-06120 Halle, Kurt-Mothes-Strasse 3, D-06120 Halle (Saale), Germany

Highly sensitive and selective FIA flow injection analysis procedures for the determination of sulfite/hydrogen sulfite/sulfur dioxide were developed on the basis of an in situ-generated o-phthalaldehyde (OPA)/ammonium reagent and fluorescence detection. The highest sensitivity was achieved at an excitation wavelength of 330 nm, an emission wavelength of 390 nm, and at pH 6.5. Sulfite concentrations between 2.5 nM and 5 µM can be determined with relative standard deviations between 10.5 and 1.0% (n ) 5, confidence level r ) 0.05) by utilization of a reagent that contains 0.2 mM OPA and 0.4 M NH4Cl in 50 mM potassium phosphate buffer. A concentration of 0.1 mM sulfite can be selectively detected in the presence of thiosulfate, thioglycolate, tetrathionate, cysteine, and ascorbate. The fluorometric sulfite detection was combined with a membrane gas diffusion step to improve the selectivity with respect to nonvolatile fluorescing substances. The total sulfite content can be quantitatively separated as sulfur dioxide into an acceptor solution before its flow detection. Between 40 nM and 0.1 mM sulfite can be determined. After 1000-fold dilution, the total sulfite content can be determined in white and red wines. Although many different sensors and determination procedures have been described, there is still a great demand for fast, sensitive, and selective procedures for the determination of hydrogen sulfite/sulfite/sulfur dioxide in foods and in environmental samples. In the following, the term S(IV) is used instead of the sum of sulfite/hydrogen sulfite/sulfur dioxide. S(IV) is widely used as a preservative in the food industry, where it is needed as a reactive antimicrobial agent. S(IV) is one of the most widely distributed environmental pollutants, which is generated mostly by internal combustion machines and thermal power stations. * Corresponding author: (tel) ++49-345-55 24 866; (fax) ++49-345-55 27 013; (e-mail) spohn.biochemtech.uni-halle.de. 10.1021/ac001049q CCC: $20.00 Published on Web 05/25/2001

© 2001 American Chemical Society

Many investigations on the electrochemical detection of sulfite, e.g., amperometric1-4 and potentiometric,5-8 have been performed. However, there are many optical detection procedures providing higher sensitivity, signal-to-noise ratio, and reliability. Sulfite can be determined photometrically after its reaction with 5,5′-dithiobis(2-nitrobenzoic acid),9-12 4,4′-dithiodipyridine,13 pararosaniline/ formaldehyde,14-17 and phthalaldehyde/ammonia18 at concentrations seldom lower than 10 µM. To improve the selectivity, both membrane11,13,16 and air gap5 gas diffusion, distillation,10 and pervaporation steps14 were used to separate sulfur dioxide from the sample matrix into acceptor solutions. Chemiluminescence detection19-23 opens up a way to the detection of nanomolar concentrations but utilizing highly reactive (1) Wygant, M. B.; Statler, J. A.; Henshall, A. J. Assoc. Off. Anal. Chem. 1997, 80, 1374-1380. (2) Azevedo, C. M. N.; Araki, K.; Toma, H. E.; Angnes, L. Anal. Chim. Acta 1999, 387, 175-180. (3) Benov, V. P.; Atanasov, B. P. Anal. Lett. 1993, 26, 2061-2069. (4) Cardwell, T. J.; Cattrall, R. W.; Chen, G. N.; Scollary, G. R.; Hamilton, I. C. J. Assoc. Off. Anal. Chem. 1993, 76, 1389-1393. (5) Marshall, G.B.; Midgley, D. Analyst 1983, 108, 701-711. (6) Badr, I. H. A.; Meyerhoff, M. E.; Hassan, S. S. M. Anal. Chim. Acta 1995, 310, 211-221. (7) Hutchins, R. S.; Molina, P.; Alajarin, M.; Vidal, A.; Bachas, L. G. Anal. Chem. 1994, 66, 3188-3192. (8) Ibrahim, I.; Cemal, Y.; Humeyra, B. Analyst 1996, 121, 1873-1876. (9) Humphrey, R. E.; Hinze; W. L.; Ward, W. H. Anal. Chem. 1970, 42, 698703. (10) Maquieira, A.; Casamayor, F.; Puchades, R.; Sagrado, S. Anal. Chim. Acta 1993, 283, 401-407. (11) Garcia Prieto, A. M.; Perez Pavon, J. L.; Moreno Cordero, B. Analyst 1994, 119, 2447-2452. (12) Gonzales, V.; Moreno, B.; Sicilia, D.; Rubio, S.; Perez-Bendito, D. Anal. Chem. 1993, 65, 1897-1902. (13) Frenzel, W.; Hillmann, B. Chem. Anal. (Warsaw) 1995, 40, 619-630. (14) Mataix, E.; Luque de Castro, M. D. Analyst 1998, 123, 1547-1549. (15) Richter, P.; Luque de Castro, M. D.; Valcarcel, M. Anal. Chim. Acta 1993, 283, 408-413. (16) Linares, P.; Luque de Castro, M. D.; Valcarcel, M. Anal. Chim. Acta 1989, 225, 443-448. (17) Fernandez, S. M. V.; Rangel, A. O. S. S.; Lima, J. L. F. J. Inst. Brew. 1998, 104, 203-205. (18) Abdel-Latif, M. S. Anal. Lett. 1994, 27, 2601-2614. (19) Koh, T.; Okabe, K. Analyst 1994, 119, 2457-2461.

Analytical Chemistry, Vol. 73, No. 13, July 1, 2001 3187

oxidants, which cause strongly restricted selectivity and operational stability. In comparison to the most widely used photometric detection, fluorometric detection often provides an improved signal-to-noise ratio combined with a higher selectivity. Fluorometric detection was applied to the determination of sulfite, after its derivatization with N-(9-acridinyl) maleimide,24-26 with monobromobimane,27 with phthalaldehyde only,28 and with phthalaldehyde/ethanolamine.29 The reaction of N-(9-acridinyl) maleimide with sulfite takes a relatively long reaction time. Because many other nucleophiles, e.g., hydrogen sulfide, are also reacting, the selectivity is strongly restricted. Monobromobimane reacts with many thiols and needs the use of organic solvents. Saltzman et al.29 determined SO2 in ambient air after its preconcentration in an absorber solution. A detection limit of 7 pptv was achieved with a measuring period of 8 min/sample corresponding to micromolar S(IV) concentrations in the absorber solution by using a methanolcontaining reagent at pH 9.0. Abdel-Latif and Guilbault30 demonstrated that o-phthalaldehyde (OPA), sulfite, and ammonia form a strongly fluorescing 1-sulfonatoisoindole derivative in alkaline media at pH 11 and used this principle to detect ammonia with high sensitivity. The aim of this paper was to develop fast, sensitive, and selective flow injection analysis (FIA) procedures for the fluorometric determination of sulfite in water samples and wine. EXPERIMENTAL SECTION Reagents. All chemicals were of analytical reagent grade and were used as received. Ammonium sulfate and OPA were purchased from Fluka (Buchs, Switzerland). All other chemicals were from Merck (Darmstadt, Germany). The standard, reagent, and buffer solutions were prepared from deionized and double distilled water. A stock solution of ∼10 mM S(IV) was prepared by dissolution of 630 mg of Na2SO3 in 500 mL of 1.0 mM NaOH/1 mM Na2EDTA degassed and equilibrated with nitrogen. The concentration was determined by back-titration of triiodide with a 10 mM thiosulfate standard solution with amperometric endpoint detection. Standard solutions with sulfite concentration lower than 0.1 mM were prepared in situ by dilution with a oxygen free solution of 1.0 mM NaOH containing 1 mM Na2EDTA. Apparatus. Figure 1 shows the different configurations of the applied FIA setups I-IV. All solutions were propelled by peristaltic pumps (Gilson Minipuls 3, Abimed, Langenfeld, Germany) with carefully calibrated pump channels. The injection valve (Rheodyne 9010, Cotati, CA) was equipped with defined loops of Teflon tubing (20) Yamada, M.; Nakada, T.; Suzuki, S. Anal. Chim. Acta 1983, 147, 401404. (21) Al-Tamrah, S. A.; Townsend, A.; Wheatley, A. R. Analyst 1987, 112, 883886. (22) Pauls, D. A.; Townsend, A. Analyst 1995, 120, 467-469. (23) He, Z. K.; Wu, F.; Meng, H.; Yuan, L.; Song, G.; Zeng, Y-E. Anal. Sci. 1998, 14, 737-740. (24) Dasgupta, P. K.; Yang, H.-Ch. Anal. Chem. 1986, 58, 2839-2844. (25) Meguro, H.; Takahashi, C.; Matsui, S.; Ohrui, H. Anal. Lett. 1983, 16, 1625-1632. (26) Akasaka, K.; Suzuki, T.; Ohrui, H.; Meguro, H. Anal. Sci. 1986, 2, 443446. (27) Ji, A. J.; Savon, S. R.; Jacobsen, D. W. Clin. Chem. 1995, 41, 897-903. (28) Takadate, A.; Fujino, H.; Obasa, M.; Goya, S. Chem. Pharm. Bull. 1986, 34, 1172-1175. (29) Saltzman, E. S.; Yvon, S. A.; Matrai, P. A. J. Atmos. Chem. 1993, 17, 7390. (30) Abdel-Latif, M. S.; Guilbault, G. G. J. Biotechnol. 1990, 14, 53-61.

3188 Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

Figure 1. FIA setups for the fluorometric determination of S(IV): I, simple setup; II, I with a continuously working gas dialysis cell; III, setup with a pulsed gas dialysis step; IV, with a gas dialysis cell in the injection loop of the injection valve; P1-6, channels of the peristaltic pumps; MC, mixing coil; RC, reaction coil; CC, conditioning coil; GDC, membrane gas diffusion cell.

(inner diameter di ) 0.3 mm). The FIA setups II-IV contain an earlier described gas diffusion cell (GDC)31 with a membrane exchange area of 385 mm2 and meander grooves with a depth of 0.1 mm and a width of 1.0 mm. Microporous Teflon membranes with a mean pore diameter of 0.2 µm (SM 11807, Sartorius, Go¨ttingen, Germany) were tightly placed between the mirror (31) Steube, K.; Spohn, U. J. Biotechnol. 1994, 33, 221-231.

symmetric meander plates, which were manufactured from plexiglass plates. The fluorescence detector F 1050 (Merck-Hitachi, Darmstadt, Germany) equipped with a 12-µL quarz cell is used as the flow detector. The peak signals are recorded and evaluated with respect to peak heights by the integrator D-2500 (MerckHitachi). The coils RC, MC, and CC were tightly knotted Teflon tubes with an inner diameter of 0.5 mm and lengths of 53, 20, and 10 cm, respectively. RESULTS AND DISCUSSION Fluorometric Detection of Sulfite. Preliminary investigations showed that hydrogen sulfite reacts with OPA in the presence of ammonium forming the highly fluorescing isoindole-1-sulfonate also in neutral and weakly acid solutions. At pH 6.5, the photometric spectrum shows a very broad range of similarly high absorption between 300 and 600 nm with absorption maximums at 330, 430, and 570 nm. The maximum fluorescence was measured at an excitation wavelength of 330 nm and an emission wavelength of 390 nm. The reagent was optimized under FIA conditions for an injected sulfite concentration of 0.1 mM by using the measuring setup I shown in Figure 1. The fluorescence reaches a maximum at pH 6.5 and between the phosphate concentrations of 50 and 80 mM. At pH 6.5 and a phosphate concentration of 75 mM, the fluorescence intensity increases to OPA concentration of 4 mM in the presence of 40 mM ammonium. Increasing the ammonium concentration, the fluorescence reaches a saturation level at 40 mM. It should be noted that this nearly optimum reagent I composition can only be adjusted in situ by 1 to 1 mixing of 8.0 mM OPA with 80 mM ammonium solutions, both prepared in 75 mM phosphate buffer and adjusted to pH 6.5. The resulting solution is unstable. Blackblue precipitates are formed slowly, which can be prevented by the use of lower OPA concentrations. At the optimum residence time of 22 s between the confluence point and the detector D, S(IV) was determined in the range between 0.1 µM and 0.1 mM. A calibration line was calculated by eq 1 from the concentrations Co and the measured peak heights h with r2 ) 0.9989 (confidence level R ) 0.05, number of repeated measurements n ) 4, number of measuring points m ) 10). The limit of detection was 50 nM.

lg h ) 1.00 lg(Co/µM) + 2.96

(1)

To detect even lower sulfite concentrations, the OPA content was decreased to 0.2 mM to increase the signal-to-noise ratio by decreasing the background fluorescence at least 10-fold. The ammonium concentration must be increased to 0.4 M to accelerate the reaction rate with hydrogen sulfite. This reagent II allows the detection of nanomolar concentrations of S(IV). Figure 2 shows a fiagram recorded for the detection of low sulfite concentrations in the range between 2.5 nM and 1.5 µM. The peak height can be correlated with the sulfite concentrations by eq 2 under FIA conditions with r2 ) 0.9996 (n ) 4, m ) 17). At least 20 determinations can be performed per hour.

lg h ) 0.99 lg(Co/nM) + 1.71

(2)

Sulfite Detection after Its Separation by Gas Diffusion through a Microporous Teflon Membrane. Because of the low excitation wavelength of 330 nm, there are many possible

fluoresceing interferences in real sample solutions taken from surface water and groundwater reservoirs, liquid foods, wine, and beer. Therefore, the application of an integrated separation step, which works according to the principle of gas diffusion across a hydrophobic and microporous membrane, was investigated thouroughly. As demonstrated earlier for the quantitative membrane separation of ammonia,32-34 carbon dioxide,34 and hydrogen sulfide,35 the possibility of complete separation also of S(IV) from acidified donor solutions and enrichment into the acceptor solution was investigated by using FIA setup II (Figure 1). Panels a and b of Figure 3 show the dependencies of the separation efficiency on the pH value in the donor stream and on the donor flow rate, respectively. The separation efficiency ΨGD is defined by eq 3 as the ratio between the S(IV) concentration [S(IV)]ao at the outlet of the acceptor channel of the GDC to the concentration [S(IV)]id at the inlet of the donor stream and increases with increasing conversion degree γSO2 ) [SO2]/[S(IV)] with the concentration [SO2] of the volatile sulfur dioxide and therefore with decreasing pH in the donor stream (Figure 3a). As shown by Figure 3b, ΨGD approaches 1 with decreasing donor flow rates, that means complete separation of the sulfite into the acceptor stream consisting of 0.1 M NaOH. Even though at equilibrium (25 °C) only ∼74% of S(IV) is present as SO2/H2SO3, a nearly complete separation is achieved

ΨGD ) [S(IV)]ao/[S(IV)]id

(3)

at flow rates smaller than 0.1 mL min-1, just at pH values lower than 1.3 in the temperature range between 25 and 55 °C. This result can be explained by the continuous elimination of SO2 from the donor solution and the continuous reequilibration converting nonvolatile S(IV) continuously into SO2/H2SO3. In Figure 3b it can also be seen that the influence of temperature changes increases with decreasing separation efficiencies. Because of the independence of the temperature at low flow rates, a significantly improved precision can be achieved in comparison to incompletely working separation cells. To implement a rapidly working FIA procedure, the separation cell was inserted between the injection valve and the flow detector as shown in Figure 1 (III). The in situ-generated OPA reagent served as the trapping acceptor solution. By using reagent II, sulfite can be determined in the range between 40 nM and 0.01 mM with excellent linearity of the calibration graph, which can be described by the double logarithmic eq 4 with r2 ) 0.9992 (R ) 0.05, m ) 17, n ) 4). A detection limit of 20 nM was calculated at a peak signal height that is 3-fold higher than the averaged noise amplitude. Up to 45 injected samples can be analyzed per hour. By using the FIA setup IV, sulfite can be enriched from the sample solution into the acceptor

lg h ) 1.03 lg(Co/nM) + 0.45

(4)

solution consisting of 0.1 M NaOH. During the first 10 min, the (32) Mana, H.; Spohn, U. Anal. Chim. Acta 1996, 325, 93-104. (33) Mana, H.; Spohn, U. Fresenius J. Anal. Chem. 2000, 366, 825-829. (34) Fuhrmann, B.; Spohn, U.; Mohr, K.-H. Biosens. Bioelectron. 1992, 7, 653660. (35) Becker, M.; Fuhrmann, B.; Spohn, U. Anal. Chim. Acta 1996, 324, 115123.

Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

3189

Figure 2. Fiagram for the fluorometric detection of 0-100 nM S(IV) at ATT 3 and 200-1500 nM S(IV) at ATT 7 (FIA setup I). Reagent II: 0.2 mM OPA, 0.4 M NH4Cl, 50 mM potassium phosphate. Injection volume: 78.6 µL, pH 6.5, Vcarrier ) Vreagent ) 0.16 mL min-1.

degree of enrichment increases linearly resulting in a ∼12-fold increase of the sensitivity. Selectivity of the Fluorometric Sulfite Detection under FIA Conditions. Table 1 summarizes the relative responses of some potential interferences of fluorometric sulfite detection both without and with separation by gas diffusion. Fluorometric sulfite detection shows an excellent selectivity against many sulfur compounds just without the gas diffusion step. Thiosulfate releases a small amount of sulfur dioxide at the short residence times in the donor stream of the GDC, which consists of 0.1 M HCl. At longer residence times, the selectivity can be decreased especially at higher thiosulfate concentrations. However, the background 3190 Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

fluorescence of 1 to 10 diluted red and white wine samples could be totally suppressed by the gas diffusion step. Although white wine samples could be analyzed for total S(IV) concentration, red wines caused an increasing blocking and fouling of the separation membrane. Analysis of Wines. The total free sulfite concentration is the sum of the concentrations of sulfite, hydrogen sulfite, and sulfur dioxide and is equal to [S(IV)]. In wines, an essential part of the total sulfite is bound to aldehyde and carbonyl groups, e.g., of acetaldehyde, reducing sugars, and phenolic compounds. The actually detected concentration of free sulfite is influenced by the equilibrium between bound and nonbound sulfite, the dilution of

Figure 3. Dependencies of the separation efficiency ΨS(IV) on (a) the donor pHD value at the donor flow rate VD ) 0.062 mL min-1 and (b) on VD ) VA the acceptor flow rate at pHD ) 1.3 at different temperatures measured near by the separation membrane: (b) 25, (1) 35, (9) 45, and ([) 55 °C, measured in FIA setup II by using 0.1 M NaOH as acceptor solution. Table 1. Selectivity of the Fluorometric Determination of Sulfite under FIA Conditions, Expressed as Peak Height Ratiosa peak height ratios without gas diffusion

interference sulfite bisulfite thiosulfate tetrathionate thioglycolate reduced glutathione cysteine L-lysine glycine formaldehyde ascorbate

0.1 mM interference 1.00 1.67 0.01 0.00 0.00 0.06 0.00 0.00

with gas diffusion

0.1 mM S(IV) + 1.0 mM interference

0.1 mM S(IV) + 0.1 mM interference

3.99 1.03 1.00

2.38 1.01 0.99 1.01 1.06 1.02 1.02 0.98 0.22 0.99

1.54

0.00

0.1 mM interference

0.00 0.00 0.00 0.00

0.00

0.1 mM S(IV) + 1.0 mM interference

0.1 mM S(IV) + 0.1 mM interference

1.01 0.99 1.00 1.00 1.00 1.00 0.39 0.94

1.00 0.98 1.00 1.00 1.00 0.85 0.99

a A concentration of 0.1 mM S(IV) correponds to 1.000, detection of interferences in the absence and in the presence of 0.1 mM sulfite, FIA setups I and II, respectively. injection volume, 10 µL, Vcarrier ) Vreagent ) 0.16 mL min-1, pH 6.5; donor stream, 0.1 M HCl.

the sample, and therefore also the total sulfite concentration. Therefore, the following investigations were restricted to the total sulfite concentration Co. The high sensitivity of the proposed fluorometric sulfite determination allows dilution of the wine samples by factors up to 10 000, whereby the oxidation by dissolved oxygen should be excluded. Assuming a total sulfite concentration of 1 mM and a dissociation degree R (eq 5) of 0.99 to avoid significant contributions to an analytical error of smaller than 1%, the dissociation constant of any bound sulfite should not be lower than 0.1 M. Dilution by a factor of 10 000 decreases this level of the dissociation constant to 10-5 M. Therefore, the peak height was measured in dependence on the factor, by which a red wine was diluted with 1 mM NaOH containing 1 mM EDTA. A dilution by a

R ) ([SO2] + [HSO3-] + [SO32-])/Co ) [S(IV)]/Co (5) factor of greater than 70 leads to an almost linear relationship between the peak height and the resulting total concentration Co.

It can be concluded that there is only a very small part of Co that remains bound. To support this conclusion, the multible standard addition method, which was proposed by Tyson,36 was applied. By using FIA setup I, series of both negative and positive standards with lower and higher sulfite concentrations, respectively, were injected into a 1000-fold diluted white wine. A regression line was calculated (eq 6) with r2 ) 0.9996, from which

h ) 67.7Co/µM - 130.2

(6)

a sulfite content of 1.92 ( 0.03 mM in the wine can be calculated. This result is in agreement with the concentration of 1.90 ( 0.04 mM, which was determined by the reference method. The reference method is based on the distillative separation of sulfur dioxide from acidified wine samples into a reservoir filled with standardized iodine solution, which is back-titrated with a standard solution of thiosulfate. (36) Tyson, J. F. Fresenius’ Z. Anal. Chem. 1988, 329, 663-667.

Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

3191

Table 2. Analysis of Wines by Fluorometric FIA and Titrimetric Distillation Methods (n ) 4, r ) 0.05) mM S(IV)

Emile Durand, 1998 Toscanello, A. Viviani, 1996 K. Spl. Rheinhessen, J. Gerhardt, 1994 Spa¨tburgunder, Baden, 1996

titration after SO2 distillation

fluorometric FIA

1.85 ( 0.04 2.69 ( 0.06 3.69 ( 0.08

1.89 ( 0.02 2.84 ( 0.03 3.92 ( 0.05

1.01 ( 0.03

1.03 ( 0.01

Table 2 summarizes some further results for the analysis of different wines, which demonstrate an acceptable aggreement with the reference method. Because the 1000-fold diluted wine samples did not show any inherent fluorescence, only a more or less great loss of sulfur dioxide during the titration process can explain the small but significant positive bias. CONCLUSIONS The total concentration of sulfite, hydrogen sulfite, and sulfur dioxide can be determined fluorometrically under flow injection conditions on the basis of an optimized OPA/ammonium reagent,

3192 Analytical Chemistry, Vol. 73, No. 13, July 1, 2001

which is produced in situ. The procedure achieves a surprisingly high selectivity and a sensitivity, which is comparable or higher than that of the known procedures to determine S(IV). Nanomolar concentrations of S(IV) can be determined without preconcentration. The proposed procedures additionally open up the way to a very fast determination of the total sulfite content in wines. The total S(IV) content of both white and red wines can be determined after their dilution without any separation step. The proposed sulfite detection can be combined with membrane gas diffusion procedures, which are performed continuously, pulsed after injection into the donor stream and in an enrichment mode. At a donor pH of smaller than 1, the S(IV) species can be completely separated as sulfur dioxide into an alkaline acceptor stream. ACKNOWLEDGMENT This work was supported by Deutsche Bundesstiftung Umwelt and the Ministry of Culture and Education of Sachsen-Anhalt. Received for review September 5, 2000. Accepted March 16, 2001. AC001049Q