摘要:

Electrophoretic Mobility Study of Ion Association Between Aromatic Anions and Quaternary Ammonium Ions in Aqueous Solution Toshio Takayanagi*, Eiko Wada and Shoji Motomizu Department of Chemistry, Faculty of Science, Okayama University, 3-1-1 Tsushimanaka, Okayama 700, Japan The ion association reaction between aromatic anions and quaternary ammonium ions in aqueous solutions was investigated through mobility changes in capillary electrophoresis. The electrophoretic mobility of aromatic anions decreased with increasing amount of quaternary ammonium salt added to the migrating solutions.The change in mobility of anions due to the ion association was found and was treated with a least-squares method, giving ion association constants. The ion associability order of the isomers of aromatic anions was found to be naphthalene-2,6-dicarboxylate > naphthalene-2,3-dicarboxylate, naphthalene-2-sulfonate > naphthalene-1-sulfonate and terephthalate > isophthalate > phthalate.The order of the ion associability of naphthalene-1,5- and -2,6-disulfonate reversed depending on the alkyl chain length of the pairing cation. The order of ion associability of quaternary ammonium ions was found to be tetraamylammonium > tetrabutylammonium > tetrapropylammonium Å octyltrimethylammonium Å hexyltrimethylammonium > tetraethylammonium > tetramethylammonium, and suggested that ion association in aqueous solution was governed by the hydrophobicity of the pairing cation.Keywords: Ion association; aqueous solution; capillary electrophoresis; aromatic anions; quaternary ammonium ions Capillary electrophoresis (CE) is a powerful separation system, especially for charged molecules. The separation of positional isomers, however, is still a difficult problem. Many approaches have been made to perform such separations. Cationic polymers, 1 acid dissociation properties2 and surfactants3 have been used for the separation of naphthalenesulfonate compounds.Hydrophobic cations have also been used as modifiers of migrating solutions. The use of cationic substances such as quaternary ammonium ions has been mainly investigated in capillary ion analysis (CIA).4–6 In a CIA system, the addition of long-chain quaternary ammonium ions could be used to reverse and control the electroosmotic flow (EOF), which permitted the detection of small anionic molecules directly and indirectly by spectrophotometry. The interaction between inorganic analyte anions and the modifiers, the cations, was found to be slight in such systems.The use of hydrophobic cations, however, should cause some interaction, ion association, with relatively bulky organic anions, such as naphthalenesulfonates and naphthalenecarboxylates. The use of ion association reagents have been investigated in the CE separation of anionic metal chelates.7,8 However, the contribution and steric effect of pairing ions in ion association have never been discussed.Ion association reactions have often been used in various analytical systems, such as two-phase distributions,9 precipitation of bulky ions and reversed-phase ion-pair chromatography. 10 In two-phase distribution systems, the contribution of the hydrophobicity of pairing ions has been found and estimated by using a fragmentation method.11,12 The ion association between inorganic cations and anions has often been investigated by using a conductivity measurement method.13,14 The ion association between bulky organic ions in aqueous media is very interesting with respect of the reaction between ionic biomolecules.Ion association in aqueous media, however, has not yet been clarified, because ion association is not as stable as chelation, and the ion associates formed are difficult to dissolve in aqueous media at high concentrations. It is very important and interesting to establish the dominant factors governing ion association.In this work, a fundamental study of ion association reactions in aqueous solutions was carried out to clarify the factors affecting the stability of ion associates. Quaternary ammonium ions were used and aromatic sulfonate and carboxylate ions were adopted as organic bulky anions, because positional isomers were available and were suitable as model reagents. Experimental Apparatus An Applied Biosystems (Foster City, CA, USA) Model 270A capillary electrophoresis system with a UV detector was used.Two types of a capillary were attached to the system: a fusedsilica capillary (50 mm id) (GL Sciences, Tokyo, Japan) and a CElect-N capillary (50 mm id) (Supelco, Bellefonte, PA, USA), where silanol groups were chemically coated on the inner wall. They were cut to the required length after a detection window had been made by burning off a small portion of the polyimide coating. By using the CElect-N capillary, the EOF was more suppressed than with an uncoated capillary.The capillaries were of 72 cm total length and 50 cm effective length to the UV detector. Reagents Analyte anions, such as sodium naphthalene-1-sulfonate (1-NS), sodium naphthalene-2-sulfonate (2-NS), disodium naphthalene-1,5-disulfonate (1,5-NDS), disodium naphthalene- 2,3-disulfonate (2,6-NDS), naphthalene-2,3-dicarboxylic acid (2,3-NDC), naphthalene-2,6-dicarboxylic acid (2,6-NDC), phthalic acid (PH), isophthalic acid (i-PH), and terephthalic acid (t-PH), were purchased from Tokyo Kasei Kogyo (Tokyo, Japan) and were used as received.Carboxylic compounds were dissolved in water with 2 mol equiv of sodium hydroxide. Migrating buffer solutions were prepared with sodium tetraborate (borax), potassium dihydrogenphosphate and disodium hydrogenphosphate, which were purchased from Wako (Osaka, Japan). Quaternary ammonium salts (Q+·X2), such as tetramethylammonium bromide (TMA+·Br2), tetraethylammonium bromide (TEA+·Br2), tetrapropylammonium bromide (TPA+·Br2), tetrabutylammonium bromide (TBA+·Br2), tetraamylammonium bromide (TAA+·Br2), hexyltrimethylammonium bromide (HTMA+·Br2) and octyltrimethylammo- Analyst, January 1997, Vol. 122 (57–62) 57nium bromide (OTMA+·Br2), were purchased from Tokyo Kasei Kogyo.Distilled, de-ionized water was used throughout. CE Measurement Migrating solution was transferred into a cathodic and an anodic reservoir, and then filled into a capillary tube before each measurement with a vacuum system. Sample solution was introduced into the capillary from the anodic end with the vacuum system for 3 s (injection volume about 9 nl).A voltage of 25 kV was then applied, and separations were started. Sample solution was introduced into a CElect-N capillary from the cathodic end and a UV detector was placed at the anodic end. The capillary was held in a thermostated room at 35 °C throughout the experiment, and analyte anions were detected photometrically at 230 nm. The EOF was monitored by adding about 3% v/v of ethanol to the sample solution. The EOF could be monitored by measuring the analyte in the method using a fused-silica capillary.However, it was necessary to determine the EOF in an inverse voltage direction when using the CElect- N capillary. The apparent electrophoretic mobility of an analyte, mepA, and the electroosmotic flow, mEOF, were calculated in the usual manner, using the eqns.(1) and (2). mEOF = LTLD VtEOF (1) mep ¢ = LTLD Vt - mEOF (2) where LT and LD are the total length of the capillary and its effective length to the UV detector, respectively, V is the applied voltage and tEOF and t are the migration times of EOF and an analyte anion, respectively. Three or more measurements were carried out to obtain the mean electrophoretic mobility in each instance. Determination of Ion Association Constants Ion association constants, Kass, between quaternary ammonium ions and aromatic anions were determined by measuring the change in electrophoretic mobility with the change in pairing ion concentrations.A least-squares method was applied to the analysis of the correlation of mepA. A laboratory-written program in BASIC, which gave both mean values and standard deviations, was developed. Results and Discussion Separation of Anions With Quaternary Ammonium Ions The separation of anions in a silica capillary in the absence and presence of appropriate amounts of quaternary ammonium salts in the migrating solutions is shown in Fig. 1. Anions with the same molecular mass and charge, such as positional isomers of naphthalenesulfonates and naphthalenedisulfonates, could not be separated in the absence of an ion association reagent, Q+, whereas some of them were well separated with the additional use of Q+, as shown in Fig. 1(b) and (c). The anionic isomers were also separated with the use of other Q+, such as TEA+, TPA+, TAA+, HTMA+ and OTMA+.The separation time was much longer in the presence of TMA+, HTMA+ and OTMA+ than in the absence of Q+, because less bulky cations adsorb fairly strongly on the internal surface of the silica capillary tube by an electrostatic interaction that suppresses the EOF, the velocity of which is usually larger than the electrophoretic mobility of the analyte anions. Changes in the velocity of the EOF with increasing Q+ concentration are shown in Fig. 2. It was found that the smaller the spherical quaternary ammonium ions, or the longer the alkyl chain in trimethyl-type ammonium ions, the more effectively they suppressed the EOF. Such quaternary ammonium ions, however, are less adsorptive on the silanol surface than tetradecyltrimethylammonium ions15 and other long-chain quaternary ammonium ions,16 which were used to reverse the direction of the EOF in the determination of anions by CIA. Fig. 1 Electropherograms of anions obtained with a fused-silica capillary in the absence and presence of quaternary ammonium salts.Sample solution: 1 310–5 m anions. Migrating solutions: 10 mm borax (pH, 9.2) and a certain amount of quaternary ammonium salt (Q+·X2). CE conditions: applied voltage 25 kV, capillary temperature 35 °C, detection wavelength 230 nm. Injection volume, 3 s (about 9 nl). Q+·X– in the migrating solutions: (a) none; (b) 20 mM TMA+·Br2; and (c) 20 mM TBA+·Br2. 1, 1-NS; 2, 2-NS; 3, 2,3-NDC; 4, 2,6-NDC; 5, 1,5-NDS; 6, 2,6-NDS; 7, PH; 8, i-PH; and 9, t-PH.Fig. 2 Change in the velocity of the EOF as a function of the amount of quaternary ammonium salt added in the migrating solution. CE conditions and injection volume as in Fig. 1. The EOF was monitored with 3% v/v ethanol in the sample solution. Migrating solution: 10 mm borax (pH, 9.2) and a certain amount of Q+·X2. 2, TMA+; 5, TEA+; «, TPA+; ~, TBA+; 8, TAA+; ., HTMA+; and /, OTMA+. 58 Analyst, January 1997, Vol. 122With HTMA+ and OTMA+, the formation of cationic micelles did not need to be considered, because the critical micelle concentration of OTMA+ was about 1.4 3 1021 m.17 In OTMA+ at high concentrations, three phthalate compounds could not be detected within 1 h of measurement. For TAA+, the baseline became much worse on increasing its concentration beyond 5 mm, because TAA+ has a high ion associability and forms the precipitates in the capillary. The migration order of the two isomers of NDS reversed in TPA+, although the migration order did not change with a concentration change in each Q+.Such a reversal was not found with other positional isomers. Separation of the two naphthalenesulfonate isomers was not achieved with TMA+ and TEA+ even at 20 mm, and was not yet achieved even with TAA+, because its concentration was too low. From these results, TBA+ was found to be the optimum modifier in respect of the separation efficiency and separation time.Separation of Anions in a Coated Capillary The separation of the anions was also examined in a polymercoated capillary, CElect-N, the EOF of which was well suppressed by the coating on the internal wall. The electropherogram obtained with the coated capillary is shown in Fig. 3. In this case, the detection end was the anodic end, and therefore the migration order of analyte anions was the opposite of that obtained in the fused-silica capillary. The separation behaviour was similar in the case of the silica capillary, which indicates that analyte anions will not interact with the capillary wall. Electrophoretic Mobility Change by Addition of Quaternary Ammonium Salts Changes in the apparent electrophoretic mobility, mepA, on addition of TMA+ and TBA+ are shown in Fig. 4(a) and (b). For each analyte anion, mepA decreased more significantly when the concentration of Q+ increased and the bulkiness of Q+ increased. The change in mobility of all the anions was very slight when TMA+ was adopted as the pairing ion.This indicates that such small cations find it difficult to form ion associates with the organic anions in aqueous solutions. A smaller change in mobility with an increase in the concentration of TMA+, as shown in Fig. 4(a), also suggests that the Joule heat generated in the capillary can be released during the electrophoresis, and that the electrophoretic mobility of anions is not affected much by the change in ionic strength of Q+ under the experimental conditions used in this work.As shown in Figs. 1 and 3, mepA of isomers decreased in the order: 2-NS > 1-NS, 2,6-NDC > 2,3-NDC and t-PH > i-PH > PH. As indicated in the previous section, the migration order or the mobility of the two naphthalenedisulfonate compounds reversed in TPA+. These results suggest that the ion associability of the anions is affected by the bulkiness and other properties of Q+. Determination of Ion Association Constants Ion association constants, Kass, were determined by using the change in electrophoretic mobility.The ion association constants of a few anions with TBA+ have already been examined in our previous work,18 where the apparent electrophoretic mobility, 2mepA, was plotted against the term (2mep + mepA)/ [Q+]. Such plots were also made in this study, as shown in Fig. 5. In this method, the mobility of the ion associates between anions and Q+ and the Kass values was obtained from the intercept on the ordinate and the slope of the plots, respectively.However, the mobility of anions, mep, in the absence of Q+ must be determined very precisely, because the value of mep is used as a standard mobility of the analyte anion in each plot. As can be seen from Fig. 5, plots for anions of low ion associability were dispersed, and the values of Kass could not be determined from the slopes. Hence the determination of ion association constants was limited to relatively stable ion associates formed between TBA+ and anions such as 1,5-NDS, 2,6-NDS and 2,6-NDC.In this study, we developed another method for the determination of ion association constants, which can give more reliable constants is applicable to monovalent anions of low ion associability. Fig. 3 Electropherogram of anions in a coated capillary. Sample solution, CE conditions and injection volume as in Fig. 1. Migrating solution contained 10 mm phosphate buffer (pH 7.0) and 20 mm TBA+·Br–. 1, 1-NS; 2, 2-NS; 3, 2,3-NDC; 4, 2,6-NDC; 5, 1,5-NDS; 6, 2,6-NDS; 7, PH; 8, i-PH; and 9, t-PH. Fig. 4 Change in apparent electrophoretic mobility of anions as a function of the molar concentration of quaternary ammonium salt added to the migrating solution. Conditions, except for Q+·Br– concentrations, as in Fig. 1. (a), TMA+; and (b), TBA+. 2, 1-NS; 5, 2-NS; «, 1,5-NDS; ~, 2,6-NDS; 8, 2,3-NDC; -, 2,6-NDC; ., PH; /, i-PH; and », t-PH. Analyst, January 1997, Vol. 122 59A 1 : 1 ion associate is assumed to be formed between an analyte anion (An2) and a quaternary ammonium ion (Q+) as shown in reaction (3), with an equilibrium constant given by eqn. (4). Q++An2"Q+·An2 (3) Kass = [Q+× An-- ] [Q+ ][An-- ] (4) The apparent electrophoretic mobility of a particular anion, mepA, can be represented as follows: mep ¢ = [An-- ] [An-- ] + [Q+× An-- ] ¥ mep + [Q+× An-- ] [An-- ] + [Q+× An-- ] ¥ mepIA (5) where mep and mepIA are the mobility of the analyte anion and that of the 1 : 1 ion associate, respectively.Eqn. (5) can be converted into eqn. (6) by using the mass balance for An2 and eqn. (4): mep ¢ = 1 1 + [Q+ ]Kass ¥ mep + [Q+ ]Kass 1 + [Q+ ]Kass ¥ mepIA (6) To calculate Kass, a non-linear least-squares method was applied. In the method, known values of [Q+] and mepA were used, and the values of the constants, mep, mepIA and Kass were obtained after the calculated curves had been optimized. The values of mep, mepIA and Kass for the aromatic anions with various Q+ are summarized in Tables 1, 2 and 3, respectively.The values obtained with the CElect-N capillary are also given. The calculated results for mep were found to be almost identical with the experimental results, as shown in Table 1. This indicates that the value of mep need not be obtained experimentally. From Table 2, it was found that mepIA was less than half of mep. This agrees with the fact that the apparent charge of divalent anions decreased to 21 on association with Q+ and that of monovalent anions with Q+ should be noncharged.Further, the values of mepIA indicate that the apparent molecular mass of the anions increases on ion association with Q+. In the case of 1-NS and 2-NS, the values of Kass were also calculated by assuming mepIA to be zero. The results for the Kass values of 1-NS and 2-NS by the two calculations were similar. The results for mep values obtained with the coated capillary were slightly larger than those obtained with the silica capillary, because the EOF in the coated capillary was very small and the correction of mep affected by the EOF could not be complete. Although the mep and mepIA values obtained by using the two kinds of capillary were slightly different from each other, the Kass values obtained were almost identical.Changes in the electrophoretic mobility were too small to determine the ion association constants when TMA+ was used as a pairing ion and phthalate ion was used.When the mobility of PH along with an increase in Q+ was used as a standard, however, the corrected mobility of 1,5-NDS and 2,6-NDS gave ion association constants of 100.6 ± 0.1 and 100.5 ± 0.1, respectively. The values are smaller than those with other Q+, which also indicates less associability of TMA+. The lines shown in Fig. 4(b) are the results obtained by the calculation; they show that the experimental data and calculated results agree favourably with each other.For the other cations, except for TMA+, the experimental data and calculated results also agreed well with each other. The ion association constants Fig. 5 Plots of mepA versus (2mep + mepA)/[TBA+]. Electrophoretic mobility used was the result obtained with a fused-silica capillary, using TBA+ as a pairing ion. Slopes for 1,5-NDS, 2,6-NDS and 2,6-NDC correspond to 1/Kass, which gave ion association constants of 101.6, 101.5, and 101.5, respectively.Other constants could not be obtained. 2, 1-NS; 5, 2-NS; «, 1,5-NDS; ~, 2,6-NDS; 8, 2,3-NDC; -, 2,6-NDC; /, i-PH; and “, t- PH. Table 1 Comparison of electrophoretic mobilities of anions, 2mep, between experimental and calculated results 2mep */1024 cm2 V21 s21 Q+ 1-NS 2-NS 1,5-NDS 2,6-NDS 2,3,-NDC 2,6-NDC PH i-PH t-PH TMA+ —† — — — — — — — — (3.04) (3.04) (4.95) (4.95) (4.47) (4.47) (4.99) (4.99) (4.99) TEA+ — — 4.96 ± 0.06 4.95 ± 0.06 — 4.47 ± 0.03 — — — (3.05) (3.05) (4.97) (4.97) (4.48) (4.48) (5.01) (5.01) (5.01) TPA+ 3.04 ± 0.06 3.05 ± 0.08 4.93 ± 0.08 4.93 ± 0.08 — 4.46 ± 0.08 — 5.00 ± 0.10 4.99 ± 0.08 (3.01) (3.01) (4.90) (4.90) (4.42) (4.42) (4.94) (4.94) (4.94) TBA+ 3.12 ± 0.08 3.12 ± 0.07 5.01 ± 0.08 5.02 ± 0.09 4.54 ± 0.07 4.55 ± 0.01 — 5.07 ± 0.06 5.05 ± 0.08 (3.12) (3.12) (5.02) (5.02) (4.54) (4.54) (5.06) (5.06) (5.06) TBA+ ‡ 3.61 ± 0.14 3.62 ± 0.15 5.80 ± 0.20 5.81 ± 0.22 5.24 ± 0.23 5.26 ± 0.23 — 5.80 ± 0.27 5.77 ± 0.21 (3.44) (3.44) (5.64) (5.64) (5.12) (5.12) (5.71) (5.71) (5.68) TAA+ 3.06 ± 0.08 3.07 ± 0.06 5.01 ± 0.09 5.02 ± 0.06 4.52 ± 0.04 4.52 ± 0.04 — 5.05 ± 0.02 5.05 ± 0.03 (3.09) (3.09) (5.03) (5.03) (4.53) (4.53) (5.06) (5.06) (5.06) HTMA+ 3.17 ± 0.06 3.17 ± 0.06 5.10 ± 0.02 5.10 ± 0.02 4.62 ± 0.03 4.62 ± 0.04 — 5.15 ± 0.04 5.14 ± 0.03 (3.15) (3.15) (5.10) (5.10) (4.61) (4.61) (5.13) (5.13) (5.13) OTMA+ 3.20 ± 0.05 3.20 ± 0.06 5.16 ± 0.10 5.16 ± 0.08 4.66 ± 0.09 4.67 ± 0.07 — 5.24 ± 0.04 5.22 ± 0.06 (3.21) (3.21) (5.18) (5.18) (4.68) (4.68) (5.22) (5.22) (5.22) * Error: 3s.The values in parentheses are the experimental results. † Not obtained. ‡ Values obtained in a coated capillary. 60 Analyst, January 1997, Vol. 122obtained in this work were close to those obtained in the previous work, which were 101.3–101.4 for the ion association between TBA+ and anions such as 1,5-NDS, 2,6-NDS, and 2,6-NDC.18 The fact that the ion association constant of 2-NS is larger than that of 1-NS seems to be related to the difference in the charge density of the functional group.The acid dissociation constant (pKa) of 1-naphthol (9.34) is smaller than that of 2-naphthol (9.51), and that of 1-naphthoic acid (3.70) is smaller than that of 2-naphthoic acid (4.17), which suggests that the negative charge of the functional group on the 2-position is larger than that on the 1-position. The more basic the functional group is, the more associable the anion should be with pairing ions.Contribution of Methylene Moiety to Ion Associability The ion association constants obtained by the least-squares method were plotted against the number of carbon atoms in Q+ (Fig. 6). For each aromatic anion, Kass increased with an increase in the carbon number or the size of Q+, that is, the logarithmic values of the ion association constants increased almost linearly with increase in the bulkiness of Q+, and their mean slope was about 0.06, which represents the increase in log Kass per methylene group. This indicates that a linear free energy relationship can be valid in an aqueous solution, as well as in the case of liquid–liquid distribution system of ion associates.Fig. 6 also indicates that the ion association in aqueous media is closely related to the hydrophobicity of the pairing cations. The contribution of the methylene moiety to log Kass, 0.06 (mean value), was one tenth of that in the liquid– liquid distribution system of ion associates, 0.6.11 In the liquid– Table 2 Electrophoretic mobility of ion associates, 2mepIA: calculated results 2mepIA */1024 cm2 V21 s21 Q+ 1-NS 2-NS 1,5-NDS 2,6-NDS 2,3-NDC 2,6-NDC PH i-PH t-PH TMA+ —† — — — — — — — — TEA+ — — 1.45 ± 0.61 1.79 ± 0.74 — 1.81 ± 0.37 — — — TPA+ 0.10 ± 0.83 0.00 ± 1.03 1.78 ± 0.49 1.78 ± 0.49 — 1.48 ± 0.41 — 1.58 ± 1.58 1.33 ± 1.43 TBA+ 0.66 ± 0.56 0.63 ± 0.46 2.07 ± 0.33 2.08 ± 0.32 1.46 ± 1.19 1.50 ± 0.35 — 1.59 ± 0.73 1.52 ± 0.82 TBA+‡ 0.33 ± 1.00 0.41 ± 0.92 2.34 ± 0.62 2.34 ± 0.60 1.46 ± 2.86 1.56 ± 0.60 — 1.49 ± 2.39 1.46 ± 2.26 TAA+ 0.77 ± 1.25 0.67 ± 0.88 2.43 ± 0.70 2.39 ± 0.46 1.69 ± 1.54 1.81 ± 0.23 — 1.79 ± 0.59 1.72 ± 0.73 HTMA+ 0.49 ± 0.69 0.49 ± 0.69 2.26 ± 0.17 2.26 ± 0.15 1.96 ± 0.53 1.91 ± 0.30 — 1.42 ± 1.28 1.33 ± 0.77 OTMA+ 0.69 ± 0.57 0.50 ± 0.37 2.11 ± 0.58 2.15 ± 0.48 2.00 ± 2.12 1.71 ± 0.45 — 1.63 ± 0.90 1.79 ± 1.20 * Error: 3s.† Not obtained. ‡ Values obtained in a coated capillary.Table 3 Ion association constants, Kass, obtained by the least-squares method Log Kass * Q+ 1-NS† 2-NS† 1,5-NDS 2,6-NDS 2,3-NDC 2,6-NDC PH i-PH t-PH TMA+ —‡ — 0.6 ± 0.1§ 0.5 ± 0.1§ — 0.6 ± 0.1§ — — — TEA+ — — 0.91 ± 0.10 0.85 ± 0.13 — 0.92 ± 0.08 — — — TPA+ 0.89 ± 0.17 0.93 ± 0.21 1.17 ± 0.08 1.17 ± 0.08 — 1.25 ± 0.08 — 0.70 ± 0.29 0.70 ± 0.22 (0.77 ± 0.16) (0.81 ± 0.19) TBA+ 1.13 ± 0.14 1.18 ± 0.10 1.39 ± 0.07 1.44 ± 0.07 0.70 ± 0.23 1.48 ± 0.07 — 0.87 ± 0.11 0.90 ± 0.12 (0.98 ± 0.15) (1.06 ± 0.10) TBA+¶ 1.12 ± 0.19 1.20 ± 0.18 1.45 ± 0.12 1.49 ± 0.13 0.79 ± 0.15 1.49 ± 0.12 — 0.89 ± 0.37 0.91 ± 0.29 (1.07 ± 0.17) (1.14 ± 0.16) TAA+ 1.36 ± 0.42 1.38 ± 0.20 1.64 ± 0.16 1.69 ± 0.10 0.90 ± 0.35 1.71 ± 0.06 — 0.95 ± 0.09 1.08 ± 0.11 (1.21 ± 0.40) (1.26 ± 0.22) HTMA+ 0.91 ± 0.14 0.91 ± 0.14 1.18 ± 0.03 1.15 ± 0.03 0.78 ± 0.16 1.18 ± 0.06 — 0.64 ± 0.19 0.70 ± 0.10 (0.83 ± 0.14) (0.83 ± 0.14) OTMA+ 0.91 ± 0.13 0.91 ± 0.07 1.20 ± 0.11 1.22 ± 0.09 0.61 ± 0.75 1.20 ± 0.09 — 0.77 ± 0.14 0.84 ± 0.20 (0.79 ± 0.14) (0.83 ± 0.08) * Error: 3s.† Values in parentheses were the results when mepIA was assumed to be zero. ‡ Not obtained. § These constants were obtained by assuming that PH did not form ion associate with TMA+ and using the increase in the electrophoretic mobility of pH with increase in the electrophoretic mobility of PH with increase in the concentration of Q+ as the effect of temperature and ionic strength on the mobility of ions.¶ Values obtained in a coated capillary. Fig. 6 Relationships between the ion association constants and the carbon number of quaternary ammonium ions (Q+). 2, 1-NS; 5, 2-NS; «, 1,5-NDS; ~, 2,6-NDS; 8, 2,3-NDC; -, 2,6-NDC; /, i-PH; and “, t- PH. Analyst, January 1997, Vol. 122 61liquid distribution system, the reaction and its equilibrium constant are represented by the equations Q++A2"Q+·A(o) (7) Kex = [Q+× A-- ](o) [Q+ ][A-- ] = KassKD,ip (8) where the subscript (o) denotes the species in the organic phase, Kex is the extraction constant of a ion associate, Q+·A2, and KD,ip is the distribution coefficient of the ion associate between the aqueous and organic phase.The present results indicate that the contribution of a methylene moiety to the distribution process is a more dominant factor in the extraction equilibrium than in the ion association process in aqueous media. Conclusion This study has demonstrated the usefulness of ion association reactions in aqueous solution for the improvement of CE separations.Ion association constants were determined by measuring the electrophoretic mobility changes of anionic species. Ion associability was shown to be affected by the hydrophobicity of the pairing ion. The contribution of the hydrophobicity to ion association constants was relatively small, 0.05–0.07 for «log Kass per methylene moiety, compared with those in the solvent extraction of ion associates.Ion association combined with CE enabled us to separate anion isomers effectively. This work was supported by a Grant-in-Aid for Scientific Research (No. 07454203) from the Ministry of Education, Science and Culture, Japan. References 1 Terabe, S., and Isemura, T., Anal. Chem., 1990, 62, 650. 2 Zenki, M., Irizawa, M., and Yukutake, H., Bunseki Kagaku, 1995, 44, 227. 3 Zenki, M., Yukutake, H., and Irizawa, M., Bunseki Kagaku, 1996, 45, 181. 4 Jandik, P., and Jones, W. R., J. Chromatogr., 1991, 546, 431. 5 Jones, W. R., J. Chromatogr., 1993, 640, 387. 6 Buchberger, W., Cousins, S. M., and Haddad, P. R., Trends Anal. Chem., 1994, 13, 313. 7 Iki, N., Hoshino, H., and Yotsuyanagi, T., J. Chromatogr. A., 1993, 652, 539. 8 Motomizu, S., Kuwabara, M., and Oshima, M., Bunseki Kagaku, 1994, 43, 621. 9 Motomizu, S., Bunseki Kagaku, 1989, 38, 147. 10 Haddad, P. R., and Jackson, P. E., Journal of Chromatography Library, Vol. 46, Ion Chromatography, Elsevier, Amsterdam, 1990. 11 Motomizu, S., Hamada, S., and T�oei, K., Bunseki Kagaku, 1983, 32, 648. 12 Matsunaga, H., and Yotsuyanagi, T., Nippon Kagaku Kaishi, 1982, 785. 13 Davies, C. W., Ion Association, Butterworths, London, 1962. 14 Nancollas, G. H., Interactions in Electrolyte Solutions, Elsevier, Amsterdam, 1966. 15 Huang, X., Luckey, J. A., Gordon, M. J., and Zare, R. N., Anal. Chem., 1989, 61, 766. 16 Benz, N. J., and Fritz, J. S., J. Chromatogr. A., 1994, 671, 437. 17 Rosen, M. J., Surfactants and Interfacial Phenomena, Wiley, New York, 1989, p. 125. 18 Takayanagi, T., and Motomizu, S., Chem. Lett., 1995, 593. Paper 6/05047F Received July 22, 1996 Accepted October 15, 1996 62 Analyst, January 1997, Vol. 122 Electrophoretic Mobility Study of Ion Association Between Aromatic Anions and Quaternary Ammonium Ions in Aqueous Solution Toshio Takayanagi*, Eiko Wada and Shoji Motomizu Department of Chemistry, Faculty of Science, Okayama University, 3-1-1 Tsushimanaka, Okayama 700, Japan The ion association reaction between aromatic anions and quaternary ammonium ions in aqueous solutions was investigated through mobility changes in capillary electrophoresis.The electrophoretic mobility of aromatic anions decreased with increasing amount of quaternary ammonium salt added to the migrating solutions. The change in mobility of anions due to the ion association was found and was treated with a least-squares method, giving ion association constants. The ion associability order of the isomers of aromatic anions was found to be naphthalene-2,6-dicarboxylate > naphthalene-2,3-dicarboxylate, naphthalene-2-sulfonate > naphthalene-1-sulfonate and terephthalate > isophthalate > phthalate.The order of the ion associability of naphthalene-1,5- and -2,6-disulfonate reversed depending on the alkyl chain length of the pairing cation. The order of ion associability of quaternary ammonium ions was found to be tetraamylammonium > tetrabutylammonium > tetrapropylammonium Å octyltrimethylammonium Å hexyltrimethylammonium > tetraethylammonium > tetramethylammonium, and suggested that ion association in aqueous solution was governed by the hydrophobicity of the pairing cation.Keywords: Ion association; aqueous solution; capillary electrophoresis; aromatic anions; quaternary ammonium ions Capillary electrophoresis (CE) is a powerful separation system, especially for charged molecules.The separation of positional isomers, however, is still a difficult problem. Many approaches have been made to perform such separations. Cationic polymers, 1 acid dissociation properties2 and surfactants3 have been used for the separation of naphthalenesulfonate compounds. Hydrophobic cations have also been used as modifiers of migrating solutions. The use of cationic substances such as quaternary ammonium ions has been mainly investigated in capillary ion analysis (CIA).4–6 In a CIA system, the addition of long-chain quaternary ammonium ions could be used to reverse and control the electroosmotic flow (EOF), which permitted the detection of small anionic molecules directly and indirectly by spectrophotometry.The interaction between inorganic analyte anions and the modifiers, the cations, was found to be slight in such systems. The use of hydrophobic cations, however, should cause some interaction, ion association, with relatively bulky organic anions, such as naphthalenesulfonates and naphthalenecarboxylates.The use of ion association reagents have been investigated in the CE separation of anionic metal chelates.7,8 However, the contribution and steric effect of pairing ions in ion association have never been discussed. Ion association reactions have often been used in various analytical systems, such as two-phase distributions,9 precipitation of bulky ions and reversed-phase ion-pair chmatography. 10 In two-phase distribution systems, the contribution of the hydrophobicity of pairing ions has been found and estimated by using a fragmentation method.11,12 The ion association between inorganic cations and anions has often been investigated by using a conductivity measurement method.13,14 The ion association between bulky organic ions in aqueous media is very interesting with respect of the reaction between ionic biomolecules. Ion association in aqueous media, however, has not yet been clarified, because ion association is not as stable as chelation, and the ion associates formed are difficult to dissolve in aqueous media at high concentrations. It is very important and interesting to establish the dominant factors governing ion association.In this work, a fundamental study of ion association reactions in aqueous solutions was carried out to clarify the factors affecting the stability of ion associates. Quaternary ammonium ions were used and aromatic sulfonate and carboxylate ions were adopted as organic bulky anions, because positional isomers were available and were suitable as model reagents. Experimental Apparatus An Applied Biosystems (Foster City, CA, USA) Model 270A capillary electrophoresis system with a UV detector was used.Two types of a capillary were attached to the system: a fusedsilica capillary (50 mm id) (GL Sciences, Tokyo, Japan) and a CElect-N capillary (50 mm id) (Supelco, Bellefonte, PA, USA), where silanol groups were chemically coated on the inner wall.They were cut to the required length after a detection window had been made by burning off a small portion of the polyimide coating. By using the CElect-N capillary, the EOF was more suppressed than with an uncoated capillary. The capillaries were of 72 cm total length and 50 cm effective length to the UV detector. Reagents Analyte anions, such as sodium naphthalene-1-sulfonate (1-NS), sodium naphthalene-2-sulfonate (2-NS), disodium naphthalene-1,5-disulfonate (1,5-NDS), disodium naphthalene- 2,3-disulfonate (2,6-NDS), naphthalene-2,3-dicarboxylic acid (2,3-NDC), naphthalene-2,6-dicarboxylic acid (2,6-NDC), phthalic acid (PH), isophthalic acid (i-PH), and terephthalic acid (t-PH), were purchased from Tokyo Kasei Kogyo (Tokyo, Japan) and were used as received.Carboxylic compounds were dissolved in water with 2 mol equiv of sodium hydroxide. Migrating buffer solutions were prepared with sodium tetraborate (borax), potassium dihydrogenphosphate and disodium hydrogenphosphate, which were purchased from Wako (Osaka, Japan).Quaternary ammonium salts (Q+·X2), such as tetramethylammonium bromide (TMA+·Br2), tetraethylammonium bromide (TEA+·Br2), tetrapropylammonium bromide (TPA+·Br2), tetrabutylammonium bromide (TBA+·Br2), tetraamylammonium bromide (TAA+·Br2), hexyltrimethylammonium bromide (HTMA+·Br2) and octyltrimethylammo- Analyst, January 1997, Vol. 122 (57–62) 57nium bromide (OTMA+·Br2), were purchased from Tokyo Kasei Kogyo. Distilled, de-ionized water was used throughout. CE Measurement Migrating solution was transferred into a cathodic and an anodic reservoir, and then filled into a capillary tube before each measurement with a vacuum system. Sample solution was introduced into the capillary from the anodic end with the vacuum system for 3 s (injection volume about 9 nl). A voltage of 25 kV was then applied, and separations were started. Sample solution was introduced into a CElect-N capillary from the cathodic end and a UV detector was placed at the anodic end.The capillary was held in a thermostated room at 35 °C throughout the experiment, and analyte anions were detected photometrically at 230 nm. The EOF was monitored by adding about 3% v/v of ethanol to the sample solution. The EOF could be monitored by measuring the analyte in the method using a fused-silica capillary.However, it was necessary to determine the EOF in an inverse voltage direction when using the CElect- N capillary. The apparent electrophoretic mobility of an analyte, mepA, and the electroosmotic flow, mEOF, were calculated in the usual manner, using the eqns. (1) and (2). mEOF = LTLD VtEOF (1) mep ¢ = LTLD Vt - mEOF (2) where LT and LD are the total length of the capillary and its effective length to the UV detector, respectively, V is the applied voltage and tEOF and t are the migration times of EOF and an analyte anion, respectively.Three or more measurements were carried out to obtain the mean electrophoretic mobility in each instance. Determination of Ion Association Constants Ion association constants, Kass, between quaternary ammonium ions and aromatic anions were determined by measuring the change in electrophoretic mobility with the change in pairing ion concentrations. A least-squares method was applied to the analysis of the correlation of mepA.A laboratory-written program in BASIC, which gave both mean values and standard deviations, was developed. Results and Discussion Separation of Anions With Quaternary Ammonium Ions The separation of anions in a silica capillary in the absence and presence of appropriate amounts of quaternary ammonium salts in the migrating solutions is shown in Fig. 1. Anions with the same molecular mass and charge, such as positional isomers of naphthalenesulfonates and naphthalenedisulfonates, could not be separated in the absence of an ion association reagent, Q+, whereas some of them were well separated with the additional use of Q+, as shown in Fig. 1(b) and (c). The anionic isomers were also separated with the use of other Q+, such as TEA+, TPA+, TAA+, HTMA+ and OTMA+. The separation time was much longer in the presence of TMA+, HTMA+ and OTMA+ than in the absence of Q+, because less bulky cations adsorb fairly strongly on the internal surface of the silica capillary tube by an electrostatic interaction that suppresses the EOF, the velocity of which is usually larger than the electrophoretic mobility of the analyte anions.Changes in the velocity of the EOF with increasing Q+ concentration are shown in Fig. 2. It was found that the smaller the spherical quaternary ammonium ions, or the longer the alkyl chain in trimethyl-type ammonium ions, the more effectively they suppressed the EOF. Such quaternary ammonium ions, however, are less adsorptive on the silanol surface than tetradecyltrimethylammonium ions15 and other long-chain quaternary ammonium ions,16 which were used to reverse the direction of the EOF in the determination of anions by CIA.Fig. 1 Electropherograms of anions obtained with a fused-silica capillary in the absence and presence of quaternary ammonium salts. Sample solution: 1 310–5 m anions. Migrating solutions: 10 mm borax (pH, 9.2) and a certain amount of quaternary ammonium salt (Q+·X2). CE conditions: applied voltage 25 kV, capillary temperature 35 °C, detection wavelength 230 nm.Injection volume, 3 s (about 9 nl). Q+·X– in the migrating solutions: (a) none; (b) 20 mM TMA+·Br2; and (c) 20 mM TBA+·Br2. 1, 1-NS; 2, 2-NS; 3, 2,3-NDC; 4, 2,6-NDC; 5, 1,5-NDS; 6, 2,6-NDS; 7, PH; 8, i-PH; and 9, t-PH. Fig. 2 Change in the velocity of the EOF as a function of the amount of quaternary ammonium salt added in the migrating solution. CE conditions and injection volume as in Fig. 1. The EOF was monitored with 3% v/v ethanol in the sample solution. Migrating solution: 10 mm borax (pH, 9.2) and a certain amount of Q+·X2. 2, TMA+; 5, TEA+; «, TPA+; ~, TBA+; 8, TAA+; ., HTMA+; and /, OTMA+. 58 Analyst, January 1997, Vol. 122With HTMA+ and OTMA+, the formation of cationic micelles did not need to be considered, because the critical micelle concentration of OTMA+ was about 1.4 3 1021 m.17 In OTMA+ at high concentrations, three phthalate compounds could not be detected within 1 h of measurement. For TAA+, the baseline became much worse on increasing its concentration beyond 5 mm, because TAA+ has a high ion associability and forms the precipitates in the capillary.The migration order of the two isomers of NDS reversed in TPA+, although the migration order did not change with a concentration change in each Q+. Such a reversal was not found with other positional isomers. Separation of the two naphthalenesulfonate isomers was not achieved with TMA+ and TEA+ even at 20 mm, and was not yet achieved even with TAA+, because its concentration was too low.From these results, TBA+ was found to be the optimum modifier in respect of the separation efficiency and separation time. Separation of Anions in a Coated Capillary The separation of the anions was also examined in a polymercoated capillary, CElect-N, the EOF of which was well suppressed by the coating on the internal wall. The electropherogram obtained with the coated capillary is shown in Fig. 3. In this case, the detection end was the anodic end, and therefore the migration order of analyte anions was the opposite of that obtained in the fused-silica capillary. The separation behaviour was similar in the case of the silica capillary, which indicates that analyte anions will not interact with the capillary wall. Electrophoretic Mobility Change by Addition of Quaternary Ammonium Salts Changes in the apparent electrophoretic mobility, mepA, on addition of TMA+ and TBA+ are shown in Fig. 4(a) and (b). For each analyte anion, mepA decreased more significantly when the concentration of Q+ increased and the bulkiness of Q+ increased. The change in mobility of all the anions was very slight when TMA+ was adopted as the pairing ion. This indicates that such small cations find it difficult to form ion associates with the organic anions in aqueous solutions. A smaller change in mobility with an increase in the concentration of TMA+, as shown in Fig. 4(a), also suggests that the Joule heat generated in the capillary can be released during the electrophoresis, and that the electrophoretic mobility of anions is not affected much by the change in ionic strength of Q+ under the experimental conditions used in this work. As shown in Figs. 1 and 3, mepA of isomers decreased in the order: 2-NS > 1-NS, 2,6-NDC > 2,3-NDC and t-PH > i-PH > PH. As indicated in the previous section, the migration order or the mobility of the two naphthalenedisulfonate compounds reversed in TPA+.These results suggest that the ion associability of the anions is affected by the bulkiness and other properties of Q+. Determination of Ion Association Constants Ion association constants, Kass, were determined by using the change in electrophoretic mobility. The ion association constants of a few anions with TBA+ have already been examined in our previous work,18 where the apparent electrophoretic mobility, 2mepA, was plotted against the term (2mep + mepA)/ [Q+].Such plots were also made in this study, as shown in Fig. 5. In this method, the mobility of the ion associates between anions and Q+ and the Kass values was obtained from the intercept on the ordinate and the slope of the plots, respectively. However, the mobility of anions, mep, in the absence of Q+ must be determined very precisely, because the value of mep is used as a standard mobility of the analyte anion in each plot.As can be seen from Fig. 5, plots for anions of low ion associability were dispersed, and the values of Kass could not be determined from the slopes. Hence the determination of ion association constants was limited to relatively stable ion associates formed between TBA+ and anions such as 1,5-NDS, 2,6-NDS and 2,6-NDC. In this study, we developed another method for the determination of ion association constants, which can give more reliable constants is applicable to monovalent anions of low ion associability.Fig. 3 Electropherogram of anions in a coated capillary. Sample solution, CE conditions and injection volume as in Fig. 1. Migrating solution contained 10 mm phosphate buffer (pH 7.0) and 20 mm TBA+·Br–. 1, 1-NS; 2, 2-NS; 3, 2,3-NDC; 4, 2,6-NDC; 5, 1,5-NDS; 6, 2,6-NDS; 7, PH; 8, i-PH; and 9, t-PH. Fig. 4 Change in apparent electrophoretic mobility of anions as a function of the molar concentration of quaternary ammonium salt added to the migrating solution.Conditions, except for Q+·Br– concentrations, as in Fig. 1. (a), TMA+; and (b), TBA+. 2, 1-NS; 5, 2-NS; «, 1,5-NDS; ~, 2,6-NDS; 8, 2,3-NDC; -, 2,6-NDC; ., PH; /, i-PH; and », t-PH. Analyst, January 1997, Vol. 122 59A 1 : 1 ion associate is assumed to be formed between an analyte anion (An2) and a quaternary ammonium ion (Q+) as shown in reaction (3), with an equilibrium constant given by eqn. (4). Q++An2"Q+·An2 (3) Kass = [Q+× An-- ] [Q+ ][An-- ] (4) The apparent electrophoretic mobility of a particular anion, mepA, can be represented as follows: mep ¢ = [An-- ] [An-- ] + [Q+× An-- ] ¥ mep + [Q+× An-- ] [An-- ] + [Q+× An-- ] ¥ mepIA (5) where mep and mepIA are the mobility of the analyte anion and that of the 1 : 1 ion associate, respectively.Eqn. (5) can be converted into eqn. (6) by using the mass balance for An2 and eqn. (4): mep ¢ = 1 1 + [Q+ ]Kass ¥ mep + [Q+ ]Kass 1 + [Q+ ]Kass ¥ mepIA (6) To calculate Kass, a non-linear least-squares method was applied.In the method, known values of [Q+] and mepA were used, and the values of the constants, mep, mepIA and Kass were obtained after the calculated curves had been optimized. The values of mep, mepIA and Kass for the aromatic anions with various Q+ are summarized in Tables 1, 2 and 3, respectively. The values obtained with the CElect-N capillary are also given. The calculated results for mep were found to be almost identical with the experimental results, as shown in Table 1.This indicates that the value of mep need not be obtained experimentally. From Table 2, it was found that mepIA was less than half of mep. This agrees with the fact that the apparent charge of divalent anions decreased to 21 on association with Q+ and that of monovalent anions with Q+ should be noncharged. Further, the values of mepIA indicate that the apparent molecular mass of the anions increases on ion association with Q+.In the case of 1-NS and 2-NS, the values of Kass were also calculated by assuming mepIA to be zero. The results for the Kass values of 1-NS and 2-NS by the two calculations were similar. The results for mep values obtained with the coated capillary were slightly larger than those obtained with the silica capillary, because the EOF in the coated capillary was very small and the correction of mep affected by the EOF could not be complete. Although the mep and mepIA values obtained by using the two kinds of capillary were slightly different from each other, the Kass values obtained were almost identical.Changes in the electrophoretic mobility were too small to determine the ion association constants when TMA+ was used as a pairing ion and phthalate ion was used. When the mobility of PH along with an increase in Q+ was used as a standard, however, the corrected mobility of 1,5-NDS and 2,6-NDS gave ion association constants of 100.6 ± 0.1 and 100.5 ± 0.1, respectively.The values are smaller than those with other Q+, which also indicates less associability of TMA+. The lines shown in Fig. 4(b) are the results obtained by the calculation; they show that the experimental data and calculated results agree favourably with each other. For the other cations, except for TMA+, the experimental data and calculated results also agreed well with each other. The ion association constants Fig. 5 Plots of mepA versus (2mep + mepA)/[TBA+].Electrophoretic mobility used was the result obtained with a fused-silica capillary, using TBA+ as a pairing ion. Slopes for 1,5-NDS, 2,6-NDS and 2,6-NDC correspond to 1/Kass, which gave ion association constants of 101.6, 101.5, and 101.5, respectively. Other constants could not be obtained. 2, 1-NS; 5, 2-NS; «, 1,5-NDS; ~, 2,6-NDS; 8, 2,3-NDC; -, 2,6-NDC; /, i-PH; and “, t- PH. Table 1 Comparison of electrophoretic mobilities of anions, 2mep, between experimental and calculated results 2mep */1024 cm2 V21 s21 Q+ 1-NS 2-NS 1,5-NDS 2,6-NDS 2,3,-NDC 2,6-NDC PH i-PH t-PH TMA+ —† — — — — — — — — (3.04) (3.04) (4.95) (4.95) (4.47) (4.47) (4.99) (4.99) (4.99) TEA+ — — 4.96 ± 0.06 4.95 ± 0.06 — 4.47 ± 0.03 — — — (3.05) (3.05) (4.97) (4.97) (4.48) (4.48) (5.01) (5.01) (5.01) TPA+ 3.04 ± 0.06 3.05 ± 0.08 4.93 ± 0.08 4.93 ± 0.08 — 4.46 ± 0.08 — 5.00 ± 0.10 4.99 ± 0.08 (3.01) (3.01) (4.90) (4.90) (4.42) (4.42) (4.94) (4.94) (4.94) TBA+ 3.12 ± 0.08 3.12 ± 0.07 5.01 ± 0.08 5.02 ± 0.09 4.54 ± 0.07 4.55 ± 0.01 — 5.07 ± 0.06 5.05 ± 0.08 (3.12) (3.12) (5.02) (5.02) (4.54) (4.54) (5.06) (5.06) (5.06) TBA+ ‡ 3.61 ± 0.14 3.62 ± 0.15 5.80 ± 0.20 5.81 ± 0.22 5.24 ± 0.23 5.26 ± 0.23 — 5.80 ± 0.27 5.77 ± 0.21 (3.44) (3.44) (5.64) (5.64) (5.12) (5.12) (5.71) (5.71) (5.68) TAA+ 3.06 ± 0.08 3.07 ± 0.06 5.01 ± 0.09 5.02 ± 0.06 4.52 ± 0.04 4.52 ± 0.04 — 5.05 ± 0.02 5.05 ± 0.03 (3.09) (3.09) (5.03) (5.03) (4.53) (4.53) (5.06) (5.06) (5.06) HTMA+ 3.17 ± 0.06 3.17 ± 0.06 5.10 ± 0.02 5.10 ± 0.02 4.62 ± 0.03 4.62 ± 0.04 — 5.15 ± 0.04 5.14 ± 0.03 (3.15) (3.15) (5.10) (5.10) (4.61) (4.61) (5.13) (5.13) (5.13) OTMA+ 3.20 ± 0.05 3.20 ± 0.06 5.16 ± 0.10 5.16 ± 0.08 4.66 ± 0.09 4.67 ± 0.07 — 5.24 ± 0.04 5.22 ± 0.06 (3.21) (3.21) (5.18) (5.18) (4.68) (4.68) (5.22) (5.22) (5.22) * Error: 3s.The values in parentheses are the experimental results. † Not obtained. ‡ Values obtained in a coated capillary. 60 Analyst, January 1997, Vol. 122obtained in this work were close to those obtained in the previous work, which were 101.3–101.4 for the ion association between TBA+ and anions such as 1,5-NDS, 2,6-NDS, and 2,6-NDC.18 The fact that the ion association constant of 2-NS is larger than that of 1-NS seems to be related to the difference in the charge density of the functional group. The acid dissociation constant (pKa) of 1-naphthol (9.34) is smaller than that of 2-naphthol (9.51), and that of 1-naphthoic acid (3.70) is smaller than that of 2-naphthoic acid (4.17), which suggests that the negative charge of the functional group on the 2-position is larger than that on the 1-position.The more basic the functional group is, the more associable the anion should be with pairing ions. Contribution of Methylene Moiety to Ion Associability The ion association constants obtained by the least-squares method were plotted against the number of carbon atoms in Q+ (Fig. 6). For each aromatic anion, Kass increased with an increase in the carbon number or the size of Q+, that is, the logarithmic values of the ion association constants increased almost linearly with increase in the bulkiness of Q+, and their mean slope was about 0.06, which represents the increase in log Kass per methylene group. This indicates that a linear free energy relationship can be valid in an aqueous solution, as well as in the case of liquid–liquid distribution system of ion associates.Fig. 6 also indicates that the ion association in aqueous media is closely related to the hydrophobicity of the pairing cations. The contribution of the methylene moiety to log Kass, 0.06 (mean value), was one tenth of that in the liquid– liquid distribution system of ion associates, 0.6.11 In the liquid– Table 2 Electrophoretic mobility of ion associates, 2mepIA: calculated results 2mepIA */1024 cm2 V21 s21 Q+ 1-NS 2-NS 1,5-NDS 2,6-NDS 2,3-NDC 2,6-NDC PH i-PH t-PH TMA+ —† — — — — — — — — TEA+ — — 1.45 ± 0.61 1.79 ± 0.74 — 1.81 ± 0.37 — — — TPA+ 0.10 ± 0.83 0.00 ± 1.03 1.78 ± 0.49 1.78 ± 0.49 — 1.48 ± 0.41 — 1.58 ± 1.58 1.33 ± 1.43 TBA+ 0.66 ± 0.56 0.63 ± 0.46 2.07 ± 0.33 2.08 ± 0.32 1.46 ± 1.19 1.50 ± 0.35 — 1.59 ± 0.73 1.52 ± 0.82 TBA+‡ 0.33 ± 1.00 0.41 ± 0.92 2.34 ± 0.62 2.34 ± 0.60 1.46 ± 2.86 1.56 ± 0.60 — 1.49 ± 2.39 1.46 ± 2.26 TAA+ 0.77 ± 1.25 0.67 ± 0.88 2.43 ± 0.70 2.39 ± 0.46 1.69 ± 1.54 1.81 ± 0.23 — 1.79 ± 0.59 1.72 ± 0.73 HTMA+ 0.49 ± 0.69 0.49 ± 0.69 2.26 ± 0.17 2.26 ± 0.15 1.96 ± 0.53 1.91 ± 0.30 — 1.42 ± 1.28 1.33 ± 0.77 OTMA+ 0.69 ± 0.57 0.50 ± 0.37 2.11 ± 0.58 2.15 ± 0.48 2.00 ± 2.12 1.71 ± 0.45 — 1.63 ± 0.90 1.79 ± 1.20 * Error: 3s.† Not obtained. ‡ Values obtained in a coated capillary. Table 3 Ion association constants, Kass, obtained by the least-squares method Log Kass * Q+ 1-NS† 2-NS† 1,5-NDS 2,6-NDS 2,3-NDC 2,6-NDC PH i-PH t-PH TMA+ —‡ — 0.6 ± 0.1§ 0.5 ± 0.1§ — 0.6 ± 0.1§ — — — TEA+ — — 0.91 ± 0.10 0.85 ± 0.13 — 0.92 ± 0.08 — — — TPA+ 0.89 ± 0.17 0.93 ± 0.21 1.17 ± 0.08 1.17 ± 0.08 — 1.25 ± 0.08 — 0.70 ± 0.29 0.70 ± 0.22 (0.77 ± 0.16) (0.81 ± 0.19) TBA+ 1.13 ± 0.14 1.18 ± 0.10 1.39 ± 0.07 1.44 ± 0.07 0.70 ± 0.23 1.48 ± 0.07 — 0.87 ± 0.11 0.90 ± 0.12 (0.98 ± 0.15) (1.06 ± 0.10) TBA+¶ 1.12 ± 0.19 1.20 ± 0.18 1.45 ± 0.12 1.49 ± 0.13 0.79 ± 0.15 1.49 ± 0.12 — 0.89 ± 0.37 0.91 ± 0.29 (1.07 ± 0.17) (1.14 ± 0.16) TAA+ 1.36 ± 0.42 1.38 ± 0.20 1.64 ± 0.16 1.69 ± 0.10 0.90 ± 0.35 1.71 ± 0.06 — 0.95 ± 0.09 1.08 ± 0.11 (1.21 ± 0.40) (1.26 ± 0.22) HTMA+ 0.91 ± 0.14 0.91 ± 0.14 1.18 ± 0.03 1.15 ± 0.03 0.78 ± 0.16 1.18 ± 0.06 — 0.64 ± 0.19 0.70 ± 0.10 (0.83 ± 0.14) (0.83 ± 0.14) OTMA+ 0.91 ± 0.13 0.91 ± 0.07 1.20 ± 0.11 1.22 ± 0.09 0.61 ± 0.75 1.20 ± 0.09 — 0.77 ± 0.14 0.84 ± 0.20 (0.79 ± 0.14) (0.83 ± 0.08) * Error: 3s.† Values in parentheses were the results when mepIA was assumed to be zero. ‡ Not obtained. § These constants were obtained by assuming that PH did not form ion associate with TMA+ and using the increase in the electrophoretic mobility of pH with increase in the electrophoretic mobility of PH with increase in the concentration of Q+ as the effect of temperature and ionic strength on the mobility of ions. ¶ Values obtained in a coated capillary.Fig. 6 Relationships between the ion association constants and the carbon number of quaternary ammonium ions (Q+). 2, 1-NS; 5, 2-NS; «, 1,5-NDS; ~, 2,6-NDS; 8, 2,3-NDC; -, 2,6-NDC; /, i-PH; and “, t- PH.Analyst, January 1997, Vol. 122 61liquid distribution system, the reaction and its equilibrium constant are represented by the equations Q++A2"Q+·A(o) (7) Kex = [Q+× A-- ](o) [Q+ ][A-- ] = KassKD,ip (8) where the subscript (o) denotes the species in the organic phase, Kex is the extraction constant of a ion associate, Q+·A2, and KD,ip is the distribution coefficient of the ion associate between the aqueous and organic phase. The present results indicate that the contribution of a methylene moiety to the distribution process is a more dominant factor in the extraction equilibrium than in the ion association process in aqueous media. Conclusion This study has demonstrated the usefulness of ion association reactions in aqueous solution for the improvement of CE separations. Ion association constants were determined by measuring the electrophoretic mobility changes of anionic species. Ion associability was shown to be affected by the hydrophobicity of the pairing ion. The contribution of the hydrophobicity to ion association constants was relatively small, 0.05–0.07 for «log Kass per methylene moiety, compared with those in the solvent extraction of ion associates. Ion association combined with CE enabled us to separate anion isomers effectively. This work was supported by a Grant-in-Aid for Scientific Research (No. 07454203) from the Ministry of Education, Science and Culture, Japan. References 1 Terabe, S., and Isemura, T., Anal. Chem., 1990, 62, 650. 2 Zenki, M., Irizawa, M., and Yukutake, H., Bunseki Kagaku, 1995, 44, 227. 3 Zenki, M., Yukutake, H., and Irizawa, M., Bunseki Kagaku, 1996, 45, 181. 4 Jandik, P., and Jones, W. R., J. Chromatogr., 1991, 546, 431. 5 Jones, W. R., J. Chromatogr., 1993, 640, 387. 6 Buchberger, W., Cousins, S. M., and Haddad, P. R., Trends Anal. Chem., 1994, 13, 313. 7 Iki, N., Hoshino, H., and Yotsuyanagi, T., J. Chromatogr. A., 1993, 652, 539. 8 Motomizu, S., Kuwabara, M., and Oshima, M., Bunseki Kagaku, 1994, 43, 621. 9 Motomizu, S., Bunseki Kagaku, 1989, 38, 147. 10 Haddad, P. R., and Jackson, P. E., Journal of Chromatography Library, Vol. 46, Ion Chromatography, Elsevier, Amsterdam, 1990. 11 Motomizu, S., Hamada, S., and T�oei, K., Bunseki Kagaku, 1983, 32, 648. 12 Matsunaga, H., and Yotsuyanagi, T., Nippon Kagaku Kaishi, 1982, 785. 13 Davies, C. W., Ion Association, Butterworths, London, 1962. 14 Nancollas, G. H., Interactions in Electrolyte Solutions, Elsevier, Amsterdam, 1966. 15 Huang, X., Luckey, J. A., Gordon, M. J., and Zare, R. N., Anal. Chem., 1989, 61, 766. 16 Benz, N. J., and Fritz, J. S., J. Chromatogr. A., 1994, 671, 437. 17 Rosen, M. J., Surfactants and Interfacial Phenomena, Wiley, New York, 1989, p. 125. 18 Takayanagi, T., and Motomizu, S., Chem. Lett., 1995, 593. Paper 6/05047F Received July 22, 1996 Accepted October 15, 1996 62 Analyst, January 1997, Vol. 1

ISSN:0003-2654    

DOI:10.1039/a605047f

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Fluorescence-based Thin Plastic Film Ion-pair Sensors for Oxygen Andrew Mills* and Mark Thomas Department of Chemistry, University College of Swansea, Singleton Park, Swansea, UK SA2 8PP A general method of preparation of thin-film sensors for O2, incorporating the dye ion-pair tris(4,7-diphenyl-1,10-phenanthroline) ruthenium(II) ditetraphenylborate, in a variety of different thin film polymer/plasticizer matrices is described. The sensitivity of the sensor depends upon the nature of the polymer matrix and plasticizer.A detailed study of one of these systems utilising the polymer poly(methyl methacrylate), PMMA, is reported. The sensitivity of this O2 sensor depends markedly upon the plasticizer concentration and is largely independent of temperature (24.5–52.5 °C) and age (up to 30 d). When exposed to an alternating atmosphere of O2 and N2, a typical oxygen film sensor in PMMA exhibits a 0–90% response and recovery time of 0.4 and 4.5 s, respectively.Keywords: Tris(4,7-diphenyl-1,10-phenanthroline) ruthenium(II) ditetraphenylborate; oxygen sensor; fluorescence oxygen sensor; thin film; plastic film; poly(methyl methacrylate) Oxygen optical sensors (also known as optrodes) are devices which respond specifically and, usually, reversibly to molecular oxygen via a change in luminescence intensity. Such optrodes have a significant number of major advantages over the customary amperometric oxygen sensors; including no oxygen consumption and ready miniaturisation. Their compact, robust nature and ability to withstand high external pressures also contributes to their usefulness making their field of application widespread. Such sensors are likely to be inexpensive and, when combined with optical fibres, make possible the development of disposable, remote, multiple analysis micro-sensors. 1,2 The majority of the optical sensors for O2 developed so far are based on the luminescence quenching of an indicator by molecular oxygen. In such sensors the fluorophores: tris(4,7- diphenyl-1,10-phenanthroline) ruthenium(ii), tris(1,10-phenanthroline) ruthenium(ii), and tris(2,2’-bipyridyl) ruthenium( ii), abbreviated as [Ru(dpp)3]2+, [Ru(phen)3]2+, and [Ru(bpy)3]2+, respectively, dominate the field.These dyes are very photostable and have long excited-state lifetimes, high quantum yields of luminescence and are readily quenched by oxygen (see Table 1).3–9 A good, brief description of the major types of oxygen sensor, based on the ruthenium diimine complexes described above, is given elsewhere.7 One problem with these dyes is the hydrophilic nature of their common salts, such as chloride or perchlorate.As a result, incorporating these dyes in a hydrophobic medium, such as silicone rubber, poses problems, such as dye leaching by water and film fogging due to water uptake. Thus, in the early work on oxygen optical sensors, [Ru(dpp)3(ClO4)2] was incorporated into silicone rubber by using a dichloromethane solution of the complex, since dichloromethane swells up and penetrates the film.3 In this latter work, problems were encountered of film curling and water condensation and, even after these were overcome, the films were still slightly cloudy.3 It should be possible to render the dyes identified in Table 1, or any fluorescent cationic dye for that matter, largely lipophilic and, therefore, soluble in the hydrophobic medium of a polymer, by coupling the dye cation to a lipophilic, nonquenching anion, such as tetraphenyl borate, Ph4B2.In this paper we describe the results of a detailed study of a series of oxygen sensors generated by incorporating the tris(4,7-diphenyl- 1,10-phenanthroline) ruthenium(ii) ditetraphenylborate [Ru(dpp)3 2+(Ph4B2)2] dye ion-pair into a variety of different plasticized polymers. Experimental Materials Unless stated otherwise all the chemicals used in this work were obtained from Aldrich (Gillingham, Dorset, UK).The chemicals used in the synthesis of the [Ru(dpp)3 2+(Ph4B2)2] dye ionpair employed were as follows: ethylene glycol, 4,7-diphenyl- 1,10-phenanthroline, and sodium tetraphenylborate. Ruthenium trichloride trihydrate, was purchased from Merck (Lutterworth, Leicestershire, UK), diethyl ether and sodium chloride were purchased from Fisons (Loughborough, Leicestershire, UK). The polymers used in this work were as follows: poly(methyl methacrylate) (PMMA) (M = 120 000), poly(vinyl acetate) (PVAc), poly(vinyl chloride) (PVC), polystyrene (PS) and cellulose acetate butyrate (CAB) (38-% m/m butyryl content).A low molecular weight poly(methyl methacrylate) (PMMAL) (M = 33 800) was also used and was purchased from Fluka Chemicals (Gillingham, Dorset, UK). The plasticizers used in this work were as follows: tri-n-butyl phosphate (TBP) 98%, Lancaster Synthesis (Morecambe, Table 1 Photochemical characteristics of the dyes Ru(bpy)3 2+, Ru(phen)3 2+ and Ru(dpp)3 2+ pO2 (S = 1 2) in silicone Dye t°/ms FL kQ(O2/109 dm23 mol21 s21 rubber3/Torr Refs.Ru(bpy)3 2+ 0.60 0.042 3.3 376.8 4, 5 (in water) (in water) (in water) Ru(phen)3 2+ 0.92 0.080 4.2 111.3 6, 7 (in water) (in butan-2-one) (in water) Ru(dpp)3 2+ 5.34 Å 0.30 2.5 29.8 8, 9 (in methanol) (in water–ethanol) (in methanol) Analyst, January 1997, Vol. 122 (63–68) 63Lancashire, UK), 2-ethylhexyl diphenyl phosphate (also known as Sancticizer 141, and abbreviated as S141), Monsanto (Newport, Gwent, UK), tris(2-ethylhexyl)phosphate (TOP) 97%, and dioctyl phthalate (DOP) 99%.The solvents used were: AnalaR grade acetone (Fisons), tetrahydrofuran (THF) 99+%, and butan-2-one, 99+%. Useful details associated with the polymers and plasticizers used in this work are given in Table 2.10–13 The air, O2 and N2 used were of a high purity ( > 99%) and obtained from BOC. O2–N2 gas mixtures of different specified compositions, spanning the range 0–100% O2, were generated using a gas blender (Model No. 852V5-S, Signal Instruments, Standards House, Doman Road, Camberley, Surrey, UK). All chemicals were used as received. Preparation of the Dye Ion-pair Complex [Ru(dpp)3 2+ (Ph4B2)2] The non-hydroscopic dye ion-pair complex of [Ru(dpp)3 2+(Ph4B2)2] was prepared as follows: 225.9 mg of RuCl3·3H2O were dissolved in a mixture of 5 cm3 of ethylene glycol and 0.5 cm3 of water at 120 °C, followed by the addition of 862.6 mg of the ligand dpp.The resulting mixture was refluxed at 165 °C for a duration of 45 min cooled, and then 50 cm3 of acetone added. This solution was filtered and the filtrate diluted with 60 cm3 of acetone. A 100 cm3 volume of the filtrate containing the crude Ru(dpp)3Cl2 was added to 100 cm3 of a 10 mmol dm23 aqueous solution of NaPh4B and a fine orange precipitate quickly developed. Finally, 100 cm3 of a 1 mol dm23 sodium chloride solution were added to coagulate the precipitate and make its filtration easier.The orange precipitate of the dye ion-pair complex, [Ru(dpp)3 2+(Ph4B2)2], was filtered and washed four times with 20 cm3 aliquots of distilled water. Purification was achieved by recrystallization of the dye ionpair from an acetone–water mixture (80 + 20 v/v). The precipitate was washed with 20 cm3 of diethyl ether and dried in a desiccator. Preparation and Casting of Film Solutions of the Dye Ion-pair Complex [Ru(dpp)3 2+ (Ph4B2)2] The thin plastic film fluorimetric sensors for O2 used in this work had the general composition: indicator dye (9.6 3 1024 mol dm23)/polymer/plasticizer, and were supported on individual small cut quartz slides.The films were generated by casting a typical film solution of two components. Component solution I comprised 1 cm3 of the same solvent which was used to dissolve the polymer in component solution II (see Table 3) and 1.6 mg of dye. Component solution II was prepared by dissolving the polymer concerned in the solvent stipulated in Table 3.The final solution used for casting, to produce the plasticized film sensor, was made up by adding solution I to 10 g of solution II and sufficient TBP to yield a final level of 30 parts per hundred of resin (phr), i.e., 23.1% m/m. The dry, thin plastic film fluorimetric sensor for O2 containing the dye under test was generated from the final film solution by casting it through a 100 mm thick brass sheet with a rectangular hole (0.8 3 1.5 cm) onto a cut quartz slide (0.85 3 4.2 30.1 cm).The film was then placed in a desiccator, at room temperature, in the dark for a period of 24 h and used. Unless otherwise stated, the plastic film sensors were used at room temperature and were typically 20 mm thick when dried, as measured using a micrometer. In a separate set of experiments, PMMA plastic film O2 sensors containing the following Table 2 Characteristics of the different polymers and plasticizers used Solubility parameter, Abbreviation Molecular formula (Molecular weight) d (J cm23)1 2 Polymer— Cellulose acetate butyrate CAB 17.910 (30 000) Poly(vinyl acetate) PVAC (–CH2CH(O2CCH3)–)n 19.111 (113 000) Polystyrene PS (–CH2CH(C6H5)–)n 18.512 (45 000) Poly(methyl methacrylate) PMMA (–CH2C(CH3)(CO2CH3)–)n 18.812 (120 000) Poly(methyl methacrylate) PMMAL (–CH2C(CH3)(CO2CH3)–)n 18.812 (33 800) Poly(vinyl chloride) PVC (–CH2CH(Cl)–)n 19.212 (95 000) Plasticizer— Phosphates— Tri-n-butyl phosphate TBP O:P(OC4H9)3 17.512 (266.3) Tris(2-ethyl hexyl) phosphate TOP O:P(OCH2(C2H5)CH(CH2)3CH3)3 16.812 (434.7) 2-Ethylhexyl diphenyl phosphate S141 CH3(CH2)3CH(C2H5)CH2OP:O(OC6H5)2 19.012 (362.4) Phthalates— Dioctyl phthalate DOP C6H4-1,2-(CO2CH2CH(C2H5)(CH2)3CH3)2 16.013 (390.6) * R = UCOCH3, UCOCH2CH2CH3 and UOH. 64 Analyst, January 1997, Vol. 122different plasticizers were generated, i.e., TBP, TOP, DOP and S141 at a level of 33 phr, i.e., 24.8% m/m.Methods Spectrophotomeric and fluorescence measurements were carried out using a Perkin-Elmer (Beaconsfield, Buckinghamshire, UK) Lambda 9 UV/VIS/NIR spectrophotometer and a Perkin- Elmer LS-5 fluorimeter/phosphorimeter, respectively. In the majority of the work carried out using the fluorimeter, a typical [Ru(dpp)3 2+(Ph4B2)2] film, coated onto a cut quartz slide, was placed in the centre of a quartz fluorescence cell, which in turn was placed in the sample holder of the fluorimeter.Gases of different O2–N2 composition were delivered to the film sensor by means of two needles positioned at each face of the quartz slide. Two small holes in the cell cover allowed insertion of the needles into the system. Unless stated otherwise, for all the oxygen sensors tested, in the studies of the variation in luminescence intensity as a function of oxygen partial pressure, the wavelengths of excitation and emission used were 430 and 596 nm, respectively.In the measurement of the response/recovery of the [Ru(dpp)3 2+(Ph4B2)2] film, as a function of time, towards rapid variations in the level of O2 (0%–100% and vice versa), a laboratory-made optical system, with a time resolution of approximately 12 ms, was used. In this system, the light source comprised a quartz–iodide tungsten filament lamp, coupled with a blue filter, which allowed light of 430 ± 20 nm through to the plastic film sensor to electronically excite the dye.Typically, the thin film sensor was supported on a cut quartz slide, attached to a Pyrex cell thermostated at 20 °C, and held square on to the excitation light beam. The luminescence from the edge of the film passed through a red filter (which allowed light of > 580 nm through) and then a high-resolution monochromator set at 596 nm (Applied Photophysics, London, UK), both held at right angles to the film. The intensity of this luminescence was measured using a photomultiplier, the output of which was monitored continuously during a run using a storage oscilloscope (Gould).The rapid delivery to the surface of the sensor of alternating streams of nitrogen and oxygen was achieved using two solenoid-activated, pressure-actuated valves, details of which are given elsewhere.14 Theory In a homogeneous medium, such as aqueous solution, quenching of the luminescence of Ru(dpp)3 2+ by oxygen has been found to obey the Stern–Volmer equation, i.e, I0/I = 1+KSVpO2 (1) where I0 and I are the intensities of luminescence in the absence and presence of oxygen at a partial pressure pO2, respectively, and KSV is the Stern–Volmer constant.The Stern–Volmer constant depends directly upon the rate constant for diffusion of oxygen, the solubility of oxygen and the natural lifetime of the electronically-excited state of the fluorophore in the plastic medium. In contrast to homogeneous solution, it has been observed for the majority of the luminescent O2 sensors developed to date, in which a fluorescent indicator such as Ru(dpp)3 2+ is incorporated into a polymer matrix, such as silicone rubber, that the Stern– Volmer plot of I0/I versus pO2 displays a downward curvature.Two models3,15 which have been used to account for this phenomenon are the multisite model, and the non-linear solubility model. In the multisite model, it is suggested that the sensor molecule can exist in two or more sites, each with its own characteristic quenching constant.The second model assumes that the deviation from linearity is due to the non-linear solubility of oxygen in the polymer. Demas et al. have identified recently several features of typical oxygen optical sensors which provide support for their multisite model.16 Both models predict the following modified form of the Stern-Volmer equation: I0/I = 1+ApO2 + BpO2/(1 +b pO2) (2) where A, B, and b are constants related to the parameters in the kinetic and solubility equations associated with the respective models.The above equation has been used to fit successfully the observed data associated with most O2 optrodes. In all the work on oxygen optical sensors described in this paper the Stern– Volmer plots exhibit a downward curvature which is described very well by eqn. (2). Thus, Fig. 1 provides an illustration of the I/I0 versus pO2 and associated Stern–Volmer plots for a typical plastic [Ru(dpp)3 2+(Ph4B2)2] film in PMMA, containing TBP (133 phr) as the plasticizer.In both plots the solid lines represent the variations predicted by eqn. (2) using the values of A, B and b given in the figure legend. We suggest that a useful, quick, albeit crude, guide to the sensitivity of any O2 optrode towards oxygen, at low oxygen levels, is the value of pO2 when the intensity of luminescence, I, has dropped to a value of I0/2, i.e., pO2 (S = 1 2). In a homogeneous medium, where the quenching is expected to obey the Stern–Volmer equation, i.e., eqn.(1), the parameter pO2(S = 1 2) is given by: pO2 (S = 1 2) = 1/KSV (3) However, for most oxygen optical sensors, where the variation in luminescence as a function of pO2 is more likely to be described by the modified form of the Stern–Volmer equation, i.e., eqn. (2), the parameter pO2(S = 1 2) is described by the following expression: Table 3 Composition of component solution II for the different polymers Weight of Volume of polymer in solvent in solution solution Polymer (II)/g Solvent (II)/cm3 CAB 30 Acetone 100 PVAC 50 Acetone 100 PS 100 Butan-2-one 100 PMMA 33 Acetone 100 PMMAL 15 Acetone 30 PVC 12.5 THF 100 Fig. 1 Plot of the observed relative intensity (I/I0) as a function of increasing pO2 level for a typical [Ru(dpp)3 2+(Ph4B2)2] in PMMA ([TBP] = 133 phr) plastic film fluorimetric sensor. The inset diagram is a Stern–Volmer plot of the data in the main diagram. The solid line represents the least squares line of best fit and was calculated using eqn.(2) and the following, optimised values for A, B and b: A = 0.0120; B = 0.0220; b = 0.0018. Analyst, January 1997, Vol. 122 65pO2(S = 1 2) = [2(A+B2b) ±A(A+B2b)2 4Ab]/2Ab (4) In this work all the values of pO2(S = 1 2) quoted have been calculated using eqn. (4). Results and Discussion Effect on Sensitivity of Different Polymers and Plasticizers In an initial study a variety of plastic film oxygen sensors were created in which [Ru(dpp)3 2+(Ph4B2)2], as always, was the fluorophore, the plasticizer was TBP and was kept the same, at a fixed level of 30 phr, and the type of polymer used was varied.For each film, the variation of the luminescence intensity as a function of pO2 was determined, from which a Stern–Volmer plot was generated. In a similar study, a series of oxygen film sensors were generated in which the polymer was kept the same, i.e., PMMA, and the type of plasticizer, although at a fixed level of 33 phr, was varied.The Stern–Volmer plots of the results of both sets of work are illustrated in Fig. 2. A plasticizer is usually included in the formulation for most plastic film gas sensors to improve the rate of gas diffusion throughout the film, since plasticization increases the mobility of polymer segments, which leads to an increase in gas diffusion coefficients.17 Plasticization also increases the workability, flexibility and distensibility of the film.It is common practice to refer to the value of the solubility parameter, d, of the plasticizer and polymer in order to identify compatible plasticizer–polymer combinations. In general, it is suggested13 that the difference in solubility parameter [Dd = d(polymer) 2 d(plasticizer)] should be less than 3.7 J1 2 cm23 2. Table 2 lists the solubility parameter values for the different polymers and plasticizers used in this work and from this data it appears that all the individual plasticizer–polymer film combinations used to generate the data illustrated in Fig. 2 should be compatible. Fig. 3 illustrates the variation in Dd as a function of pO2(S = 1 2) when (a) the polymer was fixed and the type of plasticizer was varied and (b) when the plasticizer was fixed and the type of polymer was varied; this plot was generated using the results illustrated in Fig. 2 and the data contained in Table 2. From the two plots illustrated in Fig. 3, there appears to be a positive correlation between Dd for a plastic film oxygen sensor and its oxygen film sensitivity, as measured by the value of pO2(S = 1 2).The most sensitive films [i.e., those with the lowest values of pO2(S = 1 2)] appear to be associated with highly compatible polymer–plasticizer combinations (i.e., low Dd). This general finding is not too surprising given that the sensitivity of the oxygen optical film sensor depends not only on the rate constant for diffusion of oxygen through the film, but also upon the solubility of oxygen in the film (vide supra). Both these latter parameters are expected to be improved with plasticization, and the better the polymer–plasticizer combination, as measured by Dd, the more effective the plasticizer, as measured by pO2(S = 1 2).[Ru(dpp)3 2+ (Ph4B2)2] in PMMA From the results illustrated in Fig. 3, it appears that, of all the different plasticizers tested, TBP produced the most sensitive oxygen film sensors. In a separate study, a series of [Ru(dpp)3 2+(Ph4B2)2] in PMMA oxygen sensor films were generated and studied in which the level of the TBP plasticizer contained therein was varied.Stern–Volmer plots derived from the variation of the luminescence intensity of the [Ru(dpp)3 2+(Ph4B2)2] luminophore as a function of pO2, for these different films are illustrated in Fig. 4; in this work the level of plasticizer was varied from 0 to 133 phr. In addition, in Fig. 2 (a) Plot of I0/I versus pO2 for a series of [Ru(dpp)3 2+(Ph4B2)2] in PMMA plastic film fluorimetric sensors, in which the type of plasticizer was varied, although its level in the polymer was kept the same, 33 phr.The different plasticizers were as follows: TBP (2), TOP (~), DOP (-) and S141 (8), respectively. The solid lines represents the least squares lines of best fit and were calculated using eqn. (2) and optimised values for A, B and b. (b) Plot of Io/I versus pO2 for a series of [Ru(dpp)3 2+(Ph4B2)2] in different polymer plastic film fluorimetric sensors, with a fixed final level of TBP plasticizer, 30 phr. The different polymers used to make up the films were as follows: CAB (5), PVAC (-), PS (2), PMMA (~), PVC (8) PMMAL (»), respectively.The solid lines represents the least squares lines of best fit and were calculated using eqn. (2) and optimised values for A, B and b. Fig. 3 Plot of pO2(S = 1 2) versus the change in solubility parameter, Dd, when (a) the plasticizer was varied and the polymer, PMMA, was fixed; data from Fig. 2(a) and Table 2 and (b) the polymer was varied and the plasticizer type was fixed (TBP); data from Fig. 2(a) and Table 2. Fig. 4 Stern–Volmer plots of I0/I versus pO2 for a series of plastic film fluorimetric sensors in which the polymer was PMMA and the plasticizer, TBP, was varied over the range 0–133 phr, i.e., 0 (5), 33 (2), 40 (-), 50 (8), 60 (~), 65 (»), 80 («), 106 (“), 120 (/), 133 (3) phr TBP and neat TBP (+), respectively.The solid lines represents the least squares lines of best fit and were calculated using eqn. (2) and optimised values for A, B and b. 66 Analyst, January 1997, Vol. 122Fig. 4 the Stern–Volmer plot of the data obtained for [Ru(dpp)3 2+(Ph4B2)2] in neat TBP is illustrated. Fig. 5 illustrates the variation in the sensitivity of the oxygen sensor films, as measured by pO2(S = 1 2), calculated using eqn. (4) and the data in Fig. 4, as a function of plasticizer concentration in the film.As indicated by the results in Figs. 4 and 5, increasing the level of plasticizer in the film increases markedly the sensitivity of the film towards oxygen, e.g., by a factor of 28.7, on going from 0 to 133 TBP phr. Thus, a [Ru(dpp)3 2+(Ph4B2)2] oxygen sensor film in PMMA with 133 phr TBP has a pO2(S = 1 2) = 3.7 Torr, compared with a value of 34.9 Torr for a film with 33 phr TBP and 29.8 Torr for [Ru(dpp)3 2+(Ph4B2)2] in silicone rubber.A great deal of previous effort has gone into finding new fluorophores for oxygen sensors which will give greater sensitivity than that found using the popular existing luminophores, such as [Ru(dpp)3 2+(Ph4B2)2]. From the results of our work, in polymer films which can be plasticized, such as PMMA, a high degree of tuning of the sensitivity of the oxygen film sensor is possible through the careful choice of a compatible polymer–plasticizer combination and, most notably, the level of plasticizer present.The level of plasticization in a film also has an effect on the response and recovery time of a film. Thus, in the absence of plasticizer the 90% response and recovery times, t(90), for an oxygen optical film exposed to 100 and 0% oxygen were found to be 112 and 127 s, respectively, whilst for an oxygen sensor with a plasticizer content of 133 phr, t(90) response and recovery times of 0.4 and 4.5 s, respectively, were recorded. The measured variation in t(90) response and recovery for all the films plasticized at different levels are illustrated in Fig. 6. As with all hyperbolic-type response sensors the recovery times are always greater than the response times and this effect is expected even if the response and recovery processes are both diffusion-controlled as is likely with this type of oxygen sensor. The observed decrease in t(90) for response and recovery with increasing plasticizer content, illustrated in Fig. 6, is expected, since, as noted earlier, plasticization increases the mobility of polymer segments, which leads to an increase in the coefficient for gas diffusion. A brief study in which the 90% response and recovery times of a typical oxygen optical film were determined as a function of temperature over the range 20–55 °C. Both response and recovery times were found to decrease with increasing temperature, e.g., t(90) response and recovery values of 0.4 and 4.5 s were recorded for a film at 20 °C, and 0.3 and 3.0 s for the same film held at 55 °C.An Arrhenius plot of ln(response, or recovery, time) versus T21 gave a good straight line, from the value of the gradient of which an activation energy of 26 kJ mol21 was calculated. The observed decrease in response and recovery times for a typical oxygen plastic film sensor is expected given that the rate of diffusion of oxygen through the film, which is the likely rate determining step associated with the response and recovery process, will increase with increasing temperature. The sensitivity of a typical oxygen film sensor in PMMA ([TBP] = 133 phr), as measured by the value of pO2(S = 1 2), was studied as a function of temperature over the range 24.5–52.5 °C and found to be largely independent; at these latter temperatures, for example, pO2(S = 1 2) values of 32.7 and 36.4 Torr were found, respectively.This finding is not too surprising given that an increase in temperature will not only lead to an increase in the rate constant for diffusion, but also a decrease in the solubility of oxygen in the plastic medium, leading to a marginal overall effect on the sensitivity of the film.Other work showed that the films did not change in sensitivity with age. Thus, there appeared to be little variation in pO2(S = 1 2) with age of the film, when stored in the freezer section of a fridge (24 °C), over a period of 30 d. Thus, for a typical film, pO2(S = 1 2) values of 32.5 and 38.5 Torr were determined after 0 and 30 d, respectively, stored at 24 °C.In one experiment the variation in the relative intensity of luminescence of a typical oxygen film sensor was monitored as a function of time, as the ambient gas phase was altered periodically between 100% nitrogen and 100% oxygen over a period of 12 h; the results of this work are illustrated in Fig. 7 and shows that the response of a typical [Ru(dpp)3 2+(Ph4B2)2] Fig. 5 Plot of pO2(S = 1 2) versus increasing [TBP] for the thin film sensors based on PMMA described in Fig. 4. The values for pO2(S = 1 2) were calculated using eqn. (4). The symbol 8 represents the value for pO2(S = 1 2) obtained for the dye dissolved in neat TBP plasticizer. Fig. 6 Measured 90% response (5) and recovery (-) times for a typical [Ru(dpp)3 2+(Ph4B2)2] in PMMA oxygen film sensor in which increasing amounts of the plasticizer, TBP, was used in the film formulations, spanning the range 0–133 phr. Fig. 7 Observed variation in relative luminescence intensity as a function of time over 12 h for a typical [Ru(dpp)3 2+(Ph4B2)2] in PMMA oxygen film sensor ([TBP] = 133 phr) subjected to an atmosphere which was varied periodically between 100% N2 and 100% O2, respectively. Analyst, January 1997, Vol. 122 67in PMMA ([TBP] = 133 phr) film oxygen sensor is very stable over a 12 h period of operation. In an earlier paper Klimat and Wolfbeis reported that ionpairs based on the cationic fluorophore, Ru(dpp)3 2+, which are similar to those used in this work, could be dissolved in silicone rubber to create a range of fluorimetric oxygen sensors with novel features.7 In this latter work, the most sensitive oxygen sensor reported used dodecyl sulfate, DS2, as the hydrophobic anion and a one-component, transparent, acetic acid releasing, silicone prepolymer as the silicone rubber.This sensor was found to be: (i) very stable with time (its quenching efficiency dropped by only 20–30% over 1 year when stored in a plastic bag), (ii) very fast in response to changes in gaseous oxygen ( < 1 s), (iii) very photostable (daylight illumination over 1 year caused no measurable decomposition), and (iv) stable in aqueous solution (only a 10% drop in quenching constant was observed over a period of 2 months).However, these oxygen sensors were found to be less sensitive, pO2(S = 1 2) = 55 Torr, towards oxygen compared with those reported by others3, in which perchlorate, or chloride, was used as the anion, pO2(S = 1 2) = 29.8 Torr.In this latter work, it is believed that a different quenching mechanism is operative, since the silicone rubber used contained a silica filler, which adsorbs the salts of Ru(dpp)3 2+. From the results of the work reported here it appears that [Ru(dpp)3 2+(Ph4B2)2] ion-pair sensors in plasticized polymers exhibit many of the positive aspects listed above, i.e., (i)–(iv), for the Ru(dpp)3 2+(DS2)2 in silicone rubber oxygen sensors.However, unlike the silicone rubber based sensors, the sensitivity of the [Ru(dpp)3 2+(Ph4B2)2] sensors can be varied substantially through the choice of plasticizer type and content. Thus, as noted earlier, the [Ru(dpp)3 2+(Ph4B2)2] sensors in PMMA have pO2(S = 1 2) values of: 106, 34.9 and 3.7 Torr when the TBP plasticizer content is: 0, 33 and 133 phr, respectively. The latter, highly plasticized, Ru(dpp)3 2+(Ph4B2)2] ion-pair sensor is, therefore, much more sensitive than any of the silicone-rubber based oxygen sensors reported to date, which utilise the same cationic lumiphore.The sensitivity of the [Ru(dpp)3 2+(Ph4B2)2] ion-pair sensors is also dependent upon the nature of the polymer. Thus, [Ru(dpp)3 2+(Ph4B2)2] ion-pair based sensors with pO2(S = 1 2) values much less than 3.7 Torr can be created using a different polymer to PMMA, such as CAB. Conclusion A general method of preparation of thin plastic film, fluorescence- based optical sensors for oxygen is described. The fluorophore cation, ruthenium(ii) tris(4,7-diphenyl-1,10-phenanthroline) is readily incorporated into a variety of different plasticized polymer films by coupling it with the lipophilic anion, tetraphenylborate.The sensitivity of the optical films appears to depend upon the degree of compatibility between the plasticizer and polymer; the greater the compatibility the greater the sensitivity of the sensor.In a more detailed study of a typical film, using PMMA as the polymer and TBP as the plasticizer, the sensitivity of the film towards oxygen increased markedly with increasing level of plasticizer. Increasing the level of plasticizer also decreased markedly the response and recovery times for the sensor. The sensitivity of a typical film is largely independent of temperature and age (when stored at 24 °C). The ability to tune the sensitivity of oxygen optical sensors through the judicial choice of a polymer–plasticizer combination and level of plasticizer is novel and helps widen the possible areas of application of such sensors.The authors thank Sealed Air FPD for funding this work through an EPSRC CASE award. References 1 Seitz, W. R., CRC Crit. Rev. Anal. Chem., 1988, 19, 135. 2 Wolfbeis, O. S., in Fiber Optic Chemical Sensors, ed. Wolfbeis, O. S., CRC Press, Boca Ranton, FL, 1991, vol. II, ch. 10. 3 Carraway, K.R., Demas, J. N., DeGraff B. A., and Bacon, J. R., Anal. Chem., 1991, 63, 337. 4 Lin, C.-L., and Sutin, N., J. Phys. Chem., 1976, 80, 97. 5 van Houten J., and Watts, R. J., J. Am. Chem. Soc., 1976, 98, 4853. 6 Lin, C.-T., Bottcher, W., Chou, M., Creutz C., and Sutin, N., J. Am. Chem. Soc., 1976, 98, 6536. 7 Klim�at, I., and Wolfbeis, O. S., Anal. Chem., 1995, 67, 3160, and references cited therein. 8 Demas, J. N., Harris, E. W., and McBride, R. P., J. Am. Chem. Soc., 1977, 99, 3547. 9 Nakamaru, K., Nishio, K., and Nobe, H., Sci. Rep. Hirosaki Univ., 1979, 26, 57. 10 Burrel, H., in Polymer Handbook, ed. Brandrup, J., and Immergut, E. H., Wiley, New York, 1974, p. 556. 11 Small, P. A., J. Appl. Chem., 1953, 3, 71. 12 Concise Encyclopedia of Polymer Science and Engineering, ed. Kroschwitz, J. I., Wiley, New York, 1990, p. 736. 13 Seymour, R. B., and Carrher, C. E., Polymer Chemistry—An Introduction, Marcel Dekker, New York, 3rd edn., 1992, p. 394. 14 Mills, A., Chang, Q., and Mcmurray, N., Anal.Chem., 1992, 64, 1383. 15 Li, X.-M., Ruan, F.-C and Wong, K.-Y, Analyst, 1993, 118, 289. 16 Demas, J. N., DeGraff, B. A. and Xu, W., Anal. Chem., 1995, 67, 1377. 17 Crank, J., and Park, G. S., Diffusion in Polymers, Academic Press, London, 1968, p. 21. Paper 6/06124I Received September 5, 1996 Accepted November 1, 1996 68 Analyst, January 1997, Vol. 122 Fluorescence-based Thin Plastic Film Ion-pair Sensors for Oxygen Andrew Mills* and Mark Thomas Department of Chemistry, University College of Swansea, Singleton Park, Swansea, UK SA2 8PP A general method of preparation of thin-film sensors for O2, incorporating the dye ion-pair tris(4,7-diphenyl-1,10-phenanthroline) ruthenium(II) ditetraphenylborate, in a variety of different thin film polymer/plasticizer matrices is described.The sensitivity of the sensor depends upon the nature of the polymer matrix and plasticizer. A detailed study of one of these systems utilising the polymer poly(methyl methacrylate), PMMA, is reported. The sensitivity of this O2 sensor depends markedly upon the plticizer concentration and is largely independent of temperature (24.5–52.5 °C) and age (up to 30 d).When exposed to an alternating atmosphere of O2 and N2, a typical oxygen film sensor in PMMA exhibits a 0–90% response and recovery time of 0.4 and 4.5 s, respectively. Keywords: Tris(4,7-diphenyl-1,10-phenanthroline) ruthenium(II) ditetraphenylborate; oxygen sensor; fluorescence oxygen sensor; thin film; plastic film; poly(methyl methacrylate) Oxygen optical sensors (also known as optrodes) are devices which respond specifically and, usually, reversibly to molecular oxygen via a change in luminescence intensity.Such optrodes have a significant number of major advantages over the customary amperometric oxygen sensors; including no oxygen consumption and ready miniaturisation. Their compact, robust nature and ability to withstand high external pressures also contributes to their usefulness making their field of application widespread. Such sensors are likely to be inexpensive and, when combined with optical fibres, make possible the development of disposable, remote, multiple analysis micro-sensors. 1,2 The majority of the optical sensors for O2 developed so far are based on the luminescence quenching of an indicator by molecular oxygen.In such sensors the fluorophores: tris(4,7- diphenyl-1,10-phenanthroline) ruthenium(ii), tris(1,10-phenanthroline) ruthenium(ii), and tris(2,2’-bipyridyl) ruthenium( ii), abbreviated as [Ru(dpp)3]2+, [Ru(phen)3]2+, and [Ru(bpy)3]2+, respectively, dominate the field.These dyes are very photostable and have long excited-state lifetimes, high quantum yields of luminescence and are readily quenched by oxygen (see Table 1).3–9 A good, brief description of the major types of oxygen sensor, based on the ruthenium diimine complexes described above, is given elsewhere.7 One problem with these dyes is the hydrophilic nature of their common salts, such as chloride or perchlorate.As a result, incorporating these dyes in a hydrophobic medium, such as silicone rubber, poses problems, such as dye leaching by water and film fogging due to water uptake. Thus, in the early work on oxygen optical sensors, [Ru(dpp)3(ClO4)2] was incorporated into silicone rubber by using a dichloromethane solution of the complex, since dichloromethane swells up and penetrates the film.3 In this latter work, problems were encountered of film curling and water condensation and, even after these were overcome, the films were still slightly cloudy.3 It should be possible to render the dyes identified in Table 1, or any fluorescent cationic dye for that matter, largely lipophilic and, therefore, soluble in the hydrophobic medium of a polymer, by coupling the dye cation to a lipophilic, nonquenching anion, such as tetraphenyl borate, Ph4B2.In this paper we describe the results of a detailed study of a series of oxygen sensors generated by incorporating the tris(4,7-diphenyl- 1,10-phenanthroline) ruthenium(ii) ditetraphenylborate [Ru(dpp)3 2+(Ph4B2)2] dye ion-pair into a variety of different plasticized polymers. Experimental Materials Unless stated otherwise all the chemicals used in this work were obtained from Aldrich (Gillingham, Dorset, UK). The chemicals used in the synthesis of the [Ru(dpp)3 2+(Ph4B2)2] dye ionpair employed were as follows: ethylene glycol, 4,7-diphenyl- 1,10-phenanthroline, and sodium tetraphenylborate. Ruthenium trichloride trihydrate, was purchased from Merck (Lutterworth, Leicestershire, UK), diethyl ether and sodium chloride were purchased from Fisons (Loughborough, Leicestershire, UK).The polymers used in this work were as follows: poly(methyl methacrylate) (PMMA) (M = 120 000), poly(vinyl acetate) (PVAc), poly(vinyl chloride) (PVC), polystyrene (PS) and cellulose acetate butyrate (CAB) (38-% m/m butyryl content).A low molecular weight poly(methyl methacrylate) (PMMAL) (M = 33 800) was also used and was purchased from Fluka Chemicals (Gillingham, Dorset, UK). The plasticizers used in this work were as follows: tri-n-butyl phosphate (TBP) 98%, Lancaster Synthesis (Morecambe, Table 1 Photochemical characteristics of the dyes Ru(bpy)3 2+, Ru(phen)3 2+ and Ru(dpp)3 2+ pO2 (S = 1 2) in silicone Dye t°/ms FL kQ(O2/109 dm23 mol21 s21 rubber3/Torr Refs.Ru(bpy)3 2+ 0.60 0.042 3.3 376.8 4, 5 (in water) (in water) (in water) Ru(phen)3 2+ 0.92 0.080 4.2 111.3 6, 7 (in water) (in butan-2-one) (in water) Ru(dpp)3 2+ 5.34 Å 0.30 2.5 29.8 8, 9 (in methanol) (in water–ethanol) (in methanol) Analyst, January 1997, Vol. 122 (63–68) 63Lancashire, UK), 2-ethylhexyl diphenyl phosphate (also known as Sancticizer 141, and abbreviated as S141), Monsanto (Newport, Gwent, UK), tris(2-ethylhexyl)phosphate (TOP) 97%, and dioctyl phthalate (DOP) 99%.The solvents used were: AnalaR grade acetone (Fisons), tetrahydrofuran (THF) 99+%, and butan-2-one, 99+%. Useful details associated with the polymers and plasticizers used in this work are given in Table 2.10–13 The air, O2 and N2 used were of a high purity ( > 99%) and obtained from BOC. O2–N2 gas mixtures of different specified compositions, spanning the range 0–100% O2, were generated using a gas blender (Model No. 852V5-S, Signal Instruments, Standards House, Doman Road, Camberley, Surrey, UK). All chemicals were used as received. Preparation of the Dye Ion-pair Complex [Ru(dpp)3 2+ (Ph4B2)2] The non-hydroscopic dye ion-pair complex of [Ru(dpp)3 2+(Ph4B2)2] was prepared as follows: 225.9 mg of RuCl3·3H2O were dissolved in a mixture of 5 cm3 of ethylene glycol and 0.5 cm3 of water at 120 °C, followed by the addition of 862.6 mg of the ligand dpp. The resulting mixture was refluxed at 165 °C for a duration of 45 min cooled, and then 50 cm3 of acetone added.This solution was filtered and the filtrate diluted with 60 cm3 of acetone. A 100 cm3 volume of the filtrate containing the crude Ru(dpp)3Cl2 was added to 100 cm3 of a 10 mmol dm23 aqueous solution of NaPh4B and a fine orange precipitate quickly developed. Finally, 100 cm3 of a 1 mol dm23 sodium chloride solution were added to coagulate the precipitate and make its filtration easier. The orange precipitate of the dye ion-pair complex, [Ru(dpp)3 2+(Ph4B2)2], was filtered and washed four times with 20 cm3 aliquots of distilled water. Purification was achieved by recrystallization of the dye ionpair from an acetone–water mixture (80 + 20 v/v).The precipitate was washed with 20 cm3 of diethyl ether and dried in a desiccator. Preparation and Casting of Film Solutions of the Dye Ion-pair Complex [Ru(dpp)3 2+ (Ph4B2)2] The thin plastic film fluorimetric sensors for O2 used in this work had the general composition: indicator dye (9.6 3 1024 mol dm23)/polymer/plasticizer, and were supported on individual small cut quartz slides.The films were generated by casting a typical film solution of two components. Component solution I comprised 1 cm3 of the same solvent which was used to dissolve the polymer in component solution II (see Table 3) and 1.6 mg of dye. Component solution II was prepared by dissolving the polymer concerned in the solvent stipulated in Table 3.The final solution used for casting, to produce the plasticized film sensor, was made up by adding solution I to 10 g of solution II and sufficient TBP to yield a final level of 30 parts per hundred of resin (phr), i.e., 23.1% m/m. The dry, thin plastic film fluorimetric sensor for O2 containing the dye under test was generated from the final film solution by casting it through a 100 mm thick brass sheet with a rectangular hole (0.8 3 1.5 cm) onto a cut quartz slide (0.85 3 4.2 30.1 cm).The film was then placed in a desiccator, at room temperature, in the dark for a period of 24 h and used. Unless otherwise stated, the plastic film sensors were used at room temperature and were typically 20 mm thick when dried, as measured using a micrometer. In a separate set of experiments, PMMA plastic film O2 sensors containing the following Table 2 Characteristics of the different polymers and plasticizers used Solubility parameter, Abbreviation Molecular formula (Molecular weight) d (J cm23)1 2 Polymer— Cellulose acetate butyrate CAB 17.910 (30 000) Poly(vinyl acetate) PVAC (–CH2CH(O2CCH3)–)n 19.111 (113 000) Polystyrene PS (–CH2CH(C6H5)–)n 18.512 (45 000) Poly(methyl methacrylate) PMMA (–CH2C(CH3)(CO2CH3)–)n 18.812 (120 000) Poly(methyl methacrylate) PMMAL (–CH2C(CH3)(CO2CH3)–)n 18.812 (33 800) Poly(vinyl chloride) PVC (–CH2CH(Cl)–)n 19.212 (95 000) Plasticizer— Phosphates— Tri-n-butyl phosphate TBP O:P(OC4H9)3 17.512 (266.3) Tris(2-ethyl hexyl) phosphate TOP O:P(OCH2(C2H5)CH(CH2)3CH3)3 16.812 (434.7) 2-Ethylhexyl diphenyl phosphate S141 CH3(CH2)3CH(C2H5)CH2OP:O(OC6H5)2 19.012 (362.4) Phthalates— Dioctyl phthalate DOP C6H4-1,2-(CO2CH2CH(C2H5)(CH2)3CH3)2 16.013 (390.6) * R = UCOCH3, UCOCH2CH2CH3 and UOH. 64 Analyst, January 1997, Vol. 122different plasticizers were generated, i.e., TBP, TOP, DOP and S141 at a level of 33 phr, i.e., 24.8% m/m. Methods Spectrophotomeric and fluorescence measurements were carried out using a Perkin-Elmer (Beaconsfield, Buckinghamshire, UK) Lambda 9 UV/VIS/NIR spectrophotometer and a Perkin- Elmer LS-5 fluorimeter/phosphorimeter, respectively.In the majority of the work carried out using the fluorimeter, a typical [Ru(dpp)3 2+(Ph4B2)2] film, coated onto a cut quartz slide, was placed in the centre of a quartz fluorescence cell, which in turn was placed in the sample holder of the fluorimeter. Gases of different O2–N2 composition were delivered to the film sensor by means of two needles positioned at each face of the quartz slide.Two small holes in the cell cover allowed insertion of the needles into the system. Unless stated otherwise, for all the oxygen sensors tested, in the studies of the variation in luminescence intensity as a function of oxygen partial pressure, the wavelengths of excitation and emission used were 430 and 596 nm, respectively. In the measurement of the response/recovery of the [Ru(dpp)3 2+(Ph4B2)2] film, as a function of time, towards rapid variations in the level of O2 (0%–100% and vice versa), a laboratory-made optical system, with a time resolution of approximately 12 ms, was used.In this system, the light source comprised a quartz–iodide tungsten filament lamp, coupled with a blue filter, which allowed light of 430 ± 20 nm through to the plastic film sensor to electronically excite the dye. Typically, the thin film sensor was supported on a cut quartz slide, attached to a Pyrex cell thermostated at 20 °C, and held square on to the excitation light beam.The luminescence from the edge of the film passed through a red filter (which allowed light of > 580 nm through) and then a high-resolution monochromator set at 596 nm (Applied Photophysics, London, UK), both held at right angles to the film. The intensity of this luminescence was measured using a photomultiplier, the output of which was monitored continuously during a run using a storage oscilloscope (Gould).The rapid delivery to the surface of the sensor of alternating streams of nitrogen and oxygen was achieved using two solenoid-activated, pressure-actuated valves, details of which are given elsewhere.14 Theory In a homogeneous medium, such as aqueous solution, quenching of the luminescence of Ru(dpp)3 2+ by oxygen has been found to obey the Stern–Volmer equation, i.e, I0/I = 1+KSVpO2 (1) where I0 and I are the intensities of luminescence in the absence and presence of oxygen at a partial pressure pO2, respectively, and KSV is the Stern–Volmer constant.The Stern–Volmer constant depends directly upon the rate constant for diffusion of oxygen, the solubility of oxygen and the natural lifetime of the electronically-excited state of the fluorophore in the plastic medium. In contrast to homogeneous solution, it has been observed for the majority of the luminescent O2 sensors developed to date, in which a fluorescent indicator such as Ru(dpp)3 2+ is incorporated into a polymer matrix, such as silicone rubber, that the Stern– Volmer plot of I0/I versus pO2 displays a downward curvature.Two models3,15 which have been used to account for this phenomenon are the multisite model, and the non-linear solubility model. In the multisite model, it is suggested that the sensor molecule can exist in two or more sites, each with its own characteristic quenching constant. The second model assumes that the deviation from linearity is due to the non-linear solubility of oxygen in the polymer.Demas et al. have identified recently several features of typical oxygen optical sensors which provide support for their multisite model.16 Both models predict the following modified form of the Stern-Volmer equation: I0/I = 1+ApO2 + BpO2/(1 +b pO2) (2) where A, B, and b are constants related to the parameters in the kinetic and solubility equations associated with the respective models.The above equation has been used to fit successfully the observed data associated with most O2 optrodes. In all the work on oxygen optical sensors described in this paper the Stern– Volmer plots exhibit a downward curvature which is described very well by eqn. (2). Thus, Fig. 1 provides an illustration of the I/I0 versus pO2 and associated Stern–Volmer plots for a typical plastic [Ru(dpp)3 2+(Ph4B2)2] film in PMMA, containing TBP (133 phr) as the plasticizer. In both plots the solid lines represent the variations predicted by eqn.(2) using the values of A, B and b given in the figure legend. We suggest that a useful, quick, albeit crude, guide to the sensitivity of any O2 optrode towards oxygen, at low oxygen levels, is the value of pO2 when the intensity of luminescence, I, has dropped to a value of I0/2, i.e., pO2 (S = 1 2). In a homogeneous medium, where the quenching is expected to obey the Stern–Volmer equation, i.e., eqn. (1), the parameter pO2(S = 1 2) is given by: pO2 (S = 1 2) = 1/KSV (3) However, for most oxygen optical sensors, where the variation in luminescence as a function of pO2 is more likely to be described by the modified form of the Stern–Volmer equation, i.e., eqn.(2), the parameter pO2(S = 1 2) is described by the following expression: Table 3 Composition of component solution II for the different polymers Weight of Volume of polymer in solvent in solution solution Polymer (II)/g Solvent (II)/cm3 CAB 30 Acetone 100 PVAC 50 Acetone 100 PS 100 Butan-2-one 100 PMMA 33 Acetone 100 PMMAL 15 Acetone 30 PVC 12.5 THF 100 Fig. 1 Plot of the observed relative intensity (I/I0) as a function of increasing pO2 level for a typical [Ru(dpp)3 2+(Ph4B2)2] in PMMA ([TBP] = 133 phr) plastic film fluorimetric sensor. The inset diagram is a Stern–Volmer plot of the data in the main diagram. The solid line represents the least squares line of best fit and was calculated using eqn. (2) and the following, optimised values for A, B and b: A = 0.0120; B = 0.0220; b = 0.0018.Analyst, January 1997, Vol. 122 65pO2(S = 1 2) = [2(A+B2b) ±A(A+B2b)2 4Ab]/2Ab (4) In this work all the values of pO2(S = 1 2) quoted have been calculated using eqn. (4). Results and Discussion Effect on Sensitivity of Different Polymers and Plasticizers In an initial study a variety of plastic film oxygen sensors were created in which [Ru(dpp)3 2+(Ph4B2)2], as always, was the fluorophore, the plasticizer was TBP and was kept the same, at a fixed level of 30 phr, and the type of polymer used was varied.For each film, the variation of the luminescence intensity as a function of pO2 was determined, from which a Stern–Volmer plot was generated. In a similar study, a series of oxygen film sensors were generated in which the polymer was kept the same, i.e., PMMA, and the type of plasticizer, although at a fixed level of 33 phr, was varied. The Stern–Volmer plots of the results of both sets of work are illustrated in Fig. 2. A plasticizer is usually included in the formulation for most plastic film gas sensors to improve the rate of gas diffusion throughout the film, since plasticization increases the mobility of polymer segments, which leads to an increase in gas diffusion coefficients.17 Plasticization also increases the workability, flexibility and distensibility of the film. It is common practice to refer to the value of the solubility parameter, d, of the plasticizer and polymer in order to identify compatible plasticizer–polymer combinations.In general, it is suggested13 that the difference in solubility parameter [Dd = d(polymer) 2 d(plasticizer)] should be less than 3.7 J1 2 cm23 2. Table 2 lists the solubility parameter values for the different polymers and plasticizers used in this work and from this data it appears that all the individual plasticizer–polymer film combinations used to generate the data illustrated in Fig. 2 should be compatible. Fig. 3 illustrates the variation in Dd as a function of pO2(S = 1 2) when (a) the polymer was fixed and the type of plasticizer was varied and (b) when the plasticizer was fixed and the type of polymer was varied; this plot was generated using the results illustrated in Fig. 2 and the data contained in Table 2. From the two plots illustrated in Fig. 3, there appears to be a positive correlation between Dd for a plastic film oxygen sensor and its oxygen film sensitivity, as measured by the value of pO2(S = 1 2).The most sensitive films [i.e., those with the lowest values of pO2(S = 1 2)] appear to be associated with highly compatible polymer–plasticizer combinations (i.e., low Dd). This general finding is not too surprising given that the sensitivity of the oxygen optical film sensor depends not only on the rate constant for diffusion of oxygen through the film, but also upon the solubility of oxygen in the film (vide supra).Both these latter parameters are expected to be improved with plasticization, and the better the polymer–plasticizer combination, as measured by Dd, the more effective the plasticizer, as measured by pO2(S = 1 2). [Ru(dpp)3 2+ (Ph4B2)2] in PMMA From the results illustrated in Fig. 3, it appears that, of all the different plasticizers tested, TBP produced the most sensitive oxygen film sensors. In a separate study, a series of [Ru(dpp)3 2+(Ph4B2)2] in PMMA oxygen sensor films were generated and studied in which the level of the TBP plasticizer contained therein was varied.Stern–Volmer plots derived from the variation of the luminescence intensity of the [Ru(dpp)3 2+(Ph4B2)2] luminophore as a function of pO2, for these different films are illustrated in Fig. 4; in this work the level of plasticizer was varied from 0 to 133 phr. In addition, in Fig. 2 (a) Plot of I0/I versus pO2 for a series of [Ru(dpp)3 2+(Ph4B2)2] in PMMA plastic film fluorimetric sensors, in which the type of plasticizer was varied, although its level in the polymer was kept the same, 33 phr.The different plasticizers were as follows: TBP (2), TOP (~), DOP (-) and S141 (8), respectively. The solid lines represents the least squares lines of best fit and were calculated using eqn. (2) and optimised values for A, B and b. (b) Plot of Io/I versus pO2 for a series of [Ru(dpp)3 2+(Ph4B2)2] in different polymer plastic film fluorimetric sensors, with a fixed final level of TBP plasticizer, 30 phr.The different polymers used to make up the films were as follows: CAB (5), PVAC (-), PS (2), PMMA (~), PVC (8) PMMAL (»), respectively. The solid lines represents the least squares lines of best fit and were calculated using eqn. (2) and optimised values for A, B and b. Fig. 3 Plot of pO2(S = 1 2) versus the change in solubility parameter, Dd, when (a) the plasticizer was varied and the polymer, PMMA, was fixed; data from Fig. 2(a) and Table 2 and (b) the polymer was varied and the plasticizer type was fixed (TBP); data from Fig. 2(a) and Table 2. Fig. 4 Stern–Volmer plots of I0/I versus pO2 for a series of plastic film fluorimetric sensors in which the polymer was PMMA and the plasticizer, TBP, was varied over the range 0–133 phr, i.e., 0 (5), 33 (2), 40 (-), 50 (8), 60 (~), 65 (»), 80 («), 106 (“), 120 (/), 133 (3) phr TBP and neat TBP (+), respectively. The solid lines represents the least squares lines of best fit and were calculated using eqn.(2) and optimised values for A, B and b. 66 Analyst, January 1997, Vol. 122Fig. 4 the Stern–Volmer plot of the data obtained for [Ru(dpp)3 2+(Ph4B2)2] in neat TBP is illustrated. Fig. 5 illustrates the variation in the sensitivity of the oxygen sensor films, as measured by pO2(S = 1 2), calculated using eqn. (4) and the data in Fig. 4, as a function of plasticizer concentration in the film. As indicated by the results in Figs. 4 and 5, increasing the level of plasticizer in the film increases markedly the sensitivity of the film towards oxygen, e.g., by a factor of 28.7, on going from 0 to 133 TBP phr. Thus, a [Ru(dpp)3 2+(Ph4B2)2] oxygen sensor film in PMMA with 133 phr TBP has a pO2(S = 1 2) = 3.7 Torr, compared with a value of 34.9 Torr for a film with 33 phr TBP and 29.8 Torr for [Ru(dpp)3 2+(Ph4B2)2] in silicone rubber. A great deal of previous effort has gone into finding new fluorophores for oxygen sensors which will give greater sensitivity than that found using the popular existing luminophores, such as [Ru(dpp)3 2+(Ph4B2)2].From the results of our work, in polymer films which can be plasticized, such as PMMA, a high degree of tuning of the sensitivity of the oxygen film sensor is possible through the careful choice of a compatible polymer–plasticizer combination and, most notably, the level of plasticizer present. The level of plasticization in a film also has an effect on the response and recovery time of a film.Thus, in the absence of plasticizer the 90% response and recovery times, t(90), for an oxygen optical film exposed to 100 and 0% oxygen were found to be 112 and 127 s, respectively, whilst for an oxygen sensor with a plasticizer content of 133 phr, t(90) response and recovery times of 0.4 and 4.5 s, respectively, were recorded. The measured variation in t(90) response and recovery for all the films plasticized at different levels are illustrated in Fig. 6. As with all hyperbolic-type response sensors the recovery times are always greater than the response times and this effect is expected even if the response and recovery processes are both diffusion-controlled as is likely with this type of oxygen sensor. The observed decrease in t(90) for response and recovery with increasing plasticizer content, illustrated in Fig. 6, is expected, since, as noted earlier, plasticization increases the mobility of polymer segments, which leads to an increase in the coefficient for gas diffusion.A brief study in which the 90% response and recovery times of a typical oxygen optical film were determined as a function of temperature over the range 20–55 °C. Both response and recovery times were found to decrease with increasing temperature, e.g., t(90) response and recovery values of 0.4 and 4.5 s were recorded for a film at 20 °C, and 0.3 and 3.0 s for the same film held at 55 °C.An Arrhenius plot of ln(response, or recovery, time) versus T21 gave a good straight line, from the value of the gradient of which an activation energy of 26 kJ mol21 was calculated. The observed decrease in response and recovery times for a typical oxygen plastic film sensor is expected given that the rate of diffusion of oxygen through the film, which is the likely rate determining step associated with the response and recovery process, will increase with increasing temperature.The sensitivity of a typical oxygen film sensor in PMMA ([TBP] = 133 phr), as measured by the value of pO2(S = 1 2), was studied as a function of temperature over the range 24.5–52.5 °C and found to be largely independent; at these latter temperatures, for example, pO2(S = 1 2) values of 32.7 and 36.4 Torr were found, respectively. This finding is not too surprising given that an increase in temperature will not only lead to an increase in the rate constant for diffusion, but also a decrease in the solubility of oxygen in the plastic medium, leading to a marginal overall effect on the sensitivity of the film.Other work showed that the films did not change in sensitivity with age. Thus, there appeared to be little variation in pO2(S = 1 2) with age of the film, when stored in the freezer section of a fridge (24 °C), over a period of 30 d. Thus, for a typical film, pO2(S = 1 2) values of 32.5 and 38.5 Torr were determined after 0 and 30 d, respectively, stored at 24 °C.In one experiment the variation in the relative intensity of luminescence of a typical oxygen film sensor was monitored as a function of time, as the ambient gas phase was altered periodically between 100% nitrogen and 100% oxygen over a period of 12 h; the results of this work are illustrated in Fig. 7 and shows that the response of a typical [Ru(dpp)3 2+(Ph4B2)2] Fig. 5 Plot of pO2(S = 1 2) versus increasing [TBP] for the thin film sensors based on PMMA described in Fig. 4. The values for pO2(S = 1 2) were calculated using eqn. (4). The symbol 8 represents the value for pO2(S = 1 2) obtained for the dye dissolved in neat TBP plasticizer. Fig. 6 Measured 90% response (5) and recovery (-) times for a typical [Ru(dpp)3 2+(Ph4B2)2] in PMMA oxygen film sensor in which increasing amounts of the plasticizer, TBP, was used in the film formulations, spanning the range 0–133 phr. Fig. 7 Observed variation in relative luminescence intensity as a function of time over 12 h for a typical [Ru(dpp)3 2+(Ph4B2)2] in PMMA oxygen film sensor ([TBP] = 133 phr) subjected to an atmosphere which was varied periodically between 100% N2 and 100% O2, respectively.Analyst, January 1997, Vol. 122 67in PMMA ([TBP] = 133 phr) film oxygen sensor is very stable over a 12 h period of operation. In an earlier paper Klimat and Wolfbeis reported that ionpairs based on the cationic fluorophore, Ru(dpp)3 2+, which are similar to those used in this work, could be dissolved in silicone rubber to create a range of fluorimetric oxygen sensors with novel features.7 In this latter work, the most sensitive oxygen sensor reported used dodecyl sulfate, DS2, as the hydrophobic anion and a one-component, transparent, acetic acid releasing, silicone prepolymer as the silicone rubber.This sensor was found to be: (i) very stable with time (its quenching efficiency dropped by only 20–30% over 1 year when stored in a plastic bag), (ii) very fast in response to changes in gaseous oxygen ( < 1 s), (iii) very photostable (daylight illumination over 1 year caused no measurable decomposition), and (iv) stable in aqueous solution (only a 10% drop in quenching constant was observed over a period of 2 months).However, these oxygen sensors were found to be less sensitive, pO2(S = 1 2) = 55 Torr, towards oxygen compared with those reported by others3, in which perchlorate, or chloride, was used as the anion, pO2(S = 1 2) = 29.8 Torr.In this latter work, it is believed that a different quenching mechanism is operative, since the silicone rubber used contained a silica filler, which adsorbs the salts of Ru(dpp)3 2+. From the results of the work reported here it appears that [Ru(dpp)3 2+(Ph4B2)2] ion-pair sensors in plasticized polymers exhibit many of the positive aspects listed above, i.e., (i)–(iv), for the Ru(dpp)3 2+(DS2)2 in silicone rubber oxygen sensors.However, unlike the silicone rubber based sensors, the sensitivity of the [Ru(dpp)3 2+(Ph4B2)2] sensors can be varied substantially through the choice of plasticizer type and content. Thus, as noted earlier, the [Ru(dpp)3 2+(Ph4B2)2] sensors in PMMA have pO2(S = 1 2) values of: 106, 34.9 and 3.7 Torr when the TBP plasticizer content is: 0, 33 and 133 phr, respectively. The latter, highly plasticized, Ru(dpp)3 2+(Ph4B2)2] ion-pair sensor is, therefore, much more sensitive than any of the silicone-rubber based oxygen sensors reported to date, which utilise the same cationic lumiphore.The sensitivity of the [Ru(dpp)3 2+(Ph4B2)2] ion-pair sensors is also dependent upon the nature of the polymer. Thus, [Ru(dpp)3 2+(Ph4B2)2] ion-pair based sensors with pO2(S = 1 2) values much less than 3.7 Torr can be created using a different polymer to PMMA, such as CAB. Conclusion A general method of preparation of thin plastic film, fluorescence- based optical sensors for oxygen is described. The fluorophore cation, ruthenium(ii) tris(4,7-diphenyl-1,10-phenanthroline) is readily incorporated into a variety of different plasticized polymer films by coupling it with the lipophilic anion, tetraphenylborate. The sensitivity of the optical films appears to depend upon the degree of compatibility between the plasticizer and polymer; the greater the compatibility the greater the sensitivity of the sensor. In a more detailed study of a typical film, using PMMA as the polymer and TBP as the plasticizer, the sensitivity of the film towards oxygen increased markedly with increasing level of plasticizer. Increasing the level of plasticizer also decreased markedly the response and recovery times for the sensor. The sensitivity of a typical film is largely independent of temperature and age (when stored at 24 °C). The ability to tune the sensitivity of oxygen optical sensors through the judicial choice of a polymer–plasticizer combination and level of plasticizer is novel and helps widen the possible areas of application of such sensors. The authors thank Sealed Air FPD for funding this work through an EPSRC CASE award. References 1 Seitz, W. R., CRC Crit. Rev. Anal. Chem., 1988, 19, 135. 2 Wolfbeis, O. S., in Fiber Optic Chemical Sensors, ed. Wolfbeis, O. S., CRC Press, Boca Ranton, FL, 1991, vol. II, ch. 10. 3 Carraway, K. R., Demas, J. N., DeGraff B. A., and Bacon, J. R., Anal. Chem., 1991, 63, 337. 4 Lin, C.-L., and Sutin, N., J. Phys. Chem., 1976, 80, 97. 5 van Houten J., and Watts, R. J., J. Am. Chem. Soc., 1976, 98, 4853. 6 Lin, C.-T., Bottcher, W., Chou, M., Creutz C., and Sutin, N., J. Am. Chem. Soc., 1976, 98, 6536. 7 Klim�at, I., and Wolfbeis, O. S., Anal. Chem., 1995, 67, 3160, and references cited therein. 8 Demas, J. N., Harris, E. W., and McBride, R. P., J. Am. Chem. Soc., 1977, 99, 3547. 9 Nakamaru, K., Nishio, K., and Nobe, H., Sci. Rep. Hirosaki Univ., 1979, 26, 57. 10 Burrel, H., in Polymer Handbook, ed. Brandrup, J., and Immergut, E. H., Wiley, New York, 1974, p. 556. 11 Small, P. A., J. Appl. Chem., 1953, 3, 71. 12 Concise Encyclopedia of Polymer Science and Engineering, ed. Kroschwitz, J. I., Wiley, New York, 1990, p. 736. 13 Seymour, R. B., and Carrher, C. E., Polymer Chemistry—An Introduction, Marcel Dekker, New York, 3rd edn., 1992, p. 394. 14 Mills, A., Chang, Q., and Mcmurray, N., Anal.Chem., 1992, 64, 1383. 15 Li, X.-M., Ruan, F.-C and Wong, K.-Y, Analyst, 1993, 118, 289. 16 Demas, J. N., DeGraff, B. A. and Xu, W., Anal. Chem., 1995, 67, 1377. 17 Crank, J., and Park, G. S., Diffusion in Polymers, Academic Press, London, 1968, p. 21. Paper 6/06124I Received September 5, 1996 Accepted November 1, 1996 68 Analyst, January 1997, Vol. 1

ISSN:0003-2654    

DOI:10.1039/a606124i

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Optical Chemical Sensors for Pharmaceutical Analysis Using 1,4-Bis(1,3-benzoxazol-2-yl)benzene as Sensing Material Ying Wang†, Kemin Wang, Wanhui Liu, Guoli Shen and Ruqin Yu* Department of Chemistry and Chemical Engineering, Hunan University, Changsha, 410082, China Optical chemical sensors were prepared for the determination of ethacrynate, berberine, picric acid and picrolonic acid in aqueous solutions based on the fluorescence quenching of 1,4-bis(1,3-benzoxazol-2-yl)benzene (BBOB) in differently plasticized PVC membranes.The possible quenching mechanisms are discussed in detail. The fluorescence changes of the sensing membranes involve the formation of complexes between BBOB and the quenchers in question. The optimum pH ranges for the determination of ethacrynate and berberine are 4.0–10 and 8.0–11, respectively, whereas 1.0–2.0 and 0.90–2.3 mol l21 H2SO4 media are the best for picric acid and picrolonic acid, respectively. The complex compositions of BBOB with the four quenchers are all 1 : 1.The sensors respond linearly in the measuring ranges 8.00 3 1022–1.00 3 1024 mol l21 for ethacrynate, 6.76 3 1024–1.62 3 1025 mol l21 for berberine and picrolonic acid and 2.88 3 1024–4.36 3 1026 mol l21 for picric acid. The response times to reach a stable signal were less than 30 s. In addition to the high reproducibilities, the sensors can reversibly and selectively respond to the quenchers of interest. The proposed sensors were applied to the direct assay of ethacrynic acid and berberine hydrochloride in commercial tablets and the indirect assay of cinchonine and VB1 through quantitative precipitation with picric acid and picrolonic acid, respectively.The results were in good agreement with those obtained by pharmacopoeial and other methods. Keywords: 1,4-Bis(1,3-benzoxazol-2-yl)benzene; optical chemical sensor; fluorescence quenching; pharmaceutical analysis Chemical sensor technology involves the processes of chemical recognition of the analytes and transduction of the analytical signal. More recently, fibre-optic chemical sensors that possess an appropriate sensing layer with the ability for selective and reversible recognition of the analytes have seen a growth in interest.These kinds of sensors used to have sensing tips immobilized with appropriate chemical compounds which change their optical properties upon interaction with the analytes. In particular, a thin film of plasticized organic polymer or porous material which contains the optically sensing components is attached to the end of sensing tip.To this end, selection of the sensing components plays a crucial role in sensing design. Considerable research effort has been devoted to developing optically sensing agents such as an analogue of 4-(2-pyridylazo)resorcinol1 and fluorogenic crown ether compounds, etc.2 To our knowledge, however, no sensing components of benzoxazole compounds have been reported.Because of high quantum yields,3 benzoxazole compounds have been used as optical brightening agents,4 scintillators and active compounds in dye lasers.5,6 Remarkable progress has been made in the synthesis of benzoxazole compounds.7 Furthermore, a number of studies regarding the relationship between the optical characteristics and the structure of these compounds have been published.8,9 Owing to its excellent optical properties and high lipophilicity, it is possible for benzoxazole compounds to be used as sensing materials. 1,4-Bis(1,3-benzoxazol-2-yl)benzene (BBOB) can fluoresce in the blue–green region with the strongest penetration and the smallest attenuation in sea-water, making it especially suitable for application in submarine communication and underwater monitoring. On the other hand, the maximum fluorescence peak at 400 nm just matches the maximum absorption of blood porphyrin. Therefore, it is possible for BBOB to be used for the early detection of carninomatosis. In addition, there are also some applications in isotope separation and spectral measurements.The study of the possible use of BBOB as a sensing material in fibre-optic sensor design for pharmaceutical analysis is of considerable interest. Optical chemical sensing possesses some advantages over electrochemical sensing such as freedom from electrical noise interference and no requirement for a reference electrode. The combination of these merits has resulted in the development of optodes for the assay of vitamin B6,9 glucose,10,11 thiamine,12 penicillin,13 ephedrine hydrochloride, 14 propranolol hydrochloride,14 dopamine,15 cinchonine16 and mexiletine17 in pharmaceutical analysis.This paper reports the use of BBOB for optically sensing ethacrynate and berberine, which can cause fluorescence quenching of membranes containing BBOB. Membranes containing BBOB with different plasticizers can also be used for sensing picric acid and picrolonic acid, which can strongly quench the fluorescence of BBOB under appropriate conditions.The prepared sensors can be used for the direct determination of ethacrynic acid and berberine and the indirect determination of vitamin B1 and cinchonine. To explore the properties of BBOB as an optically sensing material, the quenching mechanisms, complex composition, measuring range, response time, stability, reproducibility, reversibility and selectivity and the effect of experimental variables were investigated.Experimental Instrumentation All fluorescence measurements were conducted on a Hitachi M- 850 spectrofluorimeter with both excitation and emission slits set at 5.0 nm. Data processing was performed on a Macintosh II computer. Elemental analysis was performed on a Heraeus C– H–O–N rapid elemental analyser. Measurements of pH were carried out with a pH-3C pH meter equipped with a combination electrode calibrated by the two-point method.Reagents BBOB was synthesized according to the literature18 and was verified by melting-point determination and elemental analysis. † On leave from the Southwest China Normal University, Chongqing, Sichuan, China. Analyst, January 1997, Vol. 122 (69–75) 69The following compounds were purchased and used without further purification: relatively high molecular mass poly(vinyl chloride) (PVC) (Zhuzhou Chemical Plant), tetrahydrofuran (THF), dibutyl phthalate, bis(2-ethylhexyl) sebacate (DOS) and didecyl phthalate (Shanghai Chemical Reagent Corporation), picric acid, natrii ethacrynas, berberine hydrochloride and sodium picrolonate (Shanghai Pharmaceutical Institute).Acetate buffer solution (pH 6.0) was prepared by dissolving 100 g of sodium acetate trihydrate in 300 ml of water, adding 13 ml of 6 mol l21 acetic acid and diluting to 500 ml. Boric acid– potassium chloride–sodium hydroxide (pH 9.0) buffer solution was obtained from a mixture of 200 ml of 0.2 mol l21 boric acid and 0.2 mol l21 KCl, 85 ml of 0.2 mol l21 NaOH and 515 ml of water. Test solutions of ethacrynate, berberine and picrolonic acid were prepared by dilution of stock standard solutions.A 1.37 3 1022 mol l21 stock solution of picric acid was prepared by dissolving an appropriate amount of picric acid in 500 ml of water. The actual concentration of picric acid was determined by titration with standard sodium hydroxide solution.Solutions of other concentrations were obtained by diluting the stock standard solution with 1.5 mol l21 H2SO4. Unless stated otherwise, all solutions were prepared from analytical-reagent grade and redistilled water. Optode Membrane Preparation Membrane composition The membrane compositions were optimized to give the maximum sensitivity to the quenchers in question. Because of the planar structure of BBOB and a strong intermolecular attraction, BBOB molecules tend to approach each other and the compound has a fairly low solubility in organic solvents.A THF solution saturated with BBOB was used in all experiments for membrane preparation. Appropriate plasticizers must be selected so as to obtain a transparent and flexible membrane, which has the maximum responses to the quenchers of interest. Sensing membranes made of different plasticizers such as tricresyl phosphate, dinonyl sebacate, bis(2-ethylhexyl) sebacate (DOS), dibutyl phthalate and didecyl phthalate were prepared.Membranes consisting of dibutyl phthalate show the best responses to berberine and ethacrynate, while membranes containing didecyl phthalate and DOS give the maximum sensitivity to picric acid and picrolonic acid, respectively. Membrane preparation Different optode membranes for picric acid, ethacrynate, berberine and picrolonic acid were prepared from a mixture of 2.0 ml THF solution saturated with BBOB, 50 mg of PVC and 100 mg of different plasticizers.By means of a spin-on device,16 a membrane of thickness approximately 4 mm was cast on a 35 mm diameter quartz plate. Measurements on Quenchers Four kinds of membranes of different compositions were used in the experiments. To determine each quencher, two identical membranes were mounted in the specially designed flowthrough measuring cell.16 About 3.4 ml of sample solution were aspirated with a syringe. All fluorescence measurement were performed under batch conditions after the dissolved oxygen in the solutions had been excluded by bubbling nitrogen.After each measurement the membrane was rinsed with blank solution until the fluorescence intensity of the optode membrane only contacted with blank solution was recovered. Sample Preparation Sample solution containing approximately 300 mg of berberine was prepared by dissolving the appropriate amount of tablets with the sugar coating carefully peeled off and ground in 150 ml of boiling water.The resulting solution was filtered and the filtrate was transferred into a calibrated flask and then diluted to 250 ml. A portion of ground tablets, approximately equivalent to 200 mg of ethacrynic acid, was acidified with 5.0 ml of 0.01 mol l21 hydrochloric acid, then extracted each time with 10 ml of methylene chloride twice. The extracts were collected in a beaker and evaporated to dryness on a water-bath. The residue was dissolved in an appropriate amount of acetic acid, then the solution was transferred into a calibrated flask and diluted to 100 ml.About 3.5 g of cinchonine were dissolved in 0.04 mol l21 H2SO4 to make 500 ml of solution. To 40.0 ml of the solution were added 80.0 ml of 1.37 3 1022 mol l21 picric acid while stirring to ensure that cinchonine was completely precipitated. The solution was filtered and the filtrate was diluted to 250 ml for fluorescence measurements. About 200 mg of ground tablets of VB1 were dissolved in 0.04 mol l21 H2SO4 at 60 °C to make 200 ml solution.To 100 ml of this solution were added 50.0 ml of 1.05 3 1022 mol l21 picrolonic acid. After VB1 had completely precipitated, the solution was filtered and the filtrate was diluted to 200 ml for fluorescence assay. Results and Discussion The fluorescence responses of the optode membrane containing BBOB to different medicaments are shown in Table 1. Only ethacrynate and berberine show strong fluorescence quenching Table 1 Fluorescence responses of BBOB optode membrane to different pharmaceutical species* Fluorescence Medicament Concentration/mol l21 intensity Reagent blank 0 70.5 Sulfadiazine 1.10 3 1022 (0.04 mol l21 NaOH) 69.2 Sulfaguanidine 1.63 3 1022 (0.03 mol l21 HCl) 69.9 Chlorphenamini maleas 1.30 3 1022 69.0 Lidocaini hydrochloridium 1.00 3 1022 69.5 Diphenhydramini hydrochloridium 1.00 3 1022 70.3 Ethacrynate 1.50 3 1022 (pH 9.0) 32.0 Berberine 1.00 3 1023 (pH 9.0) 30.5 Theophylline 1.00 3 1022 69.8 Vitamine B1 9.70 3 1023 69.5 Inositol 1.06 3 1022 70.3 Levamisoli hydrochloridium 1.16 3 1022 70.0 Amobarbitalum natricum 1.03 3 1022 70.4 Chloraminium T 2.02 3 1022 67.5 Cinchonini hydrochloridium 1.03 3 1023 70.3 Colchicinum 5.00 3 1022 69.5 Ketamini hydrochloridium 1.02 3 1022 69.2 Bendazol 1.00 3 1022 (0.02% C2H5OH) 71.5 Pentoxyverine 1.00 3 1022 71.0 Ephedrini hydrochloridium 1.00 3 1022 70.0 Sodium thiopentl 1.00 3 1022 69.5 * Optode membrane composed of BBOB, dibutyl phthalate and PVC. 70 Analyst, January 1997, Vol. 122N O O N O N O N + – hn1 hn2 I II fluorescence of the sensing membrane. This is the basis of the optically sensing device in these investigations. Possible Quenching Mechanism for the Four Quenchers BBOB, as a bis(benzoxazole) compound, can exhibit strong fluorescence emission owing to the conjugated double bond system and the high mobility of its p-electrons. There are several kinds of resonance interactions at different positions in the molecule.The possible resonance form is a quinoid structure with separation of positive and negative charges and with a planar configuration in the excited state. The process of the fluorescence emission of BBOB can be represented as follows: 19 Berberine, being a quaternary ammonium salt, can form ionassociation complexes with some anionic dyes such as Bromophenol Blue (BPB)20,21 and Bromocresol Green (BCG).22 Berberine seems to form ion-association complex with BBOB in a PVC membrane owing to the existence of a negative charge in the structure (II). Therefore, it can be assumed that the fluorescence quenching of BBOB by berberine is related to the formation of an ion-association complex, which inhibits the fluorescence emission of II.On the other hand, the positive charge in the structure II seems to provide the possibility of forming an ion associate with the carboxyl group of ethacrynate at pH 6.0, causing the reduction of the fluorescence emission of II.These processes can be represented as follows: mA+(aq.)"mA+(org.) + nB2(org.) + AmBn(org.) mE2(aq.)"mE2(org.) + nB+(org.) + EmBn(org.) The respective reactions are K1 mA+(aq.) + nB2(org.)"AmBn(org.) K2 mE2(aq.) + nB(org.)"EmBn(org.) where A, E and B represent berberine, ethacrynate and BBOB, respectively, and K1 and K2 are the equilibrium constants for reactions of A and E with B, respectively. One can describe the equilibrium involved as follows:16 an 1- a = 1 nK1CB n-1[A]m (1) an 1- a = 1 nK2CB n-1[E]m (2) where CB is the total concentration of BBOB in the membrane and [A] and [E] are the concentrations of A and E in aqueous solution.The relative fluorescence value a is the ratio of free BBOB concentration, [B], to the total concentration of BBOB present in the membrane, CB, i.e., a = [B] CB (3) and can be calculated according to the following eqn.: a = F - Fs Fb - Fs (4) where Fb is the fluorescence intensity of the optode membrane in the blank solution used for the determination of berberine or ethacrynate, Fs is the fluorescence intensity when BBOB in the membrane is completely complexed by berberine or ethacrynate Fig. 1 Graph of relative fluorescence intensity versus log (concentration) of each analyte (lex = 344 nm, lem = 390 nm) (a) M1 (PVC + BBOB + dibutyl phthalate) for ethacrynate; (b) M2 (PVC + BBOB + dibutyl phthalate) for berberine; (c) M3 (PVC + BBOB + didecyl phthalate) for picric acid; (d) M4 (PVC + BBOB + DOS) for picrolonic acid.Fig. 2 Membrane responses to ethacrynate and berberine in solutions of different pH. 1, M1 for 2.00 3 1023 mol l21 ethacrynate; 2, M2 for 4.40 3 1024 mol l21 berberine. Analyst, January 1997, Vol. 122 71and F is the fluorescence intensity of the optode membrane actually measured when contacting with solutions of different concentrations of berberine or ethacrynate. It is apparent from eqns. (1) and (2) that when the stoichiometric ratios of the complexes change, the relative fluorescence values (a) have different functional relationships with the concentrations of berberine or ethacrynate. Eqns.(1) and (2) provide the basis for the determination of berberine or ethacrynate. In addition to normal ion associates, there is also the possibility of other forms of interaction between BBOB and the fluorescence quenchers involved. BBOB is a nitrogen- and oxygen-containing heterocyclic compound and the electronegativity of the oxygen atom is stronger than that of nitrogen atom.When the quencher species of interest do not contain charged groups, such as with picric and picrolonic acids in H2SO4 solution, the interaction between BBOB and the fluorescence quenchers involved and the formation of associates can be accomplished through, say, hydrogen bonding with the phenolic group in the quencher molecules and the structure of II, causing an energy decrease of II and fluorescence quenching.On the supposition that BBOB in the plasticized PVC membrane phase (org.) with picric acid and picrolonic acid in the H2SO4 solutions (aq.) forms m–n associates, one can express the equilibrium as K3 mP(aq.) + nB2(org.)"PmBn(org.) K4 mS(aq.) + nB(org.)"SmBn(org.) where P, S and B represent picric acid, picrolonic acid and BBOB, respectively, and K3 and K4 are the equilibrium Fig. 3 Membrane responses to picric acid and picrolonic acid in H2SO4 media of different concentrations. 1, M3 for 1.40 3 1025 mol l21 picric acid; 2, M4 for 5.00 3 1025 mol l21 picrolonic acid. Fig. 4 Fitting the experimental data to eqns. (1), (2), (5) and (6). (a) Berberine and (d) picrolonic acid: 1, m: n = 2 : 1, K1(K4) = 1 3 108; 2, m: n = 2 : 2, K1(K4) = 3 3 1011; 3, m: n = 1 : 1, K1(K4) = 1000 (best fit); 4, m: n = 1 : 2, K1(K4) = 3 3 107. (b) Ethacrynate 1, m: n = 2 : 1, K2 = 1.3 3 104; 2, m: n = 2 : 2, K2 = 2.5 3106; 3, m: n = 1 : 1, K2 = 120 (best fit); 4, m: n = 1 : 2, K2 = 2.5 3104.(c) Picric acid 1, m: n = 2 : 1, K3 = 1 3109; 2, m: n = 2 : 2, K3 = 3 3 1012; 3, m: n = 1 : 1, K3 = 30 000 (best fit); 4, m: n = 1 : 2, K3 = 9 3 107. 72 Analyst, January 1997, Vol. 122constants for reactions of P and S with B, respectively. Similarly to the above, one has an 1- a = 1 nK3CB n-1[P]m (5) an 1- a = 1 nK4CB n-1[S]m (6) where [P] and [S] are the concentrations of P and S in aqueous acidic sample solutions. The relative fluorescence value a can be calculated as mentioned above.Obviously, eqns. (5) and (6) provide the basis for the determination of picric acid and picrolonic acid, respectively. Characteristics of the Optical Chemical Sensor Fluorescence quenching When the BBOB optode membranes corresponding to the quenchers of interest were exposed to solutions containing different concentrations of the four quenchers, the fluorescence intensities were recorded at lex = 344 nm and lem = 390 nm.The fluorescence intensities of the optode membranes decreased as the concentrations of the quenchers increased (Fig. 1). This illustrates that the optode membranes can be used for the assay of quenchers involved in sample solutions. Influence of acidity Although there are nitrogen and oxygen atoms in the BBOB molecule, the blank fluorescence of the sensing membrane itself was found to be independent of the acidity of the medium used in the experiment. Fig. 2 shows the dependence of the a values of the optode membranes contacted with a series of 2.00 31023 mol l21 ethacrynate and 4.40 3 1024 mol l21 berberine solutions at different pH values.It is clear that the a values for ethacrynate at pH > 4 (at pH < 4, ethacrynate would precipitate) and for berberine at pH 8–11 were nearly constant. In this work, pH 6.0 HOAc–NaOAc and pH 9.0 H3BO3–NaOH buffer solutions were used for the determination of ethacrynate and berberine, respectively. Fig. 3 shows the influence of H2SO4 concentration on the a values of the membranes in contact with 1.40 3 1025 mol l21 picric acid and 5.00 3 1025 mol l21 picrolonic acid solutions.With increase in H2SO4 concentration, the ionization of picric acid and picrolonic acid decreases and consequently the quenchers in question exist mainly in molecular form. Therefore, the ability to form hydrogen bonds increases and the ratio of the concentration of free BBOB to the total concentration of BBOB, a, decreases.In other words, the ability for quenching increased. This can also be confirmed by our experiments on the influence of increasing pH from 1 to 12, which decreases the ability of fluorescence quenching of the optode membranes, that is, strongly acidic media are the best for the determination of picric and picrolonic acid. However, we also discovered that when the concentration of H2SO4 is higher than 2.50 mol l21, BBOB is slowly dissolved out from the membrane.It can be seen from Fig. 3 that the optode membranes have more sensitive and more stable responses to picric acid and picrolonic acid in 1.0–2.0 and 0.90–2.3 mol l21 H2SO4, respectively. In subsequent experiments, 1.5 mol l21 H2SO4 was used for the determination of picric acid and picrolonic acid. Complex composition and measuring range As mentioned above, when the stoichiometric ratios of the complexes change, the a values have different functional relationships with the concentrations of the quenchers in question.The experimental data were fitted to eqns. (1), (2), (5) and (6) by changing the ratio of m to n and adjusting the over-all equilibrium constant K. Fig. 4(a), (b), (c) and (d) show the best fitted curves to represent the experimental data for berberine, ethacrynate, picric acid and picrolonic acid, respectively. It is interesting that the complex ratios of the quenchers in question to BBOB in the optode membrane are all 1 : 1.The best fitted curves can serve as calibration curves for the determinations of the quenchers of interest. Log(concentration) can be converted to concentration from Fig. 4; satisfactory responses are obtained from 8.00 3 1022 to 1.00 3 1024 mol l21 for ethacrynate, from 6.7631024 to 1.62 31025 mol l21 for berberine and picrolonic acid and from 2.88 3 1024 to 4.36 3 1026 mol l21 for picric acid. Reproducibility and reversibility The reproducibility and reversibility of the optode membranes corresponding to the quenchers of interest were studied using repeated determinations (n = 12) for each sample solution.After each measurement, the membrane was rinsed with blank solution until the fluorescence intensity (Fb) was recovered. Table 2 gives the experimental results. The mean fluorescence intensities ± standard deviations are 24.9 ± 0.6, 35.9 ± 0.6, 22.8 ± 0.6, 24.5 ± 0.7 for ethacrynate, berberine, picric acid and picrolonic acid, respectively. These results indicate that the optode membranes have good reproducibility and reversibility.Response time The optode membranes M1, M2, M3 and M4 were immersed first in a low concentration and then a high concentration sample Table 2 Reproducibility and reversibility of optode membranes corresponding to different quenchers Quencher Membrane concentration/ Fluorescence Membrane composition Quencher mol l21 intensity Mean s M1 PVC+ Ethacrynate 7.50 3 1023 25.8, 25.6, 25.4, 25.3, 24.9 0.6 BBOB+ 25.0, 24.9, 24.9, 24.7, dibutyl phthalate 24.6, 24.3, 24.0, 24.0 M2 PVC+ Berberine 1.32 3 1025 36.6, 36.5, 36.4, 36.4, 35.9 0.6 BBOB+ 36.3, 36.0, 36.0, 35.8, dibutyl phthalate 35.5, 35.3, 35.1, 34.9 M3 PVC+ Picric acid 8.22 3 1026 23.8, 23.5, 23.3, 23.0, 22.8 0.6 BBOB+ 23.0, 22.9, 22.8, 22.7, didecyl phthalate 22.4, 22.0, 22.3, 21.9 M4 PVC+ Picrolonic acid 3.06 3 1025 25.5, 25.5, 25.3, 25.0, 24.5 0.7 BBOB+ 24.8, 24.5, 24.3, 24.3, DOS 24.0, 23.8, 23.5, 23.4 Analyst, January 1997, Vol. 122 73solution. All response times to reach equilibrium between the sensing membranes and the corresponding quenchers or to reach the steady-state fluorescence signal were found to be less than 30 s. Short-term stability In order to monitor the dissolving out of BBOB from the membrane into the sample solution, the fluorescence signal at 390 nm for the optode membranes in contact with the quenchers in question were recorded over a period of 10 h with an interval of 30 min (n = 21).Fig. 5 shows a graph of fluorescence intensities versus time for each optode. It illustrates that BBOB is not significantly permeable from the sensing membrane during this period of time. Different approaches for the preservation of the optode membranes such as in redistilled water, in air and in redistilled water with the dissolved oxygen removed by bubbling nitrogen were tried. The fluorescence intensities of the optode membrane in air and in redistilled water decrease slightly with time, whereas those in redistilled water with the dissolved oxygen removed remain nearly constant.Apparently, the decrease in fluorescence intensities is due to oxygen, so the membrane should be preserved in redistilled water saturated with nitrogen when not in use. Selectivity Any substance that partitions into the membranes and quenches the fluorescence of the membranes or reacts with the analytes will interfere with the measurement. No significant interferences were observed if a less than a ±5% relative error was tolerated for the determination of 5.00 3 1023 mol l21 ethacrynate, 4.40 3 1024 mol l21 berberine, 5.50 3 1025 mol l21 picric acid and 4.90 3 1025 mol l21 picrolonic acid.These results are summarized in Table 3. Obviously, the optode membranes based on BBOB can be used for the selective determination of the quenchers in question. Applications The practical applications of the present sensors were tested on the direct and indirect determination of drugs.The direct methods involve the assay of ethacrynic acid and berberine hydrochloride in commercial tablets. The sample solutions for ethacrynic acid and berberine hydrochloride were diluted with buffer solutions of pH 6.0 and 9.0, respectively, and analysed using the described optode membranes M1 and M2, respec- Fig. 5 Change in fluorescence intensity with time. (a) 1, M1 for 3.5 3 1023 mol l21 ethacrynate; 2, M3 for 1.37 3 1025 mol l21 picric acid.(b) 1, M2 for 3.85 3 1025 mol l21 berberine; 2, M4 for 5.0 3 1025 mol l21 picrolonic acid. Table 3 Maximum tolerable concentrations of interferents for the assay of different analytes. The concentration of the four analytes of interest are 5.0 31023, 4.4 3 1024, 5.5 3 1025 and 4.9 3 1025 mol l21 for ethacrynate, berberine, picric acid and picrolonic acid, respectively. Interferent Tolerable concentration/mol l21 Type Compound M1 M2 M3 M4 Alkali metal and alkaline earth metal salts NaCl 0.10 0.010 0.050 0.050 NaOAc 0.10 0.50 0.050 0.050 NaNO3 0.10 0.010 0.010 0.050 MgCl2 0.050 0.010 0.010 0.010 CaCl2 1.2 3 1023 1.2 3 1024 1.2 3 1024 1.2 3 1024 Organic acids and their salts Sodium citrate 0.05 0.10 0.080 0.10 Malic acid 7.0 3 1023 4.2 3 1023 7.0 3 1023 7.0 3 1023 Sodium oxalate 2.8 3 1022 2.5 3 1022 2.5 3 1022 2.8 3 1022 Sodium tartrate 5.0 3 1022 0.10 5.0 3 1022 5.0 3 1022 Phenylformic acid 1.5 3 1022 1.5 3 1022 1.5 3 1022 1.5 3 1022 Phenols Phenol 3.3 3 1022 6.5 3 1022 6.5 3 1022 6.5 3 1022 Vitamins VB1 5.0 3 1023 5.0 3 1023 5.0 3 1023 5.0 3 1023 Inositol 5.0 3 1023 1.0 3 1023 1.0 3 1023 1.0 3 1023 Piperazines Levamisole 5.0 3 1023 1.2 3 1023 1.2 3 1023 1.2 3 1023 Xanthine bases Theophylline 5.0 3 1023 1.9 3 1022 1.9 3 1023 5.0 3 1023 Table 4 Results for the determination of four medicaments using the optode membranes and the pharmacopoeial and other methods Recovery obtained Recovery obtained by pharmacopoeial Sample with the optode or other Medicament no.membrane* (%) method* (%) Ethacrynate 1 96.6 ± 0.8 95.8 ± 0.5 2 94.6 ± 1.0 94.6 ± 0.8 3 98.0 ± 0.5 97.9 ± 0.6 Berberine 1 73.5 ± 1.0 73.5 ± 0.8 2 74.2 ± 1.5 74.3 ± 2.0 3 70.8 ± 0.5 70.8 ± 1.3 Cinchonine 1 98.5 ± 0.5 98.5 ± 1.0 2 97.4 ± 1.0 97.0 ± 0.8 3 96.5 ± 0.8 96.8 ± 1.2 VB1 1 93.4 ± 2.0 93.0 ± 1.5 2 90.5 ± 1.2 90.0 ± 1.0 3 95.0 ± 0.8 95.6 ± 0.5 * Mean values ± s of three determinations. 74 Analyst, January 1997, Vol. 122tively. The measured results are in correspondence with the results obtained by pharmacopoeial methods.23,24 (Table 4). The indirect methods for the determination of cinchonine and VB1 are based on quantitative precipitation with picric acid and picrolonic acid, respectively. The filtrates containing the excess of picric acid and picrolonic acid were diluted with an appropriate amount of 1.5 mol l21 H2SO4 and determined using the described optode membranes M3 and M4, respectively.The moles of quencher reacted are equivalent to the moles of the analyte present. These results are in agreement with the results obtained by other methods25,26 (Table 4). This work was supported by the National Natural Science Foundation and the Foundation for PhD Thesis Research of the National Education Commission of China. References 1 Wang, K.M., Seilr, K., Rusterholz, B., and Simon, W., Analyst, 1992, 117, 57. 2 Blair, T. L., Cynkowski, T., and Bachas, L.G., Anal. Chem., 1993, 65, 945. 3 Reiser, A., Leeyshon, L. J., Saunders, D., Mijovic, M. V., Bright, A., and Bogie, A., J. Am. Chem. Soc., 1992, 94, 2414. 4 Tanaka, T., Jpn. Pat., 1966, 3515; Chem. Abstr., 1967, 65, 7328f. 5 Liphardt, B., Liphardt, B., and Luettke, W., Chem. Ber., 1982, 115, 2997. 6 Arient, J., Collect. Czech. Chem. Commun., 1980, 45, 3160. 7 Zhou, Y.-M., Wu, Y.-M., and Gao, Z.-H., Chem. J. Chin. Univ., 1989, 10, 724. 8 Qin, Y.-X., Wang, M.-Z., and Gao, Z.-H., Chem.J. Chin. Univ., 1990, 11, 22. 9 Chen, D.-H., Luque de Castro, M. D., and Valcarcel, M., Anal. Chim. Acta, 1992, 261, 269. 10 Narayanaswamy, R., and Sevilla, F., III, Anal. Lett., 1988, 21, 1165. 11 Meadows, D., and Schultz, J. S., Talanta, 1988, 35, 1450. 12 He, H.-R., Uray, G., and Wolfbeis, O. S., Anal. Lett., 1992, 25, 405. 13 Luo, S.-F., and Walt, D. R., Anal. Chem., 1989, 61, 1069. 14 He, H.-R., Uray, G., and Wolfbeis, O. S., Anal. Chim. Acta, 1991, 246, 251. 15 Wang, K.-M., Huo, X.-Q., Chan, Y.-K., and Li, H.-M., J.Hunan Univ., 1995, 22, 55. 16 Zheng, H.-H., Wang, K.-M., Liu, C.-L., and Yu, R.-Q., Talanta, 1993, 40, 1569. 17 Wang, K.-M., Ou, Y.-W., and Bian, K.-J., Chem. J. Chin. Univ., 1992, 13, 1529. 18 Zhao, Y.-M., Zhang, D.-Y., and Gao, Z.-H., Chem. J. Chin. Univ., 1984, 5, 488. 19 Nyilas, E., and Pinter, J. L., J. Am. Chem. Soc., 1960, 82, 609. 20 Auerbach, E., Ind. Eng. Chem., Anal. Ed., 1943, 51, 492. 21 Kaneda, Y., and Iwaida, M., Eisei Kagaku, 1976, 22, 370. 22 Irving, H. M. N. H., and Markham, J. J., Anal. Chim. Acta, 1967, 39, 7. 23 Pharmacopoeia of the People’s Republic of China, Part III, Chemical Engineering Press, Beijing, 1990, p. 235. 24 Pharmacopoeia of the People’s Republic of China, Part III, Chemical Engineering Press, Beijing, 1990, p. 437. 25 Pharmacopoeia of the People’s Republic of China, Part III, Chemical Engineering Press, Beijing, 1990, p. 634. 26 Shen, K.-Y., Wang, M.-X., and Hau, P.-Y., in Handbook of Practical Analysis of Pharmaceuticals, People’s Medical Press, Beijing, 1986, p. 355. Paper 6/05031J Received July 18, 1996 Accepted September 25, 1996 Analyst, January 1997, Vol. 122 75 Optical Chemical Sensors for Pharmaceutical Analysis Using 1,4-Bis(1,3-benzoxazol-2-yl)benzene as Sensing Material Ying Wang†, Kemin Wang, Wanhui Liu, Guoli Shen and Ruqin Yu* Department of Chemistry and Chemical Engineering, Hunan University, Changsha, 410082, China Optical chemical sensors were prepared for the determination of ethacrynate, berberine, picric acid and picrolonic acid in aqueous solutions based on the fluorescence quenching of 1,4-bis(1,3-benzoxazol-2-yl)benzene (BBOB) in differently plasticized PVC membranes.The possible quenching mechanisms are discussed in detail. The fluorescence changes of the sensing membranes involve the formation of complexes between BBOB and the quenchers in question. The optimum pH ranges for the determination of ethacrynate and berberine are 4.0–10 and 8.0–11, respectively, whereas 1.0–2.0 and 0.90–2.3 mol l21 H2SO4 media are the best for picric acid and picrolonic acid, respectively. The complex compositions of BBOB with the four quenchers are all 1 : 1.The sensors respond linearly in the measuring ranges 8.00 3 1022–1.00 3 1024 mol l21 for ethacrynate, 6.76 3 1024–1.62 3 1025 mol l21 for berberine and picrolonic acid and 2.88 3 1024–4.36 3 1026 mol l21 for picric acid.The response times to reach a stable signal were less than 30 s. In addition to the high reproducibilities, the sensors can reversibly and selectively respond to the quenchers of interest. The proposed sensors were applied to the direct assay of ethacrynic acid and berberine hydrochloride in commercial tablets and the indirect assay of cinchonine and VB1 through quantitative precipitation with picric acid and picrolonic acid, respectively. The results were in good agreement with those obtained by pharmacopoeial and other methods.Keywords: 1,4-Bis(1,3-benzoxazol-2-yl)benzene; optical chemical sensor; fluorescence quenching; pharmaceutical analysis Chemical sensor technology involves the processes of chemical recognition of the analytes and transduction of the analytical signal. More recently, fibre-optic chemical sensors that possess an appropriate sensing layer with the ability for selective and reversible recognition of the analytes have seen a growth in interest.These kinds of sensors used to have sensing tips immobilized with appropriate chemical compounds which change their optical properties upon interaction with the analytes. In particular, a thin film of plasticized organic polymer or porous material which contains the optically sensing components is attached to the end of sensing tip. To this end, selection of the sensing components plays a crucial role in sensing design. Considerable research effort has been devoted to developing optically sensing agents such as an analogue of 4-(2-pyridylazo)resorcinol1 and fluorogenic crown ether compounds, etc.2 To our knowledge, however, no sensing components of benzoxazole compounds have been reported.Because of high quantum yields,3 benzoxazole compounds have been used as optical brightening agents,4 scintillators and active compounds in dye lasers.5,6 Remarkable progress has been made in the synthesis of benzoxazole compounds.7 Furthermore, a number of studies regarding the relationship between the optical characteristics and the structure of these compounds have been published.8,9 Owing to its excellent optical properties and high lipophilicity, it is possible for benzoxazole compounds to be used as sensing materials. 1,4-Bis(1,3-benzoxazol-2-yl)benzene (BBOB) can fluoresce in the blue–green region with the strongest penetration and the smallest attenuation in sea-water, making it especially suitable for application in submarine communication and underwater monitoring.On the other hand, the maximum fluorescence peak at 400 nm just matches the maximum absorption of blood porphyrin. Therefore, it is possible for BBOB to be used for the early detection of carninomatosis. In addition, there are also some applications in isotope separation and spectral measurements. The study of the possible use of BBOB as a sensing material in fibre-optic sensor design for pharmaceutical analysis is of considerable interest.Optical chemical sensing possesses some advantages over electrochemical sensing such as freedom from electrical noise interference and no requirement for a reference electrode. The combination of these merits has resulted in the development of optodes for the assay of vitamin B6,9 glucose,10,11 thiamine,12 penicillin,13 ephedrine hydrochloride, 14 propranolol hydrochloride,14 dopamine,15 cinchonine16 and mexiletine17 in pharmaceutical analysis. This paper reports the use of BBOB for optically sensing ethacrynate and berberine, which can cause fluorescence quenching of membranes containing BBOB.Membranes containing BBOB with different plasticizers can also be used for sensing picric acid and picrolonic acid, which can strongly quench the fluorescence of BBOB under appropriate conditions. The prepared sensors can be used for the direct determination of ethacrynic acid and berberine and the indirect determination of vitamin B1 and cinchonine.To explore the properties of BBOB as an optically sensing material, the quenching mechanisms, complex composition, measuring range, response time, stability, reproducibility, reversibility and selectivity and the effect of experimental variables were investigated. Experimental Instrumentation All fluorescence measurements were conducted on a Hitachi M- 850 spectrofluorimeter with both excitation and emission slits set at 5.0 nm. Data processing was performed on a Macintosh II computer.Elemental analysis was performed on a Heraeus C– H–O–N rapid elemental analyser. Measurements of pH were carried out with a pH-3C pH meter equipped with a combination electrode calibrated by the two-point method. Reagents BBOB was synthesized according to the literature18 and was verified by melting-point determination and elemental analysis. † On leave from the Southwest China Normal University, Chongqing, Sichuan, China. Analyst, January 1997, Vol. 122 (69–75) 69The following compounds were purchased and used without further purification: relatively high molecular mass poly(vinyl chloride) (PVC) (Zhuzhou Chemical Plant), tetrahydrofuran (THF), dibutyl phthalate, bis(2-ethylhexyl) sebacate (DOS) and didecyl phthalate (Shanghai Chemical Reagent Corporation), picric acid, natrii ethacrynas, berberine hydrochloride and sodium picrolonate (Shanghai Pharmaceutical Institute). Acetate buffer solution (pH 6.0) was prepared by dissolving 100 g of sodium acetate trihydrate in 300 ml of water, adding 13 ml of 6 mol l21 acetic acid and diluting to 500 ml.Boric acid– potassium chloride–sodium hydroxide (pH 9.0) buffer solution was obtained from a mixture of 200 ml of 0.2 mol l21 boric acid and 0.2 mol l21 KCl, 85 ml of 0.2 mol l21 NaOH and 515 ml of water. Test solutions of ethacrynate, berberine and picrolonic acid were prepared by dilution of stock standard solutions. A 1.37 3 1022 mol l21 stock solution of picric acid was prepared by dissolving an appropriate amount of picric acid in 500 ml of water.The actual concentration of picric acid was determined by titration with standard sodium hydroxide solution. Solutions of other concentrations were obtained by diluting the stock standard solution with 1.5 mol l21 H2SO4. Unless stated otherwise, all solutions were prepared from analytical-reagent grade and redistilled water. Optode Membrane Preparation Membrane composition The membrane compositions were optimized to give the maximum sensitivity to the quenchers in question.Because of the planar structure of BBOB and a strong intermolecular attraction, BBOB molecules tend to approach each other and the compound has a fairly low solubility in organic solvents. A THF solution saturated with BBOB was used in all experiments for membrane preparation. Appropriate plasticizers must be selected so as to obtain a transparent and flexible membrane, which has the maximum responses to the quenchers of interest.Sensing membranes made of different plasticizers such as tricresyl phosphate, dinonyl sebacate, bis(2-ethylhexyl) sebacate (DOS), dibutyl phthalate and didecyl phthalate were prepared. Membranes consisting of dibutyl phthalate show the best responses to berberine and ethacrynate, while membranes containing didecyl phthalate and DOS give the maximum sensitivity to picric acid and picrolonic acid, respectively. Membrane preparation Different optode membranes for picric acid, ethacrynate, berberine and picrolonic acid were prepared from a mixture of 2.0 ml THF solution saturated with BBOB, 50 mg of PVC and 100 mg of different plasticizers.By means of a spin-on device,16 a membrane of thickness approximately 4 mm was cast on a 35 mm diameter quartz plate. Measurements on Quenchers Four kinds of membranes of different compositions were used in the experiments. To determine each quencher, two identical membranes were mounted in the specially designed flowthrough measuring cell.16 About 3.4 ml of sample solution were aspirated with a syringe.All fluorescence measurement were performed under batch conditions after the dissolved oxygen in the solutions had been excluded by bubbling nitrogen. After each measurement the membrane was rinsed with blank solution until the fluorescence intensity of the optode membrane only contacted with blank solution was recovered. Sample Preparation Sample solution containing approximately 300 mg of berberine was prepared by dissolving the appropriate amount of tablets with the sugar coating carefully peeled off and ground in 150 ml of boiling water.The resulting solution was filtered and the filtrate was transferred into a calibrated flask and then diluted to 250 ml. A portion of ground tablets, approximately equivalent to 200 mg of ethacrynic acid, was acidified with 5.0 ml of 0.01 mol l21 hydrochloric acid, then extracted each time with 10 ml of methylene chloride twice.The extracts were collected in a beaker and evaporated to dryness on a water-bath. The residue was dissolved in an appropriate amount of acetic acid, then the solution was transferred into a calibrated flask and diluted to 100 ml. About 3.5 g of cinchonine were dissolved in 0.04 mol l21 H2SO4 to make 500 ml of solution. To 40.0 ml of the solution were added 80.0 ml of 1.37 3 1022 mol l21 picric acid while stirring to ensure that cinchonine was completely precipitated.The solution was filtered and the filtrate was diluted to 250 ml for fluorescence measurements. About 200 mg of ground tablets of VB1 were dissolved in 0.04 mol l21 H2SO4 at 60 °C to make 200 ml solution. To 100 ml of this solution were added 50.0 ml of 1.05 3 1022 mol l21 picrolonic acid. After VB1 had completely precipitated, the solution was filtered and the filtrate was diluted to 200 ml for fluorescence assay.Results and Discussion The fluorescence responses of the optode membrane containing BBOB to different medicaments are shown in Table 1. Only ethacrynate and berberine show strong fluorescence quenching Table 1 Fluorescence responses of BBOB optode membrane to different pharmaceutical species* Fluorescence Medicament Concentration/mol l21 intensity Reagent blank 0 70.5 Sulfadiazine 1.10 3 1022 (0.04 mol l21 NaOH) 69.2 Sulfaguanidine 1.63 3 1022 (0.03 mol l21 HCl) 69.9 Chlorphenamini maleas 1.30 3 1022 69.0 Lidocaini hydrochloridium 1.00 3 1022 69.5 Diphenhydramini hydrochloridium 1.00 3 1022 70.3 Ethacrynate 1.50 3 1022 (pH 9.0) 32.0 Berberine 1.00 3 1023 (pH 9.0) 30.5 Theophylline 1.00 3 1022 69.8 Vitamine B1 9.70 3 1023 69.5 Inositol 1.06 3 1022 70.3 Levamisoli hydrochloridium 1.16 3 1022 70.0 Amobarbitalum natricum 1.03 3 1022 70.4 Chloraminium T 2.02 3 1022 67.5 Cinchonini hydrochloridium 1.03 3 1023 70.3 Colchicinum 5.00 3 1022 69.5 Ketamini hydrochloridium 1.02 3 1022 69.2 Bendazol 1.00 3 1022 (0.02% C2H5OH) 71.5 Pentoxyverine 1.00 3 1022 71.0 Ephedrini hydrochloridium 1.00 3 1022 70.0 Sodium thiopentl 1.00 3 1022 69.5 * Optode membrane composed of BBOB, dibutyl phthalate and PVC. 70 Analyst, January 1997, Vol. 122N O O N O N O N + – hn1 hn2 I II fluorescence of the sensing membrane. This is the basis of the optically sensing device in these investigations. Possible Quenching Mechanism for the Four Quenchers BBOB, as a bis(benzoxazole) compound, can exhibit strong fluorescence emission owing to the conjugated double bond system and the high mobility of its p-electrons.There are several kinds of resonance interactions at different positions in the molecule. The possible resonance form is a quinoid structure with separation of positive and negative charges and with a planar configuration in the excited state. The process of the fluorescence emission of BBOB can be represented as follows: 19 Berberine, being a quaternary ammonium salt, can form ionassociation complexes with some anionic dyes such as Bromophenol Blue (BPB)20,21 and Bromocresol Green (BCG).22 Berberine seems to form ion-association complex with BBOB in a PVC membrane owing to the existence of a negative charge in the structure (II).Therefore, it can be assumed that the fluorescence quenching of BBOB by berberine is related to the formation of an ion-association complex, which inhibits the fluorescence emission of II.On the other hand, the positive charge in the structure II seems to provide the possibility of forming an ion associate with the carboxyl group of ethacrynate at pH 6.0, causing the reduction of the fluorescence emission of II. These processes can be represented as follows: mA+(aq.)"mA+(org.) + nB2(org.) + AmBn(org.) mE2(aq.)"mE2(org.) + nB+(org.) + EmBn(org.) The respective reactions are K1 mA+(aq.) + nB2(org.)"AmBn(org.) K2 mE2(aq.) + nB(org.)"EmBn(org.) where A, E and B represent berberine, ethacrynate and BBOB, respectively, and K1 and K2 are the equilibrium constants for reactions of A and E with B, respectively.One can describe the equilibrium involved as follows:16 an 1- a = 1 nK1CB n-1[A]m (1) an 1- a = 1 nK2CB n-1[E]m (2) where CB is the total concentration of BBOB in the membrane and [A] and [E] are the concentrations of A and E in aqueous solution. The relative fluorescence value a is the ratio of free BBOB concentration, [B], to the total concentration of BBOB present in the membrane, CB, i.e., a = [B] CB (3) and can be calculated according to the following eqn.: a = F - Fs Fb - Fs (4) where Fb is the fluorescence intensity of the optode membrane in the blank solution used for the determination of berberine or ethacrynate, Fs is the fluorescence intensity when BBOB in the membrane is completely complexed by berberine or ethacrynate Fig. 1 Graph of relative fluorescence intensity versus log (concentration) of each analyte (lex = 344 nm, lem = 390 nm) (a) M1 (PVC + BBOB + dibutyl phthalate) for ethacrynate; (b) M2 (PVC + BBOB + dibutyl phthalate) for berberine; (c) M3 (PVC + BBOB + didecyl phthalate) for picric acid; (d) M4 (PVC + BBOB + DOS) for picrolonic acid.Fig. 2 Membrane responses to ethacrynate and berberine in solutions of different pH. 1, M1 for 2.00 3 1023 mol l21 ethacrynate; 2, M2 for 4.40 3 1024 mol l21 berberine. Analyst, January 1997, Vol. 122 71and F is the fluorescence intensity of the optode membrane actually measured when contacting with solutions of different concentrations of berberine or ethacrynate.It is apparent from eqns. (1) and (2) that when the stoichiometric ratios of the complexes change, the relative fluorescence values (a) have different functional relationships with the concentrations of berberine or ethacrynate. Eqns. (1) and (2) provide the basis for the determination of berberine or ethacrynate. In addition to normal ion associates, there is also the possibility of other forms of interaction between BBOB and the fluorescence quenchers involved.BBOB is a nitrogen- and oxygen-containing heterocyclic compound and the electronegativity of the oxygen atom is stronger than that of nitrogen atom. When the quencher species of interest do not contain charged groups, such as with picric and picrolonic acids in H2SO4 solution, the interaction between BBOB and the fluorescence quenchers involved and the formation of associates can be accomplished through, say, hydrogen bonding with the phenolic group in the quencher molecules and the structure of II, causing an energy decrease of II and fluorescence quenching.On the supposition that BBOB in the plasticized PVC membrane phase (org.) with picric acid and picrolonic acid in the H2SO4 solutions (aq.) forms m–n associates, one can express the equilibrium as K3 mP(aq.) + nB2(org.)"PmBn(org.) K4 mS(aq.) + nB(org.)"SmBn(org.) where P, S and B represent picric acid, picrolonic acid and BBOB, respectively, and K3 and K4 are the equilibrium Fig. 3 Membrane responses to picric acid and picrolonic acid in H2SO4 media of different concentrations. 1, M3 for 1.40 3 1025 mol l21 picric acid; 2, M4 for 5.00 3 1025 mol l21 picrolonic acid. Fig. 4 Fitting the experimental data to eqns. (1), (2), (5) and (6). (a) Berberine and (d) picrolonic acid: 1, m: n = 2 : 1, K1(K4) = 1 3 108; 2, m: n = 2 : 2, K1(K4) = 3 3 1011; 3, m: n = 1 : 1, K1(K4) = 1000 (best fit); 4, m: n = 1 : 2, K1(K4) = 3 3 107.(b) Ethacrynate 1, m: n = 2 : 1, K2 = 1.3 3 104; 2, m: n = 2 : 2, K2 = 2.5 3106; 3, m: n = 1 : 1, K2 = 120 (best fit); 4, m: n = 1 : 2, K2 = 2.5 3104. (c) Picric acid 1, m: n = 2 : 1, K3 = 1 3109; 2, m: n = 2 : 2, K3 = 3 3 1012; 3, m: n = 1 : 1, K3 = 30 000 (best fit); 4, m: n = 1 : 2, K3 = 9 3 107. 72 Analyst, January 1997, Vol. 122constants for reactions of P and S with B, respectively.Similarly to the above, one has an 1- a = 1 nK3CB n-1[P]m (5) an 1- a = 1 nK4CB n-1[S]m (6) where [P] and [S] are the concentrations of P and S in aqueous acidic sample solutions. The relative fluorescence value a can be calculated as mentioned above. Obviously, eqns. (5) and (6) provide the basis for the determination of picric acid and picrolonic acid, respectively. Characteristics of the Optical Chemical Sensor Fluorescence quenching When the BBOB optode membranes corresponding to the quenchers of interest were exposed to solutions containing different concentrations of the four quenchers, the fluorescence intensities were recorded at lex = 344 nm and lem = 390 nm.The fluorescence intensities of the optode membranes decreased as the concentrations of the quenchers increased (Fig. 1). This illustrates that the optode membranes can be used for the assay of quenchers involved in sample solutions. Influence of acidity Although there are nitrogen and oxygen atoms in the BBOB molecule, the blank fluorescence of the sensing membrane itself was found to be independent of the acidity of the medium used in the experiment.Fig. 2 shows the dependence of the a values of the optode membranes contacted with a series of 2.00 31023 mol l21 ethacrynate and 4.40 3 1024 mol l21 berberine solutions at different pH values. It is clear that the a values for ethacrynate at pH > 4 (at pH < 4, ethacrynate would precipitate) and for berberine at pH 8–11 were nearly constant. In this work, pH 6.0 HOAc–NaOAc and pH 9.0 H3BO3–NaOH buffer solutions were used for the determination of ethacrynate and berberine, respectively.Fig. 3 shows the influence of H2SO4 concentration on the a values of the membranes in contact with 1.40 3 1025 mol l21 picric acid and 5.00 3 1025 mol l21 picrolonic acid solutions. With increase in H2SO4 concentration, the ionization of picric acid and picrolonic acid decreases and consequently the quenchers in question exist mainly in molecular form.Therefore, the ability to form hydrogen bonds increases and the ratio of the concentration of free BBOB to the total concentration of BBOB, a, decreases. In other words, the ability for quenching increased. This can also be confirmed by our experiments on the influence of increasing pH from 1 to 12, which decreases the ability of fluorescence quenching of the optode membranes, that is, strongly acidic media are the best for the determination of picric and picrolonic acid.However, we also discovered that when the concentration of H2SO4 is higher than 2.50 mol l21, BBOB is slowly dissolved out from the membrane. It can be seen from Fig. 3 that the optode membranes have more sensitive and more stable responses to picric acid and picrolonic acid in 1.0–2.0 and 0.90–2.3 mol l21 H2SO4, respectively. In subsequent experiments, 1.5 mol l21 H2SO4 was used for the determination of picric acid and picrolonic acid.Complex composition and measuring range As mentioned above, when the stoichiometric ratios of the complexes change, the a values have different functional relationships with the concentrations of the quenchers in question. The experimental data were fitted to eqns. (1), (2), (5) and (6) by changing the ratio of m to n and adjusting the over-all equilibrium constant K. Fig. 4(a), (b), (c) and (d) show the best fitted curves to represent the experimental data for berberine, ethacrynate, picric acid and picrolonic acid, respectively. It is interesting that the complex ratios of the quenchers in question to BBOB in the optode membrane are all 1 : 1.The best fitted curves can serve as calibration curves for the determinations of the quenchers of interest. Log(concentration) can be converted to concentration from Fig. 4; satisfactory responses are obtained from 8.00 3 1022 to 1.00 3 1024 mol l21 for ethacrynate, from 6.7631024 to 1.62 31025 mol l21 for berberine and picrolonic acid and from 2.88 3 1024 to 4.36 3 1026 mol l21 for picric acid.Reproducibility and reversibility The reproducibility and reversibility of the optode membranes corresponding to the quenchers of interest were studied using repeated determinations (n = 12) for each sample solution. After each measurement, the membrane was rinsed with blank solution until the fluorescence intensity (Fb) was recovered. Table 2 gives the experimental results.The mean fluorescence intensities ± standard deviations are 24.9 ± 0.6, 35.9 ± 0.6, 22.8 ± 0.6, 24.5 ± 0.7 for ethacrynate, berberine, picric acid and picrolonic acid, respectively. These results indicate that the optode membranes have good reproducibility and reversibility. Response time The optode membranes M1, M2, M3 and M4 were immersed first in a low concentration and then a high concentration sample Table 2 Reproducibility and reversibility of optode membranes corresponding to different quenchers Quencher Membrane concentration/ Fluorescence Membrane composition Quencher mol l21 intensity Mean s M1 PVC+ Ethacrynate 7.50 3 1023 25.8, 25.6, 25.4, 25.3, 24.9 0.6 BBOB+ 25.0, 24.9, 24.9, 24.7, dibutyl phthalate 24.6, 24.3, 24.0, 24.0 M2 PVC+ Berberine 1.32 3 1025 36.6, 36.5, 36.4, 36.4, 35.9 0.6 BBOB+ 36.3, 36.0, 36.0, 35.8, dibutyl phthalate 35.5, 35.3, 35.1, 34.9 M3 PVC+ Picric acid 8.22 3 1026 23.8, 23.5, 23.3, 23.0, 22.8 0.6 BBOB+ 23.0, 22.9, 22.8, 22.7, didecyl phthalate 22.4, 22.0, 22.3, 21.9 M4 PVC+ Picrolonic acid 3.06 3 1025 25.5, 25.5, 25.3, 25.0, 24.5 0.7 BBOB+ 24.8, 24.5, 24.3, 24.3, DOS 24.0, 23.8, 23.5, 23.4 Analyst, January 1997, Vol. 122 73solution. All response times to reach equilibrium between the sensing membranes and the corresponding quenchers or to reach the steady-state fluorescence signal were found to be less than 30 s. Short-term stability In order to monitor the dissolving out of BBOB from the membrane into the sample solution, the fluorescence signal at 390 nm for the optode membranes in contact with the quenchers in question were recorded over a period of 10 h with an interval of 30 min (n = 21).Fig. 5 shows a graph of fluorescence intensities versus time for each optode. It illustrates that BBOB is not significantly permeable from the sensing membrane during this period of time. Different approaches for the preservation of the optode membranes such as in redistilled water, in air and in redistilled water with the dissolved oxygen removed by bubbling nitrogen were tried.The fluorescence intensities of the optode membrane in air and in redistilled water decrease slightly with time, whereas those in redistilled water with the dissolved oxygen removed remain nearly constant. Apparently, the decrease in fluorescence intensities is due to oxygen, so the membrane should be preserved in redistilled water saturated with nitrogen when not in use.Selectivity Any substance that partitions into the membranes and quenches the fluorescence of the membranes or reacts with the analytes will interfere with the measurement. No significant interferences were observed if a less than a ±5% relative error was tolerated for the determination of 5.00 3 1023 mol l21 ethacrynate, 4.40 3 1024 mol l21 berberine, 5.50 3 1025 mol l21 picric acid and 4.90 3 1025 mol l21 picrolonic acid. These results are summarized in Table 3.Obviously, the optode membranes based on BBOB can be used for the selective determination of the quenchers in question. Applications The practical applications of the present sensors were tested on the direct and indirect determination of drugs. The direct methods involve the assay of ethacrynic acid and berberine hydrochloride in commercial tablets. The sample solutions for ethacrynic acid and berberine hydrochloride were diluted with buffer solutions of pH 6.0 and 9.0, respectively, and analysed using the described optode membranes M1 and M2, respec- Fig. 5 Change in fluorescence intensity with time. (a) 1, M1 for 3.5 3 1023 mol l21 ethacrynate; 2, M3 for 1.37 3 1025 mol l21 picric acid. (b) 1, M2 for 3.85 3 1025 mol l21 berberine; 2, M4 for 5.0 3 1025 mol l21 picrolonic acid. Table 3 Maximum tolerable concentrations of interferents for the assay of different analytes. The concentration of the four analytes of interest are 5.0 31023, 4.4 3 1024, 5.5 3 1025 and 4.9 3 1025 mol l21 for ethacrynate, berberine, picric acid and picrolonic acid, respectively.Interferent Tolerable concentration/mol l21 Type Compound M1 M2 M3 M4 Alkali metal and alkaline earth metal salts NaCl 0.10 0.010 0.050 0.050 NaOAc 0.10 0.50 0.050 0.050 NaNO3 0.10 0.010 0.010 0.050 MgCl2 0.050 0.010 0.010 0.010 CaCl2 1.2 3 1023 1.2 3 1024 1.2 3 1024 1.2 3 1024 Organic acids and their salts Sodium citrate 0.05 0.10 0.080 0.10 Malic acid 7.0 3 1023 4.2 3 1023 7.0 3 1023 7.0 3 1023 Sodium oxalate 2.8 3 1022 2.5 3 1022 2.5 3 1022 2.8 3 1022 Sodium tartrate 5.0 3 1022 0.10 5.0 3 1022 5.0 3 1022 Phenylformic acid 1.5 3 1022 1.5 3 1022 1.5 3 1022 1.5 3 1022 Phenols Phenol 3.3 3 1022 6.5 3 1022 6.5 3 1022 6.5 3 1022 Vitamins VB1 5.0 3 1023 5.0 3 1023 5.0 3 1023 5.0 3 1023 Inositol 5.0 3 1023 1.0 3 1023 1.0 3 1023 1.0 3 1023 Piperazines Levamisole 5.0 3 1023 1.2 3 1023 1.2 3 1023 1.2 3 1023 Xanthine bases Theophylline 5.0 3 1023 1.9 3 1022 1.9 3 1023 5.0 3 1023 Table 4 Results for the determination of four medicaments using the optode membranes and the pharmacopoeial and other methods Recovery obtained Recovery obtained by pharmacopoeial Sample with the optode or other Medicament no.membrane* (%) method* (%) Ethacrynate 1 96.6 ± 0.8 95.8 ± 0.5 2 94.6 ± 1.0 94.6 ± 0.8 3 98.0 ± 0.5 97.9 ± 0.6 Berberine 1 73.5 ± 1.0 73.5 ± 0.8 2 74.2 ± 1.5 74.3 ± 2.0 3 70.8 ± 0.5 70.8 ± 1.3 Cinchonine 1 98.5 ± 0.5 98.5 ± 1.0 2 97.4 ± 1.0 97.0 ± 0.8 3 96.5 ± 0.8 96.8 ± 1.2 VB1 1 93.4 ± 2.0 93.0 ± 1.5 2 90.5 ± 1.2 90.0 ± 1.0 3 95.0 ± 0.8 95.6 ± 0.5 * Mean values ± s of three determinations. 74 Analyst, January 1997, Vol. 122tively. The measured results are in correspondence with the results obtained by pharmacopoeial methods.23,24 (Table 4). The indirect methods for the determination of cinchonine and VB1 are based on quantitative precipitation with picric acid and picrolonic acid, respectively. The filtrates containing the excess of picric acid and picrolonic acid were diluted with an appropriate amount of 1.5 mol l21 H2SO4 and determined using the described optode membranes M3 and M4, respectively. The moles of quencher reacted are equivalent to the moles of the analyte present. These results are in agreement with the results obtained by other methods25,26 (Table 4). This work was supported by the National Natural Science Foundation and the Foundation for PhD Thesis Research of the National Education Commission of China. References 1 Wang, K.M., Seilr, K., Rusterholz, B., and Simon, W., Analyst, 1992, 117, 57. 2 Blair, T. L., Cynkowski, T., and Bachas, L. G., Anal. Chem., 1993, 65, 945. 3 Reiser, A., Leeyshon, L. J., Saunders, D., Mijovic, M. V., Bright, A., and Bogie, A., J. Am. Chem. Soc., 1992, 94, 2414. 4 Tanaka, T., Jpn. Pat., 1966, 3515; Chem. Abstr., 1967, 65, 7328f. 5 Liphardt, B., Liphardt, B., and Luettke, W., Chem. Ber., 1982, 115, 2997. 6 Arient, J., Collect. Czech. Chem. Commun., 1980, 45, 3160. 7 Zhou, Y.-M., Wu, Y.-M., and Gao, Z.-H., Chem. J. Chin. Univ., 1989, 10, 724. 8 Qin, Y.-X., Wang, M.-Z., and Gao, Z.-H., Chem. J. Chin. Univ., 1990, 11, 22. 9 Chen, D.-H., Luque de Castro, M. D., and Valcarcel, M., Anal. Chim. Acta, 1992, 261, 269. 10 Narayanaswamy, R., and Sevilla, F., III, Anal. Lett., 1988, 21, 1165. 11 Meadows, D., and Schultz, J. S., Talanta, 1988, 35, 1450. 12 He, H.-R., Uray, G., and Wolfbeis, O. S., Anal. Lett., 1992, 25, 405. 13 Luo, S.-F., and Walt, D. R., Anal. Chem., 1989, 61, 1069. 14 He, H.-R., Uray, G., and Wolfbeis, O. S., Anal. Chim. Acta, 1991, 246, 251. 15 Wang, K.-M., Huo, X.-Q., Chan, Y.-K., and Li, H.-M., J. Hunan Univ., 1995, 22, 55. 16 Zheng, H.-H., Wang, K.-M., Liu, C.-L., and Yu, R.-Q., Talanta, 1993, 40, 1569. 17 Wang, K.-M., Ou, Y.-W., and Bian, K.-J., Chem. J. Chin. Univ., 1992, 13, 1529. 18 Zhao, Y.-M., Zhang, D.-Y., and Gao, Z.-H., Chem. J. Chin. Univ., 1984, 5, 488. 19 Nyilas, E., and Pinter, J. L., J. Am. Chem. Soc., 1960, 82, 609. 20 Auerbach, E., Ind. Eng. Chem., Anal. Ed., 1943, 51, 492. 21 Kaneda, Y., and Iwaida, M., Eisei Kagaku, 1976, 22, 370. 22 Irving, H. M. N. H., and Markham, J. J., Anal. Chim. Acta, 1967, 39, 7. 23 Pharmacopoeia of the People’s Republic of China, Part III, Chemical Engineering Press, Beijing, 1990, p. 235. 24 Pharmacopoeia of the People’s Republic of China, Part III, Chemical Engineering Press, Beijing, 1990, p. 437. 25 Pharmacopoeia of the People’s Republic of China, Part III, Chemical Engineering Press, Beijing, 1990, p. 634. 26 Shen, K.-Y., Wang, M.-X., and Hau, P.-Y., in Handbook of Practical Analysis of Pharmaceuticals, People’s Medical Press, Beijing, 1986, p. 355. Paper 6/05031J Received July 18, 1996 Accepted September 25, 1996 Analyst, January 1997, Vol. 122 75

ISSN:0003-2654    

DOI:10.1039/a605031j

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Optical Biosensing of Nitrate Ions Using a Sol–Gel Immobilized Nitrate Reductase Jonathan W. Aylotta, David J. Richardsonb and David A. Russell*a a School of Chemical Sciences, University of East Anglia, Norwich, UK NR4 7TJ b School of Biological Sciences, University of East Anglia, Norwich, UK NR4 7TJ. E-mail: d.russell@uea.ac.uk The coupling of enzymes with sol–gel technology creates exciting possibilities for biosensing. Enzymes can be highly selective and will only respond to specific analytes.Sol–gels provide a unique matrix in which various biomaterials can be immobilized without any loss of enzyme activity. These two components have been combined for the optical biosensing of nitrate ions. The periplasmic nitrate reductase (Nap) extracted from the denitrifying bacterium Thiosphaera pantotropha reacts specifically with the nitrate (NO32) anion. The encapsulation of this enzyme in a sol–gel structure for the optical biosensing of nitrate ions is reported.The reduction of nitrate by periplasmic nitrate reductase results in a characteristic change in the UV/VIS absorption spectrum of the nitrate reductase. This spectroscopic change has been quantitatively calibrated against nitrate concentration. The nitrate biosensing system is fully reversible and is highly sensitive and selective to nitrate ions. The results obtained show that the activity of the enzyme is not affected by the sol–gel matrix, even after a storage period of up to six months.As no leaching of the Nap from the sol–gel matrix was observed, it is clear that the encapsulation of this nitrate sensitive enzyme in a sol–gel medium represents an ideal anionic recognition element of an optical biosensor for the detection of nitrate ions in the mmol l21 range. Keywords: Sol–gel; enzyme; nitrate reductase; optical biosensing; nitrate The nitrate anion is an important environmental and human health analyte and thus its detection and quantification is considered essential.As nitrate fertilisers are increasingly used for arable farming the quantity of nitrate leaching from the fields into rivers and reservoirs is increasing cumulatively. Nitrate concentrations in water fluctuate, depending on the season, weather conditions and locality. Consequently, the levels of nitrate build up in freshwater and marine aquatic ecosystems which can lead to algal blooms and eutrophication.1 Largely on the grounds of human health the European Union (EU) has imposed a limit for nitrate in domestic water supplies of 0.8 mmol l21 (50 mg l21).2 Another area in which quantitative analysis of nitrate is required is concerned with human health.Nitrate is not toxic to humans, however, once nitrate has entered the body it can be reduced by bacteria in the stomach to nitrite and further converted to N-nitrosamine compounds which are suspected to be carcinogenic.3 In addition to possible links with cancer, there is concern over the levels of nitrate in drinking water because of its relationship with the disease methaemoglobinaema, commonly known as ‘blue-baby syndrome’.4 More recently, the concentration of nitrate in human serum has become a significant analytical parameter due to the discovery that NO is a major messenger molecule with many physiological functions.Nitrogen monoxide is difficult to measure in biological fluids but it is readily oxidized to nitrite and further to nitrate, hence the nitrate levels in human serum are directly related to NO concentration.5 It has been found that the concentration of nitrate in serum ( Å 20–120 mmol l21) can be used as a noninvasive marker to predict whether a cardiac transplant patient will accept or reject the transplanted organ.6 At present nitrate concentrations are determined using a standard method7 whereby nitrate is reduced to nitrite with a copper activated cadmium catalyst and the nitrite concentration is determined colorimetrically by the quantitative formation of a purple–red azo dye.In addition to this standard technique, ion chromatography can be used to determine nitrate concentrations. 8 However the standard method suffers from poor sensitivity and the ion chromatography technique cannot resolve mmol l21 concentrations of nitrate when large concentrations of chloride or phosphate are present. An optical sensor for the detection of nitrate would be advantageous because little or no sample preparation would be necessary, the response time of the sensor should be fast and the device could be portable for in situ measurements.In order to develop a sensor, a molecular recognition species is required that is specific to the analyte of interest, i.e., nitrate. A naturally occurring class of compounds, that are known to be highly specific towards a particular species, are enzymes. The periplasmic nitrate reductase (Nap) enzyme, which can be isolated from the bacterium Thiosphaera pantotropha, reacts specifically with nitrate, reducing it to nitrite.9 This enzyme is ideal for use as the ionic recognition component of an optical sensor as there is a consequent change in the visible absorption spectrum of the Nap when the enzyme reduces nitrate to nitrite.There have been a number of reports in the literature of electrochemical10,11 and optical12,13 chemical sensors as well as an electrochemically based biosensor14 for the monitoring of nitrate ions in a variety of media.While the chemical sensors have the sensitivity required to detect nitrate in water supplies it is evident that the lower limits of detection required to quantify nitrate concentrations in blood could only be achieved by the biosensor. The nitrate biosensor was based on the coupling of a nitrate reductase (from Escherichia coli) with an electrode via an electron transfer mediator.14 However, the nitrate reductase used to produce the bioelectrode was a membrane-bound enzyme which is not specific to nitrate as it will react with a variety of other oxyanions.15–17 In addition, this cytoplasmic enzyme could not be utilized for optical sensing purposes as it does not possess the necessary chromophoric groups.To develop an optical biosensor a means of immobilization of the molecular recognition species is required. Biological molecules and especially enzymes are prone to inactivity if any structural alteration occurs during the immobilization process and tend to be sensitive to changes of temperature and pH.The sol–gel process18 is used to produce microporous silica matrices in which an enzyme can be encapsulated. The formation of a sol–gel is a low temperature technique which can be modified to work at mild pH conditions, yielding optically transparent porous glasses. The key advantage of sol–gels is that there is Analyst, January 1997, Vol. 122 (77–80) 77little or no structural alteration of the encapsulated species.Sol– gels encapsulate the biomaterial in a growing covalent network and hold it in a solid structure whilst allowing the immobilized enzyme to behave as if it were in solution. In this work periplasmic nitrate reductase was immobilized in a sol–gel using a method developed from the procedure first reported by Ellerby et al.19 and subsequently adapted by Blyth et al.20 This is a two step method in which a buffered protein solution (pH 6) was added to a hydrolysed sol mixture (pH 2).The rise in pH after the two were added initiated the condensation reaction and the gel formed after a few minutes. The sol–gel was then left to age and subsequently dry, producing a novel solid state glassy biomaterial. Modification of these previously reported methods was required due to the large size of the Nap enzyme (109 kDa) in comparison with the previously immobilized proteins which ranged from 12 to 64 kDa in size. Once the successful encapsulation of the enzyme had been accomplished the biosensing capability of the system towards nitrate was established.In this paper we report on the formation of a nitrate anion sensitive sol–gel glass and consider factors such as the working range for the enzyme response to nitrate, selectivity and sensitivity in relation to potential biosensor development. Experimental Reagents and Instrumentation All chemicals were of analytical-reagent grade (Aldrich, Gillingham, Kent, UK) and were used as received.Aqueous solutions were prepared in doubly distilled, de-ionized water. UV/VIS absorption spectra were recorded on a Hewlett- Packard (Stockport, Cheshire, UK) 8452A diode array spectrometer. Purification of Periplasmic Nitrate Reductase (Nap) Nap was isolated using the procedures described by Berks et al.15 from the genetically altered mutant (strain M-6) of Thiosphaera pantotropha. The cells were grown in anaerobic batch culture (180 l) at 34 °C in the mineral salts medium of Burnell et al.21 containing 50 mmol l21 sodium chlorate, 100 mmol l21 kanamycin, with 30 mmol l21 potassium acetate as electron donor and carbon source and 100 mmol l21 potassium nitrate as electron acceptor.The cells were harvested at the early stationary phase (A650 approximately 0.75) by cross-flow ultrafiltration (Sartorius, Epsom, Surrey, UK). The concentrate ( Å 6 l) was then centrifuged at 5600 rpm for 30 min to produce a pellet.The pellet was then re-suspended in spheroplastic buffer (3 l, 0.5 mmol l21 sucrose, 3 mmol l21 EDTA, 100 mmol l21 TRIS-HCl of pH 8). The periplasm was extracted by the addition of lysozyme (3.5 g) to the resuspended cells. The cells were lysed at 35 °C under gentle stirring for 1 h. The suspension was then centrifuged at 12 000 rpm for 20 min to yield a pellet of spheroplasts and a supernatant of periplasm. The following purification steps were carried out at 4 °C.The periplasmic extract from the 180 l cell culture was loaded onto a DEAE-Sepharose CL-6B column (400 3 40 mm) previously equilibrated with 100 mmol l21 TRIS-HCl of pH 8. The column was washed with two column volumes of the equilibrium buffer and developed with a gradient of 0-400 mmol l21 NaCl in 800 ml of equilibrium buffer. Nap eluted at Å 150 mmol l21 NaCl and was identified by UV/VIS spectroscopy. Further purification was achieved by elution of the Nap through the anion exchange column twice so that the sample of Nap had a 410/280 nm purity ratio greater than 0.5.The sample was then concentrated by ultrafiltration (Amicon, Gloucestershire, UK; 50 ml, 3 kDa membrane) and dialysed with 25 mmol l21 phosphate buffer, pH 7, in preparation for sol–gel immobilization. Preparation of Enzyme Immobilized Sol–Gels The concentrated, purified periplasmic nitrate reductase was immobilized in a sol–gel using a two-step method.Firstly, an acid catalysed silica sol was prepared by sonication in an ice bath of tetramethylorthosilicate (TMOS, 0.92 ml), de-ionized water (0.54 ml) and HCl (0.05 mol l21, 12.5 ml) for 30 min. The sol–gel was made by mixing the silica sol precursor mixture with the concentrated, buffered Nap solution in a 50 : 50 ratio. The sol–gel solution (0.2-0.4 ml depending on the thickness of the gel required) was transferred to the side of a horizontally orientated cuvette which had parafilm barriers arranged so that a gel of dimensions 15 3 10 mm was produced on the optical face of the cuvette.Typically the thickness of the Nap encapsulated sol–gel was approximately 1 mm. The sol–gel solution took Å10 min to gel and once it had set was surrounded by phosphate buffer (50 mmol l21, pH 7.6) and allowed to age at 4 °C for a week with a change of buffer after the first and fourth days. While the sol–gels formed using this process can withstand drying, the Nap sol–gels were stored in phosphate buffer at 4 °C.Quantitative Analysis of Nitrate Using Nap in Both Solution and Encapsulated in a Sol–Gel The quantitative determination of nitrate involved two steps: (i) the stoichiometric reduction of the Nap enzyme; (ii) the addition of known concentrations of nitrate and monitoring the spectroscopic response of the enzyme in the visible region. Both the solution of periplasmic nitrate reductase and the buffer solution surrounding the sol–gel immobilized Nap were de-oxygenated by bubbling with argon to remove any oxidants from each solution.The enzyme was then reduced by the addition of ml quantities of sodium dithionite solution. Sodium dithionite has a characteristic absorption band at 318 nm when present in excess. This band falls with time when this electron donor is consumed. When the spectrum shows that the 318 nm band of the dithionite has reached the baseline level and also indicates the Nap to be fully reduced (as shown in Fig. 1) it can be assumed that there are no excess reductants in the system. This technique for reducing Nap was used when the enzyme was in solution and sol–gel encapsulated. Potassium nitrate (0.5 mmol l21) solution, which had previously been de-oxygenated by bubbling with argon, was introduced to the reduced enzyme system by successive Fig. 1 Oxidized (dashed line) and reduced (solid line) spectra of the nitrate reductase (Nap) enzyme immobilized in a sol–gel matrix. 78 Analyst, January 1997, Vol. 122additions of 1ml (equivalent to 0.5 nmol) aliquots. The change in the spectrum of Nap was measured after 2 min and 15 min for solution and sol–gel, respectively. The effect of interferent ions (e.g. chlorate, sulfate, carbonate, phosphate and chloride) on the spectroscopic characteristics of the encapsulated Nap were tested by injecting, by microsyringe, a large excess (equivalent to 10 000 3 concentration of nitrate analyte) of the respective de-oxygenated aqueous solutions to the sealed system; i.e.a solution of each interferent at a concentration of Å 500 mmol l21 was prepared and 5 mmol aliquots were injected as compared to the 0.5 nmol of nitrate injected. The possibility of the Nap enzyme leaching from the sol–gel was investigated by storing a sample of the sol–gel encapsulated Nap surrounded by a buffer solution and periodically measuring the absorption spectrum of the buffer solution. Similarly, to assess the activity of the enzyme over time a sol–gel encapsulated Nap sample was stored at 4 °C and the ability to reduce and interact the immobilized enzyme with nitrate ions was measured after a period of six months.Results and Discussion Chemistry and Spectroscopy of Nap Nap was originally identified in the wildtype strain of Thiosphaera pantotropha in 19909 but was only present in small quantities. However, a genetically altered mutant of Thiosphaera pantotropha (strain M-6)22 has been engineered that over-expresses Nap when grown anaerobically.Periplasmic nitrate reductase is light brown in colour and is a heterodimer. The enzyme consists of two sub-units: a 93 kDa sub-unit, the site of nitrate reduction, and a 16 kDa c-type cytochrome containing a di-haem centre. The nitrate binds to a molybdopterin cofactor situated in the 93 kDa sub-unit and is reduced according to the following equation: NO32 + 2H+ + 2e2 ? NO22 + H2O The reduction of nitrate is initiated by the transfer of electrons from the reduced haem centres contained within the c-type cytochrome sub-unit. The reduction of nitrate by Nap is a fast reaction of the order of 50 ms.17 While the exact mechanism of electron transfer has not yet been elucidated it is thought that an iron-sulfur cluster [(4Fe-4S)2+,1+] mediates the electron transfer between the cytochromes and the molybdopterin cofactor.17 To develop a nitrate sensor based on Nap there is a requirement for the enzyme to be reduced by a stoichiometric amount of reductant (sodium dithionite) ensuring that there are no excess electrons in the system.When reduced Nap binds nitrate, the spectral change that occurs is derived from the changing oxidation state of the haem groups situated in the cytochrome c-552 sub-unit. Upon reduction of the enzyme both the haem centres exist in the ferrous form. When nitrate is introduced to the system it is bound to the molybdopterin cofactor and electrons pass from the haem groups via the iron sulfur cluster to reduce the nitrate to nitrite.The movement of electrons from the haem centres to the site of nitrate reduction causes the ferrous iron to be oxidized to the ferric form, and because there are no excess reductants available to re-reduce the haem centres there is an associated spectral change. Preparation of Nap Immobilized Sol–Gels Sol-gel entrapment of proteins and enzymes has previously been shown to have no deleterious effect on the activity or reactivity of the encapsulated species.19,20,23 A similar result was obtained for the sol–gel encapsulation of periplasmic nitrate reductase, even though the size of Nap (109 kDa) is much larger than the species previously immobilized using sol– gel techniques.The large size of the enzyme prevented the preparation of sol–gels using previously reported procedures because there was precipitation of the immobilized enzyme leading to the sol–gels becoming highly light scattering.In order to overcome this precipitation the water content of the hydrolysed sol mixture was increased. This had the effect of initially increasing the size of pores in the sol–gel so that the enzyme could be encapsulated, producing a light brown coloured, optically transparent sol–gel. Although the increase in water content of the sol–gel mixture enlarged the pores of the sol–gel, this did not result in any subsequent leaching of the Nap enzyme.Once the Nap enzyme had been encapsulated in a sol– gel it retained its activity and was able to reduce nitrate to nitrite even after a storage period of six months. Quantitative Binding of Nitrate to Nap Immobilized in a Sol–Gel Matrix Fig. 1 shows the oxidized and reduced spectra of periplasmic nitrate reductase immobilized in a sol–gel structure. These spectra are in good agreement with the spectra for reduced and oxidized Nap in solution15 and demonstrate the successful encapsulation of the enzyme allied to its continued reactivity upon entrapment.The reaction of nitrate with Nap is reversible so by re-reducing the Nap immobilized within the sol–gel the sensor is re-usable. Spectra identical to those obtained for the reaction of nitrate with Nap in solution were obtained from the successive additions of 1 ml (0.5 nmol) aliquots of nitrate to the 2 ml phosphate buffer solution surrounding the sol–gel immobilized Nap and confirm that the activity of the enzyme has been retained.While the Soret band at approximately 418 nm significantly blue-shifts when the Nap is oxidized, a linear relationship between the spectral change and increasing nitrate concentration was not observed at this wavelength. However, the change in absorbance of the absorption band centred at 550 nm with increasing nitrate concentration (Fig. 2) shows a linear relationship from which the calibration curve shown in Fig. 3 was constructed. With the current manufacture process of the Nap encapsulated sol–gels a degree of variation in gel thickness is apparent. This requires the construction of separate calibration curves for each sensor prepared. However, the slope of each calibration curve, i.e., the sensitivity of the Nap sensing systems, is comparable. The nitrate sensitive glass gives a linear response to nitrate over the range 0–1.5 mmol l21 with a detection limit of 0.125 mmol l21. Fig. 2 Decrease of absorption band intensity at 550 nm of the encapsulated reduced Nap with increasing nitrate concentration.Analyst, January 1997, Vol. 122 79Selectivity of the Sol–Gel Immobilized Nap It has previously been reported that the Nap enzyme shows very high substrate specificity and that it does not reduce a variety of oxo-compounds including chlorate, nitrite and sulfate.15 In addition to these species we have found that Nap does not react with phosphate, carbonate, hydrogen carbonate (HCO32) or chloride anions and that metal ions have no effect on the reactivity of the enzyme.These species do not react with the enzyme whether it is in solution or entrapped in a sol–gel. The system reported in this paper has the ability to detect mmol l21 quantities of nitrate in matrices with high oxyanion concentrations. The system’s lack of interferents is significant for the analysis of nitrate in complex media such as serum and seawater. The enzyme is active over a broad pH range (pH values 6.5–10)15 although measurements should be carried out at a constant pH value within the pH range 7–9 to obtain the optimum results.Conclusions The results presented in this paper demonstrate that the enzyme periplasmic nitrate reductase is an excellent anionic recognition centre for the biosensing of nitrate and that optically transparent sol–gel glasses are ideal immobilization matrices. The enzyme has been encapsulated within a sol–gel with no structural alteration and full retention of activity.Nap is highly specific towards nitrate and does not react with any commonly encountered oxyanions. In addition to oxyanions the enzyme does not respond to other species such as metal ions and chloride. The sensitivity of the biosensing system is exceptional with ability to detect mmol l21 concentrations of nitrate. Additionally, the Nap has been shown not to leach from the sol– gel matrix and retains its activity, within the sol–gel, even after a six month storage period.It is clear therefore that the immobilization of periplasmic nitrate reductase within a sol–gel matrix offers excellent potential for the optical biosensing of nitrate. The authors acknowledge the Trustees of the Analytical Chemistry Trust Fund of the Royal Society of Chemistry for the award of an SAC Research Studentship to J.W.A. The authors would also like to thank Dr. P. Dobbin and A. Reilly for assistance with the purification of the periplasmic nitrate reductase and also R.J. Chapman for experimental assistance. References 1 Addiscott, T. M., Whitmore, A. P., and Powlson, D. S., Farming, Fertilizers and the Nitrate Problem, CAB International Oxford, 1991. 2 Briggs, D. J., The State of the Environment in the European Community 1986, Office for Official Publications of the European Communities, Luxembourg, 1987. 3 Odashima, S., Oncology, 1980, 37, 282. 4 Cornblath, M., and Hartmann, A.F., J. Pediatr., 1948, 33, 421. 5 Stichtenoth, D. O., Wollenhaupt, J., Andersone, D., Zeidler, H., and Frolich, J. C., Br. J. Rheumatol., 1995, 34, 616. 6 Benvenuti, C., Bories, P. N., and Loisance, D., Transplantation, 1996, 61, 745. 7 Greenberg, A. E., Trussel, R. R., and Clesceri, L. S., Standard Methods for the Examination of Water and Waste Water, American Public Health Association, Washington D.C., USA, 1985. 8 Dejong, P., and Burggraaf, M., Clin. Chem. Acta, 1983, 132, 63. 9 Bell, L. C., Richardson, D. J., and Ferguson, S. J., FEBS Lett., 1990, 265, 85. 10 Kang, S. C., Lee, K.-S., Kim, J.-D., and Kim, K.-J., Bull. Korean Chem. Soc., 1990, 11, 124. 11 Stanley, M. A., Maxwell, J., Forrestal, M., Doherty, A. P., MacCraith, B. D., Diamond, D., and Vos, J. G., Anal. Chim. Acta, 1994, 299, 81. 12 Lumpp, R., Reichert, J., and Ache, H. J., Sens. Actuators, B., 1992, 7, 473. 13 Mohr, G. J., and Wolfbeis, O. S., Anal. Chim. Acta, 1995, 316, 239. 14 Cosnler, S., Innocent, C., and Jouanneau, Y., Anal.Chem., 1994, 66, 3198. 15 Berks, B. C., Richardson, D. J., Robinson, C., Reilly, A., Alpin, R. T., and Ferguson, S. J., Eur. J. Biochem., 1994, 220, 117. 16 Bennett, B., Berks, B. C., Ferguson, S. J., Thomson, A. J., and Richardson, D. J., Eur. J. Biochem., 1994, 226, 789. 17 Berks, B. C., Ferguson, S. J., Moir, J. W. B., and Richardson, D. J., Biochim. Biophys. Acta, 1995, 1232, 97. 18 Brinker, C. J., and Scherer, G. W., Sol–Gel Science, Academic Press, San Diego, 1990. 19 Ellerby, L.M., Nishida, C. R., Nishida, F., Yamanka, S. A., Dunn, B., Valentine, J. S., and Zink, J. I., Science, 1992, 255, 1113. 20 Blyth, D. J., Aylott, J. W., Richardson, D. J., and Russell, D. A., Analyst, 1995, 120, 2725. 21 Burnell, J. N., John, P., and Whatley, F. R., Biochem. J., 1975, 150, 527. 22 Bell, L. C., Page, M. D., Berks, B. C., Richardson, D. J., and Ferguson, S. J., J. Gen. Microbiol., 1993, 139, 3205. 23 Dave, B.C., Dunn, B., Selverstone Valentine J., and Zink, J. I., Anal. Chem., 1994, 66, 1120A. Paper 6/06146J Received September 6, 1996 Accepted September 27, 1996 Fig. 3 Calibration curve for the decrease of 550 nm absorption band of the immobilized Nap enzyme with increasing nitrate concentration. 80 Analyst, January 1997, Vol. 122 Optical Biosensing of Nitrate Ions Using a Sol–Gel Immobilized Nitrate Reductase Jonathan W. Aylotta, David J. Richardsonb and David A.Russell*a a School of Chemical Sciences, University of East Anglia, Norwich, UK NR4 7TJ b School of Biological Sciences, University of East Anglia, Norwich, UK NR4 7TJ. E-mail: d.russell@uea.ac.uk The coupling of enzymes with sol–gel technology creates exciting possibilities for biosensing. Enzymes can be highly selective and will only respond to specific analytes. Sol–gels provide a unique matrix in which various biomaterials can be immobilized without any loss of enzyme activity. These two components have been combined for the optical biosensing of nitrate ions.The periplasmic nitrate reductase (Nap) extracted from the denitrifying bacterium Thiosphaera pantotropha reacts specifically with the nitrate (NO32) anion. The encapsulation of this enzyme in a sol–gel structure for the optical biosensing of nitrate ions is reported. The reduction of nitrate by periplasmic nitrate reductase results in a characteristic change in the UV/VIS absorption spectrum of the nitrate reductase.This spectroscopic change has been quantitatively calibrated against nitrate concentration. The nitrate biosensing system is fully reversible and is highly sensitive and selective to nitrate ions. The results obtained show that the activity of the enzyme is not affected by the sol–gel matrix, even after a storage period of up to six months. As no leaching of the Nap from the sol–gel matrix was observed, it is clear that the encapsulation of this nitrate sensitive enzyme in a sol–gel medium represents an ideal anionic recognition element of an optical biosensor for the detection of nitrate ions in the mmol l21 range.Keywords: Sol–gel; enzyme; nitrate reductase; optical biosensing; nitrate The nitrate anion is an important environmental and human health analyte and thus its detection and quantification is considered essential. As nitrate fertilisers are increasingly used for arable farming the quantity of nitrate leaching from the fields into rivers and reservoirs is increasing cumulatively.Nitrate concentrations in water fluctuate, depending on the season, weather conditions and locality. Consequently, the levels of nitrate build up in freshwater and marine aquatic ecosystems which can lead to algal blooms and eutrophication.1 Largely on the grounds of human health the European Union (EU) has imposed a limit for nitrate in domestic water supplies of 0.8 mmol l21 (50 mg l21).2 Another area in which quantitative analysis of nitrate is required is concerned with human health.Nitrate is not toxic to humans, however, once nitrate has entered the body it can be reduced by bacteria in the stomach to nitrite and further converted to N-nitrosamine compounds which are suspected to be carcinogenic.3 In addition to possible links with cancer, there is concern over the levels of nitrate in drinking water because of its relationship with the disease methaemoglobinaema, commonly known as ‘blue-baby syndrome’.4 More recently, the concentration of nitrate in human serum has become a significant analytical parameter due to the discovery that NO is a major messenger molecule with many physiological functions.Nitrogen monoxide is difficult to measure in biological fluids but it is readily oxidized to nitrite and further to nitrate, hence the nitrate levels in human serum are directly related to NO concentration.5 It has been found that the concentration of nitrate in serum ( Å 20–120 mmol l21) can be used as a noninvasive marker to predict whether a cardiac transplant patient will accept or reject the transplanted organ.6 At present nitrate concentrations are determined using a standard method7 whereby nitrate is reduced to nitrite with a copper activated cadmium catalyst and the nitrite concentration is determined colorimetrically by the quantitative formation of a purple–red azo dye.In addition to this standard technique, ion chromatography can be used to determine nitrate concentrations. 8 However the standard method suffers from poor sensitivity and the ion chromatography technique cannot resolve mmol l21 concentrations of nitrate when large concentrations of chloride or phosphate are present.An optical sensor for the detection of nitrate would be advantageous because little or no sample preparation would be necessary, the response time of the sensor should be fast and the device could be portable for in situ measurements.In order to develop a sensor, a molecular recognition species is required that is specific to the analyte of interest, i.e., nitrate. A naturally occurring class of compounds, that are known to be highly specific towards a particular species, are enzymes. The periplasmic nitrate reductase (Nap) enzyme, which can be isolated from the bacterium Thiosphaera pantotropha, reacts specifically with nitrate, reducing it to nitrite.9 This enzyme is ideal for use as the ionic recognition component of an optical sensor as there is a consequent change in the visible absorption spectrum of the Nap when the enzyme reduces nitrate to nitrite.There have been a number of reports in the literature of electrochemical10,11 and optical12,13 chemical sensors as well as an electrochemically based biosensor14 for the monitoring of nitrate ions in a variety of media. While the chemical sensors have the sensitivity required to detect nitrate in water supplies it is evident that the lower limits of detection required to quantify nitrate concentrations in blood could only be achieved by the biosensor.The nitrate biosensor was based on the coupling of a nitrate reductase (from Escherichia coli) with an electrode via an electron transfer mediator.14 However, the nitrate reductase used to produce the bioelectrode was a membrane-bound enzyme which is not specific to nitrate as it will react with a variety of other oxyanions.15–17 In addition, this cytoplasmic enzyme could not be utilized for optical sensing purposes as it does not possess the necessary chromophoric groups.To develop an optical biosensor a means of immobilization of the molecular recognition species is required. Biological molecules and especially enzymes are prone to inactivity if any structural alteration occurs during the immobilization process and tend to be sensitive to changes of temperature and pH. The sol–gel process18 is used to produce microporous silica matrices in which an enzyme can be encapsulated.The formation of a sol–gel is a low temperature technique which can be modified to work at mild pH conditions, yielding optically transparent porous glasses. The key advantage of sol–gels is that there is Analyst, January 1997, Vol. 122 (77–80) 77little or no structural alteration of the encapsulated species. Sol– gels encapsulate the biomaterial in a growing covalent network and hold it in a solid structure whilst allowing the immobilized enzyme to behave as if it were in solution.In this work periplasmic nitrate reductase was immobilized in a sol–gel using a method developed from the procedure first reported by Ellerby et al.19 and subsequently adapted by Blyth et al.20 This is a two step method in which a buffered protein solution (pH 6) was added to a hydrolysed sol mixture (pH 2). The rise in pH after the two were added initiated the condensation reaction and the gel formed after a few minutes.The sol–gel was then left to age and subsequently dry, producing a novel solid state glassy biomaterial. Modification of these previously reported methods was required due to the large size of the Nap enzyme (109 kDa) in comparison with the previously immobilized proteins which ranged from 12 to 64 kDa in size. Once the successful encapsulation of the enzyme had been accomplished the biosensing capability of the system towards nitrate was established.In this paper we report on the formation of a nitrate anion sensitive sol–gel glass and consider factors such as the working range for the enzyme response to nitrate, selectivity and sensitivity in relation to potential biosensor development. Experimental Reagents and Instrumentation All chemicals were of analytical-reagent grade (Aldrich, Gillingham, Kent, UK) and were used as received. Aqueous solutions were prepared in doubly distilled, de-ionized water.UV/VIS absorption spectra were recorded on a Hewlett- Packard (Stockport, Cheshire, UK) 8452A diode array spectrometer. Purification of Periplasmic Nitrate Reductase (Nap) Nap was isolated using the procedures described by Berks et al.15 from the genetically altered mutant (strain M-6) of Thiosphaera pantotropha. The cells were grown in anaerobic batch culture (180 l) at 34 °C in the mineral salts medium of Burnell et al.21 containing 50 mmol l21 sodium chlorate, 100 mmol l21 kanamycin, with 30 mmol l21 potassium acetate as electron donor and carbon source and 100 mmol l21 potassium nitrate as electron acceptor.The cells were harvested at the early stationary phase (A650 approximately 0.75) by cross-flow ultrafiltration (Sartorius, Epsom, Surrey, UK). The concentrate ( Å 6 l) was then centrifuged at 5600 rpm for 30 min to produce a pellet. The pellet was then re-suspended in spheroplastic buffer (3 l, 0.5 mmol l21 sucrose, 3 mmol l21 EDTA, 100 mmol l21 TRIS-HCl of pH 8).The periplasm was extracted by the addition of lysozyme (3.5 g) to the resuspended cells. The cells were lysed at 35 °C under gentle stirring for 1 h. The suspension was then centrifuged at 12 000 rpm for 20 min to yield a pellet of spheroplasts and a supernatant of periplasm. The following purification steps were carried out at 4 °C. The periplasmic extract from the 180 l cell culture was loaded onto a DEAE-Sepharose CL-6B column (400 3 40 mm) previously equilibrated with 100 mmol l21 TRIS-HCl of pH 8.The column was washed with two column volumes of the equilibrium buffer and developed with a gradient of 0-400 mmol l21 NaCl in 800 ml of equilibrium buffer. Nap eluted at Å 150 mmol l21 NaCl and was identified by UV/VIS spectroscopy. Further purification was achieved by elution of the Nap through the anion exchange column twice so that the sample of Nap had a 410/280 nm purity ratio greater than 0.5.The sample was then concentrated by ultrafiltration (Amicon, Gloucestershire, UK; 50 ml, 3 kDa membrane) and dialysed with 25 mmol l21 phosphate buffer, pH 7, in preparation for sol–gel immobilization. Preparation of Enzyme Immobilized Sol–Gels The concentrated, purified periplasmic nitrate reductase was immobilized in a sol–gel using a two-step method. Firstly, an acid catalysed silica sol was prepared by sonication in an ice bath of tetramethylorthosilicate (TMOS, 0.92 ml), de-ionized water (0.54 ml) and HCl (0.05 mol l21, 12.5 ml) for 30 min.The sol–gel was made by mixing the silica sol precursor mixture with the concentrated, buffered Nap solution in a 50 : 50 ratio. The sol–gel solution (0.2-0.4 ml depending on the thickness of the gel required) was transferred to the side of a horizontally orientated cuvette which had parafilm barriers arranged so that a gel of dimensions 15 3 10 mm was produced on the optical face of the cuvette. Typically the thickness of the Nap encapsulated sol–gel was approximately 1 mm.The sol–gel solution took Å10 min to gel and once it had set was surrounded by phosphate buffer (50 mmol l21, pH 7.6) and allowed to age at 4 °C for a week with a change of buffer after the first and fourth days. While the sol–gels formed using this process can withstand drying, the Nap sol–gels were stored in phosphate buffer at 4 °C. Quantitative Analysis of Nitrate Using Nap in Both Solution and Encapsulated in a Sol–Gel The quantitative determination of nitrate involved two steps: (i) the stoichiometric reduction of the Nap enzyme; (ii) the addition of known concentrations of nitrate and monitoring the spectroscopic response of the enzyme in the visible region. Both the solution of periplasmic nitrate reductase and the buffer solution surrounding the sol–gel immobilized Nap were de-oxygenated by bubbling with argon to remove any oxidants from each solution.The enzyme was then reduced by the addition of ml quantities of sodium dithionite solution.Sodium dithionite has a characteristic absorption band at 318 nm when present in excess. This band falls with time when this electron donor is consumed. When the spectrum shows that the 318 nm band of the dithionite has reached the baseline level and also indicates the Nap to be fully reduced (as shown in Fig. 1) it can be assumed that there are no excess reductants in the system. This technique for reducing Nap was used when the enzyme was in solution and sol–gel encapsulated.Potassium nitrate (0.5 mmol l21) solution, which had previously been de-oxygenated by bubbling with argon, was introduced to the reduced enzyme system by successive Fig. 1 Oxidized (dashed line) and reduced (solid line) spectra of the nitrate reductase (Nap) enzyme immobilized in a sol–gel matrix. 78 Analyst, January 1997, Vol. 122additions of 1ml (equivalent to 0.5 nmol) aliquots. The change in the spectrum of Nap was measured after 2 min and 15 min for solution and sol–gel, respectively.The effect of interferent ions (e.g. chlorate, sulfate, carbonate, phosphate and chloride) on the spectroscopic characteristics of the encapsulated Nap were tested by injecting, by microsyringe, a large excess (equivalent to 10 000 3 concentration of nitrate analyte) of the respective de-oxygenated aqueous solutions to the sealed system; i.e. a solution of each interferent at a concentration of Å 500 mmol l21 was prepared and 5 mmol aliquots were injected as compared to the 0.5 nmol of nitrate injected.The possibility of the Nap enzyme leaching from the sol–gel was investigated by storing a sample of the sol–gel encapsulated Nap surrounded by a buffer solution and periodically measuring the absorption spectrum of the buffer solution. Similarly, to assess the activity of the enzyme over time a sol–gel encapsulated Nap sample was stored at 4 °C and the ability to reduce and interact the immobilized enzyme with nitrate ions was measured after a period of six months.Results and Discussion Chemistry and Spectroscopy of Nap Nap was originally identified in the wildtype strain of Thiosphaera pantotropha in 19909 but was only present in small quantities. However, a genetically altered mutant of Thiosphaera pantotropha (strain M-6)22 has been engineered that over-expresses Nap when grown anaerobically. Periplasmic nitrate reductase is light brown in colour and is a heterodimer.The enzyme consists of two sub-units: a 93 kDa sub-unit, the site of nitrate reduction, and a 16 kDa c-type cytochrome containing a di-haem centre. The nitrate binds to a molybdopterin cofactor situated in the 93 kDa sub-unit and is reduced according to the following equation: NO32 + 2H+ + 2e2 ? NO22 + H2O The reduction of nitrate is initiated by the transfer of electrons from the reduced haem centres contained within the c-type cytochrome sub-unit. The reduction of nitrate by Nap is a fast reaction of the order of 50 ms.17 While the exact mechanism of electron transfer has not yet been elucidated it is thought that an iron-sulfur cluster [(4Fe-4S)2+,1+] mediates the electron transfer between the cytochromes and the molybdopterin cofactor.17 To develop a nitrate sensor based on Nap there is a requirement for the enzyme to be reduced by a stoichiometric amount of reductant (sodium dithionite) ensuring that there are no excess electrons in the system.When reduced Nap binds nitrate, the spectral change that occurs is derived from the changing oxidation state of the haem groups situated in the cytochrome c-552 sub-unit. Upon reduction of the enzyme both the haem centres exist in the ferrous form. When nitrate is introduced to the system it is bound to the molybdopterin cofactor and electrons pass from the haem groups via the iron sulfur cluster to reduce the nitrate to nitrite.The movement of electrons from the haem centres to the site of nitrate reduction causes the ferrous iron to be oxidized to the ferric form, and because there are no excess reductants available to re-reduce the haem centres there is an associated spectral change. Preparation of Nap Immobilized Sol–Gels Sol-gel entrapment of proteins and enzymes has previously been shown to have no deleterious effect on the activity or reactivity of the encapsulated species.19,20,23 A similar result was obtained for the sol–gel encapsulation of periplasmic nitrate reductase, even though the size of Nap (109 kDa) is much larger than the species previously immobilized using sol– gel techniques.The large size of the enzyme prevented the preparation of sol–gels using previously reported procedures because there was precipitation of the immobilized enzyme leading to the sol–gels becoming highly light scattering. In order to overcome this precipitation the water content of the hydrolysed sol mixture was increased. This had the effect of initially increasing the size of pores in the sol–gel so that the enzyme could be encapsulated, producing a light brown coloured, optically transparent sol–gel.Although the increase in water content of the sol–gel mixture enlarged the pores of the sol–gel, this did not result in any subsequent leaching of the Nap enzyme. Once the Nap enzyme had been encapsulated in a sol– gel it retained its activity and was able to reduce nitrate to nitrite even after a storage period of six months.Quantitative Binding of Nitrate to Nap Immobilized in a Sol–Gel Matrix Fig. 1 shows the oxidized and reduced spectra of periplasmic nitrate reductase immobilized in a sol–gel structure. These spectra are in good agreement with the spectra for reduced and oxidized Nap in solution15 and demonstrate the successful encapsulation of the enzyme allied to its continued reactivity upon entrapment. The reaction of nitrate with Nap is reversible so by re-reducing the Nap immobilized within the sol–gel the sensor is re-usable.Spectra identical to those obtained for the reaction of nitrate with Nap in solution were obtained from the successive additions of 1 ml (0.5 nmol) aliquots of nitrate to the 2 ml phosphate buffer solution surrounding the sol–gel immobilized Nap and confirm that the activity of the enzyme has been retained. While the Soret band at approximately 418 nm significantly blue-shifts when the Nap is oxidized, a linear relationship between the spectral change and increasing nitrate concentration was not observed at this wavelength. However, the change in absorbance of the absorption band centred at 550 nm with increasing nitrate concentration (Fig. 2) shows a linear relationship from which the calibration curve shown in Fig. 3 was constructed. With the current manufacture process of the Nap encapsulated sol–gels a degree of variation in gel thickness is apparent.This requires the construction of separate calibration curves for each sensor prepared. However, the slope of each calibration curve, i.e., the sensitivity of the Nap sensing systems, is comparable. The nitrate sensitive glass gives a linear response to nitrate over the range 0–1.5 mmol l21 with a detection limit of 0.125 mmol l21. Fig. 2 Decrease of absorption band intensity at 550 nm of the encapsulated reduced Nap with increasing nitrate concentration.Analyst, January 1997, Vol. 122 79Selectivity of the Sol–Gel Immobilized Nap It has previously been reported that the Nap enzyme shows very high substrate specificity and that it does not reduce a variety of oxo-compounds including chlorate, nitrite and sulfate.15 In addition to these species we have found that Nap does not react with phosphate, carbonate, hydrogen carbonate (HCO32) or chloride anions and that metal ions have no effect on the reactivity of the enzyme.These species do not react with the enzyme whether it is in solution or entrapped in a sol–gel. The system reported in this paper has the ability to detect mmol l21 quantities of nitrate in matrices with high oxyanion concentrations. The system’s lack of interferents is significant for the analysis of nitrate in complex media such as serum and seawater. The enzyme is active over a broad pH range (pH values 6.5–10)15 although measurements should be carried out at a constant pH value within the pH range 7–9 to obtain the optimum results.Conclusions The results presented in this paper demonstrate that the enzyme periplasmic nitrate reductase is an excellent anionic recognition centre for the biosensing of nitrate and that optically transparent sol–gel glasses are ideal immobilization matrices. The enzyme has been encapsulated within a sol–gel with no structural alteration and full retention of activity. Nap is highly specific towards nitrate and does not react with any commonly encountered oxyanions.In addition to oxyanions the enzyme does not respond to other species such as metal ions and chloride. The sensitivity of the biosensing system is exceptional with ability to detect mmol l21 concentrations of nitrate. Additionally, the Nap has been shown not to leach from the sol– gel matrix and retains its activity, within the sol–gel, even after a six month storage period. It is clear therefore that the immobilization of periplasmic nitrate reductase within a sol–gel matrix offers excellent potential for the optical biosensing of nitrate.The authors acknowledge the Trustees of the Analytical Chemistry Trust Fund of the Royal Society of Chemistry for the award of an SAC Research Studentship to J.W.A. The authors would also like to thank Dr. P. Dobbin and A. Reilly for assistance with the purification of the periplasmic nitrate reductase and also R. J. Chapman for experimental assistance. References 1 Addiscott, T.M., Whitmore, A. P., and Powlson, D. S., Farming, Fertilizers and the Nitrate Problem, CAB International Oxford, 1991. 2 Briggs, D. J., The State of the Environment in the European Community 1986, Office for Official Publications of the European Communities, Luxembourg, 1987. 3 Odashima, S., Oncology, 1980, 37, 282. 4 Cornblath, M., and Hartmann, A. F., J. Pediatr., 1948, 33, 421. 5 Stichtenoth, D. O., Wollenhaupt, J., Andersone, D., Zeidler, H., and Frolich, J. C., Br. J. Rheumatol., 1995, 34, 616. 6 Benvenuti, C., Bories, P. N., and Loisance, D., Transplantation, 1996, 61, 745. 7 Greenberg, A. E., Trussel, R. R., and Clesceri, L. S., Standard Methods for the Examination of Water and Waste Water, American Public Health Association, Washington D.C., USA, 1985. 8 Dejong, P., and Burggraaf, M., Clin. Chem. Acta, 1983, 132, 63. 9 Bell, L. C., Richardson, D. J., and Ferguson, S. J., FEBS Lett., 1990, 265, 85. 10 Kang, S. C., Lee, K.-S., Kim, J.-D., and Kim, K.-J., Bull. Korean Chem. Soc., 1990, 11, 124. 11 Stanley, M. A., Maxwell, J., Forrestal, M., Doherty, A. P., MacCraith, B. D., Diamond, D., and Vos, J. G., Anal. Chim. Acta, 1994, 299, 81. 12 Lumpp, R., Reichert, J., and Ache, H. J., Sens. Actuators, B., 1992, 7, 473. 13 Mohr, G. J., and Wolfbeis, O. S., Anal. Chim. Acta, 1995, 316, 239. 14 Cosnler, S., Innocent, C., and Jouanneau, Y., Anal. Chem., 1994, 66, 3198. 15 Berks, B. C., Richardson, D. J., Robinson, C., Reilly, A., Alpin, R. T., and Ferguson, S. J., Eur. J. Biochem., 1994, 220, 117. 16 Bennett, B., Berks, B. C., Ferguson, S. J., Thomson, A. J., and Richardson, D. J., Eur. J. Biochem., 1994, 226, 789. 17 Berks, B. C., Ferguson, S. J., Moir, J. W. B., and Richardson, D. J., Biochim. Biophys. Acta, 1995, 1232, 97. 18 Brinker, C. J., and Scherer, G. W., Sol–Gel Science, Academic Press, San Diego, 1990. 19 Ellerby, L. M., Nishida, C. R., Nishida, F., Yamanka, S. A., Dunn, B., Valentine, J. S., and Zink, J. I., Science, 1992, 255, 1113. 20 Blyth, D. J., Aylott, J. W., Richardson, D. J., and Russell, D. A., Analyst, 1995, 120, 2725. 21 Burnell, J. N., John, P., and Whatley, F. R., Biochem. J., 1975, 150, 527. 22 Bell, L. C., Page, M. D., Berks, B. C., Richardson, D. J., and Ferguson, S. J., J. Gen. Microbiol., 1993, 139, 3205. 23 Dave, B. C., Dunn, B., Selverstone Valentine J., and Zink, J. I., Anal. Chem., 1994, 66, 1120A. Paper 6/06146J Received September 6, 1996 Accepted September 27, 1996 Fig. 3 Calibration curve for the decrease of 550 nm absorption band of the immobilized Nap enzyme with increasing nitrate concentration. 80 Analyst, January 1997, Vol. 122

ISSN:0003-2654    

DOI:10.1039/a606146j

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Optical Sensor for Oxygen Using a Porphyrin-doped Sol–Gel Glass Sang-Kyung Lee and Ichiro Okura* Department of Bioengineering, Tokyo Institute of Technology, Nagatsuta 4259, Yokohama 226, Japan Organic dye-doped glasses, viz., platinum octaethylporphyrin incorporated in silica matrices by the sol–gel method, were prepared and their photochemical properties were investigated. The optical sensing of oxygen using these materials is based on the change in the phosphorescence intensity of the oxygen-sensitive dye.This approach allows the development of a photostable, chemically durable optical oxygen sensor with a fast response time (5 s). There was no significant bleaching after 5 months in the absence of light. Keywords: Optical sensor; organic dye-doped glass; phosphorescence; sol–gel method; oxygen sensor; platinum octaethylporphyrin In the last 10 years a variety of oxygen sensors based on luminescence quenching by oxygen have been developed and reported in environmental, clinical and analytical chemistry and also in other fields.1–6 In particular, surface oxygen pressure measurements using oxygen-sensitive luminescent materials are currently under development for aerodynamic tests in wind tunnels and in in-flight environments because they can potentially provide pressure data with high spatial resolution and data rates that are orders-of-magnitude greater than those of conventional transducers.7–9 This technique is based on a class of luminescent materials for which the luminescence is quenched by the presence of oxygen (the Stern–Volmer relationship).Using a gas of fixed composition (air), the oxygen concentration is linearly proportional to the gas pressure, and the luminescence intensity becomes a measure of the surface pressure. Many luminescent dyes have been tested as oxygen sensing probes. Among them, transition metal complexes1–5,10 and metalloporphyrins6,11–13 have been frequently utilized as oxygen-sensitive dyes owing to their visible light absorption, high sensitivity to oxygen and large Stokes’ shifts.However, most oxygen-sensitive dyes exhibit low photostability, which is a major obstacle in the commercialization of optical oxygen sensors.11 In an attempt to increase the photostability and lifetime of optical sensors, several methods have been adopted and can be classified into three categories. (1) Modification of the sensor system. Klimant and Wolfbeis14 have proposed optical isolation of the sensing layer to minimize photobleaching of the dye.Black silicone, black Teflon, and titania were used as isolation layers, which made their sensor more photostable than those with only a sensing layer. (2) Modification of the existing dyes. For example, Papkovsky et al.12 observed that PtII and PdII octaethylporphyrin ketones showed improved photostability compared with the corresponding octaethylporphyrins. However, only a few attempts have been made so far at dye modification in an attempt to enhance the photostability of optical sensors, possibly because there is little information on the relationship between the stability of a dye and its structure. (3) Modification of the support matrices.Generally, oxygen sensors, including pressure-sensitive paints (PSP) for aerodynamic applications, are constructed by dispersing or dissolving a luminescent dye in an oxygen-permeable binder material, such as a polymer matrix.However, a polymer matrix has shortcomings when used in wind tunnels and other flow facilities. Gouterman et al.11 demonstrated the effects of photobleaching of the PSP coating. The I0/I value in the Stern– Volmer relationship decreased by approximately 40–45% for different excitation wavelengths when illuminated for 1 h.11 The purpose of the present investigation was to devise a convenient route to a photostable sensor in order to use this sensor in surface pressure measurements by matrix modification. Ceramic materials such as glasses have been widely used as supports for optical sensors owing to their superior properties over organic polymers.15–22 As solid supports for chemical reagents including dyes, inorganic glass sensors have distinct advantages: firstly, glass materials and inorganic ceramics prepared by the sol–gel process are transparent, making them highly suitable for quantitative spectrophotometric tests;23,24 secondly, the glasses are chemically inert, photostable and thermally stable, compared with, for example, polymer matrices, making them highly suitable for applications in harsh environments; and thirdly, the preparation of the doped glasses is technically simple.Thus, the trapping procedure is straightforward and non-specific as compared with covalent binding of a reagent to a solid support. The reagents are generally nonleachable, thus offering clear advantages over reagent adsorption techniques.Finally, glass materials of various shapes, as well as thin films, are easily prepared. This paper describes the fabrication of porphyrin-doped sol– gel glasses and the optical oxygen sensing properties of these materials. Other sensor characteristics such as photostability and storage stability are also discussed. Experimental Materials Tetraethyl orthosilicate (TEOS) was obtained from Aldrich (Milwaukee, WI, USA) and was used as received. Platinum octaethylporphyrin (PtOEP; structure shown in Fig. 1) and Triton X-100 (scintillation grade) were obtained from Porphyrin Products (Logan, UT, USA) and Eastman-Kodak (Rochester, NY, USA), respectively. All the solvents used were of spectroscopic grade and dried over molecular sieves before use. Water was purified with a Milli-Q plus system (Millipore, Bedford, MA, USA). Fig. 1 Structure of PtOEP. Analyst, January 1997, Vol. 122 (81–84) 81Sample Preparation PtOEP-doped silica gel glasses were prepared by basic hydrolysis of TEOS, as follows.A 30 ml volume (0.135 mol) of TEOS and 31 ml (0.53 mol) of ethanol were mixed in a sample vial. Aqueous ammonia (ammonia/H2O; 40 ml/38 ml) was added over a short time while stirring. The mixture was then stirred for 2 h at room temperature. At this stage, Triton X-100 (0.2 ml) was also added in order to improve the homogeneity of the silica sol and to give a crack-free monolith. Finally, 1.2 ml of an acetone solution of PtOEP (4.72 3 1025 mol l21) were added to 2 ml of the silica sol and the resulting mixture was stirred for a further 12 h.After a gelation time of 2–3 weeks, the wet gel was dried at room temperature for 2 weeks and kept in the dark until use. In order to investigate the effect of temperature on the sensor characteristics of the PtOEP-doped glasses, xerogels were dried at room temperature, 150 and 180 °C. The highest drying temperature was chosen to be just below the melting-point of PtOEP.A PtOEP-adsorbed sample was prepared for comparison with the PtOEP-doped sample in terms of photostability. The initial PtOEP / acetone solution was mixed with commercially available silica particles and thoroughly stirred for at least 5 h. The resulting material was washed five times with deionized water, dried under vacuum for 24 h at room temperature and kept in the dark until use. A PtOEP-adsorbed silica disc was prepared using a conventional KBr IR press method.Instrumentation Absorption spectra of the films were measured using a Hitachi (Shibuya, Tokyo, Japan) Model U-2000 spectrometer. Steadystate phosphorescence intensity spectra were recorded on a Hitachi Model F-4010 fluorescence spectrometer using visible excitation (150 W xenon lamp). All the samples were excited at 535 nm, which was the wavelength producing maximum emission intensity at 646 nm. A gas flow meter was utilized to measure the relative flow rates of oxygen and nitrogen. Mixtures containing different concentrations of oxygen were obtained by mixing the gases using a 2 m long tube and were then introduced into a 1 3 1 cm cell capped with a septum and wrapped with Parafilm in which PtOEP incorporated in the silica film was exposed to the mixed gas stream for at least 10 min.For all the measurements, the sample film was placed diagonally relative to the cell. The oxygen concentration was calculated by dividing the oxygen flow rate by the sum of the total flow rate of the mixed gases.In order to investigate the effect of Triton X-100 on the distribution of PtOEP in the prepared sensor materials and the homogeneity of the phosphorescence from these materials, fluorescence microscopy and Nomarski differential-interference- contrast microscopy were used. Results and Discussion Sample Preparation and Effect of Surfactant The glasses prepared by the method described above were optically transparent. The homogeneity and reproducibility of the resulting monoliths was tested by optical absorption.For all PtOEP–silica samples prepared at room temperature, the absorbance was 0.19 ± 0.03. As shown in Fig. 2, the addition of Triton X-100 resulted in greatly improved homogeneity. Surfactant-free samples [Fig. 2(a)] formed aggregates and usually showed severe cracking. These aggregate-containing samples showed no phosphorescence. On the other hand, Fig. 2(b) shows that PtOEP was homogeneously dispersed and immobilized in the silica matrix when Triton X-100 was added; the resulting phosphorescence was emitted homogeneously from the silica matrix.The high homogeneity on addition of Triton X-100 can be explained by the following effects. Polymerization in the presence of a surfactant increases the molecular mass,24 and improves the homogeneity of the polymer. Hence, the complexity of the three-dimensional network is increased during the sol– gel transition.Consequently, inner tensions in the glass formed during condensation and polycondensation decrease and cracks do not appear. As a result, the addition of a surfactant provides a convenient preparation of the organic dye-doped monolith with improved homogeneity. When the oxygen-sensitive dye PtOEP is also added, it is completely trapped within the silica cages of the resulting glass and can be used as a chemically inert and more stable oxygen sensor material. It was possible to immobilize water-insoluble PtOEP in inorganic supports by using Triton X-100 in the sol–gel method, without affecting either the optical transparency of the glass host or the homogeneity of the dye guest.It was found that this preparation method can also be applied to various water-soluble and -insoluble sensor dyes (for example, ruthenium complexes) from our previous studies.26 Photophysical Properties Absorption maxima of the PtOEP-doped silica glass show a redshift, compared with the initial acetone solution (absorption maxima = 536, 502 and 380 nm in glass; 533, 500 and 377 nm in acetone solution; not shown).This red-shift can be explained by the lower polarity of the silica matrix, which consists of siloxane (Si–O–Si) and silanol (Si–OH) groups.16,27,28 Phosphorescence spectra of the PtOEP-doped silica glass under de-oxygenated, ambient, and oxygenated conditions are shown in Fig. 3, indicating effective quenching of the phosphorescence intensity by oxygen.The emission spectrum of the silica matrix showed no red (or blue)-shift and no differences in peak shape, compared with that of the acetone solution (646 nm in both instances). The emission maxima intensities increase strongly on going from oxygenated to ambient and de-oxygenated conditions. Hence, the sol–gel process can provide a good matrix with no interactions between PtOEP and silica during the optical sensing of oxygen. Oxygen Sensing Properties The phosphorescence intensity from the excited state of PtOEP immobilized in the silica matrix is quenched by oxygen according to the Stern–Volmer equation: Fig. 2 Images of PtOEP immobilized in silica matrices, (a) without and (b) with Triton X-100, obtained by Nomarski differential-interferencecontrast microscopy. 82 Analyst, January 1997, Vol. 122I0/I = 1+KSV [O2] where I0 and I are the phosphorescence intensities in the absence and presence of oxygen, respectively and KSV is the Stern–Volmer quenching constant.In order to minimize photodecomposition of the samples when plotting the Stern–Volmer curves, the samples were kept in the dark except during scanning, which was carried out between 640 and 650 nm. Fig. 4 shows the sensitivity (I0/I) and linearity of the sol–gel sensors prepared at different drying temperatures. The sensors that were prepared by drying at room temperature (sample 1) and 150 °C (sample 2) exhibited much higher sensitivities than the sensor prepared by drying at 180 °C (sample 3).These results indicate that ‘sintering step’ is not necessary. The roomtemperature dried sample shows good sensitivity to oxygen. The linearity is also different. Sample 1 shows an upward curvature, whereas samples 2 and 3 show a downward curvature. All three samples are non-linear although the best sensitivity is obtained with sample 2. The effect of the drying temperature on the sensitivity and linearity of the sensors described here might be explained by several factors or by a combination of factors.(1) Oxygen permeability and diffusion through the sol–gel matrix. Oxygen permeability and transport through the sol–gel matrix will greatly affect the sensitivity and linearity as well as the response time of the sensor. During the ageing and drying process, changes in the physical properties of the gel occur. It is wellknown that ageing and thermal treatment result in an increase in pore size.24 Since the pores can act as an oxygen channel to dye molecules immobilized in the sol–gel matrix, the increase in pore size results in increased molar fractions of oxygenquenchable dye and hence the resulting sensitivity and linearity of the sensor are modified.This explains why the linearity and sensitivity of sample 2 are superior to those of sample 1. However, this explanation cannot account for the low sensitivity and marked deviation from linearity of sample 3. (2) Dye molecule distribution in the sol–gel matrix.If the distribution of the dye in the sol–gel matrix is changed with drying temperature by the migration of the dye molecules into different sites within the matrix and/or by partial diminishing, this could affect the sensitivity and linearity of the sensor. This suggests that the micro-environment of the sensor materials varies with the drying temperature. In fact, the optical transparency and colour of the samples were found to change with increasing temperature.The completely transparent pink sample 1 (roomtemperature dried), kept its transparency up to 150 °C, but became slightly opaque and yellow when dried at 180 °C. This indicates that there is a change in the micro-environment of the doped sol–gel. It is well-known that the quenching behaviour of photoluminescent molecules is affected by the structure of the matrix and by the local composition in which the molecule is situated. In heterogeneous systems, it is believed that there are two or more sites with different quenching constants: quencher-easy accessible and quencher-difficult accessible sites.Therefore, the Stern–Volmer plot deviates from linearity because of the different relative contributions which originate from different quenching sites. Response Time and Reversibility Fig. 5 demonstrates the typical dynamic response of the sensor when switching between fully oxygenated and fully deoxygenated atmospheres.The response times of the sensor are about 5 s on going from nitrogen to oxygen and 10 s on going from oxygen to nitrogen. The times required to achieve 90% of the ultimate response, t (90%), for both cases are less than 5 s. These are very short response times, and are similar to those reported by McEvoy et al.10 for their optical sensor. The fast response obviously comes from the microporous texture of the sol–gel matrix. Such fast response times are not common among polymer matrix-derived optical sensors.As is also shown in Fig. 5, stable and reproducible signals were obtained with the sensor. The fact that the inorganic matrix prepared by the sol–gel method is responsible for the enhanced stability compared with polymer matrix and other sensor production methods was verified by the stability test (vide infra). Stability and Sensor Lifetime The photostability was tested by placing the PtOEP-doped silica glass in front of a UV xenon lamp, with continuous irradiation under ambient conditions.Fig. 6 shows the photostability of the sensor (sample 1) under different atmospheric conditions. For comparison, a PtOEP-adsorbed silica sample was prepared (see under Sample Preparation). The PtOEP-doped silica glass exhibited improved photostability, compared with the PtOEPadsorbed silica sample [Fig. 5(b) and (c)]. Although a 10% Fig. 3 Room-temperature phosphorescence spectra of PtOEP-doped sol– gel glass under different atmospheric conditions: N2 (top), ambient conditions (middle) and O2 (bottom). Excitation wavelength, 535 nm.Fig. 4 Stern–Volmer plots of relative phosphorescence intensity for PtOEP-doped sol–gel glasses with different drying temperatures: top (sample 2), middle (sample 1) and bottom (sample 3). Fig. 5 Response time, relative intensity change and reproducibility for sample 1 on switching between 100% nitrogen (a) and 100% oxygen (b). Analyst, January 1997, Vol. 122 83decrease in intensity was observed in a fully oxygenated environment, no decrease in intensity occurred in inert and ambient atmospheres even after a measurement time of 100 min.It should be noted that it is necessary to revert to an inert atmosphere after carrying out measurements under oxygenated conditions. A further stability test was also performed. With continuous illumination for 24 h in an ambient atmosphere, the decrease in the original luminescence intensity was < 15%.These results are comparable to those of a previous photobleaching test by Papkovsky et al.,12 who reported that a porphyrin ketone/polymer sensor showed an 88% recovery of the initial intensity under similiar conditions; the results also show that the proposed sensor is more stable than that of Gouterman et al.11 The long-term stability of the proposed sensor system was not tested. However, since the results of a periodical check over 1 month and of storage for more than 5 months in the absence of lights gave no evidence of instability, we believe that the proposed sensor is likely to be stable for 5 months.Photostability is one of the main advantages of using an inorganic material as a sensor matrix. Unfortunately, the mechanism of the photodecomposition of sensor dyes is currently not well understood. Chemical durability was also tested using water, methanol and ethanol. After storage in each solvent under ambient conditions for at least 24 h, the PtOEPdoped silica was stable and no interference effects were observed.However, the effect of other gases on sensor performance was not studied, taking into consideration the content of these gases in air and because the proposed sensor is intended to be applied to aerofoils. The PtOEP-doped sol–gel sensor exhibits no significant changes in response, intensity and sensitivity after 5 months in the absence of light. Conclusions An optical oxygen sensor was prepared using the surfactant/sol– gel method, incorporating the water- and alcohol-insoluble compound PtOEP in a silica matrix.This method can provide a general support matrix for almost any water-soluble and -insoluble oxygen-sensitive dye and relatively high concentrations of dye can be immobilized in the glass without significantly affecting either the optical transparency of the glass host or the homogeneity of the dye guest. The sol–gel optical sensor which is exemplified here should compete well with traditional polymer matrices, particularly because of its enhanced photostability and physical and thermal properties.The proposed sensor possesses good operational stability and reproducibility and a fast response time (5 s). These results suggest that it should be possible to develop a fast and photostable pressure-sensitive coating for aerodynamic applications. Studies involving the aerodynamic testing of the described sensor, and the preparation and characterization of a ruthenium complex/sol–gel sensor are currently in progress in this laboratory.The authors are grateful to Dr. K. Asai (National Aerospace Laboratory, Japan) for supplying the PtOEP and for useful discussions. Sang-Kyung Lee would like to thank Gil-Jung Kim (Nishida Lab., T.I.T.) for performing Nomarski microscopy. This work was supported, in part, by a Grant-in-Aid for Scientific Research on Priority-Areas-Research on ‘Photoreaction Dynamics’ from the Ministry of Education, Science, Sports, and Culture of Japan.References 1 Singer, E., Duveneck, G. L., Ehrat, M., and Widmer, H. M., Sens. Actuators A., 1994, 41–42, 542. 2 Hartmann, P., Leiner, M. J. P., and Lippitsch, M. E., Anal. Chem. 1995, 67, 88. 3 Xu, W., Schmidt, R., Whaley, M., Demas, J. N., DeGraff, B. A., Karikari, E. K., and Farmer, B. L., Anal. Chem., 1995, 67, 3172. 4 Bacon, J. R., and Demas, J. N., Anal. Chem., 1987, 59, 2780. 5 Carraway, E. R., Demas, J.N., DeGraff, B. A., and Bacon, J. R., Anal. Chem., 1991, 63, 332. 6 Vanderkooi, J. M., Maniara, G., Green, T. J., and Wilson, D. F., J. Biol. Chem., 1987, 262, 5476. 7 Troyanovsky, I., Sadovskii, N., Kuzmin, M., Mosharov, V., Orlov, A., Radchenko, V., and Phonov, S., Sens. Actuators B, 1993, 11, 201. 8 Morris, M. J., Donovan, J. F., Kegelman, J. T., Schwab, S. D., Levy, R. L., and Crites, R. C., AIAA J., 1993, 31, 419. 9 McLachlan, B. G., and Bell, J. H., Exp. Therm. Fluid Sci., 1995, 10, 470. 10 McEvoy, A. K., McDonagh, C. M., and MacCraith, B. D., Analyst, 1996, 121, 785. 11 Gouterman, M., Callis, J., Burns, D., Kavandi, J., Gallery, J., Khalil, G., Green, E., McLachlan, B., and Crowder, J., Proceedings of the ONR/NASA Workshop on Quantitative Flow Visualization, Purdue University, West Lafayette, IN. 12 Papkovsky, D. B., Ponomarev, G. V., Trettnak, W., and O’Leary, P., Anal. Chem., 1995, 67, 4112. 13 Papkovsky, D. B., Sens. Actuators B, 1993, 11, 293. 14 Klimant, I., and Wolfbeis, O. S., Anal. Chem., 1995, 67, 3160. 15 Samuel, J., Strinkovsk, A., Shalom, S., Lieberman, K., Ottolenghi, M., Avnir, D., and Lewis, A., Mater. Lett., 1994, 21, 431. 16 Avinir, D., Levy, D., and Reisfeld, R., J. Phys. Chem., 1984, 88, 5956. 17 Zusman, R., Rottman, C., Ottolenghi, M., and Avnir, D., J. Non- Cryst. Solids, 1990, 122, 107. 18 Badini, G. E., Grattan, K. T. V., and Tseung, A. C. C., Analyst, 1995, 120, 1025. 19 Kuselman, I., and Lev, O., Talanta, 1993, 40, 749. 20 Yang, L., and Saavedra, S. S., Anal. Chem., 1995, 67, 1307. 21 Grattan, K. T. V., Badini, G. E., Palmer, A. W., and Tseung, A. C. C., Sens. Actuators A, 1991, 25–27, 483. 22 Narang, U., Prasad, P. N., Bright, F. V., Ramanathan, K., Kumar, N. D., Malhotra, B. D., Kamalasanan, M. N., and Chandra, S., Anal. Chem., 1994, 66, 3139. 23 Better Ceramics Through Chemistry II, ed. Brinker, C. J., Clark, D. E., and Ulrich, D. R., MRS Symposium, vol. 73, Elsevier, New York, 1986. 24 Brinker C. J., and Scherer, G. W., Sol–Gel Science, The Physics and Chemistry of Sol–Gel Processing, Academic Press, New York, 1990. 25 Lissant, J. K., Emulsions and Emulsion Technology, Marcel Dekker, New York, 1974. 26 Lee, S.-K., and Okura, I., unpublished results. 27 Grauer, Z., Avinir, D., and Yariv, S., Can. J. Chem., 1984, 62, 1889. 28 Avinir, D., Kaufman, V. R., and Reisfeld, R., J. Non-Cryst. Solids, 1985, 74, 395. Paper 6/04885D Received July 11, 1996 Accepted September 23, 1996 Fig. 6 Photostability of PtOEP-doped silica glass (sample 1) in inert (a), ambient (b) and oxygenated atmosphere (d) and adsorbed silica sample (c). (b) and (c) were normalized to allow a comparison of the decrease in intensity in each instance. 84 Analyst, January 1997, Vol. 122 Optical Sensor for Oxygen Using a Porphyrin-doped Sol–Gel Glass Sang-Kyung Lee and Ichiro Okura* Department of Bioengineering, Tokyo Institute of Technology, Nagatsuta 4259, Yokohama 226, Japan Organic dye-doped glasses, viz., platinum octaethylporphyrin incorporated in silica matrices by the sol–gel method, were prepared and their photochemical properties were investigated.The optical sensing of oxygen using these materials is based on the change in the phosphorescence intensity of the oxygen-sensitive dye. This approach allows the development of a photostable, chemically durable optical oxygen sensor with a fast response time (5 s). There was no significant bleaching after 5 months in the absence of light.Keywords: Optical sensor; organic dye-doped glass; phosphorescence; sol–gel method; oxygen sensor; platinum octaethylporphyrin In the last 10 years a variety of oxygen sensors based on luminescence quenching by oxygen have been developed and reported in environmental, clinical and analytical chemistry and also in other fields.1–6 In particular, surface oxygen pressure measurements using oxygen-sensitive luminescent materials are currently under development for aerodynamic tests in wind tunnels and in in-flight environments because they can potentially provide pressure data with high spatial resolution and data rates that are orders-of-magnitude greater than those of conventional transducers.7–9 This technique is based on a class of luminescent materials for which the luminescence is quenched by the presence of oxygen (the Stern–Volmer relationship).Using a gas of fixed composition (air), the oxygen concentration is linearly proportional to the gas pressure, and the luminescence intensity becomes a measure of the surface pressure.Many luminescent dyes have been tested as oxygen sensing probes. Among them, transition metal complexes1–5,10 and metalloporphyrins6,11–13 have been frequently utilized as oxygen-sensitive dyes owing to their visible light absorption, high sensitivity to oxygen and large Stokes’ shifts. However, most oxygen-sensitive dyes exhibit low photostability, which is a major obstacle in the commercialization of optical oxygen sensors.11 In an attempt to increase the photostability and lifetime of optical sensors, several methods have been adopted and can be classified into three categories.(1) Modification of the sensor system. Klimant and Wolfbeis14 have proposed optical isolation of the sensing layer to minimize photobleaching of the dye. Black silicone, black Teflon, and titania were used as isolation layers, which made their sensor more photostable than those with only a sensing layer.(2) Modification of the existing dyes. For example, Papkovsky et al.12 observed that PtII and PdII octaethylporphyrin ketones showed improved photostability compared with the corresponding octaethylporphyrins. However, only a few attempts have been made so far at dye modification in an attempt to enhance the photostability of optical sensors, possibly because there is little information on the relationship between the stability of a dye and its structure.(3) Modification of the support matrices. Generally, oxygen sensors, including pressure-sensitive paints (PSP) for aerodynamic applications, are constructed by dispersing or dissolving a luminescent dye in an oxygen-permeable binder material, such as a polymer matrix. However, a polymer matrix has shortcomings when used in wind tunnels and other flow facilities. Gouterman et al.11 demonstrated the effects of photobleaching of the PSP coating.The I0/I value in the Stern– Volmer relationship decreased by approximately 40–45% for different excitation wavelengths when illuminated for 1 h.11 The purpose of the present investigation was to devise a convenient route to a photostable sensor in order to use this sensor in surface pressure measurements by matrix modification. Ceramic materials such as glasses have been widely used as supports for optical sensors owing to their superior properties over organic polymers.15–22 As solid supports for chemical reagents including dyes, inorganic glass sensors have distinct advantages: firstly, glass materials and inorganic ceramics prepared by the sol–gel process are transparent, making them highly suitable for quantitative spectrophotometric tests;23,24 secondly, the glasses are chemically inert, photostable and thermally stable, compared with, for example, polymer matrices, making them highly suitable for applications in harsh environments; and thirdly, the preparation of the doped glasses is technically simple.Thus, the trapping procedure is straightforward and non-specific as compared with covalent binding of a reagent to a solid support. The reagents are generally nonleachable, thus offering clear advantages over reagent adsorption techniques. Finally, glass materials of various shapes, as well as thin films, are easily prepared. This paper describes the fabrication of porphyrin-doped sol– gel glasses and the optical oxygen sensing properties of these materials.Other sensor characteristics such as photostability and storage stability are also discussed. Experimental Materials Tetraethyl orthosilicate (TEOS) was obtained from Aldrich (Milwaukee, WI, USA) and was used as received. Platinum octaethylporphyrin (PtOEP; structure shown in Fig. 1) and Triton X-100 (scintillation grade) were obtained from Porphyrin Products (Logan, UT, USA) and Eastman-Kodak (Rochester, NY, USA), respectively.All the solvents used were of spectroscopic grade and dried over molecular sieves before use. Water was purified with a Milli-Q plus system (Millipore, Bedford, MA, USA). Fig. 1 Structure of PtOEP. Analyst, January 1997, Vol. 122 (81–84) 81Sample Preparation PtOEP-doped silica gel glasses were prepared by basic hydrolysis of TEOS, as follows. A 30 ml volume (0.135 mol) of TEOS and 31 ml (0.53 mol) of ethanol were mixed in a sample vial. Aqueous ammonia (ammonia/H2O; 40 ml/38 ml) was added over a short time while stirring.The mixture was then stirred for 2 h at room temperature. At this stage, Triton X-100 (0.2 ml) was also added in order to improve the homogeneity of the silica sol and to give a crack-free monolith. Finally, 1.2 ml of an acetone solution of PtOEP (4.72 3 1025 mol l21) were added to 2 ml of the silica sol and the resulting mixture was stirred for a further 12 h. After a gelation time of 2–3 weeks, the wet gel was dried at room temperature for 2 weeks and kept in the dark until use. In order to investigate the effect of temperature on the sensor characteristics of the PtOEP-doped glasses, xerogels were dried at room temperature, 150 and 180 °C.The highest drying temperature was chosen to be just below the melting-point of PtOEP. A PtOEP-adsorbed sample was prepared for comparison with the PtOEP-doped sample in terms of photostability. The initial PtOEP / acetone solution was mixed with commercially available silica particles and thoroughly stirred for at least 5 h.The resulting material was washed five times with deionized water, dried under vacuum for 24 h at room temperature and kept in the dark until use. A PtOEP-adsorbed silica disc was prepared using a conventional KBr IR press method. Instrumentation Absorption spectra of the films were measured using a Hitachi (Shibuya, Tokyo, Japan) Model U-2000 spectrometer. Steadystate phosphorescence intensity spectra were recorded on a Hitachi Model F-4010 fluorescence spectrometer using visible excitation (150 W xenon lamp).All the samples were excited at 535 nm, which was the wavelength producing maximum emission intensity at 646 nm. A gas flow meter was utilized to measure the relative flow rates of oxygen and nitrogen. Mixtures containing different concentrations of oxygen were obtained by mixing the gases using a 2 m long tube and were then introduced into a 1 3 1 cm cell capped with a septum and wrapped with Parafilm in which PtOEP incorporated in the silica film was exposed to the mixed gas stream for at least 10 min.For all the measurements, the sample film was placed diagonally relative to the cell. The oxygen concentration was calculated by dividing the oxygen flow rate by the sum of the total flow rate of the mixed gases. In order to investigate the effect of Triton X-100 on the distribution of PtOEP in the prepared sensor materials and the homogeneity of the phosphorescence from these materials, fluorescence microscopy and Nomarski differential-interference- contrast microscopy were used.Results and Discussion Sample Preparation and Effect of Surfactant The glasses prepared by the method described above were optically transparent. The homogeneity and reproducibility of the resulting monoliths was tested by optical absorption. For all PtOEP–silica samples prepared at room temperature, the absorbance was 0.19 ± 0.03. As shown in Fig. 2, the addition of Triton X-100 resulted in greatly improved homogeneity. Surfactant-free samples [Fig. 2(a)] formed aggregates and usually showed severe cracking. These aggregate-containing samples showed no phosphorescence. On the other hand, Fig. 2(b) shows that PtOEP was homogeneously dispersed and immobilized in the silica matrix when Triton X-100 was added; the resulting phosphorescence was emitted homogeneously from the silica matrix. The high homogeneity on addition of Triton X-100 can be explained by the following effects.Polymerization in the presence of a surfactant increases the molecular mass,24 and improves the homogeneity of the polymer. Hence, the complexity of the three-dimensional network is increased during the sol– gel transition. Consequently, inner tensions in the glass formed during condensation and polycondensation decrease and cracks do not appear. As a result, the addition of a surfactant provides a convenient preparation of the organic dye-doped monolith with improved homogeneity.When the oxygen-sensitive dye PtOEP is also added, it is completely trapped within the silica cages of the resulting glass and can be used as a chemically inert and more stable oxygen sensor material. It was possible to immobilize water-insoluble PtOEP in inorganic supports by using Triton X-100 in the sol–gel method, without affecting either the optical transparency of the glass host or the homogeneity of the dye guest.It was found that this preparation method can also be applied to various water-soluble and -insoluble sensor dyes (for example, ruthenium complexes) from our previous studies.26 Photophysical Properties Absorption maxima of the PtOEP-doped silica glass show a redshift, compared with the initial acetone solution (absorption maxima = 536, 502 and 380 nm in glass; 533, 500 and 377 nm in acetone solution; not shown). This red-shift can be explained by the lower polarity of the silica matrix, which consists of siloxane (Si–O–Si) and silanol (Si–OH) groups.16,27,28 Phosphorescence spectra of the PtOEP-doped silica glass under de-oxygenated, ambient, and oxygenated conditions are shown in Fig. 3, indicating effective quenching of the phosphorescence intensity by oxygen. The emission spectrum of the silica matrix showed no red (or blue)-shift and no differences in peak shape, compared with that of the acetone solution (646 nm in both instances). The emission maxima intensities increase strongly on going from oxygenated to ambient and de-oxygenated conditions.Hence, the sol–gel process can provide a good matrix with no interactions between PtOEP and silica during the optical sensing of oxygen. Oxygen Sensing Properties The phosphorescence intensity from the excited state of PtOEP immobilized in the silica matrix is quenched by oxygen according to the Stern–Volmer equation: Fig. 2 Images of PtOEP immobilized in silica matrices, (a) without and (b) with Triton X-100, obtained by Nomarski differential-interferencecontrast microscopy. 82 Analyst, January 1997, Vol. 122I0/I = 1+KSV [O2] where I0 and I are the phosphorescence intensities in the absence and presence of oxygen, respectively and KSV is the Stern–Volmer quenching constant. In order to minimize photodecomposition of the samples when plotting the Stern–Volmer curves, the samples were kept in the dark except during scanning, which was carried out between 640 and 650 nm.Fig. 4 shows the sensitivity (I0/I) and linearity of the sol–gel sensors prepared at different drying temperatures. The sensors that were prepared by drying at room temperature (sample 1) and 150 °C (sample 2) exhibited much higher sensitivities than the sensor prepared by drying at 180 °C (sample 3). These results indicate that ‘sintering step’ is not necessary. The roomtemperature dried sample shows good sensitivity to oxygen.The linearity is also different. Sample 1 shows an upward curvature, whereas samples 2 and 3 show a downward curvature. All three samples are non-linear although the best sensitivity is obtained with sample 2. The effect of the drying temperature on the sensitivity and linearity of the sensors described here might be explained by several factors or by a combination of factors. (1) Oxygen permeability and diffusion through the sol–gel matrix. Oxygen permeability and transport through the sol–gel matrix will greatly affect the sensitivity and linearity as well as the response time of the sensor.During the ageing and drying process, changes in the physical properties of the gel occur. It is wellknown that ageing and thermal treatment result in an increase in pore size.24 Since the pores can act as an oxygen channel to dye molecules immobilized in the sol–gel matrix, the increase in pore size results in increased molar fractions of oxygenquenchable dye and hence the resulting sensitivity and linearity of the sensor are modified.This explains why the linearity and sensitivity of sample 2 are superior to those of sample 1. However, this explanation cannot account for the low sensitivity and marked deviation from linearity of sample 3. (2) Dye molecule distribution in the sol–gel matrix. If the distribution of the dye in the sol–gel matrix is changed with drying temperature by the migration of the dye molecules into different sites within the matrix and/or by partial diminishing, this could affect the sensitivity and linearity of the sensor.This suggests that the micro-environment of the sensor materials varies with the drying temperature. In fact, the optical transparency and colour of the samples were found to change with increasing temperature. The completely transparent pink sample 1 (roomtemperature dried), kept its transparency up to 150 °C, but became slightly opaque and yellow when dried at 180 °C.This indicates that there is a change in the micro-environment of the doped sol–gel. It is well-known that the quenching behaviour of photoluminescent molecules is affected by the structure of the matrix and by the local composition in which the molecule is situated. In heterogeneous systems, it is believed that there are two or more sites with different quenching constants: quencher-easy accessible and quencher-difficult accessible sites. Therefore, the Stern–Volmer plot deviates from linearity because of the different relative contributions which originate from different quenching sites.Response Time and Reversibility Fig. 5 demonstrates the typical dynamic response of the sensor when switching between fully oxygenated and fully deoxygenated atmospheres. The response times of the sensor are about 5 s on going from nitrogen to oxygen and 10 s on going from oxygen to nitrogen. The times required to achieve 90% of the ultimate response, t (90%), for both cases are less than 5 s.These are very short response times, and are similar to those reported by McEvoy et al.10 for their optical sensor. The fast response obviously comes from the microporous texture of the sol–gel matrix. Such fast response times are not common among polymer matrix-derived optical sensors. As is also shown in Fig. 5, stable and reproducible signals were obtained with the sensor. The fact that the inorganic matrix prepared by the sol–gel method is responsible for the enhanced stability compared with polymer matrix and other sensor production methods was verified by the stability test (vide infra).Stability and Sensor Lifetime The photostability was tested by placing the PtOEP-doped silica glass in front of a UV xenon lamp, with continuous irradiation under ambient conditions. Fig. 6 shows the photostability of the sensor (sample 1) under different atmospheric conditions. For comparison, a PtOEP-adsorbed silica sample was prepared (see under Sample Preparation).The PtOEP-doped silica glass exhibited improved photostability, compared with the PtOEPadsorbed silica sample [Fig. 5(b) and (c)]. Although a 10% Fig. 3 Room-temperature phosphorescence spectra of PtOEP-doped sol– gel glass under different atmospheric conditions: N2 (top), ambient conditions (middle) and O2 (bottom). Excitation wavelength, 535 nm. Fig. 4 Stern–Volmer plots of relative phosphorescence intensity for PtOEP-doped sol–gel glasses with different drying temperatures: top (sample 2), middle (sample 1) and bottom (sample 3).Fig. 5 Response time, relative intensity change and reproducibility for sample 1 on switching between 100% nitrogen (a) and 100% oxygen (b). Analyst, January 1997, Vol. 122 83decrease in intensity was observed in a fully oxygenated environment, no decrease in intensity occurred in inert and ambient atmospheres even after a measurement time of 100 min.It should be noted that it is necessary to revert to an inert atmosphere after carrying out measurements under oxygenated conditions. A further stability test was also performed. With continuous illumination for 24 h in an ambient atmosphere, the decrease in the original luminescence intensity was < 15%. These results are comparable to those of a previous photobleaching test by Papkovsky et al.,12 who reported that a porphyrin ketone/polymer sensor showed an 88% recovery of the initial intensity under similiar conditions; the results also show that the proposed sensor is more stable than that of Gouterman et al.11 The long-term stability of the proposed sensor system was not tested.However, since the results of a periodical check over 1 month and of storage for more than 5 months in the absence of lights gave no evidence of instability, we believe that the proposed sensor is likely to be stable for 5 months. Photostability is one of the main advantages of using an inorganic material as a sensor matrix. Unfortunately, the mechanism of the photodecomposition of sensor dyes is currently not well understood.Chemical durability was also tested using water, methanol and ethanol. After storage in each solvent under ambient conditions for at least 24 h, the PtOEPdoped silica was stable and no interference effects were observed. However, the effect of other gases on sensor performance was not studied, taking into consideration the content of these gases in air and because the proposed sensor is intended to be applied to aerofoils.The PtOEP-doped sol–gel sensor exhibits no significant changes in response, intensity and sensitivity after 5 months in the absence of light. Conclusions An optical oxygen sensor was prepared using the surfactant/sol– gel method, incorporating the water- and alcohol-insoluble compound PtOEP in a silica matrix. This method can provide a general support matrix for almost any water-soluble and -insoluble oxygen-sensitive dye and relatively high concentrations of dye can be immobilized in the glass without significantly affecting either the optical transparency of the glass host or the homogeneity of the dye guest.The sol–gel optical sensor which is exemplified here should compete well with traditional polymer matrices, particularly because of its enhanced photostability and physical and thermal properties. The proposed sensor possesses good operational stability and reproducibility and a fast response time (5 s).These results suggest that it should be possible to develop a fast and photostable pressure-sensitive coating for aerodynamic applications. Studies involving the aerodynamic testing of the described sensor, and the preparation and characterization of a ruthenium complex/sol–gel sensor are currently in progress in this laboratory. The authors are grateful to Dr. K. Asai (National Aerospace Laboratory, Japan) for supplying the PtOEP and for useful discussions.Sang-Kyung Lee would like to thank Gil-Jung Kim (Nishida Lab., T.I.T.) for performing Nomarski microscopy. This work was supported, in part, by a Grant-in-Aid for Scientific Research on Priority-Areas-Research on ‘Photoreaction Dynamics’ from the Ministry of Education, Science, Sports, and Culture of Japan. References 1 Singer, E., Duveneck, G. L., Ehrat, M., and Widmer, H. M., Sens. Actuators A., 1994, 41–42, 542. 2 Hartmann, P., Leiner, M. J. P., and Lippitsch, M. E., Anal. Chem. 1995, 67, 88. 3 Xu, W., Schmidt, R., Whaley, M., Demas, J. N., DeGraff, B. A., Karikari, E. K., and Farmer, B. L., Anal. Chem., 1995, 67, 3172. 4 Bacon, J. R., and Demas, J. N., Anal. Chem., 1987, 59, 2780. 5 Carraway, E. R., Demas, J. N., DeGraff, B. A., and Bacon, J. R., Anal. Chem., 1991, 63, 332. 6 Vanderkooi, J. M., Maniara, G., Green, T. J., and Wilson, D. F., J. Biol. Chem., 1987, 262, 5476. 7 Troyanovsky, I., Sadovskii, N., Kuzmin, M., Mosharov, V., Orlov, A., Radchenko, V., and Phonov, S., Sens.Actuators B, 1993, 11, 201. 8 Morris, M. J., Donovan, J. F., Kegelman, J. T., Schwab, S. D., Levy, R. L., and Crites, R. C., AIAA J., 1993, 31, 419. 9 McLachlan, B. G., and Bell, J. H., Exp. Therm. Fluid Sci., 1995, 10, 470. 10 McEvoy, A. K., McDonagh, C. M., and MacCraith, B. D., Analyst, 1996, 121, 785. 11 Gouterman, M., Callis, J., Burns, D., Kavandi, J., Gallery, J., Khalil, G., Green, E., McLachlan, B., and Crowder, J., Proceedings of the ONR/NASA Workshop on Quantitative Flow Visualization, Purdue University, West Lafayette, IN. 12 Papkovsky, D. B., Ponomarev, G. V., Trettnak, W., and O’Leary, P., Anal. Chem., 1995, 67, 4112. 13 Papkovsky, D. B., Sens. Actuators B, 1993, 11, 293. 14 Klimant, I., and Wolfbeis, O. S., Anal. Chem., 1995, 67, 3160. 15 Samuel, J., Strinkovsk, A., Shalom, S., Lieberman, K., Ottolenghi, M., Avnir, D., and Lewis, A., Mater. Lett., 1994, 21, 431. 16 Avinir, D., Levy, D., and Reisfeld, R., J. Phys. Chem., 1984, 88, 5956. 17 Zusman, R., Rottman, C., Ottolenghi, M., and Avnir, D., J. Non- Cryst. Solids, 1990, 122, 107. 18 Badini, G. E., Grattan, K. T. V., and Tseung, A. C. C., Analyst, 1995, 120, 1025. 19 Kuselman, I., and Lev, O., Talanta, 1993, 40, 749. 20 Yang, L., and Saavedra, S. S., Anal. Chem., 1995, 67, 1307. 21 Grattan, K. T. V., Badini, G. E., Palmer, A. W., and Tseung, A. C. C., Sens. Actuators A, 1991, 25–27, 483. 22 Narang, U., Prasad, P. N., Bright, F. V., Ramanathan, K., Kumar, N. D., Malhotra, B. D., Kamalasanan, M. N., and Chandra, S., Anal. Chem., 1994, 66, 3139. 23 Better Ceramics Through Chemistry II, ed. Brinker, C. J., Clark, D. E., and Ulrich, D. R., MRS Symposium, vol. 73, Elsevier, New York, 1986. 24 Brinker C. J., and Scherer, G. W., Sol–Gel Science, The Physics and Chemistry of Sol–Gel Processing, Academic Press, New York, 1990. 25 Lissant, J. K., Emulsions and Emulsion Technology, Marcel Dekker, New York, 1974. 26 Lee, S.-K., and Okura, I., unpublished results. 27 Grauer, Z., Avinir, D., and Yariv, S., Can. J. Chem., 1984, 62, 1889. 28 Avinir, D., Kaufman, V. R., and Reisfeld, R., J. Non-Cryst. Solids, 1985, 74, 395. Paper 6/04885D Received July 11, 1996 Accepted September 23, 1996 Fig. 6 Photostability of PtOEP-doped silica glass (sample 1) in inert (a), ambient (b) and oxygenated atmosphere (d) and adsorbed silica sample (c). (b) and (c) were normalized to allow a comparison of the decrease in intensity in each instance. 84 Analyst, January 1997, Vol. 122

ISSN:0003-2654    

DOI:10.1039/a604885d

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Simultaneous Automatic Determination of Trace Amounts of Copper and Cobalt by Use of a Flow-through Sensor and First-derivative Spectrometry Elisa Veredaa, Angel Riosb and Miguel Valcarcel*b a Department of Analytical Chemistry, Faculty of Sciences, University of M�alaga, 29071 M�alaga, Spain b Department of Analytical Chemistry, Faculty of Sciences, University of C�ordoba, 14004 C�ordoba, Spain A straightforward method for the simultaneous determination of cobalt and copper in binary mixtures by use of a diode-array detector accommodated in a flow-through spectrophotometric sensor and pyridoxal-4-phenylthiosemicarbazone as reagent is proposed.The coloured complexes formed give extensively overlapped spectra that can only be resolved from their first derivative. The method relies on integrated preconcentration and detection in the flow cell that allows a high sensitivity in the determination of the two metal ions with detection limits of 0.03 mg ml21 for both and an RSD (n = 11; P = 0.05) of 2.0% for cobalt and 0.9% for copper, at a rate of 36 samples h21.The method was applied to the determination of cobalt and copper in several steel samples. Keywords: Flow injection; spectrophotometry; flow-through sensor; cobalt; copper; steels Immobilization in flow injection (FI) has usually been related to enzymes1 and less often to other reagents, and then mainly for preconcentration2 and/or matrix removal,3 or reaction with the analyte4 and immunoassay.5 The use of immobilization in FI offers many advantages: convenience because manipulation is simpler than that of solutions; simplified manifolds with fewer channels; increased sensitivity; increased stability and selectivity6 resulting from the micro-environment in which the reaction takes place; and reagent economy and compatibility with automated systems.The current trend in applications of immobilization is towards placement of the reaction unit in the detection systems, which measures the change in the immobilized reagent or in the surrounding solution on reaction with the analyte.To ensure proper performance, the reaction system must meet several requirements: (a) reversibility (or at least a high capacity if the reagent is consumed); (b) stability of the immobilized reagent in the reaction medium; (c) fast kinetics; and (d) compatibility between the support and the detection system used. Thiosemicarbazones and phenylthiosemicarbazones generally react as chelating ligands for transition metal ions by bonding through the sulfur and hydrazine nitrogen atoms.7,8 As a rule, phenylthiosemicarbazones are more effective than thiosemicarbazones because their complexes have much higher absorptivities9 (çPPT–Co = 2.17 3 104, çPPT–Cu = 1.35 3 104 l mol21 cm21).This paper reports on the simultaneous determination of CoII and CuII with pyridoxal-4-phenyl-3-thiosemicarbazone (PPT) immobilized on a sorbent material, C18.The method relies on previous work,10 a very selective method where PPT was used for the FI spectrophotometric determination of cobalt based on the formation of a highly stable complex between cobalt and PPT in a strongly acidic medium; the method was not very sensitive and was subject to perturbation from copper, which interfered up to a copper-to-cobalt ratio of 3. This drawback was exploited for the simultaneous determination of cobalt and copper. Since the coloured complexes formed give extensively overlapped spectra, we used derivative spectrophotometry to resolve them from their first derivatives.In this work, in addition to resolving the binary mixture of interest, we improved the sensitivity (determination limits 0.3 and 0.1 mg ml21 for cobalt and copper, respectively) by use of a flow manifold integrating preconcentration and detection in a sorbent material packed in a flow-cell. The improved method relies on the recently developed flow-through sensor technology, 11,12 by which retention and detection are integrated in an FI system.13,14 In this case, successive passage of the complexes, previously formed in the flowing stream, and eluent through the flow cell and continuous photometric monitoring of the process provided the analytical information needed to determine cobalt and the copper.The proposed method was used for the automatic on-line determination of cobalt and copper in steels, with satisfactory results.Experimental Reagents All chemicals used were of analytical-reagent grade and solutions were prepared using high-purity water from a Millipore (Bedford, MA, USA) Milli-Q Water purification system. Stocks standard solutions of cobalt(ii) and copper(ii) were prepared from the nitrate and sulfate, respectively, and standardized complexometrically. Working standards solutions were prepared by appropiate dilution daily as required. A 0.025% solution of PPT was prepared by dissolving 0.025 g of the reagent in 30 ml of N,N-dimethylformamide (DMF) and diluting to 100 ml with water.The reagent was synthesized following a procedure described elsewhere.15 A 30% solution of perchloric acid in DMF–water (3 + 7 v/v) was used as eluent. A 3 m perchloric acid solution was also used. C18 material from a Sep-Pak C18 cartridge (Millipore Waters, Milford, MA, USA) was used for the retention and detection of cobalt and copper complexes in the flow cell. Instrumentation and Procedure A Hewlett-Packard (Avondale, PA, USA) Model 8452A diodearray detector interfaced to a Vectra ES/12 computer which delivered results through an HP Think-Jet printer was used.The flow manifold consisted of a Gilson (Worthington, OH, USA) Minipuls-3 peristaltic pump, a Rheodyne (Cotari, CA, USA) Type 50 six-port rotary valve and a Hellma (Jamaica, NY, USA) OS 0.200 flow cell with a 2 mm optical path length (Fig. 1). A schematic diagram of the FI system is shown in Fig 1.It operated as follows: a stream of standard or sample flowing at Analyst, January 1997, Vol. 122 (85–88) 85ensure complete elution of the complexes. The influence of the perchloric acid concentration was also studied. The results showed that the measured signals remained constant over the range 1.5–4.5 m perchloric acid. A concentration of 3 m was finally used in order to prevent interferences from foreign ions. Obviously, once the different streams have merged, the acid concentration in the reaction plug will have decreased by dilution.The influence of the flow injection variables (flow rates, reactor lengths) was studied; the optimum values obtained are summarized in Table 1. Under these final experimental conditions, a sampling rate of 35 h21 was achieved. Calibration graph, Sensitivity and Precision Under the optimum experimental conditions, a linear calibration graph was obtained for the individual determination of each ion.The figures of merit of these calibration graphs are summarized in Table 2. The detection and determination limits defined as the concentrations of analyte giving signals equivalent to three and ten times, respectively, the standard deviation of the blank plus the net blank intensity were calculated. The values found for each metal are also given in Table 2. The precision (RSD, n = 11) achieved in the resolution of a mixture of 2 mg ml21 CoII and 5 mg ml21 CuII was 2.0 and 0.9%, respectively. In order to check for potential synergistic effects of the mixtures of two ions, various synthetic mixtures were readily resolved by using pertinent calibration graphs (Table 3).As can be seen, mixtures of the two ions can be resolved with satisfactory results. The average errors obtained were about 4.6 and 8.8% for cobalt and copper, respectively. These errors can be considered as acceptable for simultaneous determinations involving neither sophisticated nor expensive instrumentation (e.g., atomic optical techniques).Interferences The potential interference of foreign ions in the determination of cobalt and copper by the proposed method was studied by using a fixed concentration of 2 mg ml21 of both CoII and CuII and various concentrations of each foreign ion up to a maximum ion to CoII or CuII ratio of 225 : 1. A given species was considered to interfere if it resulted in a ±5ion of the average FI signal, calculated at the established levels of cobalt and copper.The results obtained are given in Table 4. The inferference of FeIII can be significantly lowered by adding sodium fluoride to the medium. This allows one to determine cobalt and copper in iron-containing materials. Determination of Cobalt and Copper in Steels In order to evaluate the effectiveness of the method, a series of recovery experiments were carried out on various steel samples supplied by Acerinox (Algeciras, Spain), with the nominal concentrations (as determined by XRF, with calibration lines obtained using NBS standard steels) given in Table 5. The samples were prepared as described above.The results of the analyses of steels A–D are given in Table 6 as the averages of three individual determinations. Table 1 Optimum values of flow injection variables Sample flow rate 0.90 ml min21 Reagent flow rate 0.35 ml min21 HClO4 flow rate 0.30 ml min21 Eluent flow rate 0.70 ml min21 R1 dimensions 140 cm 3 0.5 mm id R2 dimensions 60 cm 3 0.5 mm id R3 dimensions 35 cm 3 0.5 mm id Table 2 Analytical figures of merit for the determination of cobalt and copper by the proposed method Linear range/ Detection limit/ Determination Analyte lopt/nm Calibration equation* r mg ml21 mg ml21 limit/mg ml21 CoII 292 y = 27.46 3 1024–3.54 3 1023x 0.999 0.3–6 0.03 0.3 CuII 420–442 y = 23.60 3 1024–5.00 3 1023x 0.997 0.1–15 0.03 0.1 * y = dA/dl (A = absorbance); x = ionic concentration of the analyte in mg ml21. Table 3 Resolution of synthetic CoII–CuII mixtures by use of the proposed method Concentration added/ Concentration found/ mg ml21 mg ml21 Sample no.CoII CuII CoII CuII 1 0.5 0.5 0.5 0.4 2 0.5 1.0 0.5 1.0 3 0.5 2.0 0.5 1.9 4 1.0 0.5 1.0 0.6 5 1.0 1.0 1.1 1.1 6 1.0 2.0 1.1 2.0 7 1.0 4.0 1.1 4.0 8 2.0 0.5 2.0 0.4 9 2.0 1.0 2.2 1.2 10 2.0 2.0 2.2 2.2 11 2.0 4.0 2.2 3.7 12 3.0 1.0 3.1 0.9 13 3.0 2.0 3.0 1.8 14 3.0 3.0 2.9 3.0 15 3.0 4.0 3.1 3.7 16 3.0 5.0 3.0 4.7 17 4.0 1.0 4.2 1.1 18 4.0 3.0 3.8 3.3 19 4.0 4.0 3.7 3.9 Table 4 Interferences in the proposed method Tolerated interferent-toanalyte Foreign species molar ratio NaI, KI, BaII, CaII, MgII, PbII, MnII, fluoride, bromide, iodide, chloride, thiocyanate, nitrate, sulfate, thiosulphate, oxalate, FeIII* > 225 CdII, glycine 100 AlIII, ZnII, CrIII 25 NiII 20 MoVI 3 FeIII 2 * With 2000 times more fluoride than CoII and CuII.Analyst, January 1997, Vol. 122 870.90 ml min21 reached reactor R1 (140 cm 30.5 mm id), where it merged with a stream consisting of 0.025% PPT in DMF– water (3 + 7 v/v) circulated at 0.35 ml min21. The mixture reached reactor R2 (60 cm 3 0.5 mm id) and was merged with a stream of 3 m HClO4 flowing at 0.30 ml min21, where the complexes of other metal ions are destroyed. The cobalt and copper complexes were then passed through reactor R3 (35 cm 3 0.5 mm id) in order to increase the reproducibility, and reached the flow cell, where they were retained on C18 packed in the flow cell (preconcentration step).When C18 was saturated (35 s), the first-derivative spectrum was recorded over the range 200–500 nm (detection step) then the switching valve (Vs) was actuated and a stream of 30% HClO4 in DMF–water (3 + 7 v/v) at 0.7 ml min21 was passed through the flow cell to elute the complexes from the C18, which was thus made ready for the next sample. A blank was prepared by replacing the sample with water and its first derivative spectrum was recorded over the same range.Mixing tubes and reactors were made of Teflon and pump tubes of vinyl. Determination of Cobalt in Steels An accurately weighed sample (0.05–0.10 g) was added to about 10 ml of hydrochloric acid (1 + 1) and refluxed gently until complete dissolution. A volume of 3.4 ml of nitric acid (1 + 1) was then added and boiling continue until the acid was removed. Next, the solution was diluted to an appropriate volume (100–250 ml).Aliquots (10–12.5 ml) of this solution were analysed by using the recommended procedure. The standard addition method was used and the results were obtained by extrapolation. Results and Discussion Cobalt and copper form two yellow 1:2 complexes with PPT (l = 430 nm, e = 1.35 3 104 l mol21 cm21 for CoII and l = 440 nm, e = 2.17 3 104 l mol21 cm21 for CuII).15 The cobalt complex is stable in perchloric acid up to a concentration of 1.2 m (12%), conditions under which most of the complexes of PPT with other cations are not formed.The optimum order of addition was cobalt, reagent and perchloric acid. This reaction can be carried out with the reagent or with the complex retained on a sorbent material. Thus, in preliminary assays, several sorbent materials were tested in order to find that most appropriate for this purpose. PPT complexes are in cationic form in an acidic medium, so cation-exchange resins (Dowex, Amberlita, SP-Sephadex) were tried; however, neither the reagent nor the complexes were retained.Other sorbent materials, such as aluminium oxide, silica gel and C18, were tested, but only C18 effectively retained both the reagent and the complexes. Of the buffered eluents tested, 30% perchloric acid in DMF–water proved the most efficient for removing the complexes from the C18. The next task was to select the most advantageous experimental procedure. If the reagent was retained on C18 first and then metal ions (in perchloric acid) were passed through it, the manifold provided a poor response because the C18 was saturated with the reagent, so the subsequent passage of the sample through the sensor gave an inadequately sensitive signal.The complexes were thus formed in the flow and then retained on C18. The reactant addition sequence providing the highest sensitivity was sample, PPT and perchloric acid (Fig. 1). A switching valve was used instead of an injection valve, in order to a ensure a higher reproducibility and sampling rate and lower sample consumption.Spectral Features Because absorbance wavelength scans did not allow resolution of the signals provided by the reagent and the complexes, preliminary assays were carried out in order to determine the most suitable derivative order for this purpose. The maximum resolution was achieved by using first-derivative spectra (Fig. 2). Fig. 2 shows the effect of the concentration of the two ions on the first-derivative spectra, and also the simultaneous effect of the two ions in a mixture.No synergistic effect from the mixture was observed, so their responses were independent of the concentration of each other. The maximum signal measured corresponded to the saturation time for the C18 (35 s), which was adopted for analytical measurements. Derivative scans were carried out over the range 200–500 nm, which proved adequate. The optimum measuring wavelengths for CoII and CuII were 290 and 430 nm, respectively; however, the sensitivity of the copper peak was poor, but could be improved by using combined signals over the range 420–442 nm (Fig. 2). Optimization of Variables The optimum reagent concentration was found to be 0.025% m/v PPT in DMF–water (3 + 7 v/v). The DMF was used to ensure complete dissolution of the reagent; the same amount was used in the eluent in order to maximize the peak height and Fig. 1 Manifold used for the simultaneous determination of CoII and CuII with PPT.R1, R2, R3 = reactors; Vs switching valve; W1, W2, W3 = waste lines; PDA = photodiode-array detector. Fig. 2 Spectra obtained from (a) a fixed concentration of CoII (3 mg ml21) and various concentrations of CuII between 1 and 6 mg ml21; and (b) a fixed concentration of CuII (3 mg ml21) and various concentrations of CoII between 1 and 6 mg ml21. 86 Analyst, January 1997, Vol. 122Conclusions The proposed method for the automatic simultaneous determination of CoII and CuII based on flow-through sensors offers interesting features such as simplicity, rapidity, low cost and flexibility.The FI technique, in its different modes, solves the chemical problems involved, the detector provides the discrimination required for the simultaneous determination and the flow-through sensor improves the sensitivity and selectivity. The system is thus highly automated and readily adaptable to a number of other chemical systems.Financial support provided by the DGICyT (PB95-0977) is gratefully acknowledged. References 1 Ruz, J., L�azaro, F., and Luque de Castro, M. D., J. Autom. Chem., 1988, 10, 15. 2 Marshall, M. A., and Mottola, H. A., Anal. Chem., 1985, 57, 729. 3 Sakamoto-Arnold, C. M., and Johnson, K. S., Anal. Chem., 1987, 59, 1789. 4 Van Zoonen, P., Kamminga, D. A., Gooijer, C., Velthorst, N. H., and Frei, R. W., Anal. Chim. Acta, 1985, 167, 249. 5 Toyoda, T., Kuan, S. S., and Guilbault, G.G., Anal. Chem., 1985, 57, 2346. 6 Ditzler, M. A., Perre-Jacques, H., and Harrington, S. A., Anal. Chem., 1986, 58, 195. 7 Campbell, M. J. M., Coord. Chem. Rev., 1975, 15, 279. 8 Singh, R. B., Garg, B. S., and Singh, R. P., Talanta, 1978, 25, 619. 9 Cano Pav�on, J. M., Microchem. J., 1981, 26, 155. 10 Cristofol de Alcaraz, E., S�anchez Rojas, F., and Cano Pav�on, J. M., Fresenius’ J. Anal. Chem., 1991, 340, 175. 11 Valc�arcel, M., and Luque de Castro, M.D., Analyst, 1993, 118, 593. 12 Luque de Castro, M. D., and Valc�arcel, M., Trends Anal. Chem., 1991, 10, 114. 13 L�azaro, F., Luque de Castro, M. D., and Valc�arcel, M., Anal. Chim. Acta, 1988, 214, 217. 14 L�azaro, F., Luque de Castro, M. D., and Valc�arcel, M., Anal. Chim. Acta, 1989, 219, 231. 15 Cristofol de Alcaraz, E., S�anchez Rojas, F., and Cano Pav�on, J. M., Talanta, 1991, 38, 445. Paper 6/05451J Received August 5, 1996 Accepted October 7, 1996 Table 5 Compositions of steel samples Element (%) Steel A Steel B Steel C Steel D Ci 0.064 0.019 0.035 0.755 Si 0.62 0.29 0.29 0.316 Mn 1.50 1.58 1.27 0.290 Sn 0.024 0.022 0.021 — Ni 8.714 9.42 8.67 — Cu 0.287 0.493 0.424 — Cr 18.33 18.75 17.31 4.233 P 0.032 0.037 0.035 0.016 S 0.005 0.006 0.002 0.009 Mo 0.385 0.492 0.466 0.0953 Ti 0.663 0.002 0.003 — Nb 0.021 0.011 0.010 — Co 0.208 0.208 0.222 4.90 Table 6 Determination of cobalt and copper in steels Nominal content (%) Content found (%) Sample CoII CuII CoII CuII A 0.208 0.287 0.218 ± 0.001 0.320 ± 0.001 B 0.208 0.493 0.225 ± 0.0018 0.506 ± 0.001 C 0.222 0.424 0.280 ± 0.001 0.433 ± 0.004 D 4.90 — 4.63 ± 0.03 — 88 Analyst, January 1997, Vol. 122 Simultaneous Automatic Determination of Trace Amounts of Copper and Cobalt by Use of a Flow-through Sensor and First-derivative Spectrometry Elisa Veredaa, Angel Riosb and Miguel Valcarcel*b a Department of Analytical Chemistry, Faculty of Sciences, University of M�alaga, 29071 M�alaga, Spain b Department of Analytical Chemistry, Faculty of Sciences, University of C�ordoba, 14004 C�ordoba, Spain A straightforward method for the simultaneous determination of cobalt and copper in binary mixtures by use of a diode-array detector accommodated in a flow-through spectrophotometric sensor and pyridoxal-4-phenylthiosemicarbazone as reagent is proposed.The coloured complexes formed give extensively overlapped spectra that can only be resolved from their first derivative.The method relies on integrated preconcentration and detection in the flow cell that allows a high sensitivity in the determination of the two metal ions with detection limits of 0.03 mg ml21 for both and an RSD (n = 11; P = 0.05) of 2.0% for cobalt and 0.9% for copper, at a rate of 36 samples h21. The method was applied to the determination of cobalt and copper in several steel samples. Keywords: Flow injection; spectrophotometry; flow-through sensor; cobalt; copper; steels Immobilization in flow injection (FI) has usually been related to enzymes1 and less often to other reagents, and then mainly for preconcentration2 and/or matrix removal,3 or reaction with the analyte4 and immunoassay.5 The use of immobilization in FI offers many advantages: convenience because manipulation is simpler than that of solutions; simplified manifolds with fewer channels; increased sensitivity; increased stability and selectivity6 resulting from the micro-environment in which the reaction takes place; and reagent economy and compatibility with automated systems.The current trend in applications of immobilization is towards placement of the reaction unit in the detection systems, which measures the change in the immobilized reagent or in the surrounding solution on reaction with the analyte. To ensure proper performance, the reaction system must meet several requirements: (a) reversibility (or at least a high capacity if the reagent is consumed); (b) stability of the immobilized reagent in the reaction medium; (c) fast kinetics; and (d) compatibility between the support and the detection system used.Thiosemicarbazones and phenylthiosemicarbazones generally react as chelating ligands for transition metal ions by bonding through the sulfur and hydrazine nitrogen atoms.7,8 As a rule, phenylthiosemicarbazones are more effective than thiosemicarbazones because their complexes have much higher absorptivities9 (çPPT–Co = 2.17 3 104, çPPT–Cu = 1.35 3 104 l mol21 cm21).This paper reports on the simultaneous determination of CoII and CuII with pyridoxal-4-phenyl-3-thiosemicarbazone (PPT) immobilized on a sorbent material, C18. The method relies on previous work,10 a very selective method where PPT was used for the FI spectrophotometric determination of cobalt based on the formation of a highly stable complex between cobalt and PPT in a strongly acidic medium; the method was not very sensitive and was subject to perturbation from copper, which interfered up to a copper-to-cobalt ratio of 3.This drawback was exploited for the simultaneous determination of cobalt and copper. Since the coloured complexes formed give extensively overlapped spectra, we used derivative spectrophotometry to resolve them from their first derivatives. In this work, in addition to resolving the binary mixture of interest, we improved the sensitivity (determination limits 0.3 and 0.1 mg ml21 for cobalt and copper, respectively) by use of a flow manifold integrating preconcentration and detection in a sorbent material packed in a flow-cell.The improved method relies on the recently developed flow-through sensor technology, 11,12 by which retention and detection are integrated in an FI system.13,14 In this case, successive passage of the complexes, previously formed in the flowing stream, and eluent through the flow cell and continuous photometric monitoring of the process provided the analytical information needed to determine cobalt and the copper.The proposed method was used for the automatic on-line determination of cobalt and copper in steels, with satisfactory results. Experimental Reagents All chemicals used were of analytical-reagent grade and solutions were prepared using high-purity water from a Millipore (Bedford, MA, USA) Milli-Q Water purification system. Stocks standard solutions of cobalt(ii) and copper(ii) were prepared from the nitrate and sulfate, respectively, and standardized complexometrically.Working standards solutions were prepared by appropiate dilution daily as required. A 0.025% solution of PPT was prepared by dissolving 0.025 g of the reagent in 30 ml of N,N-dimethylformamide (DMF) and diluting to 100 ml with water. The reagent was synthesized following a procedure described elsewhere.15 A 30% solution of perchloric acid in DMF–water (3 + 7 v/v) was used as eluent.A 3 m perchloric acid solution was also used. C18 material from a Sep-Pak C18 cartridge (Millipore Waters, Milford, MA, USA) was used for the retention and detection of cobalt and copper complexes in the flow cell. Instrumentation and Procedure A Hewlett-Packard (Avondale, PA, USA) Model 8452A diodearray detector interfaced to a Vectra ES/12 computer which delivered results through an HP Think-Jet printer was used. The flow manifold consisted of a Gilson (Worthington, OH, USA) Minipuls-3 peristaltic pump, a Rheodyne (Cotari, CA, USA) Type 50 six-port rotary valve and a Hellma (Jamaica, NY, USA) OS 0.200 flow cell with a 2 mm optical path length (Fig. 1). A schematic diagram of the FI system is shown in Fig 1. It operated as sample flowing at Analyst, January 1997, Vol. 122 (85–88) 85ensure complete elution of the complexes. The influence of the perchloric acid concentration was also studied. The results showed that the measured signals remained constant over the range 1.5–4.5 m perchloric acid.A concentration of 3 m was finally used in order to prevent interferences from foreign ions. Obviously, once the different streams have merged, the acid concentration in the reaction plug will have decreased by dilution. The influence of the flow injection variables (flow rates, reactor lengths) was studied; the optimum values obtained are summarized in Table 1. Under these final experimental conditions, a sampling rate of 35 h21 was achieved.Calibration graph, Sensitivity and Precision Under the optimum experimental conditions, a linear calibration graph was obtained for the individual determination of each ion. The figures of merit of these calibration graphs are summarized in Table 2. The detection and determination limits defined as the concentrations of analyte giving signals equivalent to three and ten times, respectively, the standard deviation of the blank plus the net blank intensity were calculated.The values found for each metal are also given in Table 2. The precision (RSD, n = 11) achieved in the resolution of a mixture of 2 mg ml21 CoII and 5 mg ml21 CuII was 2.0 and 0.9%, respectively. In order to check for potential synergistic effects of the mixtures of two ions, various synthetic mixtures were readily resolved by using pertinent calibration graphs (Table 3). As can be seen, mixtures of the two ions can be resolved with satisfactory results.The average errors obtained were about 4.6 and 8.8% for cobalt and copper, respectively. These errors can be considered as acceptable for simultaneous determinations involving neither sophisticated nor expensive instrumentation (e.g., atomic optical techniques). Interferences The potential interference of foreign ions in the determination of cobalt and copper by the proposed method was studied by using a fixed concentration of 2 mg ml21 of both CoII and CuII and various concentrations of each foreign ion up to a maximum ion to CoII or CuII ratio of 225 : 1.A given species was considered to interfere if it resulted in a ±5% variation of the average FI signal, calculated at the established levels of cobalt and copper. The results obtained are given in Table 4. The inferference of FeIII can be significantly lowered by adding sodium fluoride to the medium. This allows one to determine cobalt and copper in iron-containing materials.Determination of Cobalt and Copper in Steels In order to evaluate the effectiveness of the method, a series of recovery experiments were carried out on various steel samples supplied by Acerinox (Algeciras, Spain), with the nominal concentrations (as determined by XRF, with calibration lines obtained using NBS standard steels) given in Table 5. The samples were prepared as described above. The results of the analyses of steels A–D are given in Table 6 as the averages of three individual determinations.Table 1 Optimum values of flow injection variables Sample flow rate 0.90 ml min21 Reagent flow rate 0.35 ml min21 HClO4 flow rate 0.30 ml min21 Eluent flow rate 0.70 ml min21 R1 dimensions 140 cm 3 0.5 mm id R2 dimensions 60 cm 3 0.5 mm id R3 dimensions 35 cm 3 0.5 mm id Table 2 Analytical figures of merit for the determination of cobalt and copper by the proposed method Linear range/ Detection limit/ Determination Analyte lopt/nm Calibration equation* r mg ml21 mg ml21 limit/mg ml21 CoII 292 y = 27.46 3 1024–3.54 3 1023x 0.999 0.3–6 0.03 0.3 CuII 420–442 y = 23.60 3 1024–5.00 3 1023x 0.997 0.1–15 0.03 0.1 * y = dA/dl (A = absorbance); x = ionic concentration of the analyte in mg ml21.Table 3 Resolution of synthetic CoII–CuII mixtures by use of the proposed method Concentration added/ Concentration found/ mg ml21 mg ml21 Sample no. CoII CuII CoII CuII 1 0.5 0.5 0.5 0.4 2 0.5 1.0 0.5 1.0 3 0.5 2.0 0.5 1.9 4 1.0 0.5 1.0 0.6 5 1.0 1.0 1.1 1.1 6 1.0 2.0 1.1 2.0 7 1.0 4.0 1.1 4.0 8 2.0 0.5 2.0 0.4 9 2.0 1.0 2.2 1.2 10 2.0 2.0 2.2 2.2 11 2.0 4.0 2.2 3.7 12 3.0 1.0 3.1 0.9 13 3.0 2.0 3.0 1.8 14 3.0 3.0 2.9 3.0 15 3.0 4.0 3.1 3.7 16 3.0 5.0 3.0 4.7 17 4.0 1.0 4.2 1.1 18 4.0 3.0 3.8 3.3 19 4.0 4.0 3.7 3.9 Table 4 Interferences in the proposed method Tolerated interferent-toanalyte Foreign species molar ratio NaI, KI, BaII, CaII, MgII, PbII, MnII, fluoride, bromide, iodide, chloride, thiocyanate, nitrate, sulfate, thiosulphate, oxalate, FeIII* > 225 CdII, glycine 100 AlIII, ZnII, CrIII 25 NiII 20 MoVI 3 FeIII 2 * With 2000 times more fluoride than CoII and CuII.Analyst, January 1997, Vol. 122 870.90 ml min21 reached reactor R1 (140 cm 30.5 mm id), where it merged with a stream consisting of 0.025% PPT in DMF– water (3 + 7 v/v) circulated at 0.35 ml min21. The mixture reached reactor R2 (60 cm 3 0.5 mm id) and was merged with a stream of 3 m HClO4 flowing at 0.30 ml min21, where the complexes of other metal ions are destroyed.The cobalt and copper complexes were then passed through reactor R3 (35 cm 3 0.5 mm id) in order to increase the reproducibility, and reached the flow cell, where they were retained on C18 packed in the flow cell (preconcentration step). When C18 was saturated (35 s), the first-derivative spectrum was recorded over the range 200–500 nm (detection step) then the switching valve (Vs) was actuated and a stream of 30% HClO4 in DMF–water (3 + 7 v/v) at 0.7 ml min21 was passed through the flow cell to elute the complexes from the C18, which was thus made ready for the next sample.A blank was prepared by replacing the sample with water and its first derivative spectrum was recorded over the same range. Mixing tubes and reactors were made of Teflon and pump tubes of vinyl. Determination of Cobalt in Steels An accurately weighed sample (0.05–0.10 g) was added to about 10 ml of hydrochloric acid (1 + 1) and refluxed gently until complete dissolution.A volume of 3.4 ml of nitric acid (1 + 1) was then added and boiling continue until the acid was removed. Next, the solution was diluted to an appropriate volume (100–250 ml). Aliquots (10–12.5 ml) of this solution were analysed by using the recommended procedure. The standard addition method was used and the results were obtained by extrapolation. Results and Discussion Cobalt and copper form two yellow 1:2 complexes with PPT (l = 430 nm, e = 1.35 3 104 l mol21 cm21 for CoII and l = 440 nm, e = 2.17 3 104 l mol21 cm21 for CuII).15 The cobalt complex is stable in perchloric acid up to a concentration of 1.2 m (12%), conditions under which most of the complexes of PPT with other cations are not formed.The optimum order of addition was cobalt, reagent and perchloric acid. This reaction can be carried out with the reagent or with the complex retained on a sorbent material. Thus, in preliminary assays, several sorbent materials were tested in order to find that most appropriate for this purpose.PPT complexes are in cationic form in an acidic medium, so cation-exchange resins (Dowex, Amberlita, SP-Sephadex) were tried; however, neither the reagent nor the complexes were retained. Other sorbent materials, such as aluminium oxide, silica gel and C18, were tested, but only C18 effectively retained both the reagent and the complexes. Of the buffered eluents tested, 30% perchloric acid in DMF–water proved the most efficient for removing the complexes from the C18.The next task was to select the most advantageous experimental procedure. If the reagent was retained on C18 first and then metal ions (in perchloric acid) were passed through it, the manifold provided a poor response because the C18 was saturated with the reagent, so the subsequent passage of the sample through the sensor gave an inadequately sensitive signal. The complexes were thus formed in the flow and then retained on C18.The reactant addition sequence providing the highest sensitivity was sample, PPT and perchloric acid (Fig. 1). A switching valve was used instead of an injection valve, in order to a ensure a higher reproducibility and sampling rate and lower sample consumption. Spectral Features Because absorbance wavelength scans did not allow resolution of the signals provided by the reagent and the complexes, preliminary assays were carried out in order to determine the most suitable derivative order for this purpose.The maximum resolution was achieved by using first-derivative spectra (Fig. 2). Fig. 2 shows the effect of the concentration of the two ions on the first-derivative spectra, and also the simultaneous effect of the two ions in a mixture. No synergistic effect from the mixture was observed, so their responses were independent of the concentration of each other. The maximum signal measured corresponded to the saturation time for the C18 (35 s), which was adopted for analytical measurements. Derivative scans were carried out over the range 200–500 nm, which proved adequate.The optimum measuring wavelengths for CoII and CuII were 290 and 430 nm, respectively; however, the sensitivity of the copper peak was poor, but could be improved by using combined signals over the range 420–442 nm (Fig. 2). Optimization of Variables The optimum reagent concentration was found to be 0.025% m/v PPT in DMF–water (3 + 7 v/v).The DMF was used to ensure complete dissolution of the reagent; the same amount was used in the eluent in order to maximize the peak height and Fig. 1 Manifold used for the simultaneous determination of CoII and CuII with PPT. R1, R2, R3 = reactors; Vs switching valve; W1, W2, W3 = waste lines; PDA = photodiode-array detector. Fig. 2 Spectra obtained from (a) a fixed concentration of CoII (3 mg ml21) and various concentrations of CuII between 1 and 6 mg ml21; and (b) a fixed concentration of CuII (3 mg ml21) and various concentrations of CoII between 1 and 6 mg ml21. 86 Analyst, January 1997, Vol. 122Conclusions The proposed method for the automatic simultaneous determination of CoII and CuII based on flow-through sensors offers interesting features such as simplicity, rapidity, low cost and flexibility. The FI technique, in its different modes, solves the chemical problems involved, the detector provides the discrimination required for the simultaneous determination and the flow-through sensor improves the sensitivity and selectivity. The system is thus highly automated and readily adaptable to a number of other chemical systems.Financial support provided by the DGICyT (PB95-0977) is gratefully acknowledged. References 1 Ruz, J., L�azaro, F., and Luque de Castro, M. D., J. Autom. Chem., 1988, 10, 15. 2 Marshall, M. A., and Mottola, H. A., Anal. Chem., 1985, 57, 729. 3 Sakamoto-Arnold, C. M., and Johnson, K. S., Anal. Chem., 1987, 59, 1789. 4 Van Zoonen, P., Kamminga, D. A., Gooijer, C., Velthorst, N. H., and Frei, R. W., Anal. Chim. Acta, 1985, 167, 249. 5 Toyoda, T., Kuan, S. S., and Guilbault, G. G., Anal. Chem., 1985, 57, 2346. 6 Ditzler, M. A., Perre-Jacques, H., and Harrington, S. A., Anal. Chem., 1986, 58, 195. 7 Campbell, M. J. M., Coord. Chem. Rev., 1975, 15, 279. 8 Singh, R. B., Garg, B. S., and Singh, R. P., Talanta, 1978, 25, 619. 9 Cano Pav�on, J. M., Microchem. J., 1981, 26, 155. 10 Cristofol de Alcaraz, E., S�anchez Rojas, F., and Cano Pav�on, J. M., Fresenius’ J. Anal. Chem., 1991, 340, 175. 11 Valc�arcel, M., and Luque de Castro, M. D., Analyst, 1993, 118, 593. 12 Luque de Castro, M. D., and Valc�arcel, M., Trends Anal. Chem., 1991, 10, 114. 13 L�azaro, F., Luque de Castro, M. D., and Valc�arcel, M., Anal. Chim. Acta, 1988, 214, 217. 14 L�azaro, F., Luque de Castro, M. D., and Valc�arcel, M., Anal. Chim. Acta, 1989, 219, 231. 15 Cristofol de Alcaraz, E., S�anchez Rojas, F., and Cano Pav�on, J. M., Talanta, 1991, 38, 445. Paper 6/05451J Received August 5, 1996 Accepted October 7, 1996 Table 5 Compositions of steel samples Element (%) Steel A Steel B Steel C Steel D Ci 0.064 0.019 0.035 0.755 Si 0.62 0.29 0.29 0.316 Mn 1.50 1.58 1.27 0.290 Sn 0.024 0.022 0.021 — Ni 8.714 9.42 8.67 — Cu 0.287 0.493 0.424 — Cr 18.33 18.75 17.31 4.233 P 0.032 0.037 0.035 0.016 S 0.005 0.006 0.002 0.009 Mo 0.385 0.492 0.466 0.0953 Ti 0.663 0.002 0.003 — Nb 0.021 0.011 0.010 — Co 0.208 0.208 0.222 4.90 Table 6 Determination of cobalt and copper in steels Nominal content (%) Content found (%) Sample CoII CuII CoII CuII A 0.208 0.287 0.218 ± 0.001 0.320 ± 0.001 B 0.208 0.493 0.225 ± 0.0018 0.506 ± 0.001 C 0.222 0.424 0.280 ± 0.001 0.433 ± 0.004 D 4.90 — 4.63 ± 0.03 — 88 Analyst

ISSN:0003-2654    

DOI:10.1039/a605451j

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Determination of Ammonia in Waste Waters by a Differential pH Method Using Flow Injection Potentiometry and a Nonactin-Based Sensor Hongda Shen, Terence J. Cardwell and Robert W. Cattrall* Centre For Scientific Instrumentation, School of Chemistry, La Trobe University, Melbourne, Victoria 3083, Australia A method is described for the determination of total ammoniacal nitrogen in waters using an ammonium ion-selective sensor based on nonactin. The flow injection method relies on the measurement of the potential of the sensor at two pH values, 6.0 and 9.4, and the use of a mathematical expression to calculate the total ammoniacal nitrogen concentration.The method also compensates for the interference from moderate concentrations of other cations such as potassium and sodium. Keywords: Ammonia determination; waste waters; nonactin sensor; flow injection The determination of dissolved ammonia is important in many clinical, industrial and environmental laboratories and the methods most often used are spectrophotometric,1–5 potentiometric6 –10 and conductimetric,11–13 with either batch procedures or flow injection.Methods using fluorimetry14,15 chemiluminescence16 and optosensing17 are also documented. Many of these methods employ a gas diffusion membrane to separate ammonia from interferences. The most common potentiometric procedure involves the ammonia probe, which is relatively interference free and can detect ammonia as low as 0.01 mg l21.18 There are, however, some limitations with the ammonia probe and one of these is the slow response at low concentrations.Also, volatile amines can pass through the membrane and cause a severe interference.9 A third difficulty involves sample size, since the probe is a macrodevice and it is difficult to use when the sample volume is limited. One of the most common flow injection techniques for ammonia involves a gas diffusion cell9 in which the sample stream is separated from the receiving stream by a gaspermeable membrane such as Teflon tape.Ammoniacal nitrogen (i.e., NH3 and NH4 +) is transferred to the receiving stream as molecular ammonia in a similar way to the ammonia probe. Detection in the receiving phase is spectrophotometric using an acid–base indicator or potentiometric using an ammonium ion sensor based on nonactin. The nonactin-based ammonium ion sensor shows high selectivity for the ammonium ion in the presence of many cations19 but suffers from severe interference from alkali metal ions, especially potassium.Hence the use of the nonactin-based sensor directly in samples at acidic pH to determine total ammoniacal nitrogen as the ammonium ion is subject to interference from these ions. The aim of this work was to develop a method for determining dissolved ammonia as the ammonium ion directly in sample solutions using the nonactin-based sensor and to employ a chemometric approach to eliminate any interference from other cations.This was achieved by making measurements at two pH values. Experimental Reagents and Chemicals Ammonium, potassium and sodium chlorides (AnalaR grade) (BDH, Poole, Dorset, UK) were used to prepare standard solutions and simulated samples. Methylamine and triethylamine (laboratory-reagent grade) (BDH) were dissolved in water for the interference studies. The pH 6 adjustment buffer (buffer 1) was prepared by mixing 8.4 g of lithium chloride (laboratory-reagent grade), (Ajax, Sydney, Australia) 4.2 g of lithium hydroxide (laboratory- reagent grade) (BDH) and 6.0 ml of glacial acetic acid (analytical-reagent grade) (Mallinckrodt, Paris, KY, USA) and diluting to 1 l.The pH 9.4 adjustment buffer (buffer 2) was made by mixing the same amounts of lithium chloride and hydroxide as above with 7.5 g of boric acid (AnalaR grade) (BDH) and diluting to 1 l. For the studies with the gas diffusion cell, 1 mol l21 sodium hydroxide (laboratory-reagent grade) (Mallinckrodt), containing 0.05 mol l21 EDTA (laboratory-reagent grade) (BDH) was used for conversion into ammonia and the receiving solution was 0.01 mol l21 barium acetate (AnalaR grade) (BDH) plus 0.01 mol l21 acetic acid.All solutions were prepared using > 15 MW cm NANO-pure water (Barnstead, Dubuque, IA, USA). Nonactin-based Sensor The conventional ammonium ion-selective electrode was prepared by casting a membrane19 using a mixture consisting of 1 mg of nonactin (Fluka, Buchs, Switzerland), 66 mg of bis(1- butylpentyl) adipate (BBPA) (Fluka) and 33 mg of poly(vinyl chloride) (Fluka).This was dissolved in the minimum amount of purified tetrahydrofuran, sonicated for 10 min and then cast in a glass ring on a glass plate. After evaporation of the solvent, the electrode was constructed by gluing a cut disc of the membrane (about 0.2 mm thick) on to a plastic barrel. The inner reference electrode was Ag/AgCl in 0.1 mol l21 ammonium chloride saturated with silver chloride.The electrode was conditioned in 0.1 mol l21 ammonium chloride for 1 d before use. Potential measurements with this electrode were made using an Orion Ionanalyzer, Model 901 (Orion Research, Boston, MA, USA) with an Orion double-junction reference electrode with an outer junction filling solution consisting of 1.0 mol l21 lithium chloride. Potentials were monitored and recorded with an interfaced Apple IIe microcomputer. pH measurements were made with an Orion Ionanalyzer, Model EA940, and a combination pH–reference electrode (Ionode, Brisbane, Australia).Flow Injection The flow injection manifold used in this study is shown in Fig. 1(A) and Fig. 1(B) shows the manifold used with the gas Analyst, January 1997, Vol. 122 (89–93) 89diffusion cell to validate the results from the differential pH procedure. Both manifolds employed a Gilson (Worthington, OH, USA) Minipuls 2 peristaltic pump and manifold A incorporated two Teflon rotary valves [Rheodyne (Cotati, CA, USA) Type 50], one of which was used for sample injection (20 ml) and the other for switching between the two pH adjustment buffers.All flow lines were Teflon tubing of 0.5 mm id. The gas diffusion cell in manifold B was similar in design to one described in a previous paper5 and used Teflon plumber’s tape as the membrane. It is essential that the flow rates on each side of the membrane are identical to avoid rupture of the membrane.Thus, the water carrier and 1.0 mol l21 sodium hydroxide (also containing 0.05 mol l21 EDTA) flow rates were each 0.9 ml min21. The receiving solution consisted of 0.01 mol l21 barium acetate plus 0.01 mol l21 acetic acid and had a flow rate of 0.18 ml min21. This receiving solution ensured that all ammonia diffusing across the membrane was converted into ammonium ion. The sample size for the gas diffusion measurements was 40 ml and the throughput was 40 samples h21.The detector in each manifold was an improved version20 of our coated-wire multi-ion potentiometric detector.5 This consists of two Perspex blocks separated by a Teflon gasket. The bottom block contains silver wire surfaces which can be coated with the membrane mixture described above to form ammonium ion-selective coated-wire sensors. Several sensors of the same type were made to allow for redundancies. This detector has a separate reference arm consisting of an Ag/AgCl reference element with a reference stream of 0.15 mol l21 lithium chloride flowing at 0.9 ml min21.The potentiometric detector was interfaced to a 486 DX-33 PC through a 12-bit A/D card (Boston Technology, Boston, MA, USA). FCS software (A-Chem Technologies, Melbourne, Australia) was used for data acquisition and processing. Theory The pKb for ammonia is about 4.7 and so at pH < 7 all ammoniacal nitrogen exists as the ammonium ion. At around pH 9.3, there is about 50% ammonia and 50% ammonium ion.At pH > 11, only ammonia is present. Because the nonactin ion-selective electrode responds only to the ammonium ion and not to ammonia, the potential in a solution will decrease with increasing pH. If interfering ions such as the potassium ion are present, the contribution to the electrode potential due to the interfering ion should not change, since the ion is unaltered with change in pH. The actual potential, E1, in a solution is given by the following equation derived from the Nicolsky equation: E1 = constant + slog([NH4 +]+ S kij pot[M+]) (1) where s is the Nernst factor or electrode slope factor (mV per concentration decade), [NH4 +] and [M+] are the concentrations of the ammonium ion and an interfering monovalent cation, respectively, and kij pot is the selectivity coefficient.The term Skij pot [M+] in eqn. (1) should not change with pH, provided that the ionic strength of the solution remains constant, but the concentration of the ammonium ion will change. In this work, the potentials of the nonactin electrode were measured in samples of constant ionic strength adjusted to two pH values, 6.0 and 9.4.At pH 6.0, the potential is given by eqn. (1) where, in this case, [NH4 +] is equal to the total concentration of ammoniacal nitrogen, [NH4 +]T. However, at pH 9.4, some of the ammonium ion will be converted into ammonia and the potential, E2, is given by the equation E2 = constant + s log(d[NH4 +]T+ Skij pot [M+]) (2) where d is the fraction of the total ammoniacal nitrogen which is present as the ammonium ion at pH 9.4.Eqns. (1) and (2) are solved simultaneously to give eqn. (3), which allows the concentration of the total ammoniacal nitrogen to be determined. [NH4 +]T = [10(E12constant)/s 2 10(E22constant)/s]/(12d) (3) Results and Discussion Batch Experiments Some initial experiments were conducted with a conventional ammonium ion-selective electrode to determine the response characteristics of the nonactin membrane in solutions of different pH in the absence of the ammonium ion.These experiments were carried out in 0.1 mol l21 lithium chloride containing 0.001 mol l21 potassium chloride and the pH was adjusted with either dilute hydrochloric acid or lithium hydroxide solution. It was found that the potential changed by less than 10 mV in the pH range 5–10. More specifically, it was found that the potential in the pH 6.0 adjustment buffer was a few mV higher than that in the pH 9.4 adjustment buffer and that this difference can be compensated for by adding 5 mmol l21 potassium chloride to the pH 9.4 buffer.This is important because it is necessary in the flow injection method to have baselines for the two buffers which are identical. The conventional electrode was calibrated using the two pH adjustment buffers by mixing equal volumes of an ammonium ion standard and the buffer.The two calibration graphs are shown in Fig. 2. The electrode shows Nernstian behaviour in each buffer over the concentration range 1021–1024 mol l21. The two calibration lines are separated by a potential value which is related to the fraction of the total ammonium ion concentration which has been converted into ammonia on going from pH 6.0 to 9.4. The fraction of the ammonium ion at pH 9.4 is obtained by solving eqns. (1) and (2) (without the terms for an interfering ion) simultaneously: d = 10(E9.42E6.0)/s (4) where E6.0 = E1 and E9.4 = E2.The value of d is calculated by averaging values for several points on the abscissa and is used in the calculation of the total ammoniacal nitrogen concentration. The other way, of course, to calculate d for a particular pH is to use the expression for Kb, but the value of Kb depends on Fig. 1 Flow injection manifolds. A, Differential pH method; and B, gas diffusion. 90 Analyst, January 1997, Vol. 122the temperature, which must be known, and the pH must be determined accurately, whereas these are not requirements for the above method. The use of eqn. (3), for a constant ionic strength, to determine the concentration of total ammoniacal nitrogen in some synthetic samples was tested in the presence of the potassium ion, which is the most severe interferent with the nonactin electrode (E1 is the potential in the sample adjusted to pH 6.0 and E2 is the potential in the sample adjusted to pH 9.4).The results are given in Table 1. It can be seen that for samples in which the potassium ion concentration is similar to the total ammoniacal nitrogen concentration, there is excellent agreement between the expected and found values. However, in a sample which contained a high excess of potassium compared with ammoniacal nitrogen, the agreement is poor. The interference from organic amines (e.g., methylamine, nbutylamine, triethylamine) was examined and found to be negligible.There are two reasons for this: the low interference of the protonated form of an organic amine with the nonactinbased sensor9 and the fact that most organic amines are stronger bases than ammonia and so only very small amounts of the free base are formed at pH 9.4. It should be mentioned that the batch procedure is time consuming and tedious, particularly for solutions which contain low concentrations of the ammonium ion, since these require several minutes to reach the steady-state potential.These results, however, were encouraging enough to extend the work to developing a flow injection procedure and to test it on some real samples. Flow Injection The manifold designed for the flow injection procedure, shown in Fig. 1(A), operates in the following way. The sample (20 ml) is injected into a water carrier stream (flow rate, 0.9 ml min21) and then mixed initially with the pH 6.0 adjustment buffer which is selected through the second rotary valve and which is flowing continuously (flow rate 0.9 ml min21) into the system.The buffer selection valve is then switched to the pH 9.4 buffer (containing 5 mmol l21 KCl), which then flows continuously through the system. A second injection of the same sample is then made and mixed with this stream. Peak height measurements are made for the sample in each buffer. This procedure achieves a sampling rate of 30 samples h21. Fig. 3 shows calibration curves for the ammonium ion for each pH adjustment buffer.Six calibrations were carried out at each pH value using triplicate injections of each standard. It can be seen that linearity is achieved down to 1024 mol l21 and the lines are separated in the same way as seen in Fig. 2 for the batch measurements. The slopes of the linear sections of the two lines are identical (58.4 ± 0.6 and 58.3 ± 0.9 mV per concentration Table 1 Results for batch method for synthetic samples E/mV Calculated* Expected Sample [NH4 +]T/ [NH4 +]T/ [K+]/ No.pH 6.0 pH 9.4 mmol l21 mmol l21 mmol l21 1 281.8 299.3 0.504 0.500 0.500 2 260.8 269.9 0.725 0.500 5.00 3 228.5 248.4 4.89 5.00 0.500 4 224.8 242.9 5.34 5.00 5.00 * Ammonium ion concentration calculated from eqn. (3). Fig. 2 Batch calibration. 8, pH 6.0; and ., pH 9.4. Table 2 Results for flow injection method for synthetic samples Peak height*/mV Calculated If uncorrected Expected Sample [NH4 +]T/ [NH4 +]T/ [NH4 +]T/ [K+]/ No.pH 6.0 pH 9.4 mmol l21† mmol l21‡ mmol l21 mmol l21 1 94.9 74.5 0.244 (3.7) 0.294 0.25 0.25 2 111.0 87.4 0.504 (3.9) 0.555 0.50 0.25 3 149.9 124.5 2.43 (2.0) 2.57 2.50 0.25 4 100.7 83.9 0.269 (1.9) 0.370 0.25 0.50 5 114.9 93.4 0.555 (3.0) 0.647 0.50 0.50 6 150.3 125.8 2.42 (3.6) 2.61 2.50 0.50 7 121.1 113.6 0.317 (9.7) 0.826 0.25 2.50 8 127.4 117.1 0.530 (6.8) 1.06 0.50 2.50 9 154.0 132.8 2.56 (3.6) 3.02 2.50 2.50 * Mean peak height of triplicate injections with relative mean deviation < 1%.† Relative mean deviation (%) in parentheses, calculated for each sample using the mean of each pH data set and comparing it with all values of the other pH data set. ‡ Calculated using the pH 6.0 data only and showing the effect of the potassium ion interference. Fig. 3 Calibration in each buffer by flow injection using manifold A. 8, pH 6.0; and ., pH 9.4. Analyst, January 1997, Vol. 122 91decade for pH 6.0 and 9.4, respectively). Points were chosen in the linear range along the abscissa for the calculation of the average value of d.Initially, some synthetic samples also containing the potassium ion were tested as with the batch method and the results are given in Table 2. The results show that the method can completely eliminate the potassium ion interference in all cases except when the potassium ion concentration is much higher than (10 times) the ammonium ion concentration. This is because the peak height (potential) difference between the two buffers is relatively small, which produces a large error in the ammonium ion concentration.A ‘rule of thumb’ approach can be applied to this since it has been observed that, if the potential difference is smaller than, say, 8 mV, the result is unreliable. Analysis of Real Samples Some industrial waste water and polluted river water samples were obtained from two laboratories and analysed using the differential pH method and the results were compared with those obtained from the gas diffusion technique and, in some cases, the ammonia probe, as shown in Table 3.If the samples were initially very acidic or alkaline, the pH was adjusted to 5–7 with hydrochloric acid or lithium hydroxide prior to the determination. As seen before, excellent agreement between results is obtained when the peak height difference between the two pH values is large, i.e., > 8 mV (samples 1, 5, 6 and 11–18). Hence the method works well with natural water samples where high concentrations of cations such as potassium are absent.Very Low Ammonium Ion Concentrations When the ammonium ion concentration is below the linear range of the calibration plot shown in Fig. 3, it has been observed that there is a linear relationship between the peak height and the concentration21 (and not the logarithm of the concentration), as shown in Fig. 4. The differential pH method was therefore tested in this case. The ammoniacal nitrogen concentration is given by the equation [NH4 +]T = (E6.0 2 E9.4)/[s(12d)] (5) The results for some synthetic samples are given in Table 4 and excellent agreement is seen between the expected and found values. These particular samples contained high concentrations of the sodium ion and any interference was eliminated up to a ratio of [Na+] to [NH4 +]T of 4000 : 1.Hence the differential pH method can be applied with ammoniacal nitrogen concentrations down to about 1026 mol l21 in the presence of high concentrations of an interfering ion such as sodium.The criterion for determining initially if the ammoniacal nitrogen concentration is in the low range is the peak height value in the pH 6.0 adjustment buffer. If this is small (i.e., < 20 mV), the Fig. 4 [NH4 +] calibration at low concentration. Table 3 Results for the analysis of real samples Peak height†/mV Calculated Gas diffusion NH3 Sample [NH4 +]T/ [NH4 +]T/ probe/ Certified/ No.* pH 6.0 pH 9.4 mg l21‡ mg l21 mg l21 mg l21 1 95.3 88.6 1.24 (5.9) 1.70 (2.0) 1.50 2§ 71.7 69.3 6.29 (16) 3.66 (2.4) 3.63 3 142.0 137.6 7.66 (10) 7.37 (1.0) 6.86 4 157.2 155.0 7.27 (24) 4.44 (1.4) 4.07 5 142.9 133.8 15.2 (5.7) 15.3 (2.3) 14.3 6§ 141.2 122.2 590 (2.9) 570 (3.9) 550 7 150.1 149.5 1.90 (120) 0.12 (4.1) 0.11 8 154.9 153.4 5.70 (17) 0.05 (11) 0.04 9 155.8 155.4 1.31 (85) 0.02 (13) 0.03 10 34.3 29.2 0.14 (8.2) 0.07 (5.8) 0.06 11 136.0 119.3 20.1 (3.4) 19.8 (4.2) 19.8 12 127.6 111.4 14.2 (3.2) 14.6 (2.0) 14.3 13 67.1 56.0 1.38 (4.6) 1.48 (3.5) 1.4 14 67.8 55.3 1.54 (4.2) 1.48 (3.9) 1.4 15 82.9 68.6 2.96 (3.8) 3.17 (2.2) 2.9 16 83.6 68.5 3.15 (4.8) 2.24 (1.9) 3.0 17 83.2 67.9 3.14 (3.0) 3.30 (2.7) 3.1 18 83.8 67.8 3.33 (3.9) 3.29 (3.4) 3.2 * Samples 1–10 were industrial waste water samples and 11–18 were polluted river water samples.† Mean peak height of triplicate injections with relative mean deviation < 1%. ‡ Relative mean deviation (%) in parentheses (n = 3).§ Samples were diluted 20-fold before analysis. Table 4 Results for synthetic samples of low ammoniacal nitrogen concentration Peak height/mV* Calculated Expected Sample [NH4 +]T/ [NH4 +]T/ [Na+]/ No. pH 6.0 pH 9.4 mmol l21† mmol l21 mmol l21 1 4.8 3.2 1.14 (9.9) 1.25 0.1 2 7.5 3.7 2.73 (5.2) 2.50 0.1 3 13.2 6.1 5.10 (4.5) 5.00 0.1 4 12.5 9.1 2.44 (3.6) 2.50 1 5 16.2 10.3 4.23 (7.9) 5.00 1 6 20.1 16.6 2.51 (1.5) 2.50 10 * Mean peak height of triplicate injections with relative mean deviation < 1 mV.† Relative mean deviation (%) in parentheses. 92 Analyst, January 1997, Vol. 122ammoniacal nitrogen concentration can be assumed to be lower than 1025 mol l21. Conclusion The proposed differential pH flow injection method for the determination of ammoniacal nitrogen using the nonactin-based ammonium ion sensor provides a simple and rapid procedure without the need to have a gas diffusion cell to separate interfering cations.The method is able to compensate for the interference from other cations provided that their concentrations are not extremely high relative to the ammoniacal nitrogen concentration. It has been demonstrated that the method is particularly reliable with natural water samples and can determine ammoniacal nitrogen concentrations as low as 1026 mol l21. We thank BHP and Monash University Water Studies Centre for providing the real samples. H.S. is grateful to La Trobe University for a Postgraduate Scholarship and we acknowledge the Australian Research Council for financial assistance.References 1 Van der Linden, W. E., Anal. Chim. Acta, 1983, 151, 359. 2 Nakata, R., Kawamula, T., Sakashita, H., and Nitta, A., Anal. Chim. Acta, 1988, 208, 81. 3 Van Son, M., Schothorst, R. C., and Den Boef, G., Anal. Chim. Acta, 1983, 153, 271. 4 Schulze, G., Liu, C. Y., Brodowski, M., Elsholz, O., Frenzel, W., and Moller, J., Anal. Chim. Acta, 1988, 214, 121. 5 Hauser, P. C., Tan, S. S., Cardwell, T. J., Cattrall, R. W., and Hamilton, I. C., Analyst, 1988, 113, 1551. 6 Alegret, S., Alonso, J., Bartroli, J., and Martinez-Fabregas, E., Analyst, 1989, 114, 1443. 7 Balconi, M. L., Sigon, F., Ferraroli, R., and Realini, F., Anal. Chim. Acta, 1988, 214, 367. 8 Meyerhoff, M. E., and Fraticelli, Y. M., Anal. Lett., 1981, 14, 415. 9 Lee, H. L., and Meyerhoff, M. E., Analyst, 1985, 110, 371. 10 Han, W., and Fan, L., Fenxi Huaxue, 1986, 14, 387. 11 Rodrigues Rohwedder, J. J., and Pasquini, C., Analyst, 1991, 116, 841. 12 Cardoso de Faria, L., and Pasquini, C., Anal. Chim. Acta, 1991, 245, 183. 13 Pasquini, C., and Cardoso de Faria, L., Anal. Chim. Acta, 1987, 193, 19. 14 Zhang, G., and Dasgupta, P. K., Anal. Chem., 1989, 61, 408. 15 Mikasa, H., Motomizu, S., and Toei, K., Bunseki Kagaku, 1985, 34, 518. 16 Kraus, P. R., and Crouch, S. R., Anal. Lett., 1987, 20, 183. 17 Jeppesen, M. T., and Hansen, E. H., Anal. Chim. Acta, 1988, 214, 147. 18 Orion Model 95-12 Ammonia Electrode Instruction Manual, Orion Research, Boston, 1990. 19 Ionophores for Ion Selective Electrodes and Optodes, Fluka, Buchs, 1991. 20 Gregorio, C. G., Cardwell, T. J., and Cattrall, R. W., Anal. Methods Instrum., submitted for publication. 21 Potentiometric Water Analysis, ed. Midgley, D., and Torrance, K., Wiley, Chichester, 2nd edn., 1991, p. 448. Paper 6/05155C Received July 23, 1996 Accepted September 9, 1996 Analyst, January 1997, Vol. 122 93 Determination of Ammonia in Waste Waters by a Differential pH Method Using Flow Injection Potentiometry and a Nonactin-Based Sensor Hongda Shen, Terence J. Cardwell and Robert W. Cattrall* Centre For Scientific Instrumentation, School of Chemistry, La Trobe University, Melbourne, Victoria 3083, Australia A method is described for the determination of total ammoniacal nitrogen in waters using an ammonium ion-selective sensor based on nonactin. The flow injection method relies on the measurement of the potential of the sensor at two pH values, 6.0 and 9.4, and the use of a mathematical expression to calculate the total ammoniacal nitrogen concentration. The method also compensates for the interference from moderate concentrations of other cations such as potassium and sodium.Keywords: Ammonia determination; waste waters; nonactin sensor; flow injection The determination of dissolved ammonia is important in many clinical, industrial and environmental laboratories and the methods most often used are spectrophotometric,1–5 potentiometric6 –10 and conductimetric,11–13 with either batch procedures or flow injection. Methods using fluorimetry14,15 chemiluminescence16 and optosensing17 are also documented.Many of these methods employ a gas diffusion membrane to separate ammonia from interferences. The most common potentiometric procedure involves the ammonia probe, which is relatively interference free and can detect ammonia as low as 0.01 mg l21.18 There are, however, some limitations with the ammonia probe and one of these is the slow response at low concentrations.Also, volatile amines can pass through the membrane and cause a severe interference.9 A third difficulty involves sample size, since the probe is a macrodevice and it is difficult to use when the sample volume is limited. One of the most common flow injection techniques for ammonia involves a gas diffusion cell9 in which the sample stream is separated from the receiving stream by a gaspermeable membrane such as Teflon tape.Ammoniacal nitrogen (i.e., NH3 and NH4 +) is transferred to the receiving stream as molecular ammonia in a similar way to the ammonia probe. Detection in the receiving phase is spectrophotometric using an acid–base indicator or potentiometric using an ammonium ion sensor based on nonactin. The nonactin-based ammonium ion sensor shows high selectivity for the ammonium ion in the presence of many cations19 but suffers from severe interference from alkali metal ions, especially potassium.Hence the use of the nonactin-based sensor directly in samples at acidic pH to determine total ammoniacal nitrogen as the ammonium ion is subject to interference from these ions. The aim of this work was to develop a method for determining dissolved ammonia as the ammonium ion directly in sample solutions using the nonactin-based sensor and to employ a chemometric approach to eliminate any interference from other cations.This was achieved by making measurements at two pH values. Experimental Reagents and Chemicals Ammonium, potassium and sodium chlorides (AnalaR grade) (BDH, Poole, Dorset, UK) were used to prepare standard solutions and simulated samples. Methylamine and triethylamine (laboratory-reagent grade) (BDH) were dissolved in water for the interference studies. The pH 6 adjustment buffer (buffer 1) was prepared by mixing 8.4 g of lithium chloride (laboratory-reagent grade), (Ajax, Sydney, Australia) 4.2 g of lithium hydroxide (laboratory- reagent grade) (BDH) and 6.0 ml of glacial acetic acid (analytical-reagent grade) (Mallinckrodt, Paris, KY, USA) and diluting to 1 l.The pH 9.4 adjustment buffer (buffer 2) was made by mixing the same amounts of lithium chloride and hydroxide as above with 7.5 g of boric acid (AnalaR grade) (BDH) and diluting to 1 l. For the studies with the gas diffusion cell, 1 mol l21 sodium hydroxide (laboratory-reagent grade) (Mallinckrodt), containing 0.05 mol l21 EDTA (laboratory-reagent grade) (BDH) was used for conversion into ammonia and the receiving solution was 0.01 mol l21 barium acetate (AnalaR grade) (BDH) plus 0.01 mol l21 acetic acid.All solutions were prepared using > 15 MW cm NANO-pure water (Barnstead, Dubuque, IA, USA). Nonactin-based Sensor The conventional ammonium ion-selective electrode was prepared by casting a membrane19 using a mixture consisting of 1 mg of nonactin (Fluka, Buchs, Switzerland), 66 mg of bis(1- butylpentyl) adipate (BBPA) (Fluka) and 33 mg of poly(vinyl chloride) (Fluka). This was dissolved in the minimum amount of purified tetrahydrofuran, sonicated for 10 min and then cast in a glass ring on a glass plate.After evaporation of the solvent, the electrode was constructed by gluing a cut disc of the membrane (about 0.2 mm thick) on to a plastic barrel. The inner reference electrode was Ag/AgCl in 0.1 mol l21 ammonium chloride saturated with silver chloride.The electrode was conditioned in 0.1 mol l21 ammonium chloride for 1 d before use. Potential measurements with this electrode were made using an Orion Ionanalyzer, Model 901 (Orion Research, Boston, MA, USA) with an Orion double-junction reference electrode with an outer junction filling solution consisting of 1.0 mol l21 lithium chloride. Potentials were monitored and recorded with an interfaced Apple IIe microcomputer.pH measurements were made with an Orion Ionanalyzer, Model EA940, and a combination pH–reference electrode (Ionode, Brisbane, Australia). Flow Injection The flow injection manifold used in this study is shown in Fig. 1(A) and Fig. 1(B) shows the manifold used with the gas Analyst, January 1997, Vol. 122 (89–93) 89diffusion cell to validate the results from the differential pH procedure. Both manifolds employed a Gilson (Worthington, OH, USA) Minipuls 2 peristaltic pump and manifold A incorporated two Teflon rotary valves [Rheodyne (Cotati, CA, USA) Type 50], one of which was used for sample injection (20 ml) and the other for switching between the two pH adjustment buffers.All flow lines were Teflon tubing of 0.5 mm id. The gas diffusion cell in manifold B was similar in design to one described in a previous paper5 and used Teflon plumber’s tape as the membrane. It is essential that the flow rates on each side of the membrane are identical to avoid rupture of the membrane.Thus, the water carrier and 1.0 mol l21 sodium hydroxide (also containing 0.05 mol l21 EDTA) flow rates were each 0.9 ml min21. The receiving solution consisted of 0.01 mol l21 barium acetate plus 0.01 mol l21 acetic acid and had a flow rate of 0.18 ml min21. This receiving solution ensured that all ammonia diffusing across the membrane was converted into ammonium ion. The sample size for the gas diffusion measurements was 40 ml and the throughput was 40 samples h21.The detector in each manifold was an improved version20 of our coated-wire multi-ion potentiometric detector.5 This consists of two Perspex blocks separated by a Teflon gasket. The bottom block contains silver wire surfaces which can be coated with the membrane mixture described above to form ammonium ion-selective coated-wire sensors. Several sensors of the same type were made to allow for redundancies. This detector has a separate reference arm consisting of an Ag/AgCl reference element with a reference stream of 0.15 mol l21 lithium chloride flowing at 0.9 ml min21.The potentiometric detector was interfaced to a 486 DX-33 PC through a 12-bit A/D card (Boston Technology, Boston, MA, USA). FCS software (A-Chem Technologies, Melbourne, Australia) was used for data acquisition and processing. Theory The pKb for ammonia is about 4.7 and so at pH < 7 all ammoniacal nitrogen exists as the ammonium ion.At around pH 9.3, there is about 50% ammonia and 50% ammonium ion. At pH > 11, only ammonia is present. Because the nonactin ion-selective electrode responds only to the ammonium ion and not to ammonia, the potential in a solution will decrease with increasing pH. If interfering ions such as the potassium ion are present, the contribution to the electrode potential due to the interfering ion should not change, since the ion is unaltered with change in pH. The actual potential, E1, in a solution is given by the following equation derived from the Nicolsky equation: E1 = constant + slog([NH4 +]+ S kij pot[M+]) (1) where s is the Nernst factor or electrode slope factor (mV per concentration decade), [NH4 +] and [M+] are the concentrations of the ammonium ion and an interfering monovalent cation, respectively, and kij pot is the selectivity coefficient.The term Skij pot [M+] in eqn. (1) should not change with pH, provided that the ionic strength of the solution remains constant, but the concentration of the ammonium ion will change.In this work, the potentials of the nonactin electrode were measured in samples of constant ionic strength adjusted to two pH values, 6.0 and 9.4. At pH 6.0, the potential is given by eqn. (1) where, in this case, [NH4 +] is equal to the total concentration of ammoniacal nitrogen, [NH4 +]T. However, at pH 9.4, some of the ammonium ion will be converted into ammonia and the potential, E2, is given by the equation E2 = constant + s log(d[NH4 +]T+ Skij pot [M+]) (2) where d is the fraction of the total ammoniacal nitrogen which is present as the ammonium ion at pH 9.4.Eqns. (1) and (2) are solved simultaneously to give eqn. (3), which allows the concentration of the total ammoniacal nitrogen to be determined. [NH4 +]T = [10(E12constant)/s 2 10(E22constant)/s]/(12d) (3) Results and Discussion Batch Experiments Some initial experiments were conducted with a conventional ammonium ion-selective electrode to determine the response characteristics of the nonactin membrane in solutions of different pH in the absence of the ammonium ion.These experiments were carried out in 0.1 mol l21 lithium chloride containing 0.001 mol l21 potassium chloride and the pH was adjusted with either dilute hydrochloric acid or lithium hydroxide solution. It was found that the potential changed by less than 10 mV in the pH range 5–10. More specifically, it was found that the potential in the pH 6.0 adjustment buffer was a few mV higher than that in the pH 9.4 adjustment buffer and that this difference can be compensated for by adding 5 mmol l21 potassium chloride to the pH 9.4 buffer.This is important because it is necessary in the flow injection method to have baselines for the two buffers which are identical. The conventional electrode was calibrated using the two pH adjustment buffers by mixing equal volumes of an ammonium ion standard and the buffer.The two calibration graphs are shown in Fig. 2. The electrode shows Nernstian behaviour in each buffer over the concentration range 1021–1024 mol l21. The two calibration lines are separated by a potential value which is related to the fraction of the total ammonium ion concentration which has been converted into ammonia on going from pH 6.0 to 9.4. The fraction of the ammonium ion at pH 9.4 is obtained by solving eqns. (1) and (2) (without the terms for an interfering ion) simultaneously: d = 10(E9.42E6.0)/s (4) where E6.0 = E1 and E9.4 = E2.The value of d is calculated by averaging values for several points on the abscissa and is used in the calculation of the total ammoniacal nitrogen concentration. The other way, of course, to calculate d for a particular pH is to use the expression for Kb, but the value of Kb depends on Fig. 1 Flow injection manifolds. A, Differential pH method; and B, gas diffusion. 90 Analyst, January 1997, Vol. 122the temperature, which must be known, and the pH must be determined accurately, whereas these are not requirements for the above method. The use of eqn. (3), for a constant ionic strength, to determine the concentration of total ammoniacal nitrogen in some synthetic samples was tested in the presence of the potassium ion, which is the most severe interferent with the nonactin electrode (E1 is the potential in the sample adjusted to pH 6.0 and E2 is the potential in the sample adjusted to pH 9.4).The results are given in Table 1. It can be seen that for samples in which the potassium ion concentration is similar to the total ammoniacal nitrogen concentration, there is excellent agreement between the expected and found values. However, in a sample which contained a high excess of potassium compared with ammoniacal nitrogen, the agreement is poor. The interference from organic amines (e.g., methylamine, nbutylamine, triethylamine) was examined and found to be negligible.There are two reasons for this: the low interference of the protonated form of an organic amine with the nonactinbased sensor9 and the fact that most organic amines are stronger bases than ammonia and so only very small amounts of the free base are formed at pH 9.4. It should be mentioned that the batch procedure is time consuming and tedious, particularly for solutions which contain low concentrations of the ammonium ion, since these require several minutes to reach the steady-state potential.These results, however, were encouraging enough to extend the work to developing a flow injection procedure and to test it on some real samples. Flow Injection The manifold designed for the flow injection procedure, shown in Fig. 1(A), operates in the following way. The sample (20 ml) is injected into a water carrier stream (flow rate, 0.9 ml min21) and then mixed initially with the pH 6.0 adjustment buffer which is selected through the second rotary valve and which is flowing continuously (flow rate 0.9 ml min21) into the system.The buffer selection valve is then switched to the pH 9.4 buffer (containing 5 mmol l21 KCl), which then flows continuously through the system. A second injection of the same sample is then made and mixed with this stream. Peak height measurements are made for the sample in each buffer. This procedure achieves a sampling rate of 30 samples h21. Fig. 3 shows calibration curves for the ammonium ion for each pH adjustment buffer. Six calibrations were carried out at each pH value using triplicate injections of each standard. It can be seen that linearity is achieved down to 1024 mol l21 and the lines are separated in the same way as seen in Fig. 2 for the batch measurements. The slopes of the linear sections of the two lines are identical (58.4 ± 0.6 and 58.3 ± 0.9 mV per concentration Table 1 Results for batch method for synthetic samples E/mV Calculated* Expected Sample [NH4 +]T/ [NH4 +]T/ [K+]/ No.pH 6.0 pH 9.4 mmol l21 mmol l21 mmol l21 1 281.8 299.3 0.504 0.500 0.500 2 260.8 269.9 0.725 0.500 5.00 3 228.5 248.4 4.89 5.00 0.500 4 224.8 242.9 5.34 5.00 5.00 * Ammonium ion concentration calculated from eqn. (3). Fig. 2 Batch calibration. 8, pH 6.0; and ., pH 9.4. Table 2 Results for flow injection method for synthetic samples Peak height*/mV Calculated If uncorrected Expected Sample [NH4 +]T/ [NH4 +]T/ [NH4 +]T/ [K+]/ No.pH 6.0 pH 9.4 mmol l21† mmol l21‡ mmol l21 mmol l21 1 94.9 74.5 0.244 (3.7) 0.294 0.25 0.25 2 111.0 87.4 0.504 (3.9) 0.555 0.50 0.25 3 149.9 124.5 2.43 (2.0) 2.57 2.50 0.25 4 100.7 83.9 0.269 (1.9) 0.370 0.25 0.50 5 114.9 93.4 0.555 (3.0) 0.647 0.50 0.50 6 150.3 125.8 2.42 (3.6) 2.61 2.50 0.50 7 121.1 113.6 0.317 (9.7) 0.826 0.25 2.50 8 127.4 117.1 0.530 (6.8) 1.06 0.50 2.50 9 154.0 132.8 2.56 (3.6) 3.02 2.50 2.50 * Mean peak height of triplicate injections with relative mean deviation < 1%.† Relative mean deviation (%) in parentheses, calculated for each sample using the mean of each pH data set and comparing it with all values of the other pH data set. ‡ Calculated using the pH 6.0 data only and showing the effect of the potassium ion interference. Fig. 3 Calibration in each buffer by flow injection using manifold A. 8, pH 6.0; and ., pH 9.4. Analyst, January 1997, Vol. 122 91decade for pH 6.0 and 9.4, respectively).Points were chosen in the linear range along the abscissa for the calculation of the average value of d. Initially, some synthetic samples also containing the potassium ion were tested as with the batch method and the results are given in Table 2. The results show that the method can completely eliminate the potassium ion interference in all cases except when the potassium ion concentration is much higher than (10 times) the ammonium ion concentration. This is because the peak height (potential) difference between the two buffers is relatively small, which produces a large error in the ammonium ion concentration.A ‘rule of thumb’ approach can be applied to this since it has been observed that, if the potential difference is smaller than, say, 8 mV, the result is unreliable. Analysis of Real Samples Some industrial waste water and polluted river water samples were obtained from two laboratories and analysed using the differential pH method and the results were compared with those obtained from the gas diffusion technique and, in some cases, the ammonia probe, as shown in Table 3. If the samples were initially very acidic or alkaline, the pH was adjusted to 5–7 with hydrochloric acid or lithium hydroxide prior to the determination.As seen before, excellent agreement between results is obtained when the peak height difference between the two pH values is large, i.e., > 8 mV (samples 1, 5, 6 and 11–18). Hence the method works well with natural water samples where high concentrations of cations such as potassium are absent.Very Low Ammonium Ion Concentrations When the ammonium ion concentration is below the linear range of the calibration plot shown in Fig. 3, it has been observed that there is a linear relationship between the peak height and the concentration21 (and not the logarithm of the concentration), as shown in Fig. 4. The differential pH method was therefore tested in this case.The ammoniacal nitrogen concentration is given by the equation [NH4 +]T = (E6.0 2 E9.4)/[s(12d)] (5) The results for some synthetic samples are given in Table 4 and excellent agreement is seen between the expected and found values. These particular samples contained high concentrations of the sodium ion and any interference was eliminated up to a ratio of [Na+] to [NH4 +]T of 4000 : 1. Hence the differential pH method can be applied with ammoniacal nitrogen concentrations down to about 1026 mol l21 in the presence of high concentrations of an interfering ion such as sodium. The criterion for determining initially if the ammoniacal nitrogen concentration is in the low range is the peak height value in the pH 6.0 adjustment buffer. If this is small (i.e., < 20 mV), the Fig. 4 [NH4 +] calibration at low concentration. Table 3 Results for the analysis of real samples Peak height†/mV Calculated Gas diffusion NH3 Sample [NH4 +]T/ [NH4 +]T/ probe/ Certified/ No.* pH 6.0 pH 9.4 mg l21‡ mg l21 mg l21 mg l21 1 95.3 88.6 1.24 (5.9) 1.70 (2.0) 1.50 2§ 71.7 69.3 6.29 (16) 3.66 (2.4) 3.63 3 142.0 137.6 7.66 (10) 7.37 (1.0) 6.86 4 157.2 155.0 7.27 (24) 4.44 (1.4) 4.07 5 142.9 133.8 15.2 (5.7) 15.3 (2.3) 14.3 6§ 141.2 122.2 590 (2.9) 570 (3.9) 550 7 150.1 149.5 1.90 (120) 0.12 (4.1) 0.11 8 154.9 153.4 5.70 (17) 0.05 (11) 0.04 9 155.8 155.4 1.31 (85) 0.02 (13) 0.03 10 34.3 29.2 0.14 (8.2) 0.07 (5.8) 0.06 11 136.0 119.3 20.1 (3.4) 19.8 (4.2) 19.8 12 127.6 111.4 14.2 (3.2) 14.6 (2.0) 14.3 13 67.1 56.0 1.38 (4.6) 1.48 (3.5) 1.4 14 67.8 55.3 1.54 (4.2) 1.48 (3.9) 1.4 15 82.9 68.6 2.96 (3.8) 3.17 (2.2) 2.9 16 83.6 68.5 3.15 (4.8) 2.24 (1.9) 3.0 17 83.2 67.9 3.14 (3.0) 3.30 (2.7) 3.1 18 83.8 67.8 3.33 (3.9) 3.29 (3.4) 3.2 * Samples 1–10 were industrial waste water samples and 11–18 were polluted river water samples. † Mean peak height of triplicate injections with relative mean deviation < 1%.‡ Relative mean deviation (%) in parentheses (n = 3).§ Samples were diluted 20-fold before analysis. Table 4 Results for synthetic samples of low ammoniacal nitrogen concentration Peak height/mV* Calculated Expected Sample [NH4 +]T/ [NH4 +]T/ [Na+]/ No. pH 6.0 pH 9.4 mmol l21† mmol l21 mmol l21 1 4.8 3.2 1.14 (9.9) 1.25 0.1 2 7.5 3.7 2.73 (5.2) 2.50 0.1 3 13.2 6.1 5.10 (4.5) 5.00 0.1 4 12.5 9.1 2.44 (3.6) 2.50 1 5 16.2 10.3 4.23 (7.9) 5.00 1 6 20.1 16.6 2.51 (1.5) 2.50 10 * Mean peak height of triplicate injections with relative mean deviation < 1 mV.† Relative mean deviation (%) in parentheses. 92 Analyst, January 1997, Vol. 122ammoniacal nitrogen concentration can be assumed to be lower than 1025 mol l21. Conclusion The proposed differential pH flow injection method for the determination of ammoniacal nitrogen using the nonactin-based ammonium ion sensor provides a simple and rapid procedure without the need to have a gas diffusion cell to separate interfering cations. The method is able to compensate for the interference from other cations provided that their concentrations are not extremely high relative to the ammoniacal nitrogen concentration. It has been demonstrated that the method is particularly reliable with natural water samples and can determine ammoniacal nitrogen concentrations as low as 1026 mol l21. We thank BHP and Monash University Water Studies Centre for providing the real samples. H.S. is grateful to La Trobe University for a Postgraduate Scholarship and we acknowledge the Australian Research Council for financial assistance. References 1 Van der Linden, W. E., Anal. Chim. Acta, 1983, 151, 359. 2 Nakata, R., Kawamula, T., Sakashita, H., and Nitta, A., Anal. Chim. Acta, 1988, 208, 81. 3 Van Son, M., Schothorst, R. C., and Den Boef, G., Anal. Chim. Acta, 1983, 153, 271. 4 Schulze, G., Liu, C. Y., Brodowski, M., Elsholz, O., Frenzel, W., and Moller, J., Anal. Chim. Acta, 1988, 214, 121. 5 Hauser, P. C., Tan, S. S., Cardwell, T. J., Cattrall, R. W., and Hamilton, I. C., Analyst, 1988, 113, 1551. 6 Alegret, S., Alonso, J., Bartroli, J., and Martinez-Fabregas, E., Analyst, 1989, 114, 1443. 7 Balconi, M. L., Sigon, F., Ferraroli, R., and Realini, F., Anal. Chim. Acta, 1988, 214, 367. 8 Meyerhoff, M. E., and Fraticelli, Y. M., Anal. Lett., 1981, 14, 415. 9 Lee, H. L., and Meyerhoff, M. E., Analyst, 1985, 110, 371. 10 Han, W., and Fan, L., Fenxi Huaxue, 1986, 14, 387. 11 Rodrigues Rohwedder, J. J., and Pasquini, C., Analyst, 1991, 116, 841. 12 Cardoso de Faria, L., and Pasquini, C., Anal. Chim. Acta, 1991, 245, 183. 13 Pasquini, C., and Cardoso de Faria, L., Anal. Chim. Acta, 1987, 193, 19. 14 Zhang, G., and Dasgupta, P. K., Anal. Chem., 1989, 61, 408. 15 Mikasa, H., Motomizu, S., and Toei, K., Bunseki Kagaku, 1985, 34, 518. 16 Kraus, P. R., and Crouch, S. R., Anal. Lett., 1987, 20, 183. 17 Jeppesen, M. T., and Hansen, E. H., Anal. Chim. Acta, 1988, 214, 147. 18 Orion Model 95-12 Ammonia Electrode Instruction Manual, Orion Research, Boston, 1990. 19 Ionophores for Ion Selective Electrodes and Optodes, Fluka, Buchs, 1991. 20 Gregorio, C. G., Cardwell, T. J., and Cattrall, R. W., Anal. Methods Instrum., submitted for publication. 21 Potentiometric Water Analysis, ed. Midgley, D., and Torrance, K., Wiley, Chichester, 2nd edn., 1991, p. 448. Paper 6/05155C Received July 23, 1996 Accepted September 9, 1996 Analyst, January 1997, Vol. 122 93

ISSN:0003-2654    

DOI:10.1039/a605155c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

p Ka Values of the Opened Form of a Thieno-1,2,4-triazolo-1,4-diazepine in Water B�eatrice Legouin and Jean-Louis Burgot UFR des Sciences Pharmaceutiques et Biologiques, D�epartement d’ � Etudes Physicochimiques et Biocin�etiques des Pharmacosyst`emes, Laboratoire de Chimie Analytique, 2 Av. du Pr. L�eon-Bernard, 35043 Rennes-Cedex, France The pKa values of the freely water-soluble opened form of a thieno-1,2,4-triazolo-1,4-diazepine, namely 2-(3-aminomethyl- 5-methyl-1,2,4-triazolo)-3-(2-chlorobenzoyl)- [2,3-c]thienopiperidine, were determined. pH measurements were made in the range 1 < pH < 10.Two of the three pKa values found in this range significantly overlapped; they quantify the ionization of both the aminomethyl residue and the piperidino moiety of the molecule. The third corresponds exclusively to the ionization of the triazolo nucleus. The pKa value of the thienophenone was obtained by UV spectrophotometry using acidity functions. All results were extracted from experimental data by a curve-fitting method.Keywords: 1,4-Diazepine derivative; ultraviolet spectrophotometry; Potentiometry; pKa; ionization microconstants; curve-fitting; ionization constant The 4,5-azomethine bond of benzo- and thieno-1,4-diazepines undergoes hydrolysis in acidic media to give the corresponding aminobenzo- or aminothienophenone, which reversibly cyclizes into the original closed form.1–10 As a result, benzo- and thieno-1,4-diazepines and their respective opened forms exist in such aqueous media as pH-dependent equilibrated mixtures.Given the fact that these compounds generally possess multibasic sites, this hydrolytic process is accompanied by acid–base reactions. Hence the knowledge of the different ionization constants is of utmost importance both for carrying out mechanistic studies concerning the hydrolytic process of 1,4-diazepines and for firmly explaining their pharmacological properties. Hydrolysis and its reverse reaction has been the subject of several studies,1–10 most of which were performed in alcohol–water mixtures.As we decided to study the kinetics of such reactions in strictly aqueous media for comparison purposes, 4H-1-methyl-6-(2-chlorophenyl)- [4A,3A,4,5]piperidino-[ 3,2-f]thieno-1,2,4[4,3-a]triazolo-1,4-diazepine (NHPTT) and its opened form 2-(3-aminomethyl-5-methyl-1,2,4-triazolo)- 3-(2-chlorobenzoyl)-[2,3-c]thienopiperidine (NHPTO), which are freely soluble in water, appeared to be good candidates for model compounds.As we had previously determined the ionization constants of NHPTT in water,11 we were faced with the determination of the pKa values of NHPTO. So far, no pKa values of opened form of thieno- or benzo-1,4-diazepines have been published. Experimental Apparatus pH was measured with a Tacussel LPH 430 T pH meter that was calibrated daily with six NBS buffers (commercial buffers manufactured according to the NIST recommendations) and an Ingold Model 9811 (pH 0–14) glass electrode was used.All UV/VIS spectra were recorded by using a Uvikon Model 930 spectrophotomer with 1 cm silica cells. Reagents The water used throughout was de-ionized on a set of ion exchange columns (Bioblock Scientific, Illkirch, France) to r > 2 MW cm21. NHPTT was kindly supplied by Ipsen-Beaufour Industry (Paris, France). NHPTO was prepared from NHPTT according to our procedure.12 It was isolated as the trihydrochloride form; it crystallized with 2.5 H2O.For the concentrated acid solutions (buffers of H0 < 25, where H0 is the Hammett acidity function), Bascombe and Bell’s acidity functions (aqueous sulfuric acid solutions) were used.13,14 Methods As a rule, owing to the results obtained with NHPTT, three acid–base sites had to be considered in water: the piperidino, the triazolo and the aminomethyl groups. We also determined the pKa value of the protonated form of the carbonyl function for the sake of a complete study, although its protonation in mildly acidic aqueous media was very unlikely.The expected values were about 8 and 2 for the first two sites and about 9 for the aminomethyl residue, which can be considered, at first glance, as a benzylamine.15 For the carbonyl function, a value between 25 and 28 was expected.13,16,17 The ionization scheme is shown in Scheme 1. Whatever the pH of the solution, a preliminary study had indicated that the UV spectra quickly evolved with time, owing to the closure of the diazepine ring.The chief difficulty was overcoming the problem of the rapid establishment of the NHPTO–NHPTT equilibrium, once NHPTO had dissolved. This difficulty was solved by processing as rapidly as possible; the preparation of the solution and measurement did not take more than 1 min. Under these conditions, UV spectra recorded from around H0 = 22 to pH = 10 were not significantly different. Such was not the case for the range 210 < H0 < 27.As a result, we determined the ionization constant of the carbonyl function by UV spectrophotometry and the three others by potentiometry. pH determinations were carried out by measurement of the pH values of NHPTO solutions under carefully chosen experimental conditions. The pKa values were extracted from these experimental results by a non-linear least-squares method. To overcome the problem of the estabishment of the NHPTO– NHPTT equilibrium, the same weighed amount of NHPTO was dissolved in different volumes of sodium hydroxide solution followed by an immediate pH measurement.This allowed us to work in the range 1 < pH < 10. We had checked that, under these conditions, the polarographic wave of NHPTT did not appear.11 Curve fitting by the non-linear least-squares method required the calculation of the pH values for each of the experimental conditions. This was done using a mathematical model Analyst, January 1997, Vol. 122 (95–100) 95where pHi mean is the mean of the m experimental pH values corresponding to the volume Vi of sodium hydroxide solution added. The ratio U/U0 quantifies the lack of fit.21 Usually, the U cost function must be within 1–3 times the U0 cost function value. The search for physically convenient roots of eqn. (1) and the search for the set of thermodynamic constants Ka1, Ka2 and Ka3 which minimize the least-squares function were performed through an algorithm, developed by ourselves, which has been validated elsewhere.22 In this algorithm, the variances–covariances matrix, which is calculated during the search for the best set of parameters, gives the standard deviations of Ka1, Ka2 and Ka3.For the determination of the ionization constant of the protonated form of the carbonyl, we used the classical equation23 Acalc = eBH3 3+ C0Ka4 Ka4 + |H+ | + eBH4 4+ C0|H+ | Ka4 + |H+ | (2) to obtain the calculated absorbances, which were fitted to the experimental values.In this equation, C0 is initial concentration of NHPTO, Ka4 is the ionization constant, eBH3 3+ and eBH4 4+ are the molar absorptivities of the species BH3 3+ and BH4 4+ and ýH+ý is the concentration of H+ ions. We deliberately equated activities with concentrations owing to the high acidity of the media which required the use of acidity functions in the range 28.50 < H0 < 25.18. The buffers used for the determination of Ka4 were such that H0 = 25.18, 25.48, 25.92, 26.50, 27.00, 27.50, 28.00 and 28.50.The initial concentration of NHPTO was C0 = 7.38 31025 mol dm23. For comparison purpose, two analytical wavelengths (l1 = 260 nm and l2 = 300 nm), for which the maximum variation in absorption was observed, were chosen. Results and Discussion Acid–Base Pairs BH3 3+–BH2 2+, BH2 2+–BH+ and BH+–B pKa3, pKa2 and pKa1 values, determined by potentiometry, are given in Table 1 together with the confidence limits and with parameters which quantify the accuracy of the fit.Experimental data are given in Tables 2 and 3. The titration curve obtained with m2 = 15 mg is given in Fig. 1 together with the theoretical curve, calculated with pKa1, pKa2 and pKa3 values. The pKa2 and pKa1 values were self-consistent whichever experimental conditioed (m1 = 10 mg and m2 = 15 mg), even if the confidence limits did not overlap. Without any further data concerning the accuracy of the results obtained with each of the samples, the averages of the two pKa2 and of the two pKa1 values were considered to be the true values.Their confidence limits were the extreme values of each of the confidence limits encountered under both conditions (m1 and m2): pKa2 = 6.36 (6.25–6.45) and pKa1 = 7.50 (7.33–7.67). The close pKa1 and pKa2 values preclude their assignment to definite acid–base sites, was the case above when introducing the matter (Scheme 1). pKa2 must be considered as a macroscopic constant which quantifies ionization of the diprotonated NHPTO on the piperidino and aminomethyl residues, BH2 2+, to give the monoprotonated forms B1H+ and B2H+.Likewise, pKa1 is the macroscopic constant which quantifies the ionization of both monoprotonated forms B1H+ and B2H+ to give the fully basic NHPTO, B (Scheme 2). We did not consider that both pKa values (pKa2 = 6.36 and pKa1 = 7.50) could involve the ionization of the conjugated acid of the triazolo nucleus owing to the results obtained with NHPTT.Such a hypothesis would involve the assigment of the pKa value of 0.85 (pKa3, Table 1) to the piperidino or the aminomethyl residue. Owing to the value found for the pKa of the piperidino residue in NHPTT (pKa = 8.10), one can Table 3 Determination of the pKa values of the piperidino, aminomethyl and triazolo groups: pH (m2 = 15 mg) No. V*/cm3 1 2 3 4 5 6 Average Variance 1 1 1.718 1.694 1.728 1.683 1.773 1.736 1.722 1.03 3 1023 2 2 1.981 1.980 2.003 1.971 2.031 2.020 1.998 5.88 3 1024 3 3 2.164 2.161 2.170 2.132 2.208 2.168 2.167 5.92 3 1024 4 4 2.271 2.292 2.308 2.259 2.328 2.297 2.293 6.21 3 1024 5 5 2.373 2.395 2.428 2.374 2.554 2.402 2.421 4.66 3 1023 6 6 2.474 2.486 2.496 2.463 2.473 2.495 2.481 1.77 3 1024 7 8 2.630 2.624 2.632 2.620 2.634 2.627 2.628 2.74 3 1025 8 9 2.674 2.713 2.704 2.686 2.674 2.700 2.692 2.67 3 1024 9 10 2.744 2.782 2.755 2.750 2.757 2.773 2.760 2.09 3 1024 10 11 2.799 2.820 2.835 2.827 2.810 2.835 2.821 2.07 3 1024 11 12 2.861 2.893 2.878 2.876 2.912 2.882 2.884 2.99 3 1024 12 14 3.013 3.003 2.994 2.996 3.002 3.004 3.002 4.52 3 1025 13 16 3.096 3.145 3.124 3.141 3.128 3.100 3.122 4.18 3 1024 14 18 3.226 3.267 3.250 3.256 3.261 3.260 3.253 2.11 3 1024 15 20 3.402 3.395 3.493 3.394 3.401 3.379 3.411 1.69 3 1023 16 22 3.541 3.531 3.511 3.521 3.579 3.602 3.548 1.26 3 1023 17 24 3.735 3.732 3.731 3.732 3.822 3.768 3.753 1.34 3 1023 18 28 4.288 4.425 4.459 4.399 4.849 4.371 4.465 3.87 3 1022 19 30 5.174 5.188 5.062 5.128 4.960 5.077 5.098 7.11 3 1023 20 35 5.723 5.717 5.756 5.776 5.686 5.649 5.718 2.12 3 1023 21 40 6.059 6.032 6.043 6.073 6.060 5.976 6.041 1.20 3 1023 22 45 6.217 6.245 6.232 6.248 6.267 6.138 6.225 2.07 3 1023 23 50 6.438 6.465 6.464 6.487 6.439 6.363 6.443 1.86 3 1023 24 55 6.892 6.556 6.688 6.671 6.748 6.641 6.699 1.29 3 1022 25 58 6.874 6.788 6.849 6.879 6.868 6.717 6.829 4.13 3 1023 26 60 6.899 6.912 6.937 7.001 6.942 6.761 6.909 6.47 3 1023 27 63 7.136 7.003 7.086 7.071 7.169 6.909 7.062 8.89 3 1023 28 65 7.259 7.404 7.231 7.074 7.248 7.072 7.215 1.59 3 1022 29 68 7.318 7.251 7.266 7.169 7.314 7.193 7.252 3.75 3 1023 30 70 7.271 7.298 7.249 7.248 7.394 7.286 7.291 2.94 3 1023 * Volume of sodium hydroxide solution. 98 Analyst, January 1997, Vol. 122represented by eqn. (1), obtained from the relationships which quantify equilibria in solution and mass and charge balance laws.18.|H+ |5 + (Ka3 + C)|H+ |4 + Ka3 Ka2 + CKa3 - m MVi Ka3 - Kw Ê Ë Á � � � |H+ |3 + Ka3 Ka2 Ka1 + CKa3 Ka2 - 2 m MVi Ka3 Ka2 - KwKa3 Ê Ë Á � � � |H+ |2 + Ka3 Ka2 Ka1 C - 3 m MVi Ka3 Ka2 Ka1 - KwKa3 Ka2 Ê Ë Á � � � |H+ |- KwKa3 Ka2 Ka1 = 0 (1) Where ýH+ý stands successively for activity and concentration (see below), Ka3, Ka2 and Ka1 are the searched for ionization constants introduced above (Scheme 1), m the weighed amount of NHPTO, M the molecular mass of NHPTO, Vi is the added volume of sodium hydroxide solution, C is the concentration of the sodium hydroxide solution and Kw is the ionic product of water.pH values being defined in terms of activity, the problem of the ionic strengths of the solutions had to be taken into account. This was performed by a classical iterative process. 18 For each set of chosen thermodynamic Ka1, Ka2 and Ka3 values, eqn. (1) was solved for ýH+ý several times. At the beginning of the process, a first estimation of ýH+ý was calculated by solving eqn.(1). At this stage, the calculated ýH+ý value was neither an activity nor a concentration value because of the necessary initial mixing of activities (to which Ka1, Ka2, Ka3 and Kw values refer) with concentrations C and m/MVi, to which equations of mass and charge balances pertained. However, the first obtained value of ýH+ý permitted the Scheme 1 Ionization scheme for NHPTO and nomenclature. 96 Analyst, January 1997, Vol. 122calculation of a first pseudo-ionic strength, which, in turn, allowed the estimation of a first set of activity coefficients of all species.With these activity coefficients, a set of equilibrium concentration constants Ka1, Ka2, Ka3 and Kw was therefore computed from the thermodynamic values. A new value of ýH+ý was obtained, and the process was repeated until the ionic strength was constant. During the whole process, the Ka1, Ka2, Ka3 and Kw constants were hence successively thermodynamic constants and then, more and more, equilibrium concentration constants, while ýH+ý became more and more a concentration value.After convergence of the ionic strength, the [H+] concentration was transformed back to {H+} activity with the help of the last activity coefficient of the proton. The activity coefficients used throughout this process were computed through the extended Debye–H�uckel relationship.19 -loggi = Azi 2 I 1+ Ba I or with the Davies relationship when the ionic strength was over 0.01 mol dm23.20 -loggi = Azi 2 I 1+ I + 0.1 zi 2 I For each volume Vi of sodium hydroxide solution added (the concentrations of which were C1 = 9.00 3 1024 mol dm23 and C2 = 9.35 3 1024 mol dm23), six solutions corresponding to the same amount of NHPTO (m1 = 10 mg and m2 = 15 mg) were prepared in an absolutely independent manner.Thirty different volumes from 1 to 70 cm3 were used. The least-squares function was defined as: U = i=1 n wi (pHi calc - pHij exp )2 j=1 m where n is the number of points treated, m the number of replicates (generally m = 6), i the running point, j the replicate considered at point i, and wi a weighting factor chosen as the inverse of variance spHi 2 taken over the m replicates for the same volume Vi.This U function was compared with a theoretical minimum cost function U0, which only took into account the random experimental errors: U0 = i=1 n wi (pHi mean - pHij exp )2 j=1 m Table 1 Determination of the pKa values of the piperidino, aminomethyl and triazolo groups: results obtained by treating pH data (from 30 different volumes of sodium hydroxide solution) Parameter m1 = 10 mg* m2 = 15 mg† pKa1 7.63 7.37 (7.60; 7.67) (7.33; 7.41) pKa2 6.43 6.28 (6.40; 6.45) (6.25. 6.32) pKa3 0.85 22.60 (0.71; 1.05) (22.87; 21.89) U0 150 150 U 503 518 * CNaOH = 9.00 3 1024 mol dm23. † CNaOH = 9.35 3 1024 mol dm23. Table 2 Determination of the pKa values of the piperidino, aminomethyl and triazolo groups: pH data (m1 = 10 mg) No.V*/cm3 1 2 3 4 5 6 Average Variance 1 1 1.971 1.947 1.864 1.860 1.978 1.918 1.923 2.68 3 1023 2 2 2.249 2.249 2.198 2.130 2.219 2.199 2.207 1.95 3 1023 3 3 2.433 2.453 2.369 2.341 2.427 2.367 2.398 2.00 3 1023 4 4 2.565 2.589 2.504 2.485 2.537 2.497 2.530 1.71 3 1023 5 5 2.678 2.705 2.621 2.601 2.699 2.672 2.663 1.79 3 1023 6 6 2.778 2.805 2.700 2.693 2.774 2.745 2.749 2.03 3 1023 7 8 2.966 2.990 2.914 2.850 2.980 2.923 2.937 2.76 3 1023 8 9 3.067 3.092 2.962 2.971 3.064 3.023 3.030 2.90.219 3.057 3.081 3.138 3.097 3.125 3.45 3 1023 10 11 3.245 3.314 3.171 3.162 3.195 3.170 3.210 3.45 3 1023 11 12 3.366 3.393 3.295 3.157 3.306 3.273 3.298 6.85 3 1023 12 14 3.569 3.587 3.488 3.466 3.580 3.450 3.523 3.85 3 1023 13 16 3.875 3.903 3.852 3.801 3.786 3.675 3.815 6.67 3 1023 14 18 4.138 4.354 4.185 4.210 4.420 4.002 4.218 2.27 3 1023 15 20 5.004 5.191 4.959 5.285 5.214 4.892 5.091 2.55 3 1023 16 22 5.485 5.480 5.528 5.625 5.605 5.522 5.541 3.71 3 1023 17 24 5.747 5.839 5.855 5.829 5.895 5.768 5.822 3.06 3 1023 18 25 5.946 5.967 5.890 5.845 5.897 5.807 5.892 3.60 3 1023 19 30 6.240 6.243 6.238 6.341 6.418 6.276 6.293 5.31 3 1023 20 35 6.669 6.705 6.624 6.618 6.714 6.634 6.661 1.74 3 1023 21 40 6.845 7.101 7.054 6.987 7.005 6.879 6.979 9.84 3 1023 22 45 7.337 7.420 7.288 7.298 7.459 7.355 7.360 4.60 3 1023 23 50 7.596 7.802 7.561 7.454 7.593 7.692 7.616 1.41 3 1022 24 55 7.905 8.116 7.930 7.960 7.887 7.702 7.917 1.78 3 1022 25 58 8.456 8.384 8.401 8.491 8.611 8.500 8.474 6.70 3 1023 26 60 8.501 8.634 8.554 8.662 8.696 8.665 8.619 5.65 3 1023 27 63 8.901 9.179 8.890 9.432 9.024 9.010 9.073 4.19 3 1022 28 65 9.563 9.512 9.032 9.559 9.559 9.231 9.409 5.06 3 1022 29 68 9.644 9.798 9.701 9.593 9.711 9.604 9.675 5.97 3 1023 30 70 9.944 9.905 9.746 9.794 9.759 9.774 9.820 6.92 3 1023 * Volume of sodium hydroxide solution.Analyst, January 1997, Vol. 122 97consider that the higher value found for pKa2 or pKa1, i.e., 7.50, is representative mostly of the ionization of the piperidino moiety in NHPTO and hence the value of 6.36 for the ionization mostly of the conjugated acid of the amino group. These assumptions are in agreement with the finding of Sayer et al.,24 who noted that pKa values of the conjugated acids of imines Fig. 1 Titration curve obtained with m2 = 15 mg and CNaOH = 9.35 3 1024 mol dm23 and theoretical curve calculated with pKa3 = 22.60, pKa2 = 6.28 and pKa1 = 7.37.Scheme 2 Microforms of NHPTO. Table 5 Determination of the pKa values of the carbonyl function: spectrophotometric data for H0 values ranging from 28.50 to 25.18 No. H0 1 2 3 4 5 6 Average Variance l2 = 300 nm 1 25. 18 0.3721 0.3751 0.3760 0.3830 0.3754 0.3667 0.3747 2.84 3 1025 2 25. 48 0.3873 0.3812 0.3870 0.3914 0.3720 0.3790 0.3830 4.91 3 1025 3 25. 92 0.4015 0.4079 0.3961 0.3962 0.4068 0.3984 0.4012 2.70 3 1025 4 26. 50 0.4136 0.4185 0.4312 0.4143 0.4169 0.4072 0.4170 6.38 3 1025 5 27. 00 0.4451 0.4365 0.4570 0.4424 0.4597 0.4500 0.4485 7.85 3 1025 6 27. 50 0.4718 0.4662 0.4742 0.4772 0.4747 0.4637 0.4713 2.78 3 1025 7 28. 00 0.5107 0.5318 0.5428 0.5070 0.5010 0.5324 0.5210 2.85 3 1024 8 28. 50 0.5412 0.5505 0.5600 0.5462 0.5464 0.5537 0.5497 4.36 3 1025 l1 = 260 nm 1 25. 18 0.7773 0.7669 0.7968 0.7968 0.7753 0.7754 0.7814 1.55 3 1024 2 25. 48 0.7695 0.7514 0.7349 0.7663 0.7582 0.7567 0.7562 1.52 3 1024 3 25. 92 0.7542 0.7424 0.7245 0.7471 0.7395 0.7545 0.7437 1.25 3 1024 4 26. 50 0.7250 0.7316 0.7475 0.7369 0.7352 0.7380 0.7357 5.55 3 1025 5 27. 00 0.7011 0.6898 0.7364 0.7101 0.7070 0.7224 0.7111 2.68 3 1024 6 27. 50 0.6772 0.6736 0.6814 0.6886 0.6879 0.6746 0.6806 4.29 3 1025 7 28. 00 0.6465 0.6347 0.6698 0.6407 0.6337 0.6692 0.6491 2.71 3 1024 8 28. 50 0.6054 0.5987 0.6270 0.6142 0.5618 0.6253 0.6054 5.77 3 1024 Table 4 Determinaion of the pKa values of the carbonyl function: results obtained by treating spectrophotometric data for H0 values ranging from 28.50 to 25.18 Parameter l2 = 300 nm l1 = 260 nm pKa4 27.45 27.47 (27.49; 27.39) (27.61; 27.28) eBH4 4+ * 5250 10300 (5200; 5300) (10200; 10400) eBH33+* 7580 8250 (7520; 7640) (8000; 8550) U0 40 40 U 147 78 eBH4 4+* 4690 12250 (experimental) eBH3 3+* 7980 7150 (experimental) * e in dm3 mol21 cm21.Analyst, January 1997, Vol. 122 99were seven pH units lower than those of the corresponding amines.We found for NHPTT a value of pKa = 20.24, which could be mostly assigned to the protonated imine. Assigning the pKa = 6.36 value rather to the aminomethyl residue is also consistent with the hypothesis of Konishi et al.,10 who suggested a value pKa = 6.50 for this group in the opened form of triazolam. This value was given by analogy with that found for 2-aminoacetamido-5-chlorobenzophenone. It is worth noting that both values, 6.36 and 7.50, are undoubtedly weak values for benzylamine groups.The pKa3 values were not consistent. The lack of agreement between them can be explained by the fact that the U function is very insensitive to the Ka3 paremeter, i.e., a large variation in Ka3 only induces very small changes in U. From a chemical standpoint, this means that in the range of the lowest pH values, triazole is already in its basic form. Ideally, such a low pKa3 value would require the use of an acidity function and hence of UV spectrophotometric techniques.As data obtained in this way would not be usable because of the lack of variation of the UV spectra in the range of concern, potentiometry was used. As a result, the pKa3 values, given in Table 1, must be considered only qualitative and only indicative of the strong acidity of the protonated triazolo nucleus. However, owing to the difference between the pKa3 value and the of pKa2 and pKa1 values on the one hand and the difference between the pKa3 and pKa4 values (see below) on the other, pKa3 is only representative of the ionization of the triazolo ring.We can note that the ionization constant of the conjugated acid of the triazolo nucleus is much lower than that usually encountered with 1,2,4-triazoles25–28 and to a lesser extent with that encountered in NHPTT. The difference between NHPTT and NHPTO can be tentatively assigned to the presence of two strongly attractive groups, the carbonyl and the protonated aminomethyl residues, in the latter molecule.Acid–Base Pair BH4 4+–BH3 3+ The values found by UV spectrophotometry for pKa4 are given in Table 4. Also given are U0 and U values together with those of the molar absorptivities eBH4 4+ and eBH3 3+ obtained by curve fitting, for the sake of comparison with the experimental values. Experimental data are given in Table 5. The values obtained for both wavelengths are in very good agreement. Other values given in Table 4 are satisfactory together with the distribution of the residues (Aexp 2 Acalc), which are alternately positive and negative.The value pKa4 = 27.46 (27.61; 27.28) is that expected for a thienophenone on the H0 scale.13 The authors are grateful to G. Bouer for his assistance. References 1 Inotsume, N., and Nakano, M., J. Pharm. Sci., 1980, 69(11), 1331. 2 Pfendt, L. B., and Popovic, G. V., J. Chem. Soc., Perkin Trans. 2, 1994, 1845. 3 Maudling, H. V. Nazareno, J. P., Pearson, J. E., and Michaelis, A.F., J. Pharm. Sci., 1975, 64, 278. 4 Inotsume, N., and Nakano, M., Chem. Pharm. Bull., 1980, 28(8), 2536. 5 Vir�e, J. C., and Patriarche, G. J., J. Electroanal. Chem., 1986, 214, 275. 6 Jimenez, R. M., Alonso, R. M., Oleaga, E., Vicente, F., and Hernandez, L., Fresenius’ Z. Anal. Chem., 1987, 329(4), 468. 7 Vir�e, J. C., Gallo Hermosa, B., and Patriarche, G. J., Analusis, 1987, 15(9), 499. 8 Pfendt, L. B., Janjic, T. J., and Popovic, G. V., Analyst., 1990, 115, 1457. 9 Moro, M. E., Novillo-Fertrell, J., Velazquez, M. M., and Rodriguez, L. J., J. Pharm. Sci., 1991, 80(5), 459. 10 Konishi, M., Hirai, K., and Mori, Y., J. Pharm. Sci., 1982, 71, 1328. 11 Legouin, B., and Burgot, J. L., Analyst, 1996, 121(1), 43. 12 Legouin, B., and Burgot, J. L., in the press. 13 Rochester, C. H., in Acidity Functions, Academic Press, London, 1970, vol. 17, p. 24. 14 Bascombe, K. N., and Bell, R. P., J. Chem. Soc., 1959, 1096. 15 Blackwell, L. F., Fischer, A. Miler, J. J., Topson, R.D., and Vaughan, J., J. Chem. Soc., 1964, 3588. 16 Sayer, J. M., Prinsky, B., Schoubrum, A., and Wazshtien, W., J. Am. Chem. Soc., 1974, 96(26), 7998. 17 Rosenberg, S., Silver, S. M., Sayer, J. M., and Jencks, W. P., J. Am. Chem. Soc., 1974, 96(26), 7986. 18 Butler, J. N., in Ionic Equilibrium. A Mathematical Approach, Addison-Wesley, Reading, MA, 1964, pp. 440–458. 19 Harned, H., and Owen, B. B., in The Physical Chemistry of Electrolytic Solutions, Chapman and Hall, London, 1958, p. 66. 20 Rossotti, F., and Rosi, J. C., in The Determination of Stability Constants in Solutions, McGraw-Hill, New York, 1961, p. 30. 21 Draper, N., and Smith, H., in Applied Regression Analysis, Wiley, New York, 2nd edn., 1981, p. 33. 22 Zekri, O., Boudeville, P., Genay, P., Perly, B. Braquet, P., Jouenne, P., and Burgot, J.-L., Anal. Chem., 1996, 68(15), 2598. 23 Albert, A., and Serjeant, E. P., in The Determination of Ionization Constants, Chapman and Hall, London, 1971, p. 44. 24 Sayer, J. M., Peskin, M., and Jencks, W. P., J. Am. Chem. Soc., 1973, 95, 4277. 25 Barton, D., and Ollis, W., in Comprehensive Organic Chemistry, vol. 4, Heterocyclic Compounds, ed. Sammes, P. G., Pergamon Press, Oxford, 1979. 26 Kr�oger, C. F., and Freiberg, W., Z. Chem., 1965, 5(10), 381. 27 Kr�oger, C.-F., and Freiberg, W., Chimia, 1967, 21, p. 161. 28 Fox, J., and Jencks, W. P., J. Am. Chem. Soc., 1974, 96, 1436. Paper 6/04965F Received July 15, 1996 Accepted September 16, 1996 100 Analyst, January 1997, Vol. 122 p Ka Values of the Opened Form of a Thieno-1,2,4-triazolo-1,4-diazepine in Water B�eatrice Legouin and Jean-Louis Burgot UFR des Sciences Pharmaceutiques et Biologiques, D�epartement d’ � Etudes Physicochimiques et Biocin�etiques des Pharmacosyst`emes, Laboratoire de Chimie Analytique, 2 Av. du Pr. L�eon-Bernard, 35043 Rennes-Cedex, France The pKa values of the freely water-soluble opened form of a thieno-1,2,4-triazolo-1,4-diazepine, namely 2-(3-aminomethyl- 5-methyl-1,2,4-triazolo)-3-(2-chlorobenzoyl)- [2,3-c]thienopiperidine, were determined.pH measurements were made in the range 1 < pH < 10. Two of the three pKa values found in this range significantly overlapped; they quantify the ionization of both the aminomethyl residue and the piperidino moiety of the molecule. The third corresponds exclusively to the ionization of the triazolo nucleus. The pKa value of the thienophenone was obtained by UV spectrophotometry using acidity functions.All results were extracted from experimental data by a curve-fitting method. Keywords: 1,4-Diazepine derivative; ultraviolet spectrophotometry; Potentiometry; pKa; ionization microconstants; curve-fitting; ionization constant The 4,5-azomethine bond of benzo- and thieno-1,4-diazepines undergoes hydrolysis in acidic media to give the corresponding aminobenzo- or aminothienophenone, which reversibly cyclizes into the original closed form.1–10 As a result, benzo- and thieno-1,4-diazepines and their respective opened forms exist in such aqueous media as pH-dependent equilibrated mixtures.Given the fact that these compounds generally possess multibasic sites, this hydrolytic process is accompanied by acid–base reactions. Hence the knowledge of the different ionization constants is of utmost importance both for carrying out mechanistic studies concerning the hydrolytic process of 1,4-diazepines and for firmly explaining their pharmacological properties.Hydrolysis and its reverse reaction has been the subject of several studies,1–10 most of which were performed in alcohol–water mixtures. As we decided to study the kinetics of such reactions in strictly aqueous media for comparison purposes, 4H-1-methyl-6-(2-chlorophenyl)- [4A,3A,4,5]piperidino-[ 3,2-f]thieno-1,2,4[4,3-a]triazolo-1,4-diazepine (NHPTT) and its opened form 2-(3-aminomethyl-5-methyl-1,2,4-triazolo)- 3-(2-chlorobenzoyl)-[2,3-c]thienopiperidine (NHPTO), which are freely soluble in water, appeared to be good candidates for model compounds.As we had previously determined the ionization constants of NHPTT in water,11 we were faced with the determination of the pKa values of NHPTO. So far, no pKa values of opened form of thieno- or benzo-1,4-diazepines have been published. Experimental Apparatus pH was measured with a Tacussel LPH 430 T pH meter that was calibrated daily with six NBS buffers (commercial buffers manufactured according to the NIST recommendations) and an Ingold Model 9811 (pH 0–14) glass electrode was used.All UV/VIS spectra were recorded by using a Uvikon Model 930 spectrophotomer with 1 cm silica cells. Reagents The water used throughout was de-ionized on a set of ion exchange columns (Bioblock Scientific, Illkirch, France) to r > 2 MW cm21. NHPTT was kindly supplied by Ipsen-Beaufour Industry (Paris, France). NHPTO was prepared from NHPTT according to our procedure.12 It was isolated as the trihydrochloride form; it crystallized with 2.5 H2O.For the concentrated acid solutions (buffers of H0 < 25, where H0 is the Hammett acidity function), Bascombe and Bell’s acidity functions (aqueous sulfuric acid solutions) were used.13,14 Methods As a rule, owing to the results obtained with NHPTT, three acid–base sites had to be considered in water: the piperidino, the triazolo and the aminomethyl groups.We also determined the pKa value of the protonated form of the carbonyl function for the sake of a complete study, although its protonation in mildly acidic aqueous media was very unlikely. The expected values were about 8 and 2 for the first two sites and about 9 for the aminomethyl residue, which can be considered, at first glance, as a benzylamine.15 For the carbonyl function, a value between 25 and 28 was expected.13,16,17 The ionization scheme is shown in Scheme 1.Whatever the pH of the solution, a preliminary study had indicated that the UV spectra quickly evolved with time, owing to the closure of the diazepine ring. The chief difficulty was overcoming the problem of the rapid establishment of the NHPTO–NHPTT equilibrium, once NHPTO had dissolved. This difficulty was solved by processing as rapidly as possible; the preparation of the solution and measurement did not take more than 1 min. Under these conditions, UV spectra recorded from around H0 = 22 to pH = 10 were not significantly different.Such was not the case for the range 210 < H0 < 27. As a result, we determined the ionization constant of the carbonyl function by UV spectrophotometry and the three others by potentiometry. pH determinations were carried out by measurement of the pH values of NHPTO solutions under carefully chosen experimental conditions. The pKa values were extracted from these experimental results by a non-linear least-squares method.To overcome the problem of the estabishment of the NHPTO– NHPTT equilibrium, the same weighed amount of NHPTO was dissolved in different volumes of sodium hydroxide solution followed by an immediate pH measurement. This allowed us to work in the range 1 < pH < 10. We had checked that, under these conditions, the polarographic wave of NHPTT did not appear.11 Curve fitting by the non-linear least-squares method required the calculation of the pH values for each of the experimental conditions. This was done using a mathematical model Analyst, January 1997, Vol. 122 (95–100) 95where pHi mean is the mean of the m experimental pH values corresponding to the volume Vi of sodium hydroxide solution added.The ratio U/U0 quantifies the lack of fit.21 Usually, the U cost function must be within 1–3 times the U0 cost function value. The search for physically convenient roots of eqn. (1) and the search for the set of thermodynamic constants Ka1, Ka2 and Ka3 which minimize the least-squares function were performed through an algorithm, developed by ourselves, which has been validated elsewhere.22 In this algorithm, the variances–covariances matrix, which is calculated during the search for the best set of parameters, gives the standard deviations of Ka1, Ka2 and Ka3.For the determination of the ionization constant of the protonated form of the carbonyl, we used the classical equation23 Acalc = eBH3 3+ C0Ka4 Ka4 + |H+ | + eBH4 4+ C0|H+ | Ka4 + |H+ | (2) to obtain the calculated absorbances, which were fitted to the experimental values.In this equation, C0 is initial concentration of NHPTO, Ka4 is the ionization constant, eBH3 3+ and eBH4 4+ are the molar absorptivities of the species BH3 3+ and BH4 4+ and ýH+ý is the concentration of H+ ions. We deliberately equated activities with concentrations owing to the high acidia which required the use of acidity functions in the range 28.50 < H0 < 25.18.The buffers used for the determination of Ka4 were such that H0 = 25.18, 25.48, 25.92, 26.50, 27.00, 27.50, 28.00 and 28.50. The initial concentration of NHPTO was C0 = 7.38 31025 mol dm23. For comparison purpose, two analytical wavelengths (l1 = 260 nm and l2 = 300 nm), for which the maximum variation in absorption was observed, were chosen. Results and Discussion Acid–Base Pairs BH3 3+–BH2 2+, BH2 2+–BH+ and BH+–B pKa3, pKa2 and pKa1 values, determined by potentiometry, are given in Table 1 together with the confidence limits and with parameters which quantify the accuracy of the fit.Experimental data are given in Tables 2 and 3. The titration curve obtained with m2 = 15 mg is given in Fig. 1 together with the theoretical curve, calculated with pKa1, pKa2 and pKa3 values. The pKa2 and pKa1 values were self-consistent whichever experimental conditions were used (m1 = 10 mg and m2 = 15 mg), even if the confidence limits did not overlap.Without any further data concerning the accuracy of the results obtained with each of the samples, the averages of the two pKa2 and of the two pKa1 values were considered to be the true values. Their confidence limits were the extreme values of each of the confidence limits encountered under both conditions (m1 and m2): pKa2 = 6.36 (6.25–6.45) and pKa1 = 7.50 (7.33–7.67). The close pKa1 and pKa2 values preclude their assignment to definite acid–base sites, was the case above when introducing the matter (Scheme 1).pKa2 must be considered as a macroscopic constant which quantifies ionization of the diprotonated NHPTO on the piperidino and aminomethyl residues, BH2 2+, to give the monoprotonated forms B1H+ and B2H+. Likewise, pKa1 is the macroscopic constant which quantifies the ionization of both monoprotonated forms B1H+ and B2H+ to give the fully basic NHPTO, B (Scheme 2). We did not consider that both pKa values (pKa2 = 6.36 and pKa1 = 7.50) could involve the ionization of the conjugated acid of the triazolo nucleus owing to the results obtained with NHPTT.Such a hypothesis would involve the assigment of the pKa value of 0.85 (pKa3, Table 1) to the piperidino or the aminomethyl residue. Owing to the value found for the pKa of the piperidino residue in NHPTT (pKa = 8.10), one can Table 3 Determination of the pKa values of the piperidino, aminomethyl and triazolo groups: pH (m2 = 15 mg) No. V*/cm3 1 2 3 4 5 6 Average Variance 1 1 1.718 1.694 1.728 1.683 1.773 1.736 1.722 1.03 3 1023 2 2 1.981 1.980 2.003 1.971 2.031 2.020 1.998 5.88 3 1024 3 3 2.164 2.161 2.170 2.132 2.208 2.168 2.167 5.92 3 1024 4 4 2.271 2.292 2.308 2.259 2.328 2.297 2.293 6.21 3 1024 5 5 2.373 2.395 2.428 2.374 2.554 2.402 2.421 4.66 3 1023 6 6 2.474 2.486 2.496 2.463 2.473 2.495 2.481 1.77 3 1024 7 8 2.630 2.624 2.632 2.620 2.634 2.627 2.628 2.74 3 1025 8 9 2.674 2.713 2.704 2.686 2.674 2.700 2.692 2.67 3 1024 9 10 2.744 2.782 2.755 2.750 2.757 2.773 2.760 2.09 3 1024 10 11 2.799 2.820 2.835 2.827 2.810 2.835 2.821 2.07 3 1024 11 12 2.861 2.893 2.878 2.876 2.912 2.882 2.884 2.99 3 1024 12 14 3.013 3.003 2.994 2.996 3.002 3.004 3.002 4.52 3 1025 13 16 3.096 3.145 3.124 3.141 3.128 3.100 3.122 4.18 3 1024 14 18 3.226 3.267 3.250 3.256 3.261 3.260 3.253 2.11 3 1024 15 20 3.402 3.395 3.493 3.394 3.401 3.379 3.411 1.69 3 1023 16 22 3.541 3.531 3.511 3.521 3.579 3.602 3.548 1.26 3 1023 17 24 3.735 3.732 3.731 3.732 3.822 3.768 3.753 1.34 3 1023 18 28 4.288 4.425 4.459 4.399 4.849 4.371 4.465 3.87 3 1022 19 30 5.174 5.188 5.062 5.128 4.960 5.077 5.098 7.11 3 1023 20 35 5.723 5.717 5.756 5.776 5.686 5.649 5.718 2.12 3 1023 21 40 6.059 6.032 6.043 6.073 6.060 5.976 6.041 1.20 3 1023 22 45 6.217 6.245 6.232 6.248 6.267 6.138 6.225 2.07 3 1023 23 50 6.438 6.465 6.464 6.487 6.439 6.363 6.443 1.86 3 1023 24 55 6.892 6.556 6.688 6.671 6.748 6.641 6.699 1.29 3 1022 25 58 6.874 6.788 6.849 6.879 6.868 6.717 6.829 4.13 3 1023 26 60 6.899 6.912 6.937 7.001 6.942 6.761 6.909 6.47 3 1023 27 63 7.136 7.003 7.086 7.071 7.169 6.909 7.062 8.89 3 1023 28 65 7.259 7.404 7.231 7.074 7.248 7.072 7.215 1.59 3 1022 29 68 7.318 7.251 7.266 7.169 7.314 7.193 7.252 3.75 3 1023 30 70 7.271 7.298 7.249 7.248 7.394 7.286 7.291 2.94 3 1023 * Volume of sodium hydroxide solution. 98 Analyst, January 1997, Vol. 122represented by eqn. (1), obtained from the relationships which quantify equilibria in solution and mass and charge balance laws.18. |H+ |5 + (Ka3 + C)|H+ |4 + Ka3 Ka2 + CKa3 - m MVi Ka3 - Kw Ê Ë Á � � � |H+ |3 + Ka3 Ka2 Ka1 + CKa3 Ka2 - 2 m MVi Ka3 Ka2 - KwKa3 Ê Ë Á � � � |H+ |2 + Ka3 Ka2 Ka1 C - 3 m MVi Ka3 Ka2 Ka1 - KwKa3 Ka2 Ê Ë Á � � � |H+ |- KwKa3 Ka2 Ka1 = 0 (1) Where ýH+ý stands successively for activity and concentration (see below), Ka3, Ka2 and Ka1 are the searched for ionization constants introduced above (Scheme 1), m the weighed amount of NHPTO, M the molecular mass of NHPTO, Vi is the added volume of sodium hydroxide solution, C is the concentration of the sodium hydroxide solution and Kw is the ionic product of water.pH values being defined in terms of activity, the problem of the ionic strengths of the solutions had to be taken into account. This was performed by a classical iterative process. 18 For each set of chosen thermodynamic Ka1, Ka2 and Ka3 values, eqn. (1) was solved for ýH+ý several times.At the beginning of the process, a first estimation of ýH+ý was calculated by solving eqn. (1). At this stage, the calculated ýH+ý value was neither an activity nor a concentration value because of the necessary initial mixing of activities (to which Ka1, Ka2, Ka3 and Kw values refer) with concentrations C and m/MVi, to which equations of mass and charge balances pertained. However, the first obtained value of ýH+ý permitted the Scheme 1 Ionization scheme for NHPTO and nomenclature. 96 Analyst, January 1997, Vol. 122calculation of a first pseudo-ionic strength, which, in turn, allowed the estimation of a first set of activity coefficients of all species. With these activity coefficients, a set of equilibrium concentration constants Ka1, Ka2, Ka3 and Kw was therefore computed from the thermodynamic values. A new value of ýH+ý was obtained, and the process was repeated until the ionic strength was constant. During the whole process, the Ka1, Ka2, Ka3 and Kw constants were hence successively thermodynamic constants and then, more and more, equilibrium concentration constants, while ýH+ý became more and more a concentration value.After convergence of the ionic strength, the [H+] concentration was transformed back to {H+} activity with the help of the last activity coefficient of the proton. The activity coefficients used throughout this process were computed through the extended Debye–H�uckel relationship.19 -loggi = Azi 2 I 1+ Ba I or with the Davies relationship when the ionic strength was over 0.01 mol dm23.20 -loggi = Azi 2 I 1+ I + 0.1 zi 2 I For each volume Vi of sodium hydroxide solution added (the concentrations of which were C1 = 9.00 3 1024 mol dm23 and C2 = 9.35 3 1024 mol dm23), six solutions corresponding to the same amount of NHPTO (m1 = 10 mg and m2 = 15 mg) were prepared in an absolutely independent manner.Thirty different volumes from 1 to 70 cm3 were used.The least-squares function was defined as: U = i=1 n wi (pHi calc - pHij exp )2 j=1 m where n is the number of points treated, m the number of replicates (generally m = 6), i the running point, j the replicate considered at point i, and wi a weighting factor chosen as the inverse of variance spHi 2 taken over the m replicates for the same volume Vi. This U function was compared with a theoretical minimum cost function U0, which only took into account the random experimental errors: U0 = i=1 n wi (pHi mean - pHij exp )2 j=1 m Table 1 Determination of the pKa values of the piperidino, aminomethyl: results obtained by treating pH data (from 30 different volumes of sodium hydroxide solution) Parameter m1 = 10 mg* m2 = 15 mg† pKa1 7.63 7.37 (7.60; 7.67) (7.33; 7.41) pKa2 6.43 6.28 (6.40; 6.45) (6.25. 6.32) pKa3 0.85 22.60 (0.71; 1.05) (22.87; 21.89) U0 150 150 U 503 518 * CNaOH = 9.00 3 1024 mol dm23. † CNaOH = 9.35 3 1024 mol dm23.Table 2 Determination of the pKa values of the piperidino, aminomethyl and triazolo groups: pH data (m1 = 10 mg) No. V*/cm3 1 2 3 4 5 6 Average Variance 1 1 1.971 1.947 1.864 1.860 1.978 1.918 1.923 2.68 3 1023 2 2 2.249 2.249 2.198 2.130 2.219 2.199 2.207 1.95 3 1023 3 3 2.433 2.453 2.369 2.341 2.427 2.367 2.398 2.00 3 1023 4 4 2.565 2.589 2.504 2.485 2.537 2.497 2.530 1.71 3 1023 5 5 2.678 2.705 2.621 2.601 2.699 2.672 2.663 1.79 3 1023 6 6 2.778 2.805 2.700 2.693 2.774 2.745 2.749 2.03 3 1023 7 8 2.966 2.990 2.914 2.850 2.980 2.923 2.937 2.76 3 1023 8 9 3.067 3.092 2.962 2.971 3.064 3.023 3.030 2.90 3 1023 9 10 3.155 3.219 3.057 3.081 3.138 3.097 3.125 3.45 3 1023 10 11 3.245 3.314 3.171 3.162 3.195 3.170 3.210 3.45 3 1023 11 12 3.366 3.393 3.295 3.157 3.306 3.273 3.298 6.85 3 1023 12 14 3.569 3.587 3.488 3.466 3.580 3.450 3.523 3.85 3 1023 13 16 3.875 3.903 3.852 3.801 3.786 3.675 3.815 6.67 3 1023 14 18 4.138 4.354 4.185 4.210 4.420 4.002 4.218 2.27 3 1023 15 20 5.004 5.191 4.959 5.285 5.214 4.892 5.091 2.55 3 1023 16 22 5.485 5.480 5.528 5.625 5.605 5.522 5.541 3.71 3 1023 17 24 5.747 5.839 5.855 5.829 5.895 5.768 5.822 3.06 3 1023 18 25 5.946 5.967 5.890 5.845 5.897 5.807 5.892 3.60 3 1023 19 30 6.240 6.243 6.238 6.341 6.418 6.276 6.293 5.31 3 1023 20 35 6.669 6.705 6.624 6.618 6.714 6.634 6.661 1.74 3 1023 21 40 6.845 7.101 7.054 6.987 7.005 6.879 6.979 9.84 3 1023 22 45 7.337 7.420 7.288 7.298 7.459 7.355 7.360 4.60 3 1023 23 50 7.596 7.802 7.561 7.454 7.593 7.692 7.616 1.41 3 1022 24 55 7.905 8.116 7.930 7.960 7.887 7.702 7.917 1.78 3 1022 25 58 8.456 8.384 8.401 8.491 8.611 8.500 8.474 6.70 3 1023 26 60 8.501 8.634 8.554 8.662 8.696 8.665 8.619 5.65 3 1023 27 63 8.901 9.179 8.890 9.432 9.024 9.010 9.073 4.19 3 1022 28 65 9.563 9.512 9.032 9.559 9.559 9.231 9.409 5.06 3 1022 29 68 9.644 9.798 9.701 9.593 9.711 9.604 9.675 5.97 3 1023 30 70 9.944 9.905 9.746 9.794 9.759 9.774 9.820 6.92 3 1023 * Volume of sodium hydroxide solution.Analyst, January 1997, Vol. 122 97consider that the higher value found for pKa2 or pKa1, i.e., 7.50, is representative mostly of the ionization of the piperidino moiety in NHPTO and hence the value of 6.36 for the ionization mostly of the conjugated acid of the amino group. These assumptions are in agreement with the finding of Sayer et al.,24 who noted that pKa values of the conjugated acids of imines Fig. 1 Titration curve obtained with m2 = 15 mg and CNaOH = 9.35 3 1024 mol dm23 and theoretical curve calculated with pKa3 = 22.60, pKa2 = 6.28 and pKa1 = 7.37.Scheme 2 Microforms of NHPTO. Table 5 Determination of the pKa values of the carbonyl function: spectrophotometric data for H0 values ranging from 28.50 to 25.18 No. H0 1 2 3 4 5 6 Average Variance l2 = 300 nm 1 25. 18 0.3721 0.3751 0.3760 0.3830 0.3754 0.3667 0.3747 2.84 3 1025 2 25. 48 0.3873 0.3812 0.3870 0.3914 0.3720 0.3790 0.3830 4.91 3 1025 3 25. 92 0.4015 0.4079 0.3961 0.3962 0.4068 0.3984 0.4012 2.70 3 1025 4 26. 50 0.4136 0.4185 0.4312 0.4143 0.4169 0.4072 0.4170 6.38 3 1025 5 27. 00 0.4451 0.4365 0.4570 0.4424 0.4597 0.4500 0.4485 7.85 3 1025 6 27. 50 0.4718 0.4662 0.4742 0.4772 0.4747 0.4637 0.4713 2.78 3 1025 7 28. 00 0.5107 0.5318 0.5428 0.5070 0.5010 0.5324 0.5210 2.85 3 1024 8 28. 50 0.5412 0.5505 0.5600 0.5462 0.5464 0.5537 0.5497 4.36 3 1025 l1 = 260 nm 1 25. 18 0.7773 0.7669 0.7968 0.7968 0.7753 0.7754 0.7814 1.55 3 1024 2 25. 48 0.7695 0.7514 0.7349 0.7663 0.7582 0.7567 0.7562 1.52 3 1024 3 25. 92 0.7542 0.7424 0.7245 0.7471 0.7395 0.7545 0.7437 1.25 3 1024 4 26. 50 0.7250 0.7316 0.7475 0.7369 0.7352 0.7380 0.7357 5.55 3 1025 5 27. 00 0.7011 0.6898 0.7364 0.7101 0.7070 0.7224 0.7111 2.68 3 1024 6 27. 50 0.6772 0.6736 0.6814 0.6886 0.6879 0.6746 0.6806 4.29 3 1025 7 28. 00 0.6465 0.6347 0.6698 0.6407 0.6337 0.6692 0.6491 2.71 3 1024 8 28. 50 0.6054 0.5987 0.6270 0.6142 0.5618 0.6253 0.6054 5.77 3 1024 Table 4 Determinaion of the pKa values of the carbonyl function: results obtained by treating spectrophotometric data for H0 values ranging from 28.50 to 25.18 Parameter l2 = 300 nm l1 = 260 nm pKa4 27.45 27.47 (27.49; 27.39) (27.61; 27.28) eBH4 4+ * 5250 10300 (5200; 5300) (10200; 10400) eBH33+* 7580 8250 (7520; 7640) (8000; 8550) U0 40 40 U 147 78 eBH4 4+* 4690 12250 (experimental) eBH3 3+* 7980 7150 (experimental) * e in dm3 mol21 cm21. Analyst, January 1997, Vol. 122 99were seven pH units lower than those of the corresponding amines. We found for NHPTT a value of pKa = 20.24, which could be mostly assigned to the protonated imine. Assigning the pKa = 6.36 value rather to the aminomethyl residue is also consistent with the hypothesis of Konishi et al.,10 who suggested a value pKa = 6.50 for this group in the opened form of triazolam. This value was given by analogy with that found for 2-aminoacetamido-5-chlorobenzophenone. It is worth noting that both values, 6.36 and 7.50, are undoubtedly weak values for benzylamine groups.The pKa3 values were not consistent. The lack of agreement between them can be explained by the fact that the U function is very insensitive to the Ka3 paremeter, i.e., a large variation in Ka3 only induces very small changes in U. From a chemical standpoint, this means that in the range of the lowest pH values, triazole is already in its basic form. Ideally, such a low pKa3 value would require the use of an acidity function and hence of UV spectrophotometric techniques.As data obtained in this way would not be usable because of the lack of variation of the UV spectra in the range of concern, potentiometry was used. As a result, the pKa3 values, given in Table 1, must be considered only qualitative and only indicative of the strong acidity of the protonated triazolo nucleus. However, owing to the difference between the pKa3 value and the of pKa2 and pKa1 values on the one hand and the difference between the pKa3 and pKa4 values (see below) on the other, pKa3 is only representative of the ionization of the triazolo ring.We can note that the ionization constant of the conjugated acid of the triazolo nucleus is much lower than that usually encountered with 1,2,4-triazoles25–28 and to a lesser extent with that encountered in NHPTT. The difference between NHPTT and NHPTO can be tentatively assigned to the presence of two strongly attractive groups, the carbonyl and the protonated aminomethyl residues, in the latter molecule.Acid–Base Pair BH4 4+–BH3 3+ The values found by UV spectrophotometry for pKa4 are given in Table 4. Also given are U0 and U values together with those of the molar absorptivities eBH4 4+ and eBH3 3+ obtained by curve fitting, for the sake of comparison with the experimental values. Experimental data are given in Table 5. The values obtained for both wavelengths are in very good agreement.Other values given in Table 4 are satisfactory together with the distribution of the residues (Aexp 2 Acalc), which are alternately positive and negative. The value pKa4 = 27.46 (27.61; 27.28) is that expected for a thienophenone on the H0 scale.13 The authors are grateful to G. Bouer for his assistance. References 1 Inotsume, N., and Nakano, M., J. Pharm. Sci., 1980, 69(11), 1331. 2 Pfendt, L. B., and Popovic, G. V., J. Chem. Soc., Perkin Trans. 2, 1994, 1845. 3 Maudling, H. V. Nazareno, J. P., Pearson, J. E., and Michaelis, A. F., J. Pharm. Sci., 1975, 64, 278. 4 Inotsume, N., and Nakano, M., Chem. Pharm. Bull., 1980, 28(8), 2536. 5 Vir�e, J. C., and Patriarche, G. J., J. Electroanal. Chem., 1986, 214, 275. 6 Jimenez, R. M., Alonso, R. M., Oleaga, E., Vicente, F., and Hernandez, L., Fresenius’ Z. Anal. Chem., 1987, 329(4), 468. 7 Vir�e, J. C., Gallo Hermosa, B., and Patriarche, G. J., Analusis, 1987, 15(9), 499. 8 Pfendt, L. B., Janjic, T. J., and Popovic, G. V., Analyst., 1990, 115, 1457. 9 Moro, M. E., Novillo-Fertrell, J., Velaz, M. M., and Rodriguez, L. J., J. Pharm. Sci., 1991, 80(5), 459. 10 Konishi, M., Hirai, K., and Mori, Y., J. Pharm. Sci., 1982, 71, 1328. 11 Legouin, B., and Burgot, J. L., Analyst, 1996, 121(1), 43. 12 Legouin, B., and Burgot, J. L., in the press. 13 Rochester, C. H., in Acidity Functions, Academic Press, London, 1970, vol. 17, p. 24. 14 Bascombe, K. N., and Bell, R. P., J. Chem. Soc., 1959, 1096. 15 Blackwell, L. F., Fischer, A. Miler, J. J., Topson, R. D., and Vaughan, J., J. Chem. Soc., 1964, 3588. 16 Sayer, J. M., Prinsky, B., Schoubrum, A., and Wazshtien, W., J. Am. Chem. Soc., 1974, 96(26), 7998. 17 Rosenberg, S., Silver, S. M., Sayer, J. M., and Jencks, W. P., J. Am. Chem. Soc., 1974, 96(26), 7986. 18 Butler, J. N., in Ionic Equilibrium. A Mathematical Approach, Addison-Wesley, Reading, MA, 1964, pp. 440–458. 19 Harned, H., and Owen, B. B., in The Physical Chemistry of Electrolytic Solutions, Chapman and Hall, London, 1958, p. 66. 20 Rossotti, F., and Rossotti, J. C., in The Determination of Stability Constants in Solutions, McGraw-Hill, New York, 1961, p. 30. 21 Draper, N., and Smith, H., in Applied Regression Analysis, Wiley, New York, 2nd edn., 1981, p. 33. 22 Zekri, O., Boudeville, P., Genay, P., Perly, B. Braquet, P., Jouenne, P., and Burgot, J.-L., Anal. Chem., 1996, 68(15), 2598. 23 Albert, A., and Serjeant, E. P., in The Determination of Ionization Constants, Chapman and Hall, London, 1971, p. 44. 24 Sayer, J. M., Peskin, M., and Jencks, W. P., J. Am. Chem. Soc., 1973, 95, 4277. 25 Barton, D., and Ollis, W., in Comprehensive Organic Chemistry, vol. 4, Heterocyclic Compounds, ed. Sammes, P. G., Pergamon Press, Oxford, 1979. 26 Kr�oger, C. F., and Freiberg, W., Z. Chem., 1965, 5(10), 381. 27 Kr�oger, C.-F., and Freiberg, W., Chimia, 1967, 21, p. 161. 28 Fox, J., and Jencks, W. P., J. Am. Chem. Soc., 1974, 96, 1436. Paper 6/04965F Received July 15, 1996 Accepted September 16, 1996 100 Analyst, January 1997, Vol.

ISSN:0003-2654    

DOI:10.1039/a604965f

出版商:RSC    

年代:1997

数据来源: RSC