Determination of Kinetics of the Karl Fischer Reaction Based on

It was shown that for imidazole-based methanolic reagents exposed to a large variation .... The Use of a Custom-Built Coulometric Karl Fischer Instrum...
0 downloads 0 Views 532KB Size
Download PDF
Anal. Chem. 1996, 68, 784-791

Determination of Kinetics of the Karl Fischer Reaction Based on Coulometry and True Potentiometry Anders Cedergren

Department of Analytical Chemistry, Umeå University, S-901 87 Umeå, Sweden

A new measurement technique based on a combination of coulometry and zero-current potentiometry is described for determination of the kinetics of rapidly reacting Karl Fischer (KF) reagents. This makes it possible to determine the order as well as the rate constant for large variations in the concentrations of iodine and water present during a titration. It was shown that for imidazolebased methanolic reagents exposed to a large variation in the concentration of water, the KF reaction is first order with respect to iodine, sulfur dioxide, and water only for reagents in which the concentration of nonprotonated imidazole is very low. The rate constant determined for such a reagent (1 M imidazole, 0.8 M sulfur dioxide, 0.1 M iodine) was equal to that reported earlier in the literature. Regions showing first-order kinetics were also found for low concentrations of water when imidazole concentrations up to 2 mol/L were used, provided that these reagents had a quotient [Im]free/[ImH+] around 4. In the interval 2-8 mol/L of imidazole, the order of the reaction with respect to iodine was, in most cases, onehalf, while it was changed to between one-half and one with respect to water. The rate of the KF reaction was found to increase by nearly 5 orders of magnitude for a reagent in which the concentration of nonprotonated imidazole was increased from 0 (rate constant equal to 2.6 × 103 L2 mol-2 s-1) to about 7 mol/L. For most of these reagents, a recovery rate close to 100% was attained. A high concentration of nonprotonated imidazole in combination with a high concentration of sulfur dioxide could, however, lead to a change in stoichiometry of the KF reaction when larger amounts of water were determined (250 µg of water added to 3.4 mL of reagent solution). A reaction scheme is proposed which might explain this change in stoichiometry observed for some reagent compositions. By use of the described most rapidly reacting reagents, it was shown to be possible to carry out titrations even at such a low end-point concentration as 10-10 M of iodine within 1-2 min. At present, near-IR spectroscopy is the fastest growing technique for industrial product control of moisture. The success of this technique is, however, critically dependent on the accuracy of the analytical method used to determine the water content of the standards. The Karl Fischer (KF) method is frequently used for this purpose and has a potential of producing very accurate results, provided that interfering side reactions can be brought to a minimum. 784 Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

Surprisingly little work has been reported in the literature on the optimization of the KF reagents in situations where the sample constituents interfere with the KF reagent. Such method developments have been hindered, however, by the lack of fundamental knowledge about the kinetics and mechanisms of the different types of reagents developed during the years. Such studies are not straightforward because the KF systems are relatively unstable and atmospheric moisture as well as the strong adsorption of moisture on surfaces causes great problems. These factors might explain why it was nearly 40 years after the publication of Karl Fischer’s original work1 before any paper appeared in the literature dealing with the kinetics of the KF reaction. In 1974,2 it was shown that the KF reaction is first order with respect to iodine, sulfur dioxide, and water, and the value of the rate constant in a pyridine/methanol reagent was reported to be (1.2 ( 0.2) × 103 L2 mol-2 s-1. These results were later verified by Verhoef and Bahrendrecht,3,4 who determined the reaction rate constant in the pH interval of 2-11 and showed that it increases linearly with pH of the reagent up to pH ≈ 5 and then maintains a constant value between pH 5.5 and 8. At pH >8.5, the reaction rate increases again but to a lesser extent than in the pH range below 5. The apparent dissociation constant for the equilibrium CH3OH + SO2 T H+ + CH3SO3- was found to be 10-5.1 in the presence of 0.5 M sodium iodide.3 Recalculation of the reaction rate constant for the methyl sulfite ion instead of sulfur dioxide resulted in a constant value for the studied pH range up to pH 8.5.3 These results lead to the following proposed reaction for the KF reaction

(I2,I3-) + H2O + CH3SO3-(HB+) + 2B f CH3SO4-BH+ + 2BH+I- (1)

where B is a suitable base. Although not verified experimentally, an altenative mechanism was later suggested by Fischer,5 who proposed that it is not methyl sulfite but bisulfite which is the reactive sulfur species in the KF reaction. Moreover, positive iodine was thought to be the active iodine compound, forming an intermediate, iodosulfonic acid. The reaction was proposed to proceed in three steps: (1) Fischer, K. Angew. Chem. 1935, 48, 394-396. (2) Cedergren, A. Talanta 1974, 21, 265-271. (3) Verhoef, J. C.; Bahrendrecht, E. J. Electroanal. Chem. 1976, 71, 305315. (4) Verhoef, J. C.; Bahrendrecht, E. J. Electroanal. Chem. 1977, 75, 705717. (5) Fischer, W. Kontakte (Darmstadt) 1989, 1, 30-33. 0003-2700/96/0368-0784$12.00/0

© 1996 American Chemical Society

B

H2O + SO2 98 HSO3- + BH+ B

(2)

HSO3- + I+ f [HSO3I] 98 SO3 + HB+ + I-

(3)

SO3 + CH3OH f CH3SO4H

(4)

To judge which one of these two mechanisms is correct is difficult, due to the fact that in methanol both reactions lead to the formation of methyl sulfate. The pH dependence of reaction 2 can also be used to support the mechanism argued by Fischer.5 Verhoef and Bahrendrecht3 also showed that the reaction rate at pH ≈ 6 is the same in reagents buffered with either dichloroacetate or salicylate as in pyridine-buffered reagents. Later, Wu¨nsch and Seubert6 found the reaction rate constant in a reagent buffered with imidazole in the pH range 2-10 to be the same as that found by Verhoef and Bahrendrecht.3,4 Ora¨dd and Cedergren7 verified that the reaction rate of the KF reaction at pH ≈ 6.5 is independent of whether pyridine or imidazole is used to buffer the reagent. They also showed that the reaction rate of the KF reaction is essentially accelerated in the presence of an excess of nonprotonated imidazole. Moreover, it was shown that first-order kinetics were obtained only below pH 7 or above pH 9. For example, it was found that the rate constant increases with the square of the concentration of nonprotonated imidazole at pH 9.4 ([Im]free/[ImH+] ) 4), giving k ) {(27 400 ( 400)[Imfree]2 + (11 200 ( 700)} M-4 s-1. In the pH interval of 7-9, this relation was obtained only for the responses collected during the first 10 s (out of about 30 s) after the sample introduction. It should be pointed out that these investigations were carried out using a relatively narrow iodine excess interval (10-5-10-6 M). It should also be mentioned that the concentration of sulfur dioxide was chosen to be relatively low for the reagents tested. In this way, the rate of the KF reaction is lowered, which means that the error caused by the response time of the indicating platinum electrode could be minimized. The same measuring system was used earlier3,8 and shown to be favorable when compared to both rotating ring-disk9 electrode measurement and the more commonly used biamperometric technique.10 The kinetic techniques used so far to study the KF reaction have been based almost entirely on the response curves obtained within a relatively small range of iodine excess. In this paper, a new kinetic method is described which makes it possible to study the kinetics of extremely rapidly reacting KF reagents, like those based on the presence of high concentrations of nonprotonated imidazole. The method permits the study of the kinetics for large variations in the iodine concentration (7 orders of magnitude) in combination with a large variation in the concentration of water. In this way, new information about the order and rate of the KF titration system will be gained, and hopefully this can be used to spread some more light on the mechanism of the KF reaction. EXPERIMENTAL SECTION Chemicals. Methanol (Merck, p.a.), imidazole (Fluka, puriss p.a.), sulfur dioxide (Fluka, >99.97%), and iodine (Riedel-deHae¨n, p.a.) were used as received. (6) Wu ¨ nsch, G.; Seubert, A. Fresenius Z. Anal. Chem. 1989, 334, 16-21. (7) Ora¨dd, C.; Cedergren, A. Anal. Chem. 1994, 66, 2603-2607. (8) Cedergren, A. Talanta 1974, 21, 367-375. (9) Verhoef, J. C.; Barendrecht, E. J. J. Electroanal. Chem. 1977, 75, 705-717. (10) Cedergren, A. Talanta 1974, 21, 553-563.

Figure 1. Coulometric cell used.

Safety Considerations. Methanol is highly flammable and toxic by inhalation, in contact with skin, and if swallowed. Imidazole is harmful by inhalation, in contact with skin, and if swallowed. Sulfur dioxide is intensely irritating to the eyes and to the respiratory tract. Asbestos is carcinogenic. Reagents. The reagents were prepared by dissolving 1-8 mol/L of imidazole, 0.1-2 mol/L of sulfur dioxide, 0.1 mol/L of iodine in methanol. Coulometric determinations of sulfur dioxide (i.e., the sum of all SIV species obtained from the added sulfur dioxide) was performed iodometrically, according to the procedure described by Cedergren et al.,11 using an LKB 16300 coulometric analyzer. The titration medium for the SO2 determinations consisted of an aqueous solution containing 0.4% KI and 0.68% acetic acid. Standardization. All standard solutions were standardized using calibrated 10-500 µL syringes and reagents which are known to give 100% recovery rate, such as Hydranal Coulomat A or similar reagents. Instrumentation. The cell used for the coulometric procedure was constructed from polymethylpentene (TPX, Mitsui Petrochemical Industries Ltd.) and is shown in Figure 1. It consists of three chambers: one for the auxiliary electrode, one for the injection of the sample through a silicone rubber with both generating and indicating electrodes, and one for the reference electrode. Electrolytic contacts were made using asbestos-filled liquid junctions. All electrodes were manufactured from platinum and connected to an LKB 16300 coulometric analyzer. This instrument contains a high-impedance voltmeter, which accurately measures the voltage between the indicator and the reference electrodes. This voltage is compared with a preset potential, and any deviation is amplified. (The current which is allowed to flow between the generating electrode system is strictly proportional to the difference between the preset and the actual potentials of the solution in which the indicating electrode is positioned.) The (11) Cedergren, A.; Wikby, A.; Bergner, K. Anal. Chem. 1975, 47, 100-106.

Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

785

Figure 2. Typical response curve used to evaluate the kinetics of the reagent under investigation. Eo/mV is the measured potential of the indicating electrode versus the reference which corresponds to the background level. Eimax/mV is the redox potential of the indicating electrode at the maximum (imax) generating current.

species takes part in the reaction. At lower pH, the reaction involves both I2 and I3-, according to the work by Verhoef and Bahrendrecht,3,4 while at higher pH it seems likely that positive iodine (I+,Im2I+) and hydrolysis products of iodine are also involved.12,13 As was discussed in the introduction, l and m are normally considered to be equal to 1. [SO2]eff is a measure of the sum of sulfur compounds which react with iodine in a slightly acidic water media. For a titration of the type shown in Figure 2, d[H2O]/dt, i.e., the rate of water reaction at a certain time, is equal to the difference between the rate of iodine generation and the rate of change in the concentration of iodine, d[“I”2]/dt (i.e., the slope of a curve corresponding to the change in the concentration of iodine during the course of the titration). Therefore, we can write the rate constant as follows:

kl,m ) way in which the end point is approached could be adjusted by adjusting the gain of the instrument so that it decreased rapidly at first and then asymptotically toward the background current at the end of the titration. The current time integral can be followed on a display down to 10-11 equiv, which corresponds to 0.09 ng of water. For the kinetic experiments, a relatively low gain was selected in order to create a favorable measuring situation at the plateau of the titration curve (see Figure 2). The temperature was normally around 21 ( 0.5 °C. The potential between the indicating and reference electrodes could also be followed accurately by using a Fluke 45 dual-display multimeter connected to a computer in which the data collection program Fluke QS 45 was available. Procedure. The excess iodine in the prepared reagent was reduced to a transparent solution (about 5 × 10-3 M of iodine) by carefully adding water with a Hamilton syringe, and then 4.3 mL of this reagent was transferred to each of the cell compartments. A suitably strong water-methanol solution was then carefully added to the working compartment with a 10 µL Hamilton syringe until a suitable iodine level was achieved, e.g., 10-5 mol/L. The cell was then turned upside down to remove any moisture from the walls. The titrator was switched on and the drift value noted. Normally, 5-10 min was needed to obtain a drift value in the range 0.1-0.3 µg/min. When this value had been reached, the calibration curve, i.e., the relationship between the redox potential and the concentration of iodine in the working cell compartment, was obtained by incremental generation of iodine in combination with the measurement of the electrode potential, which typically needed a few seconds to equilibrate. The slope of the calibration curve, as obtained by regression analysis based on five electrode potential readings, did not normally deviate more than 0.2% from the theoretical value in the range 10-7-10-4 M excess iodine. For lower iodine levels, a more elaborate procedure was applied, including a compensation for the background drift. RESULTS AND DISCUSSION Description of the Kinetic Model. The rate expression for the KF reaction can be written as follows:

d[H2O]/dt ) kl,m[“I2”]l[SO2]eff[H2O]m

(5)

The use of [“I2”] reflects the uncertainty about which iodine 786

Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

(i/nF)(1000/v) - d[“I2”]/dt

(6)

[“I2”]l[SO2]eff[H2O]m

where i is in amperes, n ) 2 (2I- f I2 + 2e-), F is the Faraday constant, and v ) 4.3 mL. At imax, d[I2]/dt ) 0, since at this point the rate of water reaction is equal to the rate of iodine generation [(i/nF)(1000/4.3) M s-1]. We can then write

kl,m )

[(i/nF)(1000/v)]imax

(7)

[I“I2”]ilmax[SO2]eff[H2O]immax

[“I”2] in M is obtained by converting the electrode potential at Eimax into a concentration via the calibration curve:

[H2O]imax ) [H2O]added + [indiff]t0imax +

∫

[unreacted H2O]t)0 - ([“I2”]0 - [“I2”]imax) - [

timax

0

i dt] (8)

The term [indiff]timax is obtained simply by multiplying the drift value [(i0/nF)(1000/4.3) M s-1] by timax/s, assuming that the drift is entirely due to diffusion of water into the cell. At steady state (i.e., d[“I2”]/dt ) 0) and in accordance with the rate equation,

[unreacted H2O]t)0 )

(

)

(i0/nF)(1000/4.3) [“I2”]l0[SO2]effkl,m

1/m

(9)

To give an idea about the accuracy expected to be obtained with the proposed method, the data for the titration course shown in Figure 2 will be commented on below. For this experiment, 2.18 ( 0.02 µg of water (28.1 × 10-6 M) was added to the titration cell (4.3 mL) at time t0. The background drift was determined to be 0.12 ( 0.004 µg of water per minute (equal to 2.6 × 10-8 M s-1). The redox potential corresponding to the baseline was -86.1 ( 0.05 mV, which corresponds to an iodine concentration of 5.91 × 10-6 M. imax was 328 ( 1 µA, which corresponds to a reaction rate of water at this point of (3.95 ( 0.01) × 10-7 M s-1, and Eimax (12) Ora¨dd, C.; Cedergren, A. Anal. Chem. 1995, 67, 999-1004. (13) Cedergren, A.; Ora¨dd, C. Anal. Chem. 1994, 66, 2010-2016.

Table 1. Reagents Tested with Respect to the Ratio Taken/Found H2O for Current Densities in the Range 1-20 mA cm-2

A B C D E F G a

[Im] (Μ)

[SO2] (Μ)

[I2] (Μ)

taken/founda (mean value, %) of 2-5 titrations

1 2 3 4 5 8 8

0.1 1.0 0.1 0.2 1.6 0.4 1.4

0.05 0.10 0.05 0.10 0.05 0.05 0.05

99 ( 1 100 ( 0.5 99 ( 0.5 99 ( 0.5 99 ( 1 99 ( 1 see Figure 3

50 and 250 µg of water were tested.

Figure 3. Taken/found water as a function of the maximum current density used for reagent G. For the calculations, a 1:1 stoichiometric ratio between iodine and water was assumed.

was -209.0 ( 0.03 mV, which is equal to an iodine concentration of 3.9 × 10-10 M. The time needed to reach imax was 22 ( 2 s, which gives the concentration of water which had diffused into the cell during this time: 22 (2.6 × 10-8 M) ) 0.57 × 10-6 M. The current integral between to and timax was obtained from the integrator of the instrument, and this value corresponds to the addition of (12.8 ( 0.2) × 10-6 M iodine. The value of the unreacted water present in the titration solution at time to was for the very rapid reagent used for the experiment given in Figure 2 (2.08 M SO2eff, 8 M imidazole, 0.2 M iodide in methanol) without significance (2 × 10-10 M). This means that [H2O]imax can be estimated to 8.8 × 10-6 M. The value of klm, then, is equal to

3.95 × 10-7 M-2 s-1 ) (3.9 × 10 )(2.08)(8.8 × 10-6) -10

5.5 × 107 M-2 s-1 for l ) m ) 1 according to eq 7. Since the coulometric method

Figure 4. Test of different reaction orders (upper diagrams) for reagents containing 1 M imidazole (total). The lower diagrams give the values of the concentrations of iodine and water at imax for the different amounts of water injected. For example, C (left diagram) denotes an experiment in which 250 g of water was injected, and by locating C in the lower diagram, it can be seen that at imax the concentrations of iodine and water are 10-8 and 10-2.5 M, respectively.

used for determination of [SO2]eff gives an accuracy of about 99.9%, the error in the value of k should be less than a few percent, provided that a true steady state equilibrium state is attained at imax and that the theoretical slope of the calibration curve electrode potential versus the iodine concentration is obtained even at this very low level. Investigations of the Stoichiometry and Current Efficiency. Table 1 summarizes a number of reagent types which will later be examined with respect to their kinetic behavior. As can be seen, except for reagent G, the recoveries are close to 100% in the range 1-20 mA cm-2. The results for reagent G is given in Figure 3, and it can be seen that the same high recovery is obtained for the smaller amount of water. The fact that the quotient taken/found exceeds 100% indicates that the stoichiometry of the KF reaction is shifted from 1:1 toward 1:2 (iodine: H2O). In separate experiments it was found that the recovery could be increased to 99.5 ( 0.5% simply by including a waiting period of about 10 min before the titration was started. Evidently, slow kinetics are involved in the establishment of the equilibrium Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

787

Figure 5. Test of different reaction orders for reagents containing 2 M total concentration of imidazole. For explanation, see Figure 4.

between the free sulfur dioxide and water-forming sulfite. During the waiting period, the water available for the hydrolysis of iodine in the titration may be lowered, while at the same time the concentration of sulfite is increased. It seems likely that the possible nonstoichiometric reaction CH3OH

“HIO” + SO32-(Im H+)2 f SO3 9 8 CH3SO4-Im H+ Im

is discriminated in this way. “HIO” is used to indicate the uncertainty of the composition of the hydrolysis product. Investigations of the Kinetics. Eight different types of reagents based on imidazole/methanol were examined, and the compositions of these are given in Figures 4-7. (In the discussion of Figures 4-7, a refers to the left-hand side diagrams and b to the right.) Total concentration of 1, 2, 5, and 8 M imidazole were selected for the investigations, each in combination with a low and a high concentration of sulfur dioxide. For each reagent, firstorder kinetics was tried, the results of which are shown in the figures along with the “best fit”, i.e., the values of m and l which give the flat line. The values of the quotient [Im]free/[ImH+] were calculated on the basis of the fact that 1 mol of added sulfur dioxide gives 1 mol of acid when a reagent is properly buffered with imidazole (see introduction), and 1 mol of iodine gives 2 mol of acid when reacting with water (reaction 1). The measured pH for a solution in which [Im]free/[ImH+] is equal to 1 will be 8.812 when the following methanolic buffer is used:12 trichloroacetic acid/sodium 788 Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

Figure 6. Test of different reaction orders for reagents containing 5 M total concentration of imidazole. For explanation, see Figure 4.

hydroxide (pH ) 4.9), salicylic acid/sodium salicylate (pH ) 7.9), acetic acid/sodium acetate (pH ) 9.7). The pH value of the reagent reported in Figure 4a was earlier12 determined to be 6.6, while that in Figure 4b was determined to be 9.4, which is the same as that obtained by calculation. Considering Figure 4a, the results are as expected; i.e., under these conditions, the KF reaction is first order with respect to iodine, sulfur dioxide, and water. The value of the rate constant, 2.6 × 103 L2 mol-2 s-1, is in line with that recently reported by Ora¨dd and Cedergren7 [(2.5 ( 0.1) × 103 L2 mol-2 s-1] for a similar reagent. The value obtained for [Im]free/[ImH+] ) 4, as given in Figure 4b, is also identical to that reported earlier7 for up to 10 µg of water added (k ) 3 × 104 L2 mol-2 s-1). By inspecting the lower curve in Figure 4b, it can be seen that the region where a first-order reaction is found corresponds to a variation in the concentration of iodine in the interval 10-5-10-6 M. This measuring interval coincides with the condition used in the study of Ora¨dd and Cedergren.7 However, for larger sample amounts, the reaction order shifts from l ) 1, m ) 1 to l ) 1/2, m ) 1/2. It is interesting to see that the latter combination of exponents can be used to describe the kinetics over nearly 3 orders of magnitude. (Relative standard deviation (RSD) of the rate constants corresponding to points A-F was about 30%). Considering the results obtained for 2 M imidazole, as shown in Figure 5, only for the lower concentration of sulfur dioxide

Figure 8. Recorded titration curves for two different KF reagents using different end-point concentrations of iodine. Figure 7. Test of different reaction orders for reagents containing 8 M total concentration of imidazole. For explanation, see Figure 4.

(higher pH) it was possible to find a combination of m and l which gave a straight line. Again, m ) 1/2 and l ) 1 seem to fit relatively well (RSD 25%). When the imidazole concentration is increased further, it can be seen that for the lower quotients ([Im]free/[ImH+] ) 0.9 and 5 (Figures 6b and 7b) as compared to the quotients 15 and 16 (Figures 6a and 7a), the KF reaction is best described by using the exponent values 1/2, 1/2. It should be emphasized that the KF reaction is accelerated by nearly 5 orders of magnitude when 8 M imidazole is used as compared to the standard pyridine/ methanol reagent. The lower diagrams of Figures 4-7 give the concentration values of iodine and water at imax. The letters A, B, C, etc. correspond to the different amounts of water used in the experiments. The analytical importance of a very rapid KF reaction is demonstrated in Figure 8 for sample amounts normally considered to be troublesome to determine by means of KF titrimetry. The reagent used for the experiments shown in the lower part of the diagram is considered to be a fairly rapidly reacting reagent. In spite of this, it is difficult to accurately determine 1 µg of water using this reagent. If the concentration of iodine at the end point has to be lowered in order to minimize the drift due to iodineconsuming interferents, there is no chance to do this, as is evident from the left-hand side of this bottom figure. A low end-point concentration of iodine may lead to a discrimination of some types

Figure 9. Recorded potentiometric response curves for two different KF reagents, the compositions of which are given in the figure.

of reactions in relation to the main KF reaction. Errors due to such side reactions cannot simply be compensated for by subtracting the background drift, since the rate of the side reaction is normally less during the titration (at lower iodine concentrations), so such a procedure will lead to an overcompensation. A very favorable situation exists when the 8 M imidazole reagent is used, as can be seen in the upper part of Figure 8. For the titration carried out using 10-6 M of iodine in the end point, the KF reaction is so rapid that the factor that limits the time needed for a titration is the response time of the indicating electrode system (i.e., the time needed to establish a steady state equilibrium potential). The right-hand side of the figure shows that it is possible to use an end point as low as 10-10 M of iodine and complete the titration Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

789

Figure 10. Proposed reaction scheme for imidazole-based methanolic KF reagents.

within 2 min. This is a remarkable difference if we compare it with the level of iodine used to control commercial coulometric titrators. In the author’s opinion, high levels often lead to unfavorable measuring situations when iodine-consuming substances are present in the sample.12,15,16 This is especially true for trace water determinations. One drawback encountered with reagents containing such a high imidazole concentrations as 8 M is illustrated in Figure 9. As can be seen, the reaction time is very short (only illustrated for the 8 M reagent) when water is added (change from A to B) to an excess of iodine (in which the reactive iodine compound has had some time to form), in contrast to the case when iodine is generated by a current pulse (change from C to D). In the latter case, about 10 s is needed. For a reagent with 2 M imidazole, the response time (or the time to establish an equilibrium in the iodine chemical system) is only a few seconds. The relatively rapid electrode response at the 10-10 M iodine level (see Figure 8, upper right curve) indicates the presence of a different type of iodine complex as compared to that present at the higher iodine level. Mechanistic Aspects. In Figure 10, some of the most probable reactions are considered. The left-hand side of the diagram corresponds to the mechanisms already suggested by Verhoef and Bahrendrecht3 and Fischer.5 It should be pointed out that the existence of a complex like Im2I+ has been verified (14) Fischer, W.; Beil, S.; Krenn, K.-D. Anal. Chim. Acta 1992, 257, 165-171. (15) Kågevall, I.; Åstro¨m, O.; Cedergren, A. Anal. Chim. Acta 1981, 132, 215218. (16) Lindquist, J. J. Pharm. Biomed. Anal. 1984, 2, 37-44.

790

Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

in the presence of primary, secondary, and tertiary amines.17 Fourier transform Raman studies are in progress in our laboratory to examine the structure of different iodine species relevant to the KF reaction. The deviation from the ideal 1:1 stoichiometry of the KF reaction has been explained by the following reactions

H2O + SO2 + B f BH+HSO3-

(10)

(I3-,I2) + H2O + BH+HSO3- + 2B f (BH+)2SO42- + 2BH+I- (11) which corresponds to a stoichiometry of 1:2 between iodine and water. However, in view of the work reported by Fischer et al.,14 sulfur trioxide is an intermediate in the KF reaction (as well as in the well-known Bunsen reaction). This means that we instead should consider the following reaction sequence:

(I3-,I2) + BH+HSO3- + B f SO3 + 2BH+I-

(12)

B CH3OH

V

BH+CH3SO4In normal KF titrations, methanol is in a very large excess over water, and since the reaction between sulfur trioxide and methanol is very rapid, this reaction should be quite dominant in relation to that between water and sulfur trioxide. (17) Gu ¨ ndu ¨ g, T.; Tastekin, M. Anal. Chim. Acta 1994, 286, 247-251.

The results given in this paper give clear indications that a hydrolysis product of iodine is formed which, when reacting with bisulfite or sulfite, causes a change in the stoichiometry of the KF reaction. This complication was also reported in connection with the determination of water in active carbonyl compounds by means of imidazole-based reagents.12 It should be mentioned that this type of negative error did not show up when the standard less basic pyridine/methanol reagent was used.13 It must be emphasized that the complexity of the KF system containing imidazole as the base is very high and that the chemistry involved is far from being well understood. For example, very few of the rate constants listed in Figure 10 are known. Surprisingly, not even data for the equilibrium reaction SO2 + H2O T HSO3- + H+ are available in the literature for a methanolic solution. This is very regretable, since this equilibrium reaction seems to play a key role in determining the stoichiometry, at least for reagents containing an excess of unprotonated imidazole. The rate of the forward reaction of this equilibrium is also likely to be of great importance for the so-called bisulfite addition reaction, the rate of which should be directly related to the concentration of HSO3-, SO32- in the reagent solutions.

specific applications since the basisity of the reagent, i.e., the quotient [Im]free/[ImH+] can be varied by a factor of about 20, while at the same time the reaction rate constant is kept in the interval 106-108 L2 mol-2 s-1. Moreover, errors due to side reactions may be minimized by lowering the end point concentration of iodine in the coulometric cell. It should be emphasized that a prerequisite for full utilization of the rapidly reacting reagents discussed above is a coulometric system of the type described in this paper. Two very important characteristics of such a system are the low drift (less than one-tenth that obtained with commercial instrumentation) and the possibility to control the iodine level in the end point at concentrations as low as 10-10 M.

CONCLUSIONS By means of the information given in this paper, it is possible to optimize imidazole-containing methanolic KF reagents for

AC950552D

ACKNOWLEDGMENT The author thanks Dr. Eugen Scholz, Riedel-de Hae¨n, for reading the manuscript.

Received for review June 7, 1995. Accepted October 12, 1995.X

X

Abstract published in Advance ACS Abstracts, January 1, 1996.

Analytical Chemistry, Vol. 68, No. 5, March 1, 1996

791