Electrochemical oxidation of niazid and isoniazid at mercury electrodes. Influence of the adsorption of the reaction product on the polarographic and voltammetric curves

253

J. Electroanal. Chem., 352 (1993) 253-265 Elsevier Sequoia S.A., Lausanne J E C 02636

Electrochemical oxidation of niazid and isoniazid at mercury electrodes. Influence of the adsorption of the reaction product on the polarographic and voltammetric curves J.M. R o d r l g u e z M e l l a d o *, M. A n g u l o a n d R. M a r i n G a l v l n

Departamento de Qufraica Fisica y Termodindmica Aplicada, Facultad de Ciencias, Universidad de C6rdoba, 14004-C6rdoba (Spain) (Received 5 October 1992; in revised form 3 December 1992)

Abstract

The electrochemical oxidation of niazid and isoniazid, at mercury electrodes, was studied by dc and differential pulse polarography and linear-sweep cyclic voltammetry in the pH range 6-13. At pH > 8.5, the positive scans show a prewave, in addition to the main oxidation wave, which can be suppressed by changing experimental variables such as the concentration, temperature and ethanol content in the medium. In the absence of the prewave, Tafel slopes and reaction orders were obtained at the potentials corresponding to the foot of the polarographie waves. On the basis of polarographic, voltammetric and kinetic results and taking into account the literature data, the oxidation processes were found to be of the ECE type, where the rate-determining step was the release of an H + ion from the intermediate formed after two reversible one-electron transfers. The results obtained for the prewave agree with those expected for a process in which the product is more strongly adsorbed than the reactant. It is also shown that the adsorption follows a Langmuir isotherm for which the Gibbs energy of adsorption is potential dependent.

INTRODUCTION N i a z i d a n d isoniazid a r e t h e h y d r a z i d e s d e r i v e d f r o m 3- an d 4-pyridine carboxylic acid (nicotinic a n d isonicotinic acid respectively) w h o s e s t r u c t u r e s ar e CONHNH 2 niazid ~

CONHNH2

isoniazid ~ x ~

* To whom correspondence should be addressed. 0022-0728/93/$06.00 © 1993 - Elsevier Sequoia S.A. All rights reserved

254

Isoniazid is widely used as an antitubercular and anti-actinomycotic agent in medicine because of its antagonism for nicotinic acid and nicotinamide. The polarographic and macroscale oxidation of isoniazid was first investigated by Lund [1] who found that this compound showed one oxidation wave in the pH range 8.5-13.5. The limiting current I L of this wave was not proportional to the isoniazid concentration c and the values of IL/C diminished with this variable at the lowest pH values. This dependence was a function of the supporting electrolyte used. The height of the wave corresponded to a four-electron process at the concentrations usually employed in polarography. Controlled-potential macroelectrolysis performed at pH 11 in phosphate buffer gave 1,2-diisonicotinoyl hydrazide as the product of the bielectronic oxidation together with traces of 4-pyridine aldehyde. In strongly basic solutions, formation of isonicotinic acid was also observed. This acid was not generated through hydrolysis of any of the above compounds, as was verified experimentally. Thus the proposed mechanism was - 2e-

Py--CO--N2H2

) +2OH-

2H20 + Py--CO~N~--~-N-

L N 2 + PyCO +

[ + O H - ~ PyCOOH ~ ~ +PyCON2H2 ~ (PyCONH) 2

The relative decrease of I L with c was explained on the basis of competition between O H - and the isoniazid anion, with partial hydrolysis of the species Py-CO-N=N- explaining the presence of aldehyde. Nevertheless the value of the dissociation pK of the hydrazide group of isoniazid is 11.0 [2] and below pH 10 this compound is not dissociated. Since the anion must be generated from the neutral molecule at these pH values, the slope of the half-wave potential versus pH plot must show a change in slope, in contrast with what is found experimentally. The aim of the work reported in this paper was first to complete the study of the oxidation of niazid and isoniazid on mercury electrodes on the basis of polarographic, voltammetric and kinetic measurements, and second to show the influence of adsorption on the oxidation processes. EXPERIMENTAL

All reagents used were of Merck p.a. grade with the exception of hydrazides which were obtained from Aldrich. A buffer solution consisting of 0.05 M boric acid and sodium bicarbonate was used as the supporting electrolyte. The pH was adjusted with solid NaOH and the ionic strength was adjusted to 0.3 M with NaC1. The working concentration of hydrazides was 1 × 10 -4 M, except in experiments in which the influence of this variable was studied; in this case the range of concentrations employed was 1 × 10-5-5 × 10 -3 M. All solutions were purged

~5 with purified nitrogen and the temperature was kept at 25 ___0.1°C. Triple-distilled water and mercury were also used. An I N E L E C S A computerized assembly equipped with a PDC 1212 potentiostat and a G O T 1018 function generator was used for dc polarography together with a dropping mercury electrode with rn = 1.052 mg s -1, t ---4.1 s (open circuit) in 0.1 M HCIO 4 and h = 50 cm. The drop time was mechanically controlled at 1 s. For dc and differential pulse (DP) polarography on SMDE static mercury drop electrode (and voltammetry, an A M E L 433 polarographic analyser attached to an Intel 80386-based microcomputer was used. The pulse amplitude for DP polarograms was 10 mV and the pulse duration was 50 ms. The drop area was 0.01 cm 2 and an internal A g / A g C l / K C l ( s a t ) electrode was used as the reference. The data were obtained as ASCII computer files which were read using adequate software developed in Microsoft Quick Basic language. This software is interactive with graphical outputs of the data to either the screen or the printer. Thus the contribution of the supporting electrolyte to dc and DP polarograms was eliminated by linear extrapolation of the data obtained at potentials prior to the reduction of the reactant. DP polarograms were analysed using a curve-fitting method and equations previously described [3-5]. Logarithmic analyses of dc polarograms were made in the form of log[Ia/(IL- I)] versus E plots, where a depends on the mechanism considered [6]. Tafel curves were obtained at potentials corresponding to the foot of the dc polarographic waves, i.e. at I < 0.05I L [6]. RESULTS AND DISCUSSION

Oxidation mechanism The electrochemical oxidation of niazid and isoniazid over the p H range 6-13 has been studied using dc and DP polarography. In both cases one or two oxidation waves were obtained depending on the p H of the medium, the concentration of depolarizer, the drop time (pulse duration in DP polarography) and ethanol content in the medium. At p H < 6 the waves overlap with the oxidation of mercury. When the ethanol content in the medium is increased at constant p H and concentration values, the polarograms show only one wave (or peak). This is also observed at low p H ( < 8.5) and concentration values. The overall limiting current in dc polarography is virtually p H independent, having a value corresponding to the transfer of four electrons [1]. This is confirmed by the value of the current which is the same as that obtained for the reduction of the hydrazides in basic media and corresponds to four-electron processes [1,2,7]. The first anodic wave, which is observed at p H > 8.5, is an adsorption prewave, as will be discussed in the following section. In this section only the oxidation of the non-adsorbed hydrazides will be considered. The limiting current of the main oxidation wave in de polarography is proportional to the hydrazide concentration. At p H < 8.5 the half-wave potentials of the

256

100

>

E ElL

-100

0

ILl

-200

-300

Fig. 1. DP polarography: variation with pH of the peak potential of niazid (e) and isoniazid (o). Hydrazide concentration, 1 x 10 -4 M in 20% ethanol; pH 7.01; reference electrode, Ag/AgCl/KCl(sat).

unique wave shift towards more negative values as the pH increases. This variation is linear, with a slope of - 6 3 + 2 mV/decade. In DP polarography, the peak potentials of the corresponding peak lie on the same straight line as those obtained for the main peak above this pH value (Fig. 1). The slopes of these lines are - 62 + 2 mV/decade. Both the peak and half-wave potentials of the wave observed below pH 8.5 are independent of the concentration at low concentration values ( < 1 x 10 -4 M) and shift slightly towards more positive values at high concentrations, probably because of the influence of the adsorption of the reactant. In the same pH range, the peak potential shifts towards more negative values when the pulse duration is increased (Fig. 2) with slopes of - 1 2 mV/decade and - 1 3 mV/decade for niazid and isoniazid respectively at pH 7. Logarithmic analyses of the dc waves in the form of E versus log[I/(l L - I ) ] plots are linear under the conditions where the prewave is not observed, i.e. at pH < 8.5, low hydrazide concentrations or high ethanol content. Figure 3 shows some of these plots. As can be seen, above pH 8.5 the lines are curved because of the presence of the prewave at this concentration. Analogous results were obtained at pH 7 as the reactant concentration is varied: below 1 x 10 -4 M, the logarithmic analyses were linear and at high concentrations the curves were distorted. To study the influence of the ethanol content on the logarithmic

257

30 20 >

E 10LLJ

0-10-

1.0

1:5

2'.0

[og('Z/ms)

Fig. 2. DP polarography: dependence of the peak potentials on pulse duration ~" for niazid (e) and isoniazid (o). Hydrazide concentration, 1 × 10 -4 M in 20% ethanol; reference electrode,' Ag/AgC1/KCl(sat).

4

-z6o

-~6o

3

2

1

o

~6o E/mV

Fig. 3. Logarithmic analyses of the oxidation dc waves of 1 x 10 -4 M niazid (e) and 1 × 10 -4 M isoniazid (o): (1) pH 6.0; (2) pH 7.0; (3) pH 8.0; (4) pH 9.0. Reference electrode, Ag/AgCl/KCl(sat).

258

analysis, solutions containing equimolar quantities of CO3a- and HCO 3 ions were used (i.e. pH = 10.3) and the ethanol concentration was varied between zero and 80% in volume. In these conditions, the linearity of the logarithmic analysis increases as the ethanol content increases; thus for solutions containing 40% of ethanol or more, the logarithmic analyses were linear. In very basic media (pH > 11.5) at moderate hydrazide and ethanol concentrations (< 1 × 10 -4 M, 20%) the logarithmic analyses were also linear. In all cases where the plots were linear, the slopes of the logarithmic analyses have values of 29 + 2 mV/decade. DP polarograms were analysed by using a curve-fitting method and the following equation was derived for first-order processes [3]: L I ' = 4Ip ( 1 + L ) 2 (1)

I' = A I / A E and L = e x p [ - ( E - Ep)/b], where Ip and Ep are the peak intensity and peak potential respectively, and b is a term which coincides with the logarithmic analysis slope in dc polarography. Under the conditions in which the logarithmic analyses were linear, the DP polarograms were symmetrical. The shapes of the peaks correspond to that predicted by eqn. (1), and b values obtained from the best fits of eqn. (1) to the experimental polarograms are close to 29 + 2 m V / decade.

1

2.0 2,0

5

2

3

~ 1.5.

1.0

~ 1.0"

-100 (a)

-50

E/rnV

0

-100

-50

0

50

E/mY

(b)

Fig. 4. Tafel plots: (a) niazid at pH 7 ((1) 3 x 1 0 -4 M; (2) 2 x 1 0 -4 M; (3) 0.6×10 -4 M; (4) 0.4x 10 -4 M; (5) 0.2)<10 -4 M); (b) 1 x l 0 -4 M isoniazid ((1) pH 7.5; (2) pH 7.0; (3) pH 6.75; (4) pH 6.0). Reference electrode, Ag/AgCl/KCl(sat).

259 The kinetic parameters of the oxidation processes were obtained in the range of potentials where the influence of the mass transport on the electrochemical process can be neglected. When one only wave was observed, the log I versus E plots were linear as is shown in Fig. 4. In these cases the Tafel slopes for both hydrazides are 30 + 2 mV/decade. The electrochemical reaction orders were obtained from the variations of these curves with the concentrations of reactants. These orders were unity with respect to the hydrazide concentration and - 2 with respect to the H + ion concentration. In the absence of the prewave, the linearity of the logarithmic analyses and the shape of the DP polarograms indicate that the oxidation processes are first order with respect to the hydrazides. This is confirmed by the lack of dependence of the half-wave and peak potentials on the reactant concentration, at low concentration values, and by the value of the electrochemical reaction order. The values of both the Tafel and logarithmic analysis slopes, as well as of the b parameter in DP polarography, indicate that there is a reversible transfer of two electrons. Moreover, the dependence of the peak potentials on the pulse duration in DP polarography indicates that the process is of the EC type, i.e. the ratedetermining step must be a chemical reaction which follows two reversible oneelectron transfers [5]. The involvement of the H + ion on the electrochemical, process is evident from the variations of the half-wave and peak potentials with the pH, and from the value of the electrochemical reaction order with respect to this species. Thus on the basis of the above results and taking into account the data reported in the literature, the following reaction scheme is proposed: P y - C O - N H - N H 2 ~ Py-CO-NH=NH + 2H++ 2e-

(2)

Py-CO-NH=NH

(3)

rds P y - C O - N H = N - + H +

P y - C O - N H = N - + 2 O H - ~ PyCOO- + N 2 + HzO + 2e-

(4)

The intermediate products and end-products have been identified elsewhere using controlled-potential electrolysis [1]. Reaction (3) is proposed as the rate-determining step since it is the only one compatible with our experimental results. The I - E - t relationship for dc polarography derived from the above scheme is

[5] E=E~ +

RT

RT In(k3c ~) -

[ 3rrt +

RT

[ IL-I) In[ - - 7 - -

(5)

where cH is the concentration of the H + ion, D is the diffusion coefficient of the hydrazide and k 3 is the rate constant of reaction (3). It is evident from this equation that the plots of E versus log[I/(I L - I ) ] must be linear with slopes of 29.6 mV/decade (at 25°C). Moreover, the slope of the Ell 2 versus pH plot must be - 2 . 3 0 3 R T / F ~ 60 mV/decade. Finally, eqn. (1) can be derived from eqn. (5) under the conditions of DP polarography, with b = R T / F [3]. All these predictions agree with the experimental results.

260

."" .........i .....

.,.

.' ....

<~ I.O ;/.......... -~.

.... .gr

J

1

'dL ................3 .......... ~;'/'"3:'~""~"':

-400 (o)

-360

0 . . . . . ":/:?$"' -z,bO

E/mV

..

,'/ .':............

/(

.

...:"2!"

-i?

/

0.5 0

'

..

3".

,

-.

"...:..'. ""..":::'~......

-3()0

....

E/mY

(b)

Fig. 5. Effect of the niazid concentration on (a) dc and (b) DP polarograms at pH 12: (1) 2.2× 10 -4 M; (2) 1.2 × 10 -4 M; (3) 0.5 × 10 -4 M. Reference electrode, Ag/AgCl/KCl(sat).

The results obtained at the foot of the wave confirm the above scheme. Thus the relationship in this zone of potentials is

I-E I = 2FK2k3K'c~2cA exp(2FE/RT) (6) where K ' = exp(-2FAqbref/RT), At~ref being the potential of the reference electrode, K 2 is the equilibrium constant of reaction (2), cA is the hydrazide concentration and the rest of the symbols have their usual meanings. The electrochemical reaction orders with respect to both the hydrazides and the H ÷ ion and the values of the Tafel slope derived from this equation (1 and 2, and 29.6 mV/decade respectively) agree with the experimental values.

Adsorption wave At pH > 8.5 the overall limiting current in dc polarography is proportional to the hydrazide concentration, whereas that of the first wave depends slightly on (or does not vary with) this variable, as shown in Figs. 5 and 6. The behaviour of the peak currents in DP polarography is similar (Fig. 5(b)). This fact, together with the effect of the different experimental variables on this wave, indicates that this corresponds to the oxidation of the hydrazide in the adsorbed state. It is known that the appearance of a prewave is associated with an electrochemical process occurring between adsorbed species, where the product is more strongly adsorbed than the reactant [8]. Thus at concentrations above a given value c m, at which the film becomes compact, the relationship in d c polarography for the prewave corresponding to the reversible process R ~ O + n e - is given by

I-E

[8]

E = EO_ 2.303RT ( K_~o) + 2.303RT (-1L~ ) n---~-- log n~ log

(7)

261

~2 i-,.-*

0

i

;~

:]'

4 c,, lO'r'/M

Fig. 6. Dependence of the limiting current of the first anodic wave on the reactant concentration at oH 12: • niazid; o isoniazid. Reference electrode, Ag/AgCl/KCl(sat).

where I L depends on c m (or Fro, i.e. the maximum surface concentration of O or R) but not on the concentration of the reactants; K o and K R are the adsorption coefficients of the Langrnuir isotherm of both the oxidized and reduced species. Thus the equation for a Langmuir isotherm is [9] c i

Fi = Fm K i + c i

i = O, R

(8)

Equation (7) predicts linear logarithmic analyses with slopes of 2.303RT/nF = 60/n mV/decade. In our case the logarithmic analyses are linear, but the slopes have a mean value of 12 + 2 mV/decade. Although this could be compatible with n = 4, it seems reasonable to assume that the oxidation mechanism must be similar to that observed for the main wave. A possible explanation of the value of the slope of the logarithmic analysis involves the assumption that the Gibbs energy of adsorption varies with the potential. This dependence can be introduced in an approximate form as was done by Wopschall and Shain [9] on the basis of the previous work of Parsons [10]:

' [ °'°nF(E-E°)] K o = K o exp - R T

(9)

where o"o determines the magnitude of the influence of the field on the adsorption coefficient, which becomes important for the case of strong adsorption. Introducing eqn. (9) into eqn. (7) we obtain

E=EO

2.303RT (K~) 2.303RT (/L-~) (1 +O.o)n F log + (1 +~ro)n F log

(10)

i.e. the slope of the logarithmic analysis is changed by a factor of 1 + ~ro. By using the experimental value we have o-o = 1.5, assuming n = 2.

262

:f .....

2

:t .....

iI

:I

~ ~o ~

20

20

1

1

- 5 0t 0

-400

- 3 0m 0

-200

(0 )

-100

0

E/mV

-500 (b)

-4,00

-300

- 2 0j 0

_1100

0

E/mV

Fig. 7. Effect of the scan rate on the cyclic voltammograms for aqueous solutions of 1 x 10 -3 M hydrazides at pH 12. Scan r a t e s / V s -1 (from bottom to top): (a) 0.1, 0.2, 0.5, 1.0, 2.0; (b) 0.05, 0.5, 1.0, 2.0. Reference electrode, Ag/AgCl/KCl(sat).

l

I

I

!

I O0

80 0

E <~ _"- 6O U

-~

40

% -

-

20

0 I

I

I

I

-400

-300

-200

- 1O0 E/mY

Fig. 8. Dependence of the cyclic voltammogram current function I/vi/2c on the niazid concentration in aqueous solution at pH 12: (a) 10.0× 10 - 4 M; (b) 5.0)< 10 - 4 M; (c) 1.0)< 10 - 4 M . v = 2.0 V s - l ; reference electrode, Ag/AgC1/KCl(sat).

263

These results can be confirmed using linear sweep voltammetry. Thus at high concentrations a sharp prepeak was observed in addition to the main oxidation peak for both compounds (Fig. 7). As can be seen, the prepeak is more evident at high scan rates, in accord with Wopschall and Shain [9]. However, the effect is stronger for niazid than for isoniazid. The prepeaks are more pronounced in very basic media and the peak intensity decreases as the pH decreases. The proximity at the prepeak and the main oxidation peak prevents measurement of its half-width, which must be related to tro. Nevertheless, the value of this last parameter indicates that the prepeak must be very sharp compared with the main oxidation peak, as illustrated in Fig. 7(a) [9]. At sufficiently high scan rates, the intensities of the prepeaks are almost proportional to the scan rate, as is expected theoretically [8,9]. At low scan rates the adsorption peak current decreases and the ratio of the diffusion peak height to the adsorption peak height increases (Fig. 7). As the bulk concentration of hydrazide is increased, there is a relative decrease in the height of the adsorption peak compared with that of the diffusion peak. This behaviour is shown in Fig. 8 for niazid. Below a given concentration value, only the adsorption peak is observed, which indicates that the bulk concentration is less than c m and the film is not compact [9]. Figure 9(a) shows the dependence of the peak current of the adsorption peak on the hydrazide concentration. As can be seen, these values tend

i

I

i

i

i

4O

2O

I0 0

0

0

I 2

I 4

I e

I 8

I 10

~I04/m~ i-I

I

i

i

I

t

2

4.

6

8

10

c x l O 4" / t o o l 1-1 Fig. 9. Variation of the cyclic voltammogram peak current of the prepeak with the niazid concentration at pH 12 in aqueous solution. Reference electrode, Ag/AgCl/KCl(sat). The inset shows the linearized Langmuir isotherm (plot corresponding to eqn. 12),

264

lO

8 =t 6 i--,i

4 2 0 I

-400

!

-300

I

-200

I

- 1 O0

E/mV Fig. 10. Effect of the scanning rate on the cyclicvoltammogramsfor 1 × 10 - 4 M aqueous isoniazid at pH 12. Scan rates/V s-l: (a) 0.1; (b) 0.5; (c) 1.0; (d) 2.0. Reference electrode, Ag/AgCl/KCl(sat). to a limiting value when the concentration increases. This is due to the saturation of the electrode surface at high bulk concentration values predicted by the Langmuir isotherm. The peak area A(a) of the adsorption peak obtained by integration of the current-potential curves is related to the surface excess F R of the species R by the equation [11]

A ( a) = nFS F R

(11)

where S is the electrode area. In addition, the equation for a Langmuir isotherm can be expressed in linear form as:

cR/ra --- KR/Fm

+ CR/Fm

(12)

Figure 9(b) shows the plot corresponding to the voltammograms of Fig. 9(a). As can be seen, the experimental results agree with the theoretical predictions, although small deviations from the regression line are observed at low c R values because of the overlap between the adsorption and the diffusion peak, which makes integration of the former difficult. Finally, Fig. 10 shows the effect of the scan rate on the voltammograms of isoniazid at a concentration where only the adsorption peak is observed. In conclusion, the voltammetric results agree with the theoretical predictions derived for a process in which the product is adsorbed more strongly than the reactant.

265 ACKNOWLEDGEMENTS

Authors wish to express their acknowledgement to Junta de Andalucla for the financial support of this work. REFERENCES 1 H. Lund, Acta Chim. Scand., 17 (1963) 1077. 2 M. Angulo, R. Marin Gaivin, M. Ruiz Montoya and J.M. Rodriguez Mellado, J. Electroanal. Chem., 348 (1993) 305. 3 J.M. Rodrlguez Mellado, M. Bl~izquez, M. Dominguez and J.J. Ruiz, J. Electroanal. Chem., 195 (1985) 263. 4 J.M. Rodrlguez Mellado, M. Blfizquez and M. Domlnguez, J. Electroanal. Chem., 241 (1988) 291. 5 J.M. Rodriguez Mellado, M. Blfizquez and M. Domlnguez, Comput. Chem., 12 (1988) 257. 6 J.M. Rodrlguez Mellado, M. Bl~zquez, L. Camacho and J.J. Ruiz, J. Electroanal. Chem., 190 (1985) 47. 7 J.M. Rodriguez Mellado and R. Mar[n Galvln, Bull. Electrochem., 7 (1991) 142. 8 E. Laviron, J. Electroanal. Chem., 52 (1974) 355, and references cited therein. 9 R.H. WopschaU and I. Shain, Anal. Chem., 39 (1967) 1515. 10 R. Parsons, J. Electroanal. Chem., 5 (1963) 397; 7 (1964) 136. 11 A.J. Bard and L.R. Faulkner, Electrochemical Methods: Fundamentals and Applications, Wiley, New York, 1980.