Anal Bioanal Chem (2002) 372 : 513–518 DOI 10.1007/s00216-001-1221-7

O R I G I N A L PA P E R

C. Masalles · S. Borrós · C. Viñas · F. Teixidor

Simple PVC–PPy electrode for pH measurement and titrations

Received: 1 September 2001 / Revised: 7 November 2001 / Accepted: 12 November 2001 / Published online: 11 January 2002 © Springer-Verlag 2002

Abstract Cobaltabis(dicarbollide) [3,3′-Co(1,2-C2B9H11)]–doped polypyrrole (PPy) films have been prepared galvanostatically on glassy carbon electrodes in acetonitrile solution. The potential response behavior of the film of this new material has been investigated in some common pH buffers and in acid–base titrations. The potentiometric characteristics of the resulting films are indicative of a quasi-Nernstian response (approximately 50 mV/pH unit), a linearity range from pH 12 to 3 and correlation coefficients (r2) of approximately 0.98. The electrode is suitable for pH measurements and for monoprotic titrations of strong alkalis with strong acids, and weak bases with strong acids, but the long response time hinders the use of this electrode for multiprotic titrations. The time response has been dramatically improved by reducing the film thickness by using the template effect of a non-conducting polymer (PVC) cast over the graphite surface before PPy deposition. PPy polymerization occurs in the free channels of PVC leading to the formation of PPy wires. The morphological change of PPy does not affect the slope or linearity range. The response of the PVC–PPy electrochemical sensor is rapid and the sensor is easy to prepare, at low cost, and its performance is comparable with that of commercial glass electrodes. Keywords Conducting polymers · pH Electrode · Polypyrrole · Sensor · Titration

Introduction There has recently been much interest in the development of sensors based on intrinsically conducting polymers (ICP), C. Masalles · C. Viñas · F. Teixidor Institut de Ciencia de Materials de Barcelona (ICMAB–CSIC), Campus UAB, 08193 Bellaterra, Spain S. Borrós (✉) Institut Químic de Sarrià, Via Augusta 380, 08017, Barcelona, Spain e-mail: [email protected]

because of their easy preparation, high conductivity, and stability. Nitrate-selective electrodes [1], humidity sensors [2], fluoride electrodes [3], and devices for detection of metal ions [4, 5] based on ICP have been described. Polypyrrole (PPy) has attracted most attention because it is one of the most stable ICP known [6] and one of the easiest to synthesize, and so has tremendous technological potential. Two doping structures have been proposed to exist in PPy [7]. One has the pyrrole unit protonated at the β carbon enabling deprotonation and subsequent dedoping. Thus, it is possible that doped PPy could perform as an H+ sensor. The capacity of PPy incorporating synergetic pHdependent fragments to act as an H+ sensor has been mentioned in previous reports, but the H+ response has never been attributed to the ICP. In this sense, quinhydrone (QH) [8] was chemically incorporated in a PPy producing a pH sensor with a response of 46.0 mV/pH unit. In a similar experiment a non-Nernstian slope of 35.0 mV/pH unit was observed for a plain PPy. Also, thin films prepared from Prussian Blue and N-substituted polypyrroles have been used as pH sensors [9]. These were shown to be suitable for optical determination of pH over pH 5–9 but undergo irreversible spectral changes if exposed to pH>9. The use of an electroinactive PPy film as an insulator, on the other hand, enabled preparation of an amperometric pH sensor on the basis of a metal/insulator/metal device [10]. Although the role of the insulator PPy film can be simply that of a permselective organic polymer this may be the only example yet reported so far for which the pH sensor capacity is attributed uniquely to the PPy; the complex electrode design [10], however, makes measurement of the pH of real samples difficult. Thus the practical implementation of this device is not competitive with the glass electrode in terms of ease of use and price. Some examples of glass-free pH electrodes have been reported before [11]. In this paper, we describe a fully glassfree PPy-modified electrode that can be used as a potentiometric pH sensor, with low cost, and with performance comparable with those of commercial glass electrodes. The cobaltabis(dicarbollide) [3,3′-Co(1,2- C2B9H11)2]– anion (Fig. 1) was used as doping agent in the development

514 electrodes were rinsed with MeCN to remove unreacted monomer and dopant, and then kept in 0.1 mol L–1 NaHCO3 for 2 h for premeasurement conditioning. Conditioning is not necessary for experiments performed in organic media. All titrations were carried out with an automatic titrator (Crison Compact-D). The standard acceptance lecture criterion in this automatic titrator for conventional glass electrodes (drift less than 0.5 mV s–1) was taken for the PPy pH electrodes SEM micrographs of PPy and PPy–PVC composites were taken with a Jeol 5310 scanning electron microscope under accelerating voltage of 20 kV. To obtain higher electrical conductivities and to avoid sample metallization, which could alter the observed PPy morphology, PVC was removed from the electrode surface by immersion in THF before SEM studies.

Results and discussion

Fig. 1 The [3,3′-Co(1,2-C2B9H11)2]– anion

of these PPy-based pH electrodes. The cobaltabis(dicarbollide) anion has high chemical resistance [12, 13, 14] and, as we have described [15], the carborane-doped material has major advantages over common dopant anions. The high boron content in the deposited film, about 40% [15] of total atoms, results in a very high overoxidation resistance [15], preventing OH– attack on the polymeric chain. The use of such a high-volume anion with low charge density also prevents the dopant–anion exchange observed in alkaline media [6, 16, 17]. The development of this all plastic electrode generation will enable the construction of pH sensors without the typical problems associated to the conventional glass electrodes (fragility and response loss in organic media because of removal of water from the gel layer).

For the first PPy based pH electrodes (first generation) the PPy was deposited over bare graphite electrodes to give a homogeneous and continuous film on the electrode surface (Fig. 2). In this figure the white spots correspond to remaining cesium salt of cobaltabis(dicarbollide). Calibration plots (E against pH) for these electrodes were recorded by successive addition of 0.1 mol L–1 HCl to a 0.1 mol L–1 sodium carbonate solution (Fig. 3). The absolute pH was recorded simultaneously using a commercial glass electrode.

Experimental All reagents were obtained from Aldrich Chemicals and were of the highest grade available. Cesium cobaltabis(dicarbollide) Cs[3,3′Co(1,2-C2B9H11)2] was obtained from Katchem. Pyrrole was distilled under vacuum before use. The pH sensors were galvanostatically electrodeposited (1 mA cm–2, for 30 s) on graphite, using an EG&G PAR273A potentiostat-galvanostat, in one step, by anodic oxidation of 0.1 mol L–1 Py and 0.035 mol L–1 Cs[3,3′-Co(1,2C2B9H11)2] in acetonitrile (MeCN) solution in a conventional threeelectrode, one-compartment cell. Platinum wire and Ag/AgCl/ (Bu4N)Cl 0.1 mol L–1 CH3CN were used as counter and reference electrodes, respectively. All solutions were deoxygenated by bubbling with nitrogen for 10 min before polymerization. The ICP was deposited on a graphite electrode (0.6 cm diameter) either bare or covered with poly(vinyl chloride) (PVC). The electrodes were polished with diamond paste (Kelmet, UK; particle size 6 to 0.1 µm) before ICP deposition, in order to minimize the effect of the graphite surface on PPy morphology. The PVC pretreatment had been achieved by successive deposition (1×35 µL, 5×25 µL) of a tetrahydrofuran (THF) solution containing 8.8 mg mL–1 PVC and 4.4 mg mL–1 dioctyl phthalate (added as a plasticizer) to the electrode surface. A homogeneous and bubble-free PVC film is obtained by allowing solvent evaporation for 1 h between additions. The electrodes were then left at room temperature in the vertical position, the soaked part at the top, for 12 h. The PPy-coated

Fig. 2 SEM micrographs of PPy on a bare electrode

Fig. 3 Calibration plot (E against pH) for the PPy-modified electrode

515

Table 1 Potential response of the PPy film electrode in a variety of common buffers Buffer

0.05 mol L –1 potassium tetraoxalate 0.05 mol L –1 citrate KH tartrate (sat. at 25°C) 0.05 mol L –1 citrate 0.05 mol L –1 phthalate 0.1 mol L –1 acetic acid + 0.01 mol L–1 Sodium acetate 0.01 mol L –1 acetic acid + 0.1 mol L –1 sodium acetate 0.025 mol L –1 dihydrogen phosphate + 0.025 mol L –1 hydrogen phosphate 0.04 mol L –1 disodium hydrogen phosphate + 0.01 mol L –1 potassium dihydrogen phosphate 0.05 mol L –1 disodium tetraborate 0.025 mol L –1 sodium hydrogen carbonate + disodium carbonate Ca(OH)2 (sat. at 25°C) a

Literature pH value

Calculated potentiala (mV)

Measured potential (mV)

Measured pH

Relative error (%)

1.646 3.557 3.776 4.008 4.644 4.713 6.865

390.7 291.6 280.3 268.2 235.3 231.7 120.13

342.1 297.3 274.0 264.5 229.1 225 127.6

2.6 3.5 3.9 4.1 4.8 4.8 6.7

57.17b –2.97 3.3 1.89 2.64 2.84 –2.04

7.428

90.9

92.4

7.4

–0.32

9.182 10.012

0.0 –43.1

12.2 –51.9

9.0 10.19

–2.51 1.75

12.454

–169.6

–150.3

12.1

–2.96

Calculated from the calibration plot in Fig.2 Out of linearity range

b

All potentials are referred to a Ag/AgCl/(1 mol L–1 KCl) reference electrode. The calibration plot shows a quasiNernstian slope (practically 50 mV/pH unit) and a linearity interval from pH 12 to 3. Correlation coefficients (r2) values were >0.98. No significant differences (below 2%) were detected between different electrodes after the conditioning treatment. Calibration was confirmed by dipping the electrode in common buffers and allowing stabilization of potential. Table 1 summarizes the results obtained for the PPy electrodes compared with the values found in the literature [18] and the relative error in the pH measurement. It is apparent there is good agreement between the theoretical value described in bibliography and the pH value measured, the only difference between is for the pH 1.6 buffer, which is clearly out of the linearity range. The electrode was tested both in aqueous and organic media (MeCN). Figure 4 shows the titration plot of aqueous sodium hydroxide with acetic acid (Fig. 4a) and diethylamine with perchloric acid in acetonitrile (Fig. 4b). Results obtained for monoprotic titrations of strong alkalis with strong acids, and weak bases with strong acids are fully comparable (difference <0.5%) with results obtained with commercial glass electrodes, although a major drawback of this electrode was its slow response. The electrode reaches 90% of the total potential change in a few seconds, but requires several minutes before the system is totally stable. This slow equilibration rate becomes a problem if the electrode is used in automatic titrators, especially for multiprotic titrations. The pH change and associated potential evolution in multiprotic titrations is smaller than in the titrations described above, which is why very fast electrodes are required for these titrations. Automatic titration of sodium carbonate (0.1 mol L–1) with hydrochloric acid (0.1 mol L–1), with the bare PPy pH sensor is shown in Fig. 5a. Because of the slow response the titrations take too much time, the real

Fig. 4 Automatic titration of (a) acetic acid against sodium hydroxide in aqueous media and (b) diethylamine against perchloric acid in acetonitrile

potential change at the end point is smaller than predicted, and the first derivative is ill defined (Fig. 5c). All this complicates the detection of the inflexion point in titration curves. This slow response led us to hypothesize that the protonation of the electrode surface was adequate, but the diffusion inside the membrane was too slow. To prove this, the film thickness was decreased by reducing the deposition time. The same good slope (50 mV/pH unit) and linearity range (pH 3 to 12) were obtained for the new electrodes, and they were faster, but the improvement achieved was not enough to furnish a time response similar to that of commercial glass electrodes. The second-generation electrode was produced by using the template effect of a non-conducting polymer (PVC) cast over the graphite support. As is well known, electropolymerization on non-conducting polymer-coated electrodes is possible if the monomer and electrolyte can dif-

516 Fig. 5 Automatic NaCO3 titration (a) without PVC and (b) PVC coated. (c) and (d) are the corresponding first derivatives

fuse into the film and thereby reach the electrode surface [19]. The use of templates during polymerization leads to the formation of PPy wires [20, 21] by filling the free channels in the structure of the non-conducting polymer. This is a very easy and reproducible way of producing very thin PPy structures. The same polymerization media and intensities employed for the plain PPy were used for the PVC-coated electrodes, but the deposition time was limited to ensure the PPy was present in the PVC-induced channels only. The PVC incorporated electrodes were kept in the polymerization media for 10 min before PPy deposition was initiated. Solvent diffusion into the PVC matrix facilitates PPy polymerization by enabling the monomer to reach the conducting surface of the electrode. Figure 6 shows the potential trace during electropolymerization on to the bare (Fig. 6a) and PVC-covered (Fig. 6b) electrodes. Polymerization on a bare electrode leads to a constant potential plot, with a small potential decrease as the film grows, proving the good conducting properties of the resulting material. Completely different behavior is obtained for polymerization over PVC. The formation of the initial PPy seeds on the graphite surface, from the monomer absorbed during the pre-deposition treatment, results in potentials comparable with those obtained for the conventional electrode – when the amount of monomer on the glassy carbon surface decreases a rapid potential increase is detected (Fig. 6b). The PPy polymerization starts at these point nuclei, furnishing a non-continuous PPy deposit on the electrode surface. As the PPy wires grow, the overpotential necessary to force Py diffusion into the PVC matrix diminishes, reaching the potential obtained for the bare electrode when the wires just project over the PVC surface and directly contact the solution. At this stage the electrodeposition is interrupted to prevent the formation of the

Fig. 6 Potential evolution during PPy polymerization on (a) bare electrode and (b) PVC-covered electrode

continuous PPy film. SEM micrographs of the continuous PPy film and PPy wired electrode can be seen in Figs. 2 and 7, respectively. A continuous film of PPy on bare graphite is shown in Fig. 2a. A very homogeneous and smooth surface is apparent. In contrast the SEM micrograph obtained for the PVC precoated electrode (Fig. 7) reveals a discontinuous and rough PPy coating over the graphite electrode. It is apparent that polymerization using the template effect of PVC positively affects the PPy morphology in two different ways: 1. preventing the formation of a continuous film; and 2. increasing the effective surface area of the polymer, because of the high roughness obtained.

517 Table 2 Potentiometric selectivity coefficients (KpotH+) of a variety of interfering ions

aNo

Fig. 7 SEM micrograph of PPy on PVC-precoated electrodes

Fig. 8 Electrode potential stability in pH 4 and 7 buffer

Both would provide an increase in the rate of diffusion of H+ into the polymer matrix and, moreover, the increase in the surface roughness should result in improvement in the pH sensitivity of the polymer which compensates for the reduction in the apparent surface area because of the formation of a non-continuous film. The new calibration plots obtained with the PVC-coated electrodes have a slightly better slope (52 mV/pH unit) and linearity range (2.5 up to 12). As expected, and proving our first hypothesis, the change in PPy morphology has no clear effect on the thermodynamics of the system (similar slope and linearity range) but clearly affects the kinetics, resulting in a great improvement in the time response. The evolution of potential as a function of time for the PVC-modified electrodes in pH 4 and pH 7 buffer (potassium hydrogen phthalate and potassium dihydrogen phosphate–disodium hydrogen phosphate, respectively) shows that approximately 96% of the response occurs in less than 10 s, and once the potential is stabilized only a small drift is detectable (3 mV) even after 30 min of continuous pH measurement (Fig. 8). Long-term stability has been tested by performing the same experiment with electrodes maintained in 0.1 mol L–1 aqueous solutions of KNO3, HCl, and Na2CO3 for a month. The plots were not

detectable interference

Xn–

KpotH+

K+ Na+ Ca2+ Li+

3.2×10–5 5.1×10–5 2.1×10–4 N/Da

significantly different, even after the alkaline treatment. These results confirm that use of the cobaltabis(dicarbollide) anion results in excellent dopant retention by the polymer matrix, without the problems caused by dopant loss described for alkaline media. The problems observed with the first generation of pH electrodes in multiprotic titrations performed with an automatic titrator were totally solved with the PVC pretreated electrodes (Figs 5b and d). The short response time in response to sudden pH changes enabled acquisition of the maximum values of the first derivative at the point of inflexion and end points fully comparable with results obtained with commercial glass electrodes. The total time employed in the multiprotic titrations never exceeded 10% of that needed with a commercial electrode. The excellent behavior of the PVC-covered electrodes was also tested by repeating the experiments with addition of some common electrolytes that could act as possible interferents, e.g. LiClO4, NaCl, KCl, NaNO3, KNO3, CaCl2, LiNO3, and LiCl, in the range from 10–3 mol L–1 to 1 mol L–1. Addition of these electrolytes modified neither the slope of the calibration plots nor their linearity range. No detectable specific response was obtained for most of the ions. When sodium-, calcium-, or potassium-containing electrolytes were used at high concentration (>0.25 mol L–1) a potential displacement of the entire calibration plots was observed. For these ions the selectivity of the pH electrode was quantified by evaluation of the potentiometric selectivity coefficient (KpotH+) by the mixed solution method [22, 23], and applying the Nicolsky–Eisenman equation, KpotH+a1/nx=aH+[exp((E1–E2)F/RT)]–aH+, where E1 and E2 are, respectively, the potentials in the absence and presence of interfering ions (Xn–). Table 2 summarizes the resulting KpotH+ values. It is apparent that the electrodes are highly independent of the common aqueous ions. No effect of any accompanying anion has been observed. Interferences have only been observed when the cations are present at high concentration. They cause displacement of the calibration plot only and do not affect the slope or the linearity range. The presence of a high concentration of these interfering ions could be a problem in the measurement of pH absolute values but never in a titration.

Conclusions The applicability of a plain ICP for pH measurement has been demonstrated for the first time. No specific sensing moieties were required. The doping anion [3,3′-Co(1,2C2B9H11)2]– was used, because of its beneficial properties to the ICP; other large low-coordinating anions could, how-

518

ever, play a similar role. It seems that the most important factor in the implementation of PPy-based pH electrodes is the film thickness and morphology. The PVC pre-treatment enables easy control of film thickness and, therefore, improves the time response significantly. As expected, the system can be used in organic media for long periods without significant loss of response. Preliminary results show the high interdependence of electropolymerization current and pH linearity range. Further work will enable enhancement of electrode response, especially in the acid pH range. The effect of the non-conducting polymer pre-coating on system kinetics will be tested by changing the polymer and the plasticizer. Acknowledgements The authors gratefully acknowledge the financial support and the equipment provided by Crison Instruments S.A. and under project MAT98–0921; C. Masalles thanks the Generalitat de Catalunya for a pre-doctoral grant (1999FI00006).

References 1. Sun B, Fitch PG (1997) Electroanalysis 9:494–497 2. Ogura K, Shiigi H, Nakayama M (1996) J Electrochem Soc 143:2925–2930 3. Nicolas M, Fabre B, Simonet J (2000) Eur J Org Chem 9: 1703–1710 4. Barisci JN, Murray P, Small CJ, Wallace GG (1996) Electroanalysis 8:330–335 5. Arrigan DWM, Lowens MJ (1999) Electroanalysis 11:647–652

6. Naarman H, Strohriegel P (1992) In: H. Krichelford (ed) Handbook of polymer synthesis, vol B. Marcel Dekker, New York 7. Li Y, Jianyong J, Yang J (1995) Synth Methods 74:49–53 8. Aquino-Binay C, Kumar N, Lamb R, Pigram P (1996) Chem Mater 8:2579–2585 9. Konski R, Wolfbeis O (1998) Anal Chem 70:2544–2550 10. Osaka T, Fukuda T, Kanagawa H, Momma T, Yamauchi S (1993) Sensors Actuators B 13–14:205–208 11. Düssel H, Komorsky-Lovric S (1995) Electroanalysis 7:889– 894 12. Viñas C, Gomez S, Bertrán J, Teixidor F, Dozol JF, Rouquette H (1998) Inorg Chem 37:3640–3643 13. Viñas C, Gomez S, Bertrán J, Teixidor F, Dozol JF, Rouquette H (1998) Chem Commun 191–192 14. Viñas C, Bertrán J, Gomez S, Teixidor F, Dozol JF, Rouquette H, Kivekäs R, Sillanpää R (1998) J Chem Soc Dalton Trans 2849 –2854 15. Masalles C, Borrós S, Viñas C, Teixidor F (2000) Adv Mater 12:1199–1202 16. Pearson JF, Slater JM, Jovanovic V (1992) Analyst 117:1885– 1890 17. Jovanovic VM, Markicevic L, Stankovic S, Stankovic R, Jovanovic MS (1995) Electroanalysis 7:574–578 18. Chemical Rubber Company (1992–1993) CRC Handbook of chemistry and physics 73rd edn. CRC, Boca Raton, pp 8–34 19. Niwa O, Tamamura T (1984) J Chem Soc Chem Commun 817 20. Jêrome C, Jêrome R (1998) Angew Chem Int Ed Engl 37: 2488–2490 21. Choi S-J, Park S-M (2000) Adv Mater 12:1547–1549 22. Srinivasan K, Rechnitz GA (1969) Anal Chem 41:1203–1207 23. Minami H, Sato N, Sugarawa M, Umezawa Y (1991) Anal Sci 7:853–862