Characterization of Polymer Films of Pyrrole Derivatives for Chemical Sensing by Cyclic Voltammetry, X-ray Photoelectron Spectroscopy and Vapour Sorption Studies Zhiping Deng, David C. Stone and Michael Thompson Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, Canada M5S 3H6 Eight different conducting polymer films formed from pyrrole and N-substituted pyrrole derivatives were characterized by cyclic voltammetry, X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy.In particular, the XPS of poly[N-butylpyrrole], poly[N-(2-carboxyethyl)pyrrole], poly[N–(6–hydroxyhexyl)pyrrole] and poly[N-(6-tetrahydropyranylhexyl)pyrrole] is reported for the first time. The vapour sorption properties of these films were also examined by forming the films onto the electrodes of thickness-shear mode acoustic wave sensors. The influence of the pendant side chain is apparent in both the electrochemical behaviour, composition, doping level, morphology and the nature and extent of polymer–vapour interactions.The latter can be rationalized by consideration of vapour physical properties and solvatochromic parameters. Keywords: Cyclic voltammetry; X-ray photoelectron spectroscopy; poly(pyrrole) and derivatives; vapour sorption; thickness-shear mode; quartz crystal; acoustic wave chemical sensor Organic conducting polymers such as poly(pyrrole) have found extensive application in highly diverse fields such as biomaterials, chemistry, electronics, microfabrication, non-linear optics, sensors and textiles.1–5 In the area of analytical chemistry, for example, such materials have been used to form chemically modified electrodes,6 permeation membranes,7,8 liquid chromatographic stationary phases9,10 and chemical and biological sensors.11–16 These chemical applications are all based on the physico-chemical interactions that occur between the conducting polymer and the respective chemical species.Our own interest in these materials is their use as selective (or partially selective) coatings for thickness-shear mode (TSM) acoustic wave sensors. One reason for using conducting polymers in this way is the ease of film formation by direct electropolymerization onto the TSM electrode surface. Other advantages lie in the ease with which redox state and incorporated counter ions can be changed electrochemically, and the wide range of functionalities that can be introduced through chemical reaction with the pyrrole ring nitrogen. This in turn provides various mechanisms for modifying the extent of different reversible interactions between the polymer films and different analytes, thus affecting both the sensitivity and selectivity of the sensor system.In a series of papers,17–20 we have described the application of various N-substituted poly(pyrroles) to organic vapour sensing. These include the parent compound, poly(pyrrole) (PPY), and poly(N-methylpyrrole) (PMPY), poly(N-butylpyrrole) (PBPY), poly[N-(2-cyanoethyl)pyrrole] (PCPY), poly[N- (2-carboxyethyl)pyrrole] (PCbPY), poly(N-phenylpyrrole) (PPPY), poly[N–(6–hydroxyhexyl)pyrrole] (PHPY) and poly[N-(6-tetrahydropyranylhexyl)pyrrole] (PTHPY).In the preceding papers, we have described the preparation, characterization and effect of redox state of PCPY18,20 as well as the application of all eight polymers to the selective detection of aroma components.19 In the present paper, we provide a more detailed characterization and comparison of the eight conducting polymers, with an emphasis on their formation and application as coatings for organic vapour sensors.Experimental Reagents and Materials The solvents hexane, toluene, methanol, butanal, acetonitrile, triethylamine, ethanol, butan-1-ol, hexan-1-ol and nonan-1-ol (analytical-reagent grade, Aldrich, Milwaukee, WI, USA) were used as received. Pyrrole (98%), N-methypyrrole (99%) and N- (2-cyanoethyl)pyrrole (99%) (Aldrich) were vacuum-distilled before use.Both N-phenylpyrrole (99%) (Aldrich) and tetrabutylammonium perchlorate (TBAP) (electrochemical grade, Fluka, Buchs, Switzerland) were used as received. N-(2-Carboxyethyl) pyrrole, N-butylpyrrole, N-(6-tetrahydropyranylhexyl) pyrrole and N-(6-hydroxyhexyl)pyrrole were synthesized as described previously.19 All other chemicals were obtained from Aldrich and used as received. The piezoelectric crystals were rough or optically polished 9 MHz AT-cut quartz crystals with gold electrodes (International Crystal Manufacturing, Oklahoma City, OK, USA).A combined Pt–Ag/AgCl (3 mol l21 KCl) electrode (Metrohm, Herisau, Switzerland) was utilized as the counter and reference electrode, respectively. Polymerization Procedures One electrode of the TSM device was used as a working electrode for the electropolymerization of pyrrole and its derivatives using a cell as described previously.17 Unpolished crystals were used for vapour sorption experiments while optically polished crystals were used for the cyclic voltammetry studies.All devices were washed with acetone and dried under nitrogen before use. Polymerization was achieved using a solution containing monomer (0.1 mol l21) and TBAP (0.1 mol l21) in de-oxygenated acetonitrile using either constant potential or cyclic voltammetric deposition with a potentiostat/ galvanostat (Model 273, EG&G Princeton Applied Research, Princeton, NJ, USA).The polymerization potentials for the different monomers have been listed previously.19 The coated TSM sensors were then rinsed with acetonitrile, dried with nitrogen, washed with acetone, dried with nitrogen and finally placed in an oven for 1 h at 100 °C. The coating mass for each TSM device was determined from the in-air frequency difference before and after polymer deposition. Analyst, October 1997, Vol. 122 (1129–1138) 1129Cyclic Voltammetry Cyclic voltammograms for film deposition and growth on the electrodes of polished TSM devices were obtained for ten cycles at a sweep rate of 100 mV s21 for potentials varying between 20.8 and +1.2 V versus Ag/AgCl, depending on the monomer used.The resulting films were then studied using sweep rates of 20–100 mV s21 and potentials varying between 20.8 and +1.2 V versus Ag/AgCl using TBAP (0.1 mol l21) in acetonitrile as the supporting electrolyte.X-ray Photoelectron Spectroscopy (XPS) XPS was performed using a Leybold Max 200 XPS (Leybold, Cologne, Germany) instrument using an unmonochromatized Mg Ka source and an analysis area of 2 3 4 mm2. Survey and low resolution spectra were obtained using a pass energy of 192 eV; high resolution spectra were acquired with a pass energy of 48 eV. All spectra were satellite-subtracted and normalized using software and elemental sensitivity factors provided by the manufacturer.The binding energy scale was further calibrated to 285.0 eV for the main C (1s) feature in order to compensate for sample charging effects. Scanning Electron Microscopy Scanning electron micrographs were obtained using a Hitachi Model S-570 microscope and recorded using the Quartz PCI image capture system. In order to reduce charging effects, the coated TSM devices were sputter-coated with a thin gold film using an argon plasma vapour deposition system (Polaron Model E5100, Polaron, Watford, Hertfordshire, UK).All micrographs were acquired using an accelerating voltage of 18 kV. Vapour Sorption Studies Vapour generation was achieved by bubbling helium gas through the corresponding liquid at room temperature and pressure using the flow system described previously.18,19 The flow rates of the sample and purge streams were held at 30 ml min21 for all experiments. The frequency shift of the polymer-coated TSM sensors was measured using a universal frequency counter (Model HP5334B, Hewlett-Packard, Avondale, PA, USA).Data were collected, displayed and stored using a Macintosh II computer equipped with an IEEE 488 interface bus using software written in-house. The coated TSM sensors were initially purged using pure helium carrier gas for 1–2 h to allow the device to stabilize. Once a stable baseline had been obtained, the stream was switched to the sample vapour and returned to the purge gas once the steady-state frequency shift had been obtained.Measurements were performed in triplicate for each coating– vapour combination. Vapour concentrations were obtained using a liquid nitrogen trap for a fixed time interval and weighing the resulting condensate. The measured vapour concentrations are given in Table 1, together with the relevant physical properties for the compounds studied. Results and Discussion Cyclic Voltammetry The cyclic voltammograms for the electropolymerization of all eight monomers are fairly similar, the main differences being in the position and magnitude of the anodic and cathodic peaks.That for N-(2-carboxyethyl)pyrrole is shown in Fig. 1 as a representative example. As can be seen, the first oxidation peak occurs at a higher oxidation potential (+1.06 V) than that of pyrrole (+0.85 V). In fact, the oxidation potentials of all seven N-substituted pyrroles are higher than that of the parent compound (Table 2). This is attributable to the steric and electronic effects of the pendant side chain, the inductive effect rendering the oxidation process more difficult for the mono- Table 1 Physical data and concentration values for the test solvents and corresponding vapours Cv/mg l21 Boiling- Density/ Molecular (at 21 °C and Solvent point/°C g cm23 mass/g mol21 30 ml min21) Hexane 69 0.659 86.16 31 Toluene 110 0.865 110.6 10 Water 100 1.000 18.00 15 Acetonitrile 82 0.786 41.05 34 Triethylamine 88.8 0.726 101.19 30 Butanal 75 0.800 72.11 22 Methanol 64.7 0.791 32.04 40 Ethanol 78 0.785 46.07 27 Butan-1-ol 117.7 0.810 74.12 8.3 Hexan-1-ol 156.5 0.814 102.18 1.2 Nonan-1-ol 215 0.827 144.26 0.3 Fig. 1 Cyclic voltammetric polymerization of N-(2-carboxyethyl)pyrrole (0.1 mol l21 in 0.1 mol l21 TBAP–acetonitrile) at scan rate of 100 mV s21 versus Ag–AgCl (3 mol l21). Table 2 Anodic and cathodic peak potentials (versus Ag/AgCl) for the cyclic voltammetric polymerization of N-substituted pyrrole monomers (0.1 mol l21 monomer in 0.1 mol l21 TBAP–acetonitrile, 100 mV s21 scan rate) Epa Epa Epc Polymer (monomer)/V* (polymer)/V† (polymer)/V† PPY 0.85 20.08 20.23 PMPY 0.90 0.46 0.42 PBPY 1.08 0.65 0.47 PCPY 1.10 0.70 0.64 PCbPY 1.06 0.68 0.56 PPPY 1.19 0.67 0.58 PHPY 1.05 0.68 0.52 PTHPY 1.08 0.69 0.58 * From first scan.† From second scan. 1130 Analyst, October 1997, Vol. 122mers.21 When the potential is reversed on the first scan, the anodic current continues to increase for a short time before decreasing.The following cathodic current is also higher than that in the forward scan until the potential is +0.76 V, resulting in a loop in the voltammogram. This is indicative of a nucleation mechanism, and is commonly observed for the electropolymerization of conducting polymers.22 The effect is attributed to the fact that polymerization occurs faster at a polymeric nucleus than at the uncoated electrode surface. The anodic peak for the second scan occurs at a much lower potential (+0.68 V), and corresponds to polymer growth with incorporation of perchlorate counter ions (Table 2).The peak is broad because there will be a distribution of oligomer sizes, while diffusion of counter ions in and out of the film will be slow. Adams23 has concluded that whenever the second and subsequent sweeps of a cyclic voltammogram differ markedly from the first, a followup chemical reaction has occurred. These results, therefore, show the characteristics of an ECE reaction, in which electron transfer is followed by a chemical reaction and subsequent electron transfer reaction.24 Therefore, the overall polymerization mechanism for the pyrrole derivatives is the same as that for the parent compound.25 Another feature of the voltammograms is that, on successive scans, the anodic peak shifts to increasingly positive potential with a corresponding increase in current until a limiting value is reached.A similar effect is observed for the cathodic peak, except here the shift is towards a more negative potential.The shift is, at least in part, due to the polymer film resistance giving rise to a larger overpotential. The ratio of the cathodic to anodic peak currents is also less than one, confirming that the anodic peak consists of both reversible and irreversible contributions. The anodic peak potentials for the different films are very similar with the exception of PPY and PMPY. This may be explained by considering the effect of side-chain length on the porosity of the polymer.It is anticipated that as the length of the substituent chain increases, the polymer chains will be forced farther apart making it easier for counter ions to diffuse in and out of the film. This is confirmed by the cathodic peak potentials for the series PPY, PMPY, PBPY and PTHPY, which show an increasing shift towards positive potentials. Following film formation, the cyclic voltammogram for each polymer film was obtained at various scan rates using TBAP in acetonitrile without monomer as the supporting electrolyte.Well-defined, broad anodic and cathodic peaks were obtained in all cases (Fig. 2), which reflects the slow rate of counter ion transfer in and out of the film as well as interactions between electroactive sites and the electrochemical non-equivalence of these sites.26 The magnitude of the anodic and cathodic peak currents increases linearly with scan rate, while peak separation also increases.The polymer redox processes are, therefore, not diffusion-controlled. Table 3 summarizes the positions and separation of the anodic and cathodic peaks. Both the anodic (Epa) and cathodic (Epc) peak potentials become increasingly positive relative to the parent PPY as the chain length of the Nsubstituent increases. This trend arises from increasing distortion of inter-ring coplanarity, which would destabilize the cationic (oxidized) form of these polymers.One potential benefit of the positive shift in Epa is that films formed from Nsubstituted pyrrole derivatives will be less sensitive to atmospheric oxidation. This implies improved handling, storage and lifetime characteristics for chemical sensors fabricated using such films. Another observation is that the peak separation (DE) is considerably lower for PPY and PMPY compared with the other polymer films. In fact, the oxidation and reduction of both PPY and PMPY films are essentially reversible, while other irreversible processes must be involved for the remaining films.As an example, PHPY shows two anodic waves, the second appearing at Å 1.0 V for scan rates of 80 and 100 mV s21. This is likely associated with oxidation of the hydroxyl group, although further work is needed to confirm this. X-ray Photoelectron Spectroscopy Although the X-ray photoelectron spectra of PPY, PMPY, PPPY and PCPY films have been previously reported, 17,18,20,26–31 we have included the relevant data with those for PBPY, PCbPY, PHPY and PTHPY in Table 4 for ease of comparison. This lists the results of peak deconvolution on the C(1s), N(1s), O(1s) and Cl(2p) regions for all eight as-prepared polymer films.In addition, Fig. 3 shows the C(1s) spectrum obtained for PCbPY. All eight films show the expected principal C(1s) component at 285 eV arising from the pyrrole ring a and b carbons26 and methylene groups within the pendant side chain.These species are unresolvable under the experimental conditions used here. A second C(1s) component occurring between 286.1 and 286.4 eV can arise from a number of different sources. For PCPY, for example, it is attributed to the nitrile group as discussed previously.18 For PHPY, there is also a contribution from the side chain –CH2OH group, while PTHPY contains two ether groups. It is, therefore, not surprising that these three films show significant enhancement of the 286 eV component relative to the remaining polymers.The origin of this peak for PPY, PMPY, PBPY, PCbPY and PPPY is amenable to several interpretations. Pfluger and Street26 attribute this band in PPY to ‘disorder type’ carbons, which they define as being crosslinked, chain-terminating and non-a,aA bonded carbons as well as partially saturated rings. There will also be a small contribution from unavoidable hydrocarbon contaminants, while Atanasoska et al.28 have suggested that electrostatic interaction of ring a carbons with counter ions will also have an effect.A fourth possibility is the presence of covalently bound chlorine species, which we will consider shortly. The final C(1s) component between 288 and 289 eV is generally attributed to carbonyl or carboxyl species resulting from chain termination.21,28 This has been partly confirmed by FTIR studies of chemically polymerized PPY samples,29 which clearly show the presence of a small band at 1705 cm21 that scales in intensity with the 288 eV C(1s) component. For PCbPY, there is also the pendant carboxyl group (289.1 eV), which clearly increases the relative contribution of this component.Turning now to the N(1s) region, all the films show two peaks located at approximately 400 and 402 eV, with the major component being the lower binding energy signal. This is in general agreement with other XPS studies of different PPY species,26,28 a number of which also show a small shoulder at Å 397 eV.17,20,29,30 The main peak at 400 eV arises from the neutral pyrrole ring nitrogen, while the higher binding energy component is generally attributed to partially charged nitrogens within bipolaron sub-units.The presence of the shoulder at 397 eV depends, to some extent, on the experimental signal-to-noise ratio but also on the film preparation conditions since it is more pronounced for neutral than as-prepared or oxidized PPY and PCPY.17,20 Lei et al.29 have attributed it to –CNN– defects in the pyrrole backbone, while Vigmond et al.17 ascribed it to interchain hydrogen bonding effects.In this latter interpretation, equal intensity peaks on either side of the principal N(1s) line are expected due to electron donation from one nitrogen to another. This being the case, one would expect enhancement of the higher binding energy component in the as-prepared films for polymers in which electron donation from one ring nitrogen to an adjacent ring side chain is possible.Such an effect is, in fact, observed for all the polymers relative to the phenyl- and alkyl-substituted pyrroles (PMPY, PBPY and PPPY). The exception here is PCPY, although the situation in this case may Analyst, October 1997, Vol. 122 1131be masked by the additional nitrogen per pyrrole unit contributed by the nitrile group. The Cl(2p) spectra show peaks with the 2p3/2 component occurring at 207.6 and 200.7 eV.The higher binding energy peak is due to incorporated perchlorate ion, while the lower peak is assigned to covalently incorporated chlorine as discussed previously.20 Support for this assignment comes from the results of Kang et al.30 and Toshima and Tayanagi,31 who observed similar Cl(2p) spectra for chemically polymerized PPY samples. In this respect, it should be noted that the covalently bound chlorine will also result in an increase in the C(1s) component at 286 eV.It is possible to estimate the doping ratio of the as-prepared polymer films from the XPS data using the Cl/N ratio, although Fig. 2 Cyclic voltammograms for the polymer films in 0.1 mol l21 TBAP–acetonitrile without monomer at different scan rates (mV s21). 1132 Analyst, October 1997, Vol. 122this will not be as accurate as that calculated from the composition obtained by bulk elemental analysis since only a thin layer of an irregular surface is being analysed. The estimated doping ratio for each film is given in Table 5, together with the corresponding number of electrons per monomer unit (n) involved in polymerization and subsequent oxidation.It is also possible to estimate a value of n from the cyclic voltammetry data using Nicholson’s model for irreversible charge transfer32 Ip = nFAC*n1/2(pDanF/RT)1/2c(bt) (1) where F is Faraday’s constant, A is the area, C* is the surface concentration of electroactive species, v is the scan rate, D is the diffusion coefficient, a is the transfer coefficient and c(bt) is the current function for irreversible charge transfer.A difficulty here, however, is obtaining a reasonable estimate of the surface concentration for an amorphous, porous polymer film. One crude approach is to use the generally accepted value of n = 2.3–2.5 for PPY and use this to calculate the surface concentration, then assume an identical value for the Nsubstituted derivatives. Agreement between the XPS data and the values of n derived in this manner is surprisingly good, although it breaks down for PCPY and PCbPY.The O(1s) region can generally be deconvoluted into two components at 533.5 and 532.0 eV, representing singly and doubly bonded oxygen species, respectively. For the nonoxygen containing polymers, the only sources of oxygen in the final film will be incorporated counter ion and carboxy groups introduced by chain termination reactions.28 Note that perchlorates also show oxygen binding energies of Å 533 eV, which is unresolvable from the other species for the instrumental conditions used in this study.For these reasons, the O(1s) region has rarely been discussed in previous XPS studies of poly- (pyrroles). Simple composition calculations assuming that the perchlorate and carbonyl oxygen species overlap yield RNO: R– O ratios close to the theoretical 1 : 1 for PPY, PCbPY and PPPY, but break down for PMPY, PBPY, PCPY and PTHPY. This is anticipated since the relative amount of chlorine species other than perchlorate was not taken into account. What the results do show, however, is the expected progressive increase in R–O species for PHPY and PTHPY relative to the parent compound.One curious feature of the O(1s) deconvolution is the apparent presence of an additional oxygen species with a high binding energy of 535 eV. Such species can be observed on plasma-cleaned gold surfaces. This can be discounted, however, since no gold peaks are observed in any of the survey scans.Table 3 Anodic and cathodic peak potentials (versus Ag/AgCl) of Nsubstituted pyrrole polymers in 0.1 mol l21 TBAP–acetonitrile for a 20 mV s21 scan rate Polymer Epa/V Epc/V DE/V PPY 0.28 0.25 0.027 PMPY 0.51 0.48 0.030 PBPY 0.72 0.56 0.15 PCPY 0.82 0.72 0.10 PCbPY 0.88 0.76 0.12 PPPY 0.84 0.70 0.14 PHPY 0.81 0.64 0.16 PTHPY 0.86 0.76 0.10 Table 4 Polymer elemental composition (at.-%) and peak analysis from the XPS data C(1s) N(1s) O(1s) Cl(2p) Polymer BE*/eV at.-% BE*/eV at.-% BE*/eV at.-% BE*/eV at.-% PPY 66.9 11.6 18.2 3.3 285.0 76.9 400.2 71.3 532.4 87.5 207.6 100.0 286.3 18.4 402.1 28.7 533.7 12.5 288.1 4.7 PMPY 74.5 10.4 12.7 2.4 285.1 69.1 400.1 87.3 531.6 64.7 200.9 100.0 286.4 18.7 401.8 12.7 533.0 35.3 288.4 12.2 PBPY 71.9 13.4 12.8 1.9 285.0 75.7 398.5 87.9 531.8 52.9 201.1 78.6 286.4 16.6 400.3 12.1 533.0 47.1 208.1 21.4 288.2 7.7 PCPY 73.7 7.8 15.4 2.3 284.8 51.3 399.9 93.8 532.0 70.7 200.7 40.1 286.4 43.8 401.7 6.2 533.6 29.3 207.5 59.7 288.6 5.0 PCbPY 63.9 9.1 25.0 2.0 284.9 63.7 399.9 59.8 532.3 60.8 200.6 68.2 286.2 18.9 400.4 34.9 533.5 33.6 207.5 31.8 289.1 17.3 401.6 9.4 534.6 5.6 PPPY 73.3 8.1 16.7 1.3 284.9 75.0 400.7 92.0 532.6 63.0 200.4 100.0 286.1 21.0 402.4 8.0 533.5 37.0 287.9 4.0 PHPY 74.6 7.3 16.6 1.5 284.9 58.5 400.1 69.0 532.5 56.8 207.4 100.0 286.1 36.1 401.4 31.0 533.8 36.8 288.5 5.4 535.6 6.4 PTHPY 64.6 5.4 27.5 2.6 285.0 61.0 400.3 62.8 532.4 22.4 200.8 18.8 286.3 29.5 402.3 37.3 533.4 67.5 207.9 81.2 288.3 9.5 534.6 10.2 * BE = Binding energy.Analyst, October 1997, Vol. 122 1133There are several other possible explanations, including differential charging affecting the incorporated counter ion and the presence of an oxidized nitrogen species within the polymer. It is difficult to envisage the latter, however, since the peak is only observed for N-substituted polymers. A final possibility is that it is a fitting artifact arising from the unresolved perchlorate oxygen, since there are an infinite number of mathematical solutions that would equally well fit the O(1s) envelope.Further work is, therefore, needed to clarify this matter. Scanning Electron Microscopy In previous studies, we have found that surface morphology can have an effect on the response of the polymer-coated sensors to different vapours.17–20 Polymer morphology is, in turn, influenced to various degrees by substrate morphology, film thickness and level of oxidation, and the nature of the counter ion. PPY films are frequently described as having a ‘cauliflower- like’ appearance, which arises from the nucleation and phase growth mechanism of the electropolymerization process.This detail is not apparent on very thin films, however, which simply follow the underlying surface. It is, therefore, important to specify both substrate surface roughness and film thickness when comparing the appearance of different films.In this, and our previous studies, the films were formed on microscopically rough surfaces. This was necessary since these polymer films adhere to gold electrodes primarily through mechanical interlock, and are prone to peeling from optically flat surfaces. In particular, the cleaning procedure used in the vapour sorption studies is highly effective as a means of removing conducting polymer films from polished gold electrodes, while films formed on the unpolished devices remain intact.The scanning electron micrographs for six of the polymers are shown in Fig. 4. Micrographs of PPY and PCPY have been published elsewhere.17,18,20,25 The corresponding film thicknesses, calculated from the in-air frequency shift due to film deposition, Dfs, are listed in Table 6. Generally, films of the Nsubstituted pyrrole derivatives show similar morphologies to the parent compound, although there are some differences. Both PMPY and PPPY, for example, exhibit a much more granular surface than PPY even for the same film thickness.The films of PBPY and PHPY are the most similar to PPY, while PCPY and PCbPY again show similar topography but with deep, open channels running into the bulk of the polymer. In this context, it should be noted that these channels disappear if the PCPY films are heavily oxidized by application of a large positive potential, becoming very similar in appearance to the as-prepared PBPY and PHPY films.The most different film in terms of morphology is PTHPY which, although as thick as the PPY and PPPY films, shows no apparent differences from the underlying gold electrode. These differences in morphology have a significant influence on the response of the coated vapour sensors. This occurs firstly through variations in effective surface area, which determines the extent of initial adsorption, and secondly through the film porosity, which affects both the absolute steady-state frequency response (extent of vapour absorption) and recovery time (vapour desorption). Table 5 Doping levels and number of electrons involved in electropolymerization –oxidation from the XPS and cyclic voltammetry (CV) data n Polymer Cl/N ratio Doping ratio XPS CV PPY 0.286 0.29 2.3 2.2 PMPY 0.232 0.23 2.2 2.3 PBPY 0.136 0.14 2.1 2.1 PCPY 0.289 0.15 2.2 2.6 PCbPY 0.219 0.22 2.2 2.7 PPPY 0.154 0.15 2.2 2.2 PHPY 0.205 0.21 2.2 2.5 PTHPY 0.488 0.49 2.5 2.6 Fig. 3 X-ray photoelectron spectra of PCbPY showing the C(1s), N(1s), O(1s) and Cl(2p) regions. 1134 Analyst, October 1997, Vol. 122Vapour Sorption Studies Since the magnitude of the observed frequency shift for vapour sorption (Dfv) varies with both vapour concentration (Cv) and film thickness, it is necessary to normalize the data before meaningful comparisons can be made. In keeping with our earlier work, we therefore calculate the sensor partition coefficient,33 K f f C S = D D nr V (2) where Dfv is in Hz, Dfs (kHz) is the frequency shift due to the polymer coating, r (g cm23) is the film density and Cv is in g cm23. Alternatively, the value of Dfv may be normalized to the shift that would have been obtained for a film having Dfs = 30 kHz.A series of organic vapours were chosen as probe molecules representing a variety of structural and functional group interactions between the vapour and the polymer film (Table 1). Hexane, for example, can interact solely through dispersion forces while toluene also exhibits p-electron overlap and polarizibility interactions.The remaining vapours were chosen to represent differing degrees of hydrogen bond acidity and basicity and dipole–dipole interactions, while a series of four primary alcohols was included to study the effects of chain length and analyte volatility. The solvatochromic parameters for all these vapours are listed in Table 7. While the corresponding Fig. 4 Scanning electron micrographs of six of the conducting polymer films. (a) PMPY; (b) PBPY; (c); PCbPY; (d) PPPY; (e) PHPY; and (f) PTHPY. Table 6 Film thickness and in-air frequency shifts corresponding to electropolymerization of the pyrrole derivatives on the TSM device Polymer PPY PMPY PCbPY PBPY PCPY PPPY PHPY PTHPY Thickness/mm 0.45 0.39 2.96 1.91 0.98 0.49 1.10 0.48 Dfs/kHz 28.8 21.3 104.5 99.0 50.9 26.0 55.2 22.0 Analyst, October 1997, Vol. 122 1135coating parameters are not known for the pyrrole derivatives used in this study, these values are nonetheless useful in comparing the response behaviour of the different vapour– coating combinations.The differing abilities of these combinations to interact through various mechanisms gives rise to different partial selectivities for each coating. This is reflected in the experimentally observed log K values, which are listed in Tables 8 and 9 and summarized graphically in Fig. 5. Comparing hexane and methanol, for example, shows that hexane generally gives much lower normalized frequency shifts ( < 2 kHz) than methanol (2–8 kHz), which reflects the difference in enthalpy of vaporization, and hence vapour concentration, between these two solvents (31.56 and 37.43 kJ mol21 at 25 °C, respectively).When the effect of vapour concentration is factored out by calculating log K, however, we see that the differences in response are actually minimal for the PMPY- and PBPY-coated sensors. On the other hand, fairly large differences in response in favour of methanol are observed for those coatings capable of forming hydrogen bonds, especially PCPY, PCbPY, PHPY and PPY.Water and toluene both have significantly higher boilingpoints and lower vapour pressures than hexane and methanol. Their normalized frequency shifts generally lie between the extremes observed for hexane and methanol, reflecting the fact that while their Cv values are lower, their rate of desorption from the film will also be slower resulting in a larger steadystate equilibrium vapour concentration within the film coatings.Comparing the log K values for this pair shows a lower extent of differentiation between water and toluene although, again, a large difference in favour of water vapour is observed for the two most polar coatings (PCPY and PCbPY). PBPY, PPY and PTHPY all show larger log K values for toluene than either water or methanol vapour.This is readily explained since toluene has the largest polarizibility, dipolar and dispersion solvatochromic parameters while the side chains of PBPY, PPY and PTHPY are all essentially non-polar in nature. PHPY, which has a relatively non-polar side chain in spite of its terminal hydroxyl group, also shows very similar response behaviour towards the same vapours. Comparing the remaining vapours, methanol and acetonitrile show very similar response patterns with the greatest differentiation observed for PTHPY.Triethylamine, on the other hand, gives a markedly lower response with the most polar coatings (PCPY and PCbPY) and reduced responses with PPY, PMPY, PPPY and PHPY. Butanal shows a similar pattern to methanol and acetonitrile, showing greater log K values for PPY, PMPY and PBPY. This reflects the larger dispersion parameter for butanal compared with methanol and acetonitrile. Neither butanal, acetonitrile nor triethylamine can function as Table 7 Solvatochromic parameters for the test liquids (from refs. 34 and 35). The parameters represent polarizability (R2), dipolar interactions (p2 *), hydrogen bond acidity and basicity (a2 H and b2 H, respectively) and dispersion interactions (log L16) Solvent R2 p2 * a2 H b2 H Log L16 Hexane 0.00 0.00 0.00 0.00 2.688 Toluene 0.601 0.55 0.00 0.14 3.344 Methanol 0.278 0.40 0.37 0.41 0.922 Water 0.00 0.43 0.33 0.65 0.267 Butanal 0.187 0.65 0.00 0.40 2.270 Acetonitrile 0.237 0.75 0.09 0.44 1.560 Triethylamine 0.101 0.15 0.00 0.67 3.077 Ethanol 0.246 0.40 0.33 0.44 1.485 Butan-1-ol 0.224 0.40 0.33 0.45 2.601 Hexan-1-ol 0.210 0.40 0.33 0.45 3.610 Nonan-1-ol 0.193 0.40 0.33 0.45 5.124 Table 8 Normalized frequency shifts and corresponding partition coefficients obtained for the coated TSM sensors on exposure to different test vapours Analyte Hexane Toluene Methanol Water Polymer Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K PPY 279 ± 17 2.65 845 ± 19 3.70 2948 ± 82 3.56 2581 ± 53 4.01 PMPY 1038 ± 16 3.21 670 ± 83 3.51 1845 ± 97 3.35 1156 ± 45 3.57 PBPY 1860 ± 80 3.39 3462 ± 189 4.16 2230 ± 121 3.36 996 ± 65 3.43 PCPY 146 ± 9 2.34 83 ± 7 2.67 3870 ± 76 3.73 2863 ± 49 4.03 PCbPY 170 ± 9 2.41 281 ± 18 3.12 8019 ± 105 3.98 4252 ± 107 4.13 PPPY 987 ± 45 3.17 4030 ± 145 4.27 5331 ± 132 3.69 2954 ± 123 3.96 PHPY 441 ± 40 2.81 1461 ± 18 3.82 6757 ± 162 3.88 1824 ± 117 3.74 PTHPY 712 ± 75 2.97 2768 ± 87 4.05 4315 ± 152 3.64 1952 ± 35 3.72 Butanal Acetonitrile Triethylamine Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K 2078 ± 81 3.67 1875 ± 62 3.51 1065 ± 45 3.34 1873 ± 75 3.62 1564 ± 82 3.35 975 ± 21 3.20 3236 ± 83 3.78 2873 ± 46 3.54 2528 ± 138 3.54 3392 ± 91 3.86 4560 ± 152 3.88 573 ± 12 3.04 3775 ± 78 3.71 6556 ± 149 3.96 593 ± 42 2.97 2092 ± 11 3.65 4054 ± 52 3.75 1275 ± 63 3.30 4175 ± 39 3.93 4378 ± 121 3.76 1783 ± 103 3.43 1562 ± 50 3.46 1115 ± 16 3.13 2787 ± 33 3.58 Table 9 Normalized frequency shifts and corresponding partition coefficients obtained for the coated TSM sensors on exposure to a series of primary alcohol vapours Analyte Ethanol Butan-1-ol Hexan-1-ol Nonan-1-ol Polymer Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K PPY 1299 ± 31 3.38 827 ± 43 3.69 714 ± 19 4.47 236 ± 11 4.59 PMPY 3373 ± 56 3.78 555 ± 19 3.51 521 ± 8 4.32 201 ± 15 4.51 PBPY 2673 ± 85 3.61 1421 ± 64 3.85 2661 ± 58 4.96 530 ± 28 4.86 PCPY 3959 ± 42 3.84 522 ± 27 3.47 437 ± 15 4.23 189 ± 8 4.47 PCbPY 4819 ± 67 3.93 2756 ± 94 4.12 1202 ± 71 4.68 300 ± 8 4.68 PPPY 4947 ± 79 3.93 3732 ± 153 4.32 2611 ± 39 5.01 161 ± 5 4.40 PHPY 2250 ± 28 3.57 1650 ± 25 3.95 2103 ± 46 4.90 393 ± 13 4.77 PTHPY 3613 ± 102 3.74 2930 ± 49 4.16 1827 ± 31 4.79 143 ± 9 4.29 1136 Analyst, October 1997, Vol. 122hydrogen bond donors, so one might expect lower responses to these vapours for coatings that are also unable to function as hydrogen bond donors (e.g., PMPY, PBPY and PPPY). This is in fact seen for butanal and acetonitrile, but dispersion interactions are also a strong factor for triethylamine, resulting in much higher responses with PBPY and PTHPY.The series of primary alcohols show the expected increase in log K with increasing carbon number, although it is non-linear. This is presumably a result of concentration effects arising from the high vapour concentrations used in this study compared with those found in typical gas chromatography experiments. Ethanol shows the least variation in log K between different coatings, while hexan-1-ol shows the greatest.The solvatochromic parameters for this series differ only in the magnitude of the polarizibility and dispersion parameters, reflecting the fact that only the chain length is changing. In this respect, it is interesting that the largest log K values are observed for the coatings having longer side chains such as PBPY and PHPY. One method to characterize the vapour sorption properties of the polymer coatings across the series of test vapours is to use hierarchical cluster analysis.In a previous paper, this technique was employed in order to examine similarities in response patterns (and therefore potential redundancy) for four longchain unsaturated aroma components.19 A similar analysis was performed on the log K values measured for the current set of analyte vapours using the Euclidean distance, single linkage (nearest neighbour) method without autoscaling. The cluster analysis was also repeated for various sub-sets of the data.Fig. 6 shows the dendrogram obtained using all the vapours listed in Table 1. Some differences were observed between the current and previous study, which employed only five- to nine-carbon chain unsaturated alcohols and aldehydes.19 For example, while the most polar films (PCPY and PCbPY) clustered together in both cases, different groupings were observed for the remaining polymers. In fact, the current results demonstrate a clustering pattern that much more closely reflects both the polarity and hydrogen bonding ability of the different polymeric coatings.This is clearly seen in the fact that the hydrogen bond donors PPY and PHPY cluster together, as do the longer side chain, non-hydrogen bond donors PPPY, PTHPY and PBPY. For comparison, in the previous study (where dispersion interactions were likely more significant than for the short chain analyte vapours used here) the clustering pattern showed a strong correlation with side-chain length so that PTHPY and PHPY were grouped together. Returning to the present study, the observed clustering of the polymeric coatings was further confirmed when the analysis was repeated on sub-sets of the data containing either polar or non-polar analytes.For the purposes of analysis, these were defined in two different ways: (1) on the presence or absence of polar functional groups (hydroxyl, carbonyl, etc.); and (2) on the arbitrary condition that, for ‘polar’ analytes, logL16 < 2.5.This second definition reflects the fact that longer chain analytes have a significant non-polar component, while the first definition yielded only two ‘non-polar’ cases. The resulting dendrograms (not shown) support the view that the magnitude of the response (and, therefore, the partial selectivity) for a given coating–vapour pair is a direct consequence of the nature and strength of the initial adsorption interaction.Since these interactions are also those involved in solvation processes, solvatochromic parameters such as those listed in Table 7 provide a useful tool for the rational choice of sensor coatings, even when the corresponding coating parameters are not known. Fig. 5 Response patterns expressed as log K for the different polymers and test vapours. Fig. 6 Hierarchical cluster analysis on the log K values for the 11 vapours listed in Table 4. Actual values are listed in Tables 8 and 9.Analyst, October 1997, Vol. 122 1137Conclusions Both the formation and properties of electropolymerized Nsubstituted pyrrole films have been characterized by various methods, including cyclic voltammetry, XPS, scanning electron microscopy and vapour sorption studies. In particular, the characterization of PBPY, PCbPY, PHPY and PTHPY by XPS is described. Peak potentials for electropolymerization and subsequent redox modification are shifted to more positive values for all the derivatives relative to the parent compound, reflecting the effect of the side chain on the susceptibility to oxidation.The number of electrons involved in the electropolymerization process varies from 2.1 to 2.7 according to both the XPS and cyclic voltammetry data. The response behaviour of the coated TSM devices may be explained by consideration of the different physical mechanisms responsible for reversible interactions between different vapours and the polymer films.These are conveniently quantified by the sensor partition coefficient and the relevant solvatochromic parameters. The results obtained both here and in previous studies show that polymeric films of N-substituted pyrrole derivatives may be employed as coatings for chemical vapour sensors. In particular, the influence of the side chain on the mechanism and extent of coating–vapour interactions renders these polymers well-suited to the production of chemical sensor arrays based on acoustic wave devices.We are grateful to the Natural Sciences and Engineering Research Council (Canada) for financial assistance. We also thank Professor R. H. Morris of the University of Toronto for the use of the cyclic voltammetry system. References 1 Handbook of Conducting Polymers, ed. Skotheim, T. A., Marcel Dekker, New York, 1986, vol. 1 and 2. 2 Conducting Polymers: Special Applications, ed. Alc�acer, L., Proceedings of the workshop held at Sintra, Portugal, July 28–31, 1986, Reidel, Dordrecht, 1987. 3 Aldissi, M., Inherently Conducting Polymers: Processing, Fabrication, Applications, Limitations, Noyes Data Corporation, Park Ridge, NJ, 1989. 4 Science and Applications of Conducting Polymers, ed. Salaneck, W. R., Clark, D. T., and Samuelsen, E. J., Papers from the 6th European Physical Society Industrial Workshop, Lofthus, Norway, May 28–31, 1990, Adam Hilger, Bristol, 1991. 5 Intrinsically Conducting Polymers: An Emerging Technology, ed.Aldissi, M., NATO ASI Series E: Applied Sciences Volume 246, Kluwer Academic Publishers, Dordrecht, 1991. 6 Josowicz, M., Analyst, 1995, 120, 1019. 7 Feldheim, D. L., and Elliott, C. M., J. Membr. Sci., 1992, 70, 9. 8 Schmidt, V. M., Tegtmeyer, D., and Heitbaum, H., Adv. Mater., 1992, 4, 428. 9 Ge, H., and Wallace, G. G., Anal. Chem., 1989, 61, 2391. 10 Deinhammer, R. S., Shimazu, K., and Porter, M. D., Anal. Chem., 1991, 63, 1889. 11 Kunugi, Y., Nigorikawa, K., Harima, Y., and Yamashita, K., J.Chem. Soc., Chem. Commun., 1994, 873. 12 Slater, J. M., Paynter, J., and Watt, E. J., Analyst, 1993, 118, 379. 13 Teasdale, P. R., and Wallace, G. G., Analyst, 1993, 118, 329. 14 Topart, P., and Josowicz, M., J. Phys. Chem., 1992, 96, 8662. 15 Bartlett, P. N., and Ling-Chung, S. K., Sens. Actuators, 1989, 20, 287. 16 Amrani, M. E. H., Ibrahim, M. S., and Persaud, K. C., Mater. Sci. Eng., 1993, C:1, 17. 17 Vigmond, S. J., Kallury, K. M.R., Ghaemmaghami, V., and Thompson, M., Talanta, 1992, 39, 449. 18 Deng, Z., Stone, D. C., and Thompson, M., Can. J. Chem., 1995, 73, 1427. 19 Deng, Z., Stone, D. C., and Thompson, M., Analyst, 1996, 121, 671. 20 Deng, Z., Stone, D. C., and Thompson, M., Analyst , 1996, 121, 1341. 21 Diaz, A. F., and Bargon, J., in Handbook of Conducting Polymers., ed. Skothem, T. A., Marcel Dekker, New York, 1986, vol. 1, pp. 81– 115. 22 Zhao, Z. S., and Pickup, P. G., J. Electroanal. Chem., 1996, 404, 55. 23 Adams, R. N., Acc. Chem. Res., 1969, 2, 175. 24 Waltman, R. J., and Bargon, J., Can. J. Chem., 1986, 64, 76. 25 Diaz, A. F., and Kanazawa, K. K., in Extended Linear Chain Compounds, ed. Miller, J. S., Plenum, New York, 1983, vol. 3, p. 417–441. 26 Pfluger, P., and Street, G. B., J. Chem. Phys., 1984, 80, 544. 27 Chan, H. S. O., Kang, E. T., Neoh, K. G., and Lim, Y. K., Synth. Met., 1989, 30, 189. 28 Atanasoska, L., Naoi, K., and Smyrl, W. H., Chem. Mater., 1992, 4, 988. 29 Lei, J., Cai, Z., and Martin, C. R., Synth. Met., 1992, 46, 53. 30 Kang, E. T., Neoh, K. G., Ong, Y. K., Tan, K. L., and Tan, B. T. G., Macromolecules, 1991, 24, 2822. 31 Toshima, N., and Tayanagi, J.-I., Chem. Lett., 1990, 1369. 32 Nicholson, R. S., and Shain, I., Anal. Chem., 1964, 36, 706. 33 Grate, J. W., Athur, S., Jr., Ballantine, D. S., Wohltjen, H., Abraham, M. H., McGill, R. A., and Sasson, P., Anal. Chem., 1988, 60, 869. 34 Abraham, M. H., Whiting, G. S., Doherty, R.M., and Shuely, W. J., J. Chem. Soc., Perkin Trans. 2, 1990, 2, 1451. 35 Abraham, M. H., Whiting, G. S., Doherty, R. M., and Shuely, W. J., J. Chem. Soc., Perkin Trans. 2, 1990, 2, 1851. Paper 7/03165C Received May 8, 1997 Accepted June 30, 1997 1138 Analyst, October 1997, Vol. 122 Characterization of Polymer Films of Pyrrole Derivatives for Chemical Sensing by Cyclic Voltammetry, X-ray Photoelectron Spectroscopy and Vapour Sorption Studies Zhiping Deng, David C.Stone and Michael Thompson Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, Canada M5S 3H6 Eight different conducting polymer films formed from pyrrole and N-substituted pyrrole derivatives were characterized by cyclic voltammetry, X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy. In particular, the XPS of poly[N-butylpyrrole], poly[N-(2-carboxyethyl)pyrrole], poly[N–(6–hydroxyhexyl)pyrrole] and poly[N-(6-tetrahydropyranylhexyl)pyrrole] is reported for the first time.The vapour sorption properties of these films were also examined by forming the films onto the electrodes of thickness-shear mode acoustic wave sensors. The influence of the pendant side chain is apparent in both the electrochemical behaviour, composition, doping level, morphology and the nature and extent of polymer–vapour interactions. The latter can be rationalized by consideration of vapour physical properties and solvatochromic parameters.Keywords: Cyclic voltammetry; X-ray photoelectron spectroscopy; poly(pyrrole) and derivatives; vapour sorption; thickness-shear mode; quartz crystal; acoustic wave chemical sensor Organic conducting polymers such as poly(pyrrole) have found extensive application in highly diverse fields such as biomaterials, chemistry, electronics, microfabrication, non-linear optics, sensors and textiles.1–5 In the area of analytical chemistry, for example, such materials have been used to form chemically modified electrodes,6 permeation membranes,7,8 liquid chromatographic stationary phases9,10 and chemical and biological sensors.11–16 These chemical applications are all based on the physico-chemical interactions that occur between the conducting polymer and the respective chemical species. Our own interest in these materials is their use as selective (or partially selective) coatings for thickness-shear mode (TSM) acoustic wave sensors.One reason for using conducting polymers in this way is the ease of film formation by direct electropolymerization onto the TSM electrode surface.Other advantages lie in the ease with which redox state and incorporated counter ions can be changed electrochemically, and the wide range of functionalities that can be introduced through chemical reaction with the pyrrole ring nitrogen. This in turn provides various mechanisms for modifying the extent of different reversible interactions between the polymer films and different analytes, thus affecting both the sensitivity and selectivity of the sensor system.In a series of papers,17–20 we have described the application of various N-substituted poly(pyrroles) to organic vapour sensing. These include the parent compound, poly(pyrrole) (PPY), and poly(N-methylpyrrole) (PMPY), poly(N-butylpyrrole) (PBPY), poly[N-(2-cyanoethyl)pyrrole] (PCPY), poly[N- (2-carboxyethyl)pyrrole] (PCbPY), poly(N-phenylpyrrole) (PPPY), poly[N–(6–hydroxyhexyl)pyrrole] (PHPY) and poly[N-(6-tetrahydropyranylhexyl)pyrrole] (PTHPY).In the preceding papers, we have described the preparation, characterization and effect of redox state of PCPY18,20 as well as the application of all eight polymers to the selective detection of aroma components.19 In the present paper, we provide a more detailed characterization and comparison of the eight conducting polymers, with an emphasis on their formation and application as coatings for organic vapour sensors.Experimental Reagents and Materials The solvents hexane, toluene, methanol, butanal, acetonitrile, triethylamine, ethanol, butan-1-ol, hexan-1-ol and nonan-1-ol (analytical-reagent grade, Aldrich, Milwaukee, WI, USA) were used as received. Pyrrole (98%), N-methypyrrole (99%) and N- (2-cyanoethyl)pyrrole (99%) (Aldrich) were vacuum-distilled before use. Both N-phenylpyrrole (99%) (Aldrich) and tetrabutylammonium perchlorate (TBAP) (electrochemical grade, Fluka, Buchs, Switzerland) were used as received.N-(2-Carboxyethyl) pyrrole, N-butylpyrrole, N-(6-tetrahydropyrylhexyl) pyrrole and N-(6-hydroxyhexyl)pyrrole were synthesized as described previously.19 All other chemicals were obtained from Aldrich and used as received. The piezoelectric crystals were rough or optically polished 9 MHz AT-cut quartz crystals with gold electrodes (International Crystal Manufacturing, Oklahoma City, OK, USA). A combined Pt–Ag/AgCl (3 mol l21 KCl) electrode (Metrohm, Herisau, Switzerland) was utilized as the counter and reference electrode, respectively.Polymerization Procedures One electrode of the TSM device was used as a working electrode for the electropolymerization of pyrrole and its derivatives using a cell as described previously.17 Unpolished crystals were used for vapour sorption experiments while optically polished crystals were used for the cyclic voltammetry studies. All devices were washed with acetone and dried under nitrogen before use.Polymerization was achieved using a solution containing monomer (0.1 mol l21) and TBAP (0.1 mol l21) in de-oxygenated acetonitrile using either constant potential or cyclic voltammetric deposition with a potentiostat/ galvanostat (Model 273, EG&G Princeton Applied Research, Princeton, NJ, USA). The polymerization potentials for the different monomers have been listed previously.19 The coated TSM sensors were then rinsed with acetonitrile, dried with nitrogen, washed with acetone, dried with nitrogen and finally placed in an oven for 1 h at 100 °C.The coating mass for each TSM device was determined from the in-air frequency difference before and after polymer deposition. Analyst, October 1997, Vol. 122 (1129–1138) 1129Cyclic Voltammetry Cyclic voltammograms for film deposition and growth on the electrodes of polished TSM devices were obtained for ten cycles at a sweep rate of 100 mV s21 for potentials varying between 20.8 and +1.2 V versus Ag/AgCl, depending on the monomer used.The resulting films were then studied using sweep rates of 20–100 mV s21 and potentials varying between 20.8 and +1.2 V versus Ag/AgCl using TBAP (0.1 mol l21) in acetonitrile as the supporting electrolyte. X-ray Photoelectron Spectroscopy (XPS) XPS was performed using a Leybold Max 200 XPS (Leybold, Cologne, Germany) instrument using an unmonochromatized Mg Ka source and an analysis area of 2 3 4 mm2.Survey and low resolution spectra were obtained using a pass energy of 192 eV; high resolution spectra were acquired with a pass energy of 48 eV. All spectra were satellite-subtracted and normalized using software and elemental sensitivity factors provided by the manufacturer. The binding energy scale was further calibrated to 285.0 eV for the main C (1s) feature in order to compensate for sample charging effects. Scanning Electron Microscopy Scanning electron micrographs were obtained using a Hitachi Model S-570 microscope and recorded using the Quartz PCI image capture system.In order to reduce charging effects, the coated TSM devices were sputter-coated with a thin gold film using an argon plasma vapour deposition system (Polaron Model E5100, Polaron, Watford, Hertfordshire, UK). All micrographs were acquired using an accelerating voltage of 18 kV. Vapour Sorption Studies Vapour generation was achieved by bubbling helium gas through the corresponding liquid at room temperature and pressure using the flow system described previously.18,19 The flow rates of the sample and purge streams were held at 30 ml min21 for all experiments.The frequency shift of the polymer-coated TSM sensors was measured using a universal frequency counter (Model HP5334B, Hewlett-Packard, Avondale, PA, USA). Data were collected, displayed and stored using a Macintosh II computer equipped with an IEEE 488 interface bus using software written in-house.The coated TSM sensors were initially purged using pure helium carrier gas for 1–2 h to allow the device to stabilize. Once a stable baseline had been obtained, the stream was switched to the sample vapour and returned to the purge gas once the steady-state frequency shift had been obtained. Measurements were performed in triplicate for each coating– vapour combination. Vapour concentrations were obtained using a liquid nitrogen trap for a fixed time interval and weighing the resulting condensate.The measured vapour concentrations are given in Table 1, together with the relevant physical properties for the compounds studied. Results and Discussion Cyclic Voltammetry The cyclic voltammograms for the electropolymerization of all eight monomers are fairly similar, the main differences being in the position and magnitude of the anodic and cathodic peaks. That for N-(2-carboxyethyl)pyrrole is shown in Fig. 1 as a representative example.As can be seen, the first oxidation peak occurs at a higher oxidation potential (+1.06 V) than that of pyrrole (+0.85 V). In fact, the oxidation potentials of all seven N-substituted pyrroles are higher than that of the parent compound (Table 2). This is attributable to the steric and electronic effects of the pendant side chain, the inductive effect rendering the oxidation process more difficult for the mono- Table 1 Physical data and concentration values for the test solvents and corresponding vapours Cv/mg l21 Boiling- Density/ Molecular (at 21 °C and Solvent point/°C g cm23 mass/g mol21 30 ml min21) Hexane 69 0.659 86.16 31 Toluene 110 0.865 110.6 10 Water 100 1.000 18.00 15 Acetonitrile 82 0.786 41.05 34 Triethylamine 88.8 0.726 101.19 30 Butanal 75 0.800 72.11 22 Methanol 64.7 0.791 32.04 40 Ethanol 78 0.785 46.07 27 Butan-1-ol 117.7 0.810 74.12 8.3 Hexan-1-ol 156.5 0.814 102.18 1.2 Nonan-1-ol 215 0.827 144.26 0.3 Fig. 1 Cyclic voltammetric polymerization of N-(2-carboxyethyl)pyrrole (0.1 mol l21 in 0.1 mol l21 TBAP–acetonitrile) at scan rate of 100 mV s21 versus Ag–AgCl (3 mol l21).Table 2 Anodic and cathodic peak potentials (versus Ag/AgCl) for the cyclic voltammetric polymerization of N-substituted pyrrole monomers (0.1 mol l21 monomer in 0.1 mol l21 TBAP–acetonitrile, 100 mV s21 scan rate) Epa Epa Epc Polymer (monomer)/V* (polymer)/V† (polymer)/V† PPY 0.85 20.08 20.23 PMPY 0.90 0.46 0.42 PBPY 1.08 0.65 0.47 PCPY 1.10 0.70 0.64 PCbPY 1.06 0.68 0.56 PPPY 1.19 0.67 0.58 PHPY 1.05 0.68 0.52 PTHPY 1.08 0.69 0.58 * From first scan.† From second scan. 1130 Analyst, October 1997, Vol. 122mers.21 When the potential is reversed on the first scan, the anodic current continues to increase for a short time before decreasing. The following cathodic current is also higher than that in the forward scan until the potential is +0.76 V, resulting in a loop in the voltammogram. This is indicative of a nucleation mechanism, and is commonly observed for the electropolymerization of conducting polymers.22 The effect is attributed to the fact that polymerization occurs faster at a polymeric nucleus than at the uncoated electrode surface.The anodic peak for the second scan occurs at a much lower potential (+0.68 V), and corresponds to polymer growth with incorporation of perchlorate counter ions (Table 2). The peak is broad because there will be a distribution of oligomer sizes, while diffusion of counter ions in and out of the film will be slow.Adams23 has concluded that whenever the second and subsequent sweeps of a cyclic voltammogram differ markedly from the first, a followup chemical reaction has occurred. These results, therefore, show the characteristics of an ECE reaction, in which electron transfer is followed by a chemical reaction and subsequent electron transfer reaction.24 Therefore, the overall polymerization mechanism for the pyrrole derivatives is the same as that for the parent compound.25 Another feature of the voltammograms is that, on successive scans, the anodic peak shifts to increasingly positive potential with a corresponding increase in current until a limiting value is reached.A similar effect is observed for the cathodic peak, except here the shift is towards a more negative potential. The shift is, at least in part, due to the polymer film resistance giving rise to a larger overpotential.The ratio of the cathodic to anodic peak currents is also less than one, confirming that the anodic peak consists of both reversible and irreversible contributions. The anodic peak potentials for the different films are very similar with the exception of PPY and PMPY. This may be explained by considering the effect of side-chain length on the porosity of the polymer. It is anticipated that as the length of the substituent chain increases, the polymer chains will be forced farther apart making it easier for counter ions to diffuse in and out of the film.This is confirmed by the cathodic peak potentials for the series PPY, PMPY, PBPY and PTHPY, which show an increasing shift towards positive potentials. Following film formation, the cyclic voltammogram for each polymer film was obtained at various scan rates using TBAP in acetonitrile without monomer as the supporting electrolyte. Well-defined, broad anodic and cathodic peaks were obtained in all cases (Fig. 2), which reflects the slow rate of counter ion transfer in and out of the film as well as interactions between electroactive sites and the electrochemical non-equivalence of these sites.26 The magnitude of the anodic and cathodic peak currents increases linearly with scan rate, while peak separation also increases. The polymer redox processes are, therefore, not diffusion-controlled. Table 3 summarizes the positions and separation of the anodic and cathodic peaks.Both the anodic (Epa) and cathodic (Epc) peak potentials become increasingly positive relative to the parent PPY as the chain length of the Nsubstituent increases. This trend arises from increasing distortion of inter-ring coplanarity, which would destabilize the cationic (oxidized) form of these polymers. One potential benefit of the positive shift in Epa is that films formed from Nsubstituted pyrrole derivatives will be less sensitive to atmospheric oxidation.This implies improved handling, storage and lifetime characteristics for chemical sensors fabricated using such films. Another observation is that the peak separation (DE) is considerably lower for PPY and PMPY compared with the other polymer films. In fact, the oxidation and reduction of both PPY and PMPY films are essentially reversible, while other irreversible processes must be involved for the remaining films. As an example, PHPY shows two anodic waves, the second appearing at Å 1.0 V for scan rates of 80 and 100 mV s21.This is likely associated with oxidation of the hydroxyl group, although further work is needed to confirm this. X-ray Photoelectron Spectroscopy Although the X-ray photoelectron spectra of PPY, PMPY, PPPY and PCPY films have been previously reported, 17,18,20,26–31 we have included the relevant data with those for PBPY, PCbPY, PHPY and PTHPY in Table 4 for ease of comparison. This lists the results of peak deconvolution on the C(1s), N(1s), O(1s) and Cl(2p) regions for all eight as-prepared polymer films.In addition, Fig. 3 shows the C(1s) spectrum obtained for PCbPY. All eight films show the expected principal C(1s) component at 285 eV arising from the pyrrole ring a and b carbons26 and methylene groups within the pendant side chain. These species are unresolvable under the experimental conditions used here. A second C(1s) component occurring between 286.1 and 286.4 eV can arise from a number of different sources. For PCPY, for example, it is attributed to the nitrile group as discussed previously.18 For PHPY, there is also a contribution from the side chain –CH2OH group, while PTHPY contains two ether groups.It is, therefore, not surprising that these three films show significant enhancement of the 286 eV component relative to the remaining polymers. The origin of this peak for PPY, PMPY, PBPY, PCbPY and PPPY is amenable to several interpretations. Pfluger and Street26 attribute this band in PPY to ‘disorder type’ carbons, which they define as being crosslinked, chain-terminating and non-a,aA bonded carbons as well as partially saturated rings.There will also be a small contribution from unavoidable hydrocarbon contaminants, while Atanasoska et al.28 have suggested that electrostatic interaction of ring a carbons with counter ions will also have an effect. A fourth possibility is the presence of covalently bound chlorine species, which we will consider shortly.The final C(1s) component between 288 and 289 eV is generally attributed to carbonyl or carboxyl species resulting from chain termination.21,28 This has been partly confirmed by FTIR studies of chemically polymerized PPY samples,29 which clearly show the presence of a small band at 1705 cm21 that scales in intensity with the 288 eV C(1s) component. For PCbPY, there is also the pendant carboxyl group (289.1 eV), which clearly increases the relative contribution of this component.Turning now to the N(1s) region, all the films show two peaks located at approximately 400 and 402 eV, with the major component being the lower binding energy signal. This is in general agreement with other XPS studies of different PPY species,26,28 a number of which also show a small shoulder at Å 397 eV.17,20,29,30 The main peak at 400 eV arises from the neutral pyrrole ring nitrogen, while the higher binding energy component is generally attributed to partially charged nitrogens within bipolaron sub-units.The presence of the shoulder at 397 eV depends, to some extent, on the experimental signal-to-noise ratio but also on the film preparation conditions since it is more pronounced for neutral than as-prepared or oxidized PPY and PCPY.17,20 Lei et al.29 have attributed it to –CNN– defects in the pyrrole backbone, while Vigmond et al.17 ascribed it to interchain hydrogen bonding effects. In this latter interpretation, equal intensity peaks on either side of the principal N(1s) line are expected due to electron donation from one nitrogen to another.This being the case, one would expect enhancement of the higher binding energy component in the as-prepared films for polymers in which electron donation from one ring nitrogen to an adjacent ring side chain is possible. Such an effect is, in fact, observed for all the polymers relative to the phenyl- and alkyl-substituted pyrroles (PMPY, PBPY and PPPY).The exception here is PCPY, although the situation in this case may Analyst, October 1997, Vol. 122 1131be masked by the additional nitrogen per pyrrole unit contributed by the nitrile group. The Cl(2p) spectra show peaks with the 2p3/2 component occurring at 207.6 and 200.7 eV. The higher binding energy peak is due to incorporated perchlorate ion, while the lower peak is assigned to covalently incorporated chlorine as discussed previously.20 Support for this assignment comes from the results of Kang et al.30 and Toshima and Tayanagi,31 who observed similar Cl(2p) spectra for chemically polymerized PPY samples.In this respect, it should be noted that the covalently bound chlorine will also result in an increase in the C(1s) component at 286 eV. It is possible to estimate the doping ratio of the as-prepared polymer films from the XPS data using the Cl/N ratio, although Fig. 2 Cyclic voltammograms for the polymer films in 0.1 mol l21 TBAP–acetonitrile without monomer at different scan rates (mV s21). 1132 Analyst, October 1997, Vol. 122this will not be as accurate as that calculated from the composition obtained by bulk elemental analysis since only a thin layer of an irregular surface is being analysed. The estimated doping ratio for each film is given in Table 5, together with the corresponding number of electrons per monomer unit (n) involved in polymerization and subsequent oxidation. It is also possible to estimate a value of n from the cyclic voltammetry data using Nicholson’s model for irreversible charge transfer32 Ip = nFAC*n1/2(pDanF/RT)1/2c(bt) (1) where F is Faraday’s constant, A is the area, C* is the surface concentration of electroactive species, v is the scan rate, D is the diffusion coefficient, a is the transfer coefficient and c(bt) is the current function for irreversible charge transfer.A difficulty here, however, is obtaining a reasonable estimate of the surface concentration for an amorphous, porous polymer film.One crude approach is to use the generally accepted value of n = 2.3–2.5 for PPY and use this to calculate the surface concentration, then assume an identical value for the Nsubstituted derivatives. Agreement between the XPS data and the values of n derived in this manner is surprisingly good, although it breaks down for PCPY and PCbPY. The O(1s) region can generally be deconvoluted into two components at 533.5 and 532.0 eV, representing singly and doubly bonded oxygen species, respectively.For the nonoxygen containing polymers, the only sources of oxygen in the final film will be incorporated counter ion and carboxy groups introduced by chain termination reactions.28 Note that perchlorates also show oxygen binding energies of Å 533 eV, which is unresolvable from the other species for the instrumental conditions used in this study. For these reasons, the O(1s) region has rarely been discussed in previous XPS studies of poly- (pyrroles).Simple composition calculations assuming that the perchlorate and carbonyl oxygen species overlap yield RNO: R– O ratios close to the theoretical 1 : 1 for PPY, PCbPY and PPPY, but break down for PMPY, PBPY, PCPY and PTHPY. This is anticipated since the relative amount of chlorine species other than perchlorate was not taken into account. What the results do show, however, is the expected progressive increase in R–O species for PHPY and PTHPY relative to the parent compound.One curious feature of the O(1s) deconvolution is the apparent presence of an additional oxygen species with a high binding energy of 535 eV. Such species can be observed on plasma-cleaned gold surfaces. This can be discounted, however, since no gold peaks are observed in any of the survey scans. Table 3 Anodic and cathodic peak potentials (versus Ag/AgCl) of Nsubstituted pyrrole polymers in 0.1 mol l21 TBAP–acetonitrile for a 20 mV s21 scan rate Polymer Epa/V Epc/V DE/V PPY 0.28 0.25 0.027 PMPY 0.51 0.48 0.030 PBPY 0.72 0.56 0.15 PCPY 0.82 0.72 0.10 PCbPY 0.88 0.76 0.12 PPPY 0.84 0.70 0.14 PHPY 0.81 0.64 0.16 PTHPY 0.86 0.76 0.10 Table 4 Polymer elemental composition (at.-%) and peak analysis from the XPS data C(1s) N(1s) O(1s) Cl(2p) Polymer BE*/eV at.-% BE*/eV at.-% BE*/eV at.-% BE*/eV at.-% PPY 66.9 11.6 18.2 3.3 285.0 76.9 400.2 71.3 532.4 87.5 207.6 100.0 286.3 18.4 402.1 28.7 533.7 12.5 288.1 4.7 PMPY 74.5 10.4 12.7 2.4 285.1 69.1 400.1 87.3 531.6 64.7 200.9 100.0 286.4 18.7 401.8 12.7 533.0 35.3 288.4 12.2 PBPY 71.9 13.4 12.8 1.9 285.0 75.7 398.5 87.9 531.8 52.9 201.1 78.6 286.4 16.6 400.3 12.1 533.0 47.1 208.1 21.4 288.2 7.7 PCPY 73.7 7.8 15.4 2.3 284.8 51.3 399.9 93.8 532.0 70.7 200.7 40.1 286.4 43.8 401.7 6.2 533.6 29.3 207.5 59.7 288.6 5.0 PCbPY 63.9 9.1 25.0 2.0 284.9 63.7 399.9 59.8 532.3 60.8 200.6 68.2 286.2 18.9 400.4 34.9 533.5 33.6 207.5 31.8 289.1 17.3 401.6 9.4 534.6 5.6 PPPY 73.3 8.1 16.7 1.3 284.9 75.0 400.7 92.0 532.6 63.0 200.4 100.0 286.1 21.0 402.4 8.0 533.5 37.0 287.9 4.0 PHPY 74.6 7.3 16.6 1.5 284.9 58.5 400.1 69.0 532.5 56.8 207.4 100.0 286.1 36.1 401.4 31.0 533.8 36.8 288.5 5.4 535.6 6.4 PTHPY 64.6 5.4 27.5 2.6 285.0 61.0 400.3 62.8 532.4 22.4 200.8 18.8 286.3 29.5 402.3 37.3 533.4 67.5 207.9 81.2 288.3 9.5 534.6 10.2 * BE = Binding energy.Analyst, October 1997, Vol. 122 1133There are several other possible explanations, including differential charging affecting the incorporated counter ion and the presence of an oxidized nitrogen species within the polymer.It is difficult to envisage the latter, however, since the peak is only observed for N-substituted polymers. A final possibility is that it is a fitting artifact arising from the unresolved perchlorate oxygen, since there are an infinite number of mathematical solutions that would equally well fit the O(1s) envelope.Further work is, therefore, needed to clarify this matter. Scanning Electron Microscopy In previous studies, we have found that surface morphology can have an effect on the response of the polymer-coated sensors to different vapours.17–20 Polymer morphology is, in turn, influenced to various degrees by substrate morphology, film thickness and level of oxidation, and the nature of the counter ion. PPY films are frequently described as having a ‘cauliflower- like’ appearance, which arises from the nucleation and phase growth mechanism of the electropolymerization process.This detail is not apparent on very thin films, however, which simply follow the underlying surface. It is, therefore, important to specify both substrate surface roughness and film thickness when comparing the appearance of different films. In this, and our previous studies, the films were formed on microscopically rough surfaces. This was necessary since these polymer films adhere to gold electrodes primarily through mechanical interlock, and are prone to peeling from optically flat surfaces.In particular, the cleaning procedure used in the vapour sorption studies is highly effective as a means of removing conducting polymer films from polished gold electrodes, while films formed on the unpolished devices remain intact. The scanning electron micrographs for six of the polymers are shown in Fig. 4. Micrographs of PPY and PCPY have been published elsewhere.17,18,20,25 The corresponding film thicknesses, calculated from the in-air frequency shift due to film deposition, Dfs, are listed in Table 6.Generally, films of the Nsubstituted pyrrole derivatives show similar morphologies to the parent compound, although there are some differences. Both PMPY and PPPY, for example, exhibit a much more granular surface than PPY even for the same film thickness. The films of PBPY and PHPY are the most similar to PPY, while PCPY and PCbPY again show similar topography but with deep, open channels running into the bulk of the polymer.In this context, it should be noted that these channels disappear if the PCPY films are heavily oxidized by application of a large positive potential, becoming very similar in appearance to the as-prepared PBPY and PHPY films. The most different film in terms of morphology is PTHPY which, although as thick as the PPY and PPPY films, shows no apparent differences from the underlying gold electrode.These differences in morphology have a significant influence on the response of the coated vapour sensors. This occurs firstly through variations in effective surface area, which determines the extent of initial adsorption, and secondly through the film porosity, which affects both the absolute steady-state frequency response (extent of vapour absorption) and recovery time (vapour desorption). Table 5 Doping levels and number of electrons involved in electropolymerization –oxidation from the XPS and cyclic voltammetry (CV) data n Polymer Cl/N ratio Doping ratio XPS CV PPY 0.286 0.29 2.3 2.2 PMPY 0.232 0.23 2.2 2.3 PBPY 0.136 0.14 2.1 2.1 PCPY 0.289 0.15 2.2 2.6 PCbPY 0.219 0.22 2.2 2.7 PPPY 0.154 0.15 2.2 2.2 PHPY 0.205 0.21 2.2 2.5 PTHPY 0.488 0.49 2.5 2.6 Fig. 3 X-ray photoelectron spectra of PCbPY showing the C(1s), N(1s), O(1s) and Cl(2p) regions. 1134 Analyst, October 1997, Vol. 122Vapour Sorption Studies Since the magnitude of the observed frequency shift for vapour sorption (Dfv) varies with both vapour concentration (Cv) and film thickness, it is necessary to normalize the data before meaningful comparisons can be made.In keeping with our earlier work, we therefore calculate the sensor partition coefficient,33 K f f C S = D D nr V (2) where Dfv is in Hz, Dfs (kHz) is the frequency shift due to the polymer coating, r (g cm23) is the film density and Cv is in g cm23. Alternatively, the value of Dfv may be normalized to the shift that would have been obtained for a film having Dfs = 30 kHz.A series of organic vapours were chosen as probe molecules representing a variety of structural and functional group interactions between the vapour and the polymer film (Table 1). Hexane, for example, can interact solely through dispersion forces while toluene also exhibits p-electron overlap and polarizibility interactions. The remaining vapours were chosen to represent differing degrees of hydrogen bond acidity and basicity and dipole–dipole interactions, while a series of four primary alcohols was included to study the effects of chain length and analyte volatility.The solvatochromic parameters for all these vapours are listed in Table 7. While the corresponding Fig. 4 Scanning electron micrographs of six of the conducting polymer films. (a) PMPY; (b) PBPY; (c); PCbPY; (d) PPPY; (e) PHPY; and (f) PTHPY. Table 6 Film thickness and in-air frequency shifts corresponding to electropolymerization of the pyrrole derivatives on the TSM device Polymer PPY PMPY PCbPY PBPY PCPY PPPY PHPY PTHPY Thickness/mm 0.45 0.39 2.96 1.91 0.98 0.49 1.10 0.48 Dfs/kHz 28.8 21.3 104.5 99.0 50.9 26.0 55.2 22.0 Analyst, October 1997, Vol. 122 1135coating parameters are not known for the pyrrole derivatives used in this study, these values are nonetheless useful in comparing the response behaviour of the different vapour– coating combinations.The differing abilities of these combinations to interact through various mechanisms gives rise to different partial selectivities for each coating. This is reflected in the experimentally observed log K values, which are listed in Tables 8 and 9 and summarized graphically in Fig. 5. Comparing hexane and methanol, for example, shows that hexane generally gives much lower normalized frequency shifts ( < 2 kHz) than methanol (2–8 kHz), which reflects the difference in enthalpy of vaporization, and hence vapour concentration, between these two solvents (31.56 and 37.43 kJ mol21 at 25 °C, respectively).When the effect of vapour concentration is factored out by calculating log K, however, we see that the differences in response are actually minimal for the PMPY- and PBPY-coated sensors. On the other hand, fairly large differences in response in favour of methanol are observed for those coatings capable of forming hydrogen bonds, especially PCPY, PCbPY, PHPY and PPY.Water and toluene both have significantly higher boilingpoints and lower vapour pressures than hexane and methanol. Their normalized frequency shifts generally lie between the extremes observed for hexane and methanol, reflecting the fact that while their Cv values are lower, their rate of desorption from the film will also be slower resulting in a larger steadystate equilibrium vapour concentration within the film coatings. Comparing the log K values for this pair shows a lower extent of differentiation between water and toluene although, again, a large difference in favour of water vapour is observed for the two most polar coatings (PCPY and PCbPY).PBPY, PPY and PTHPY all show larger log K values for toluene than either water or methanol vapour. This is readily explained since toluene has the largest polarizibility, dipolar and dispersion solvatochromic parameters while the side chains of PBPY, PPY and PTHPY are all essentially non-polar in nature.PHPY, which has a relatively non-polar side chain in spite of its terminal hydroxyl group, also shows very similar response behaviour towards the same vapours. Comparing the remaining vapours, methanol and acetonitrile show very similar response patterns with the greatest differentiation observed for PTHPY. Triethylamine, on the other hand, gives a markedly lower response with the most polar coatings (PCPY and PCbPY) and reduced responses with PPY, PMPY, PPPY and PHPY.Butanal shows a similar pattern to methanol and acetonitrile, showing greater log K values for PPY, PMPY and PBPY. This reflects the larger dispersion parameter for butanal compared with methanol and acetonitrile. Neither butanal, acetonitrile nor triethylamine can function as Table 7 Solvatochromic parameters for the test liquids (from refs. 34 and 35). The parameters represent polarizability (R2), dipolar interactions (p2 *), hydrogen bond acidity and basicity (a2 H and b2 H, respectively) and dispersion interactions (log L16) Solvent R2 p2 * a2 H b2 H Log L16 Hexane 0.00 0.00 0.00 0.00 2.688 Toluene 0.601 0.55 0.00 0.14 3.344 Methanol 0.278 0.40 0.37 0.41 0.922 Water 0.00 0.43 0.33 0.65 0.267 Butanal 0.187 0.65 0.00 0.40 2.270 Acetonitrile 0.237 0.75 0.09 0.44 1.560 Triethylamine 0.101 0.15 0.00 0.67 3.077 Ethanol 0.246 0.40 0.33 0.44 1.485 Butan-1-ol 0.224 0.40 0.33 0.45 2.601 Hexan-1-ol 0.210 0.40 0.33 0.45 3.610 Nonan-1-ol 0.193 0.40 0.33 0.45 5.124 Table 8 Normalized frequency shifts and corresponding partition coefficients obtained for the coated TSM sensors on exposure to different test vapours Analyte Hexane Toluene Methanol Water Polymer Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K PPY 279 ± 17 2.65 845 ± 19 3.70 2948 ± 82 3.56 2581 ± 53 4.01 PMPY 1038 ± 16 3.21 670 ± 83 3.51 1845 ± 97 3.35 1156 ± 45 3.57 PBPY 1860 ± 80 3.39 3462 ± 189 4.16 2230 ± 121 3.36 996 ± 65 3.43 PCPY 146 ± 9 2.34 83 ± 7 2.67 3870 ± 76 3.73 2863 ± 49 4.03 PCbPY 170 ± 9 2.41 281 ± 18 3.12 8019 ± 105 3.98 4252 ± 107 4.13 PPPY 987 ± 45 3.17 4030 ± 145 4.27 5331 ± 132 3.69 2954 ± 123 3.96 PHPY 441 ± 40 2.81 1461 ± 18 3.82 6757 ± 162 3.88 1824 ± 117 3.74 PTHPY 712 ± 75 2.97 2768 ± 87 4.05 4315 ± 152 3.64 1952 ± 35 3.72 Butanal Acetonitrile Triethylamine Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K 2078 ± 81 3.67 1875 ± 62 3.51 1065 ± 45 3.34 1873 ± 75 3.62 1564 ± 82 3.35 975 ± 21 3.20 3236 ± 83 3.78 2873 ± 46 3.54 2528 ± 138 3.54 3392 ± 91 3.86 4560 ± 152 3.88 573 ± 12 3.04 3775 ± 78 3.71 6556 ± 149 3.96 593 ± 42 2.97 2092 ± 11 3.65 4054 ± 52 3.75 1275 ± 63 3.30 4175 ± 39 3.93 4378 ± 121 3.76 1783 ± 103 3.43 1562 ± 50 3.46 1115 ± 16 3.13 2787 ± 33 3.58 Table 9 Normalized frequency shifts and corresponding partition coefficients obtained for the coated TSM sensors on exposure to a series of primary alcohol vapours Analyte Ethanol Butan-1-ol Hexan-1-ol Nonan-1-ol Polymer Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K Dfv/Hz Log K PPY 1299 ± 31 3.38 827 ± 43 3.69 714 ± 19 4.47 236 ± 11 4.59 PMPY 3373 ± 56 3.78 555 ± 19 3.51 521 ± 8 4.32 201 ± 15 4.51 PBPY 2673 ± 85 3.61 1421 ± 64 3.85 2661 ± 58 4.96 530 ± 28 4.86 PCPY 3959 ± 42 3.84 522 ± 27 3.47 437 ± 15 4.23 189 ± 8 4.47 PCbPY 4819 ± 67 3.93 2756 ± 94 4.12 1202 ± 71 4.68 300 ± 8 4.68 PPPY 4947 ± 79 3.93 3732 ± 153 4.32 2611 ± 39 5.01 161 ± 5 4.40 PHPY 2250 ± 28 3.57 1650 ± 25 3.95 2103 ± 46 4.90 393 ± 13 4.77 PTHPY 3613 ± 102 3.74 2930 ± 49 4.16 1827 ± 31 4.79 143 ± 9 4.29 1136 Analyst, October 1997, Vol. 122hydrogen bond donors, so one might expect lower responses to these vapours for coatings that are also unable to function as hydrogen bond donors (e.g., PMPY, PBPY and PPPY). This is in fact seen for butanal and acetonitrile, but dispersion interactions are also a strong factor for triethylamine, resulting in much higher responses with PBPY and PTHPY. The series of primary alcohols show the expected increase in log K with increasing carbon number, although it is non-linear.This is presumably a result of concentration effects arising from the high vapour concentrations used in this study compared with those found in typical gas chromatography experiments. Ethanol shows the least variation in log K between different coatings, while hexan-1-ol shows the greatest. The solvatochromic parameters for this series differ only in the magnitude of the polarizibility and dispersion parameters, reflecting the fact that only the chain length is changing.In this respect, it is interesting that the largest log K values are observed for the coatings having longer side chains such as PBPY and PHPY. One method to characterize the vapour sorption properties of the polymer coatings across the series of test vapours is to use hierarchical cluster analysis. In a previous paper, this technique was employed in order to examine similarities in response patterns (and therefore potential redundancy) for four longchain unsaturated aroma components.19 A similar analysis was performed on the log K values measured for the current set of analyte vapours using the Euclidean distance, single linkage (nearest neighbour) method without autoscaling. The cluster analysis was also repeated for various sub-sets of the data.Fig. 6 shows the dendrogram obtained using all the vapours listed in Table 1. Some differences were observed between the current and previous study, which employed only five- to nine-carbon chain unsaturated alcohols and aldehydes.19 For example, while the most polar films (PCPY and PCbPY) clustered together in both cases, different groupings were observed for the remaining polymers. In fact, the current results demonstrate a clustering pattern that much more closely reflects both the polarity and hydrogen bonding ability of the different polymeric coatings.This is clearly seen in the fact that the hydrogen bond donors PPY and PHPY cluster together, as do the longer side chain, non-hydrogen bond donors PPPY, PTHPY and PBPY.For comparison, in the previous study (where dispersion interactions were likely more significant than for the short chain analyte vapours used here) the clustering pattern showed a strong correlation with side-chain length so that PTHPY and PHPY were grouped together. Returning to the present study, the observed clustering of the polymeric coatings was further confirmed when the analysis was repeated on sub-sets of the data containing either polar or non-polar analytes.For the purposes of analysis, these were defined in two different ways: (1) on the presence or absence of polar functional groups (hydroxyl, carbonyl, etc.); and (2) on the arbitrary condition that, for ‘polar’ analytes, logL16 < 2.5. This second definition reflects the fact that longer chain analytes have a significant non-polar component, while the first definition yielded only two ‘non-polar’ cases.The resulting dendrograms (not shown) support the view that the magnitude of the response (and, therefore, the partial selectivity) for a given coating–vapour pair is a direct consequence of the nature and strength of the initial adsorption interaction. Since these interactions are also those involved in solvation processes, solvatochromic parameters such as those listed in Table 7 provide a useful tool for the rational choice of sensor coatings, even when the corresponding coating parameters are not known.Fig. 5 Response patterns expressed as log K for the different polymers and test vapours. Fig. 6 Hierarchical cluster analysis on the log K values for the 11 vapours listed in Table 4. Actual values are listed in Tables 8 and 9. Analyst, October 1997, Vol. 122 1137Conclusions Both the formation and properties of electropolymerized Nsubstituted pyrrole films have been characterized by various methods, including cyclic voltammetry, XPS, scanning electron microscopy and vapour sorption studies.In particular, the characterization of PBPY, PCbPY, PHPY and PTHPY by XPS is described. Peak potentials for electropolymerization and subsequent redox modification are shifted to more positive values for all the derivatives relative to the parent compound, reflecting the effect of the side chain on the susceptibility to oxidation. The number of electrons involved in the electropolymerization process varies from 2.1 to 2.7 according to both the XPS and cyclic voltammetry data. The response behaviour of the coated TSM devices may be explained by consideration of the different physical mechanisms responsible for reversible interactions between different vapours and the polymer films.These are conveniently quantified by the sensor partition coefficient and the relevant solvatochromic parameters. The results obtained both here and in previous studies show that polymeric films of N-substituted pyrrole derivatives may be employed as coatings for chemical vapour sensors. In particular, the influence of the side chain on the mechanism and extent of coating–vapour interactions renders these polymers well-suited to the production of chemical sensor arrays based on acoustic wave devices. We are grateful to the Natural Sciences and Engineering Research Council (Canada) for financial assistance. We also thank Professor R. H. Morris of the University of Toronto for the use of the cyclic voltammetry system. References 1 Handbook of Conducting Polymers, ed. Skotheim, T. A., Marcel Dekker, New York, 1986, vol. 1 and 2. 2 Conducting Polymers: Special Applications, ed. Alc�acer, L., Proceedings of the workshop held at Sintra, Portugal, July 28–31, 1986, Reidel, Dordrecht, 1987. 3 Aldissi, M., Inherently Conducting Polymers: Processing, Fabrication, Applications, Limitations, Noyes Data Corporation, Park Ridge, NJ, 1989. 4 Science and Applications of Conducting Polymers, ed. Salaneck, W. R., Clark, D. T., and Samuelsen, E. J., Papers from the 6th European Physical Society Industrial Workshop, Lofthus, Norway, May 28–31, 1990, Adam Hilger, Bristol, 1991. 5 Intrinsically Conducting Polymers: An Emerging Technology, ed. Aldissi, M., NATO ASI Series E: Applied Sciences Volume 246, Kluwer Academic Publishers, Dordrecht, 1991. 6 Josowicz, M., Analyst, 1995, 120, 1019. 7 Feldheim, D. L., and Elliott, C. M., J. Membr. Sci., 1992, 70, 9. 8 Schmidt, V. M., Tegtmeyer, D., and Heitbaum, H., Adv. Mater., 1992, 4, 428. 9 Ge, H., and Wallace, G. G., Anal. Chem., 1989, 61, 2391. 10 Deinhammer, R. S., Shimazu, K., and Porter, M. D., Anal. Chem., 1991, 63, 1889. 11 Kunugi, Y., Nigorikawa, K., Harima, Y., and Yamashita, K., J. Chem. Soc., Chem. Commun., 1994, 873. 12 Slater, J. M., Paynter, J., and Watt, E. J., Analyst, 1993, 118, 379. 13 Teasdale, P. R., and Wallace, G. G., Analyst, 1993, 118, 329. 14 Topart, P., and Josowicz, M., J. Phys. Chem., 1992, 96, 8662. 15 Bartlett, P. N., and Ling-Chung, S. K., Sens. Actuators, 1989, 20, 287. 16 Amrani, M. E. H., Ibrahim, M. S., and Persaud, K. C., Mater. Sci. Eng., 1993, C:1, 17. 17 Vigmond, S. J., Kallury, K. M. R., Ghaemmaghami, V., and Thompson, M., Talanta, 1992, 39, 449. 18 Deng, Z., Stone, D. C., and Thompson, M., Can. J. Chem., 1995, 73, 1427. 19 Deng, Z., Stone, D. C., and Thompson, M., Analyst, 1996, 121, 671. 20 Deng, Z., Stone, D. C., and Thompson, M., Analyst , 1996, 121, 1341. 21 Diaz, A. F., and Bargon, J., in Handbook of Conducting Polymers., ed. Skothem, T. A., Marcel Dekker, New York, 1986, vol. 1, pp. 81– 115. 22 Zhao, Z. S., and Pickup, P. G., J. Electroanal. Chem., 1996, 404, 55. 23 Adams, R. N., Acc. Chem. Res., 1969, 2, 175. 24 Waltman, R. J., and Bargon, J., Can. J. Chem., 1986, 64, 76. 25 Diaz, A. F., and Kanazawa, K. K., in Extended Linear Chain Compounds, ed. Miller, J. S., Plenum, New York, 1983, vol. 3, p. 417–441. 26 Pfluger, P., and Street, G. B., J. Chem. Phys., 1984, 80, 544. 27 Chan, H. S. O., Kang, E. T., Neoh, K. G., and Lim, Y. K., Synth. Met., 1989, 30, 189. 28 Atanasoska, L., Naoi, K., and Smyrl, W. H., Chem. Mater., 1992, 4, 988. 29 Lei, J., Cai, Z., and Martin, C. R., Synth. Met., 1992, 46, 53. 30 Kang, E. T., Neoh, K. G., Ong, Y. K., Tan, K. L., and Tan, B. T. G., Macromolecules, 1991, 24, 2822. 31 Toshima, N., and Tayanagi, J.-I., Chem. Lett., 1990, 1369. 32 Nicholson, R. S., and Shain, I., Anal. Chem., 1964, 36, 706. 33 Grate, J. W., Athur, S., Jr., Ballantine, D. S., Wohltjen, H., Abraham, M. H., McGill, R. A., and Sasson, P., Anal. Chem., 1988, 60, 869. 34 Abraham, M. H., Whiting, G. S., Doherty, R. M., and Shuely, W. J., J. Chem. Soc., Perkin Trans. 2, 1990, 2, 1451. 35 Abraham, M. H., Whiting, G. S., Doherty, R. M., and Shuely, W. J., J. Chem. Soc., Perkin Trans. 2, 1990, 2, 1851. Paper 7/03165C Received May 8, 1997 Accepted June 30, 1997 1138 Analyst, October 1997, Vol. 1