Comparison of electronic transport in polyaniline blends, polyaniline and polypyrrole

A.B. Kaiser^a, C.-J. Liu^a, P.W. Gilberd^a, B. Chapman^a,b, N.T. Kemp^a,
B. Wessling^c, A.C. Partridge^b, W.T. Smith^d, J.S. Shapiro^d

^aPhysics Department, Victoria University of Wellington, P O Box 600, Wellington, New Zealand
^bIndustrial Research Ltd., P O Box 31 310, Lower Hutt, New Zealand
^cZipperling Kessler & Co., Postfach 1464, D-22904 Ahrensburg, Germany
^dSchool of Chemistry, Macquarie University, North Ryde, N.S.W., Australia 3207

Abstract

The conductivity of PAni/PETG copolyester blends shows the expected reduction compared to unblended PAni, but PAni/PMMA (and PAni/PVC blends at lower temperatures) show a larger conductivity. SEM images indicate that the different behaviour of the blends appears to be associated with greater granularity retained by the PAni particles dispersed in PETG. The conductivities of the more highly conducting PAni blends, like PAni-CSA and its blends with PMMA measured by other authors, show a change to metallic sign for the temperature dependence near room temperature - simple models combining hopping/tunnelling with metallic conduction are consistent with the conductivity except at very low temperatures. The thermopower of PAni blends shows a remarkable similarity for blends with different insulating polymers and widely varying conductivity. For polypyrrole samples of very different conductivity, the change to metallic sign for conductivity near room temperature does not occur, indicating a much smaller `metallic' resistance contribution. Nevertheless, the thermopower of polypyrrole, like that of PAni and PAni blends, is of metallic size and generally increases with temperature, as expected for metallic diffusion thermopower.

Keywords: polyaniline, polypyrrole, conductivity, reflection spectroscopy, scanning electron microscopy, thermopower

___________________________________________________________________________________________________

1. Introduction

The heterogeneous nature of conducting polymers revealed by their transport and other properties has been a recurring theme [1] since Park et al. [2] described transport in polyacetylene in terms of metallic strands separated by thin potential barriers. Such a picture accounts for the fact that the conductivity still retains a nonmetallic temperature dependence at lower temperatures, even when the thermopower shows remarkably linear metallic diffusion behaviour at all temperatures [2,3]. It also follows that the intrinsic conductivity is much higher than the observed macroscopic conductivity, reaching values [4] higher than that of good metals for the highest conductivity polyacetylene [5]. Thus there appears to be some mechanism in the polymers that can diminish even the intrinsic scattering of carriers by thermally-excited phonons that limits the conductivity of good metals.

The most plausible mechanism for this suppression of scattering by phonons in metallic regions is the quasi-one-dimensional nature of conduction along polymer chains [6,7]: carriers can be scattered only back along the chains, which requires phonons of relatively high energy which are not readily excited below room temperature. Hence the resistivity should rise rather sharply at temperatures corresponding to 2kF, where kF is the Fermi wavevector. One of us [4] has provided support for this mechanism by showing that the conductivity temperature dependence of a wide range of polyacetylene samples is consistent with a combination of quasi-one-dimensional metallic conduction and tunnelling/hopping conduction. In particular, the change in sign of conductivity temperature dependence often seen below room temperature is consistent with the increase of the backscattering resistivity term - the alternative explanation in terms of delocalization of carriers in a homogeneous picture appears to be ruled out by the linearity of thermopower, which would be expected to also show the effects of such a delocalization.

In this paper, we report conductivity data for polyaniline blends and polypyrrole, and analyze our data, and that for high-conductivity PAni-CSA from other authors, in terms of a similar heterogeneous picture. To help understand the unusual behaviour of the polyaniline dispersed in insulating polymers, we have made infra-red reflectance measurements and scanning electron microsope (SEM) images.

2. Polyaniline blends

PAni protonated with organic sulfonic acid (50 mol%) [8] was dispersed in PMMA, PVC and PETG copolyester [9]. Although PAni is the minority component of the dispersions, the increase of their IR reflectance at low energies (shown in Fig. 1) provides clear evidence of metallic behaviour, similar to that seen for unblended compressed PAni powder. While the PETG and PVC blends show a smaller increase than unblended PAni, the PMMA blends show a greater effect. This correlates with the greater dc conductivity at room temperature of 60% PMMA blends compared to unblended PAni, while the other blends show a smaller room temperature conductivity [10]. As might be expected, all our IR reflectances are smaller than that for high-conductivity PAni protonated with camphor sulfonic acid (CSA) [11]; the phonon features we observe between 1000 and 1700 cm^-1 are remarkably similar to those seen in PAni-CSA [11], but as expected are of larger magnitude for our lower conductivity samples.

Fig. 1. Reflectance (at at room temperature) of PAni blended with insulating polymers.

Fig. 2. Temperature dependence of normalized conductivity of PAni blends and unblended PAni; the lines show fits to Eq. (1).

Fig. 3. SEM images of fracture surfaces for PAni blends and unblended PAni (the bars represent 10 um).

The temperature dependence of conductivity [[sigma]] in Fig. 2 shows the characteristic difference [10] between PAni/PMMA and PAni/PVC on the one hand, and PAni/PETG and unblended PAni on the other; conductivity magnitudes for these particular samples range from about 10 S/cm (PMMA) to 2 S/cm (PETG). The difference in temperature dependence gives rise to much lower values of the tunnelling/hopping parameter T0 for the PMMA and PVC blends in the fits in Fig. 2 to the standard conductivity behaviour [10]

where [[rho]]m, [[rho]]0, Tm and T0 are constants. The values of T0 are 70 K and 200 K for the PMMA and PVC blends, compared to 700 K and 1200 K for the PETG blend and PAni. The second term is largest and corresponds to tunelling/hopping behaviour (for example between mesoscopic metallic islands [12]). The first term represents quasi-1D metallic conduction along polymer chains [7] as mentioned in the Introduction, with Tm corresponding to the energy of the 2kF wavevector (we take Tm ~ 1000 K, which gives reasonable fits, throughout this paper). All four fits have this metallic term contributing about 10% of the resistance at room temperature.

When we examine SEM images of PAni and the blends (Fig. 3), we notice a striking difference for the PMMA and PVC blends, namely a much smoother morphology than for PAni and the PETG blends, which show distinct granularity. The sparse granularity in the 85% PETG blend compared to the 60% PETG blend indicates that the granularity is associated with the PAni (no granularity is of course seen in pure PETG); close examination of the image of the 85% PETG blend indicates that the PAni could occur in seams, consistent with other studies [13]. The difference between the blends does not appear to be related to partial crystallinity in PETG, since we saw no evidence for any crystallinity using X-ray diffraction. It appears that blending in PMMA and PVC reduces the "barriersÓ aound PAni particles, resulting in enhanced conduction (especially at low temperatures), while blending in     

Fig. 4. Conductivity of PAni-CSA/PMMA blends from Yoon et al. [14] (not all data points are shown), fitted to Eq. (1). Also shown is the conductivity of PAni-CSA, fitted to Eq. (2).

PETG has far less effect.

Turning now to conductivity measurements by Yoon et al. [14] on high-conductivity PAni-CSA/PMMA blends (Fig. 4), we see that these blends have a somewhat similar shape to that of our PAni/PMMA blends, but with a larger peak that occurs at lower temperatures. As the fits show, this shape is also well described by our Eq. (1), but with smaller values of T0 (about 40 K) and somewhat larger metallic contributions to resistance at room temperature (about 15%).

Note that our simple picture describes the change from hopping/tunnelling to metallic temperature dependence of conductivity, but not the localization behaviour at low temperatures, which has been extensively investigated by Yoon et al. [14].

Unlike the conductivity, the thermopower of all our PAni blends [10,15] shows a striking similarity, increasing with temperature to values of 7 to 10 uv/K at room temperature, with a reduction in slope above 100 K. This thermopower also has a strong resemblance to the thermopower of high-conductivity of PAni-CSA blends [14], both as regards magnitude and temperature dependence.

3. Polyaniline

The conductivity for pure PAni [14] shown in Fig. 4 cannot be described by Eq. (1) since the conductivity shows metallic behaviour in that it remains large even as temperature tends to zero. In this case, as we found for highly-conducting polyacetylene [4], the temperature dependence above the localization regime at low temperature is well described by a quasi-1D metallic term in series with fluctuation-induced tunnelling between extended metallic regions [16]

where Tc and Ts are constants depending on the barrier parameters. This also applies to the data for highly-conducting PAni-CSA of Holland et al. [17] shown in Fig. 5,    

Fig. 5. Conductivity of PAni doped with CSA from Holland et al. [17], fitted to Eq. (2) (the percentage protonation of nitrogen sites is indicated beside each data set).

where the metallic contribution to room temperature resistance is again in the range 10% - 15%. These authors fitted their data to a model similar to that used by us earlier [4], but with a linear T term for metallic resistivity as for normal metals; this produces similar results for polyaniline but we found the faster exponential temperature variation was required for polyacetylene [4]. We note that lower conductivity PAni protonated by phosphoric acids [18] shows a flattening in conductivity near room temperature but not a peak, with conductivity tending to zero at zero temperature like the behaviour in Fig. 2 above.

4. Polypyrrole

In the case of polypyrrole, there appears to be a significant difference in the conductivity behaviour compared to polyaniline and polyacetylene, namely a less metallic temperature dependence. In fact, in the fits to our data in Fig. 6, the metallic term is absent, and in the high-conductivity polypyrrole data of Sato et al. [19] and Yoon et al. [20], it is small so that the conductivity shows no peak near room temperature. (There is also an intriguing upturn in conductivity sometimes seen at very low temperatures.)

In addition, the temperature dependence (with exponential best-fit exponent for the hopping term in Eq. (1) equal to 0.3) is closer to that for 2D variable range hopping rather than tunnelling between metallic islands or 1D hopping.

The thermopower of our polypyrrole samples, examples of which are shown in Fig. 7, is somewhat variable, but is small in magnitude and increases with temperature as for metallic diffusion thermopower.

We thank Alan MacDiarmid for helpful discussions, and the Foundation for Research, Science and Technology for support of the research in New Zealand.

Fig. 6. Conductivity of polypyrrole doped with p-toluene-sulfonate, one sample having a room temperature conductivity of about 80 S/cm, and the other (with a wrinkled morphology [21]) about 3 S/cm; the fits are to Eq. (1) with no metallic term and with the exponent of the exponential term equal to 0.3 rather than 0.5.

Fig. 7. Thermopower of our polypyrrole samples compared to that for PAni blends.