Segmentation of conducting domains in PEDOT:PSS films induced by an additive for conductivity enhancement

We investigate the relationship between the morphology and in-plane conductivity of poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS) films made from aqueous dispersions with/without ethylene glycol additive. Nanometer-scale current images of the films obtained using a conductive atomic-force microscope reveal that PEDOT-rich highly conducting domains are segmented into smaller ones — with the total area of these domains being nearly constant — for larger percentages of ethylene glycol leading to higher in-plane conductivities. The in-plane transport mechanism is found to have a strong dependence on the effective thickness of insulating barriers formed by excess PSS between neighboring highly conducting domains.

D oped conjugated polymers are promising materials that possess relatively high conductivities together with useful properties such as easy processability, optical transparency, and mechanical flexibility. 1,2) It has been reported that truly metallic charge transport is achieved in doped polyaniline films with in-plane conductivities of ∼1000 S=cm. 3,4) For a variety of applications, [5][6][7][8][9] poly(3,4-ethylenedioxythiophene) (PEDOT) doped with poly(styrene sulfonate) (PSS) is popular and has been intensively studied. Very interestingly, PEDOT:PSS is known to exhibit conductivity that increases with decreasing PSS composition, in contrast to other conjugated polymers with similar main chains, e.g., poly (3hexylthiophene), whose conductivity increases with increasing dopant density in the dc-to-THz frequency range. 10,11) This unique character of PEDOT:PSS has been ascribed to the formation of phase-separated grains; each grain consists of a PEDOT-rich core and a thin PSS-rich shell, where excess PSS works as an insulator rather than a dopant, as observed by photoelectron spectroscopy. [12][13][14] Practical methods for improving the conductivity of PEDOT:PSS films have been developed with the use of organic additives, such as dimethyl sulfoxide, sorbitol, and ethylene glycol, which can be added to PEDOT:PSS aqueous dispersions before casting or spincoating. [15][16][17][18] More details on the morphology of PEDOT:PSS films have been revealed using scanning probe microscopy and electron microscopy, [19][20][21][22][23][24][25][26] recently in combination with X-ray scattering measurements. 27,28) However, there is controversy about microscopic processes induced by additives; 24,26,28) it has not yet been established what types of morphological change provide an increase in conductivity for PEDOT:PSS films.
In this paper, we report nanometer-scale current images of PEDOT:PSS films obtained using a conductive atomicforce microscope (AFM), with an emphasis on the relationship between the morphology and in-plane conductivity. The PEDOT:PSS films were systematically made from aqueous dispersions with increasing percentages of ethylene glycol additive, which led to an increase in the in-plane conductivity by orders of magnitude. We found that highly conducting domains distributed in the current images are segmented into smaller oneswith the total area of these domains being nearly constantfor larger percentages of ethylene glycol. By estimating the effective thickness of insulating barriers between neighboring highly conducting domains in every current image, we explained the orders-of-magnitude difference in the in-plane conductivity among the films. In terms of sizes, the highly conducting domains and insulating barriers were assigned to the previously reported PEDOT-rich cores and PSS-rich shells of PEDOT:PSS grains, respectively.
The samples used in this study were PEDOT:PSS films made from aqueous dispersions that included no and three small volumes of ethylene glycol additive: 0, 1.0, 3.0, and 9.0%. The pristine PEDOT:PSS aqueous dispersion, which had PEDOT and PSS compositions of 0.5 and 0.8 wt %, respectively, and did not include ethylene glycol, was purchased from Sigma-Aldrich (product No. 483095). The addition of ethylene glycol led to an increase in the in-plane conductivity, which we evaluated by applying the four-probe method to cast films on glass slides. The obtained values of the in-plane conductivity are shown in Table I. The in-plane conductivity increased monotonically from 1.5 × 10 −1 to 1.2 × 10 2 S=cm at room temperature with increasing percentage of ethylene glycol.
To perform the conductive AFM measurements, we fabricated samples A-D on a single substrate of low-resistivity Si (7 × 2 mm 2 in size) by casting the aqueous dispersions into slightly shifted positions on the substrate. This allowed us to prepare uniform conditions for the four samples in terms of film formation processes and AFM measurements. The film thicknesses of samples A-D are listed in Table I. The AFM system used here was a JEOL SPM-4610A equipped with ultrahigh-vacuum chambers. Nanometer-scale current images and topographic images of each sample were simultaneously recorded under a vertical bias voltage of 0.3 V between the substrate and an AFM probe coated with Pt=Ir (NanoWorld CONTPt). During the measurements, samples A-D were kept in a high vacuum of ∼7 × 10 −8 Pa and scanned with the identical probe and a common contact force so that the vertical current values could be properly compared among the four samples. We obtained a spatial resolution of ∼2 nm in the current images.
Content from this work may be used under the terms of the Creative Commons Attribution 4.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
The current images of samples A-D with a scan area of 1000 × 1000 nm 2 are shown in Figs. 1(a)-1(d), respectively, where the gray scale corresponds to vertical currents ranging from zero to the maximum value observed for each sample. The maximum current increased monotonically from 131 to 891 nA with increasing percentage of ethylene glycol. In every current image, currents close to the maximum value are seen for only a few pixels. Currents somewhat below the maximum value are distributed in a much broader space; this behavior is relevant to the in-plane conductivity and is examined later. Furthermore, the current images have little correlation with the topographic images shown in Figs. 1(e)-1(h) for exactly the same scan areas, indicating that the vertical currents are rather insensitive to surface fluctuations. Figure 2 shows the current images recorded for the central areas (500 × 500 nm 2 ) of Figs. 1(a)-1(d), with the gray-scale range limited to twice the average vertical current observed for each sample. This provides sufficient brightness to visualize the spatial distribution of currents somewhat below the maximum value and allows us to define the highly conducting domains as the bright parts with currents higher than the average. As seen in Figs. 2(a)-2(d), the highly conducting domains are randomly distributed in each sample, and they become significantly finer with increasing percentage of ethylene glycol. Note that the latter feature has never been revealed by conductive AFM measurements before. 24,26) The highly conducting domains are separated from one another by extremely dark parts (see the inset), which we call insulating barriers hereafter, with currents of less than 5 nA. The remaining parts, with currents ranging from 5 nA to the average, mainly surround the highly conducting domains and are thus categorized as slightly conducting domains. Note that the average vertical current increased from 12.0 to 66.4 nA with increasing percentage of ethylene glycol, whereas the in-plane conductivity increased rapidly by orders of magnitude (see Table I).
We analyzed the number and size of the highly conducting domains by fitting approximate ellipses to the bright parts shown in Figs. 2(a)-2(d). The results of this analysis are summarized in Table I. With increasing percentage of ethylene glycol, the total number n con of highly conducting domains was found to increase from 354 (for sample A) to 617 (for sample C) and then decrease to 530 (for sample D), while they retained a nearly constant proportion, p con ∼ 30%, to the scan area of 500 × 500 nm 2 . This means that the addition of ethylene glycol induced the segmentation of  highly conducting domains into smaller ones. We furthermore evaluated the proportion p ins of insulating barriers to the scan area of 500 × 500 nm 2 for every sample. As shown in Table I, p ins changed from 39.5 to 18.1% with increasing percentage of ethylene glycol. The decrease in p ins is linked to an increase in the proportion 1 − p con − p ins of slightly conducting domains because p con is nearly constant (∼30%). More details of the highly conducting domains, that is, the histograms with respect to the major-axis length, are shown in Figs. 3(a)-3(d) for samples A-D, respectively. Here, the major-axis length can be regarded as a measure of domain size because there was a correlation between the minorand major-axis lengths, with the oblateness ranging from 0 to ∼0.7 (see Figs. S1-S3 in the online supplementary data at http://stacks.iop.org/APEX/9/051601/mmedia). The histogram for sample A has a rather wide distribution versus the major-axis length up to ∼50 nm; those for samples B-D have significantly narrower distributions with peaks shifted to smaller major-axis lengths of 10-15 nm. In particular, for sample B, as many as 270 highly conducting domains have major-axis lengths of 10-15 nm. The major-axis lengths obtained here are similar to the typical sizes of colloidal PEDOT:PSS grains in aqueous dispersions observed previously by Yan et al. through dynamic light scattering. 23) Thus, the highly conducting domains in Figs. 2(a)-2(d) can naturally be assigned to the PEDOT-rich cores of grains, whereas the insulating barriers are considered to be formed by excess PSS. Now, let us discuss how the spatial distribution of highly conducting domains yields in-plane conductivity and enhances it with increasing percentage of ethylene glycol. The segmentation of highly conducting domains described above occurs together with a reduction in the effective thickness l ins of the insulating barriers between neighboring highly conducting domains. In the most simplified situation, where highly conducting domains surrounded by slightly conducting domains have a total number n con and are separated by insulating barriers of proportion p ins in a scan area of L 2 , the effective barrier thickness l ins can be expressed as Lð1 À ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 À p ins p Þ= ffiffiffiffiffiffiffiffi n con p . A schematic illustration for deriving this expression is shown in Fig. S4 in the online supplementary data at http://stacks.iop.org/APEX/9/051601/ mmedia. The expression gives a crude estimate of l ins for the actual samples, which changes from 5.9 to 1.9 nm with increasing percentage of ethylene glycol, as shown in Table I. The value l ins = 5.9 nm before the addition of ethylene glycol is consistent with the thicknesses of the PSS shells (5-10 nm) observed previously by Lang et al. using a high-angle annular dark-field scanning transmission electron microscope. 25) To reproduce the orders-of-magnitude difference in the inplane conductivity among samples A-D, we need to consider a transport mechanism that has strong dependence on l ins . An in-plane mobility proportional to exp(−2d=d 0 ), where d is the gap between neighboring conducting domains and d 0 is the damping constant of the carrier wave functions, can be expected for hopping transport through thin barriers. 29,30) Here, we simulated the in-plane conductivity versus l ins assuming that d = l ins , d 0 = 1.23 nm, and the carrier density is constant; d 0 was treated as an adjustable parameter for reproducing the ratios of the in-plane conductivities and was then set to that value. The simulated in-plane conductivities for l ins = 4.0 and 1.9 nm were 24 times and 6.6 × 10 2 times higher, respectively, than that for l ins = 5.9 nm. These values were close to the experimental results; the measured in-plane conductivities for samples B and C were indeed 23 times and 6.6 × 10 2 times higher, respectively, than that for sample A (see Table I). The simplest simulation described above underestimates the in-plane conductivity of sample D by a factor of 0.59, suggesting that the actual carrier density or intradomain mobility was significantly higher in sample D than in sample A, perhaps owing to the conformational change in PEDOT chains. 17)  In summary, we performed conductive AFM measurements of PEDOT:PSS films made from aqueous dispersions that included no and three small volumes of ethylene glycol additive. Highly conducting domains distributed in nanometer-scale current images were found to be segmented into smaller oneswith the total area of these domains being nearly constantfor larger percentages of ethylene glycol leading to higher in-plane conductivities. By estimating the effective thickness of the insulating barriers between neighboring highly conducting domains in every current image, we were able to reproduce the observed orders-ofmagnitude difference in the in-plane conductivity among the films. The sizes of the highly conducting domains and insulating barriers observed here were consistent with those reported previously for the PEDOT-rich cores and PSS-rich shells of PEDOT:PSS grains, respectively. Thus, we have provided direct evidence of microscopic morphological changes that substantially enhance charge transport in PEDOT:PSS films.