Philipp
Stadler
*a,
Dominik
Farka
a,
Halime
Coskun
a,
Eric D.
Głowacki
a,
Cigdem
Yumusak
a,
Lisa M.
Uiberlacker
b,
Sabine
Hild
b,
Lucia N.
Leonat
c,
Markus C.
Scharber
a,
Petr
Klapetek
d,
Reghu
Menon
e and
N. Serdar
Sariciftci
a
aInstitute of Physical Chemistry, Johannes Kepler University Linz, Altenbergerstr. 69, 4040 Linz, Austria. E-mail: philipp.stadler@jku.at; Fax: +43 732 2468 1213; Tel: +43 732 2468 8770
bInstitute of Polymer Science, Johannes Kepler University Linz, Altenbergerstr. 69, 4040 Linz, Austria
cThe National Institute for Electrical Engineering, ICPE-CA, Splaiul Unirii 313, 030138 Bucharest, Romania
dCentral European Institute of Technology, Koliste 13a, 602 00 Brno, Czech Republic
eDepartment of Physics, Indian Institute of Science, Bangalore 560012, India
First published on 23rd June 2016
Weak localization describes a metallic system, where due to the presence of disorder the electrical transport is governed by inelastic electron relaxation. Therefore the theory defines a threshold of spatial and of energetic disorder, at which a metal–insulator transition takes place. To achieve a metallic state in an inherently disordered system such as a conductive polymer, one has to overcome the threshold of localization. In this work we show that the effective suppression of disorder is possible in solution-processible poly(3,4-ethylenedioxythiophene)–poly(styrene sulfonate). We grow polymer films under optimized conditions allowing self-organization in solution. Interestingly, we find the requisite threshold, at which the system becomes finally metallic. We characterize the transition using a complementary morphology and magneto-electrical transport study and find coherent electron interactions, which emerge as soon as local order exceeds the macromolecular dimensions. These insights can be used for discrete improvement in the electrical performance, in particular for tailoring conductive polymers to alternative metal-like conductors.
We start with a number of thin-film PEDOT:PSS recipes yielding 10 nm thickness and successively increase the width to access a more and more bulk phase up to 10 mm. In addition we modify the co-solvent present between 0 and 10% of the original volume dispersion. Our goal is to attain a detailed insight of the co-solvent concentration and of the final film thickness into the resulting conductivities (Fig. 1A). We consistently find excellent electrical performances at DMSO concentrations around 5%. Here we achieve peak conductivities of 1000 S cm−1 at a thickness of 55 nm as seen in the map in Fig. 1A denoted further as (1). The effect is less pronounced at lower and higher DMSO concentrations, which is in agreement with previous studies.22,23
When the film thickness expands to a bulk system greater than 1 μm we see a similar increase of the electrical performance on the DMSO concentration. Nonetheless, the bulk-films achieve a maximum of 490 S cm−1 at 5% DMSO and 8.4 μm respectively (2). So the map in Fig. 1A exhibits a consistent out-performance of thin-films over bulk films. Other detailed studied systems are 10% DMSO and 8.4 μm (3) and 2% DMSO and 8.4 μm (4).
When we repeat the mapping at 1.8 K we obtain a first insight, where in the map the local order allows the formation of a metallic state (Fig. 1B). According to Mott's minimum conductivity, regions off the metallic regime exhibit an infinite low conductivity, which is valid for the preponderant dark regions seen in the 1.8 K map. Regimes in proximity to a metal–insulator transition (MIT) appear still conductive – they are found in the same hotspots as at room temperature with an important changeover: at 1.8 K, the bulk systems outperform the thin-films. The reason for this lies in the pronounced temperature-dependence of the thin film conductivities. Meanwhile, the bulk film conductivities exhibit a moderate temperature-profile in agreement with weak localization with a flattening as the temperature approaches 0 (Fig. 1C).
In combination, the maps show us the optimum region, which lies between 2 and 5% DMSO and 8.4 μm layer thickness (2) and (4). At 5% DMSO and 1.8 K, for example, the bulk-conductivity (2) exceeds the 55 nm thin-film (1) conductivity by a factor of nearly 2. This minor temperature-sensitivity points out that the thickness plays a role in terms of local order – it relates to the different parameters in the processing and to a suppressed substrate-effect in the bulk-systems. To support our findings from the conductivity maps we view the morphologies by AFM and XRD, in particular in the optimum region (Fig. 2).
For the bulk films we plot the XRD spectra at no (ref), 2% (4) and 5% DMSO (2). A sharpened profile in terms of crystallinity correlates with the presence of DMSO during growth.25,26 (Fig. 2E). In general the increase of the electrical conductivity from Fig. 1A is reflected in a pronounced signal, which underlines the higher degree of crystallinity. The discrete peak originating from the π-stacking of PEDOT emerges at 25 deg. It is overlapped by the broad response of the PSS halo with a maximum at 20 deg. The spectral shaping through DMSO consequently supports our argument of improved crystallization in the presence of DMSO valid for both components, PEDOT and PSS, respectively. To clarify, we present a comparison of PEDOT and regio-regular poly(-3-hexylthiophene) (rr-P3HT) grown under similar conditions in the ESI.† A supplementary view of the morphological changes induced by DMSO is outlined by atomic force microscopy (AFM). As shown in the phase images in Fig. 2A–D we find different grain sizes at each representative (conductivity) area (marked as a white rectangular frame) from the conductivity map (Fig. 2E). Fig. 2D depicts the thin-film phase image revealing a randomized distribution of structured domains with different grain sizes. This is reminiscent to the bulk-film without DMSO, where polydispersity is detected again. Obviously, the presence of small amounts of DMSO (2%) and thicker films (4) improves the homogeneity a lot as shown in Fig. 2B, where the grains of the structure become smaller. To achieve a growing grain size and a homogeneous dispersion the concentration has to be further increased to 5% (2), where we find the point of merit in terms of conductivity, crystallinity and morphology in the bulk-films. We denote that we derive the bulk-film morphology from cross-sectional scans in order to allow a meaningful comparison between bulk and thin-film morphologies.
The morphology response of the different conductivity-stages reveals the importance of long-range order plus the role of dispersivity seen in growing ordered domains of similar size, which explains the improved electrical performance at low-T.
Back in Fig. 1 we outline the positive temperature coefficient with finite conductivity as T approaches 0 best seen in (4), (1) and (2). This behaviour corresponds to weak localization. To resolve the low-T regime, we use the derivative of the conductivity in the logarithmic scale
further denoted as W.27
The metallic character is expressed by the slope and magnitude of W(T). As seen in Fig. 3 the system without DMSO (ref) shows a constant W with quantities greater than 1 (insulating regime). In contrast, (4), (1) and (2) deviate − W(T) is not constant. In particular, at low temperature it illustrates that on which side of the MIT the system is situated. In the case of bulk 5% DMSO (2), the slope of W(T) becomes positive indicating a metallic state (Fig. 3B). We denote that in particular the thin-film response from (1) shows a clear non-metallic fingerprint at low-T, which we assign to the role of dispersivity.
The proximity of the metallic state can be tested by applying a high magnetic field perpendicular to the substrate plane (Bz = 9 T). The field induces localization and quenches the metallic state as shown in Fig. 3. A substantial change in W(T) by B is therefore seen below 10 K. There W splits up into a critical-metallic (B = 0 T) and non-metallic regime (B = 9 T). This is true for both (4) and (2) (only for the bulk systems), where a shallow metallic behaviour is observed at T below 10 K.
So far, we have created a system with sufficient local order. We prove the metallic behaviour in the bulk samples between 1.8 K and 10 K seen by W(T) in Fig. 3B. In addition we demonstrate that a high magnetic fields can disrupt the local order of a metallic state, in particular at low temperatures.
When we take the view to the lower B-field, interestingly we find a concomitantly stimulating effect on the metallic character in our samples of merit (2 and 4). In contrast to Fig. 3 low B-fields enhance the conductivity. A detailed view of the magnetoconductivity as a function of B shows this positive contribution (Fig. 4).
We observe the conductivity increasing up to 4% of the original value. It is followed by a significant decrease to the negative as the field is increased. The negative effect is damped at high temperatures, where the turn to the negative is indicated at high B and not seen at all at temperatures above 6 K. One explanation for magnetic stimulation relates to a electro-magnetic resonance, earlier found in systems defined by weak localization: In fact, the stimulation reflects an interaction among the electron wave functions, as soon as the inelastic mean free path of electrons λε and the Landau orbit size LD.28–30 become comparable.
![]() | (1) |
Eqn (1) depicts the relation of LD (or B) and of λε (Fig. 4). At 1.8 K, for example, the ΔσB-value increases up to 1 T (2), before it collapses to the negative by increasing B (Fig. 4C and D). The peak maximum of ΔσB corresponds to the amplitude of the local order or λε, respectively. It can be consequently read out by plotting LD instead of B. A similar but right-shifted profile is observed at 3.85 K. At higher temperatures the stimulations do not collapse meaning that λε exceeds the maximum field at Bmax = 9 T or LD = 8.1 nm according to eqn (1), as λε decreases below the maximum magnetic field as we increase the temperature. Positive magnetoconductivity has been reported solely in combination with metallic samples – it reveals a coherence among electron wave functions and B.28,30–37 In our case this is supported by the morphology study (homogeneous domains of similar size, Fig. 3). Other theories assign the stimulating magneto-effect to the spin–orbit coupling. However, in organic matter these effects can be neglected.11 We use the coherent electron interactions as a direct consequence of long-range order inside the polymer. To quantitatively evaluate the magnetoconductivity, we derive concrete values for λε statistical parameters to visualize the local order. Kawabata has posted a relation between magnetoconductivity and the relaxation time (hence the mean free path λε).28 We can derive the magnitude by plotting + σBversus B2 according to
![]() | (2) |
Label | Thickness/nm | % DMSO | σ 300K/S cm−1 | σ 1.8K/S cm−1 | Ratio σ300K/σ1.8K | λ ε/nm (at 1.8 K) |
---|---|---|---|---|---|---|
(ref) | 8400 | 0 | 7.1 | N/A | N/A | N/A |
(1) | 55 | 5 | 1000 | 101.5 | 9.83 | N/A |
(2) | 8300 | 5 | 490 | 170 | 2.54 | 35.5 |
(3) | 9200 | 10 | 275 | 80 | 3.42 | N/A |
(4) | 8300 | 2 | 155 | 50 | 2.96 | 21.8 |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6tc02129h |
This journal is © The Royal Society of Chemistry 2016 |