High Thermoelectric Performance in Solution‐Processed Semicrystalline PEDOT:PSS Films by Strong Acid–Base Treatment: Limitations and Potential

Abstract Thermoelectric (TE) generation with solution‐processable conducting polymers offers substantial potential in low‐temperature energy harvesting based on high tunability in materials, processes, and form‐factors. However, manipulating the TE and charge transport properties accompanies structural and energetic disorders, restricting the enhancement of thermoelectric power factor (PF). Here, solution‐based strong acid–base treatment techniques are introduced to modulate the doping level of poly(3,4‐ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) thin films with preserving its molecular orientation, enabling to achieve a remarkably high PF of 534.5 µW m−1 K−2. Interestingly, theoretical modeling suggested that further de‐doping can increase the PF beyond the experimental value. However, it is impossible to reach this value experimentally, even without any degradation of PEDOT crystallinity. Uncovering the underlying reason for the limitation, an analysis of the relationship among the microstructure–thermoelectric performance–charge transport property revealed that inter‐domain connectivity via tie‐chains and the resultant percolation for transport are crucial factors in achieving high TE performance, as in charge transport. It is believed that the methods and fundamental understandings in this work would contribute to the exploitation of conducting polymer‐based low‐temperature energy harvesting.


Section S2. Kang-Snyder model
Kang-Snyder modelling.σ and α can be generally characterized by the transport function σE(E) describing the capability for electrical conduction at each energy level in the units of σ. [1,2] By solving the Boltzmann transport equations, the α and σ are expressed with σE(E) as the following equations: Here, f is the Fermi-Dirac distribution function, f = 1/[1+exp{(E−EF)/(kBT)}], where q, kB and EF are a unit charge, Boltzmann constant, and chemical potential of the polymer, respectively.
The Kang-Snyder model suggests a novel form of transport function, [1] which has power law energy dependency with energy E above the transport edge Et, To fit our data using SLoT model, we employed the "SLoTModel.xlsx"file provided by the authors. [3]The modelling parameters (e.g., cmax, A0, A1, WH,max, WH,slope) and material parameters (e.g., M0, N, ρ, kBT) are detailed in Table S2.respectively. [4,5]In this model, thermally activated voltage fluctuations across insulating S-6 regions are critical for electrical conduction.Parameters derived using the FIT model are summarized in Table S3.
We also investigated the σ(T) behavior of our PEDOT:PSS films using the variable range hopping (VRH) model, σ = σ0exp(Ta/T) −1/m , which is generally adopted to explain the thermally-activated type of σ(T) of the conjugated polymers, [6] where σ 0 is the σ at the infinite T, and Ta is the characteristic temperature.Here, m is the crucial factor that determines the charge transport dimensionality.The m-values were determined from Zabrodskii plots (W = ∆lnσ/∆lnT vs. T) where the slope is equal to the value of −1/m, as illustrated in Figure S7.
Interestingly, we found that none of the m-values from our samples conformed to the standard VRH dimensions of m = 2, 3, or 4.This observation suggests that the thermally activated behaviors in our PEDOT:PSS films cannot be explained by the localized state hopping mechanisms.
Calculation of the electrical parameters extracted by the Hall effect measurements.The Hall coefficient RH, Hall mobility μH, and Hall carrier concentration nH were obtained using the following equations: The actual Hall voltage VH was corrected by subtracting both the field-independent and fielddependent offset voltages arising from the positive magnetoconductance from the measured ∆Vxy, as reported previously. [7]The extracted VH for TDAE-treated films is presented in sulfonic acid units in PSS, respectively. [8]Following the TFSA treatment, the relative peak intensity of PEDOT to PSS increased noticeably, indicating the removal of PSS.
(, ) = { 0 ( <   ) where σE0(T) and s are the transport coefficient and transport parameter, respectively.By replacing the transport function σE(E) (Equation S1 and S2) with Equation S3, the simplified form of σ and α can be obtained:  =  0 () ×  −1 () Fi(η) is the non-normalized complete Fermi-Dirac integral, and η = (EF − Et)/kBT is the reduced chemical potential.The above equations (Equation S4, S5) enable us to determine σE0, s and η from experimentally obtained σ and α. et al. presented the semi-localized transport model by modifying the transport function of the K-S model. [3]The SLoT model well captures the transition in charge transport behavior in conjugated polymers as a function of carrier concentration ratio c.The transport function of SLoT model is expressed as:   (, , ) In this equation, WH represents the localization energy, which correlates with both spatial and electrostatic effects.This function differs from the K-S model by introducing an additional term σ0exp(−WH(c)/(kBT)), representing an Arrhenius-like hopping contribution from the charge carriers.The SLoT model also sets the transport parameter s to 1.As the carrier concentration ratio c exceeds cd (the carrier ratio to achieve delocalized transport), WH converges to zero, making the transport function of the SLoT model equivalent to the s = 1 case in the K-S model.Conversely, when c falls below cd, WH increases, causing the transport function to deviate from the s = 1 fit observed in the K-S model.

5
Here, cmax is the maximum c value needed for the data set, A0 and A1 are fitting parameters, WH,max is the maximum localization energy, and WH,slope represents the rate at which localization diminishes as c increases.M0denotes the molecular weight of the monomer unit, and N is the number of sites per monomer ratio, and ρ is the density of the polymer.S-Section S4.Experimental details for measuring temperature-dependent electrical conductivity and magnetotransport propertiesDevice fabrication.Temperature-dependent electrical conductivity and magnetotransport properties were measured using Hall bar-structured devices, where PEDOT:PSS layers were precisely patterned to enable accurate measurement of local potentials.The fabrication process for the Hall bars included several key steps.Initially, a Si/SiO2 substrate was cleaned through a sonication processes in deionized water, followed by additional rinsing in acetone and isopropanol.Electrodes consisting of titanium (Ti, 7 nm) and gold (Au, 70 nm) were deposited onto the Si/SiO2 substrate through e-gun evaporation, followed by patterning using standard lift-off lithography techniques.The dimensions for the channel length (L) and width (W) were 240 μm and 60 μm, respectively.Four probes were positioned between source and drain electrodes, with the distance between two longitudinal probes along the channel length (defined as L * ) designed to be 110 μm.The width of the probe was 15 μm.After electrodes formation, the PEDOT:PSS layer was spin-coated onto the Si/SiO2 substrate, and sequential acid-base treatments were applied as described in the manuscript.Subsequently, a parylene-C layer (1 μm) was deposited onto the PEDOT:PSS films using a parylene coater (PDS2010), to protect the active layer during subsequent processing.The active layer was then patterned into precise Hall bar geometry through photolithography and oxygen plasma etching at 150 W for 5 min.Both photoresist and parylene-C were remained as a protective layer for the active channel.All fabrication steps were conducted in a cleanroom environment with controlled humidity and temperature.For the magnetotransport measurements using a PPMS system, a ball bonder was utilized for wiring the hall bar device to the PPMS puck.Temperature-dependent electrical conductivity of the PEDOT:PSS-TFSA films.The temperature-dependent electrical conductivity σ(T) was evaluated from the same hall bar geometry using the formula σ = IL * /(ΔVxxWt), where t represents the thickness of the active layer.For all PEDOT:PSS films, a negative T dependence of σ (dσ/dT < 0) was observed, indicating the metallic behaviors above the critical temperature.Below this critical temperature, the thermally activated type of σ (dσ/dT > 0) is observed.The σ(T) of our samples at low temperature aligns well with the fluctuation-induced tunneling (FIT) model: σ = σ0exp{−T1/(T + T0)}, where σ0 is the σ at infinite T, and T0 is the characteristic temperature,

Figure S6 .
Figure S6.Change in localization energy WH as a function of TDAE treatment time.The values of WH for the samples were extracted by the SLoT model.When the WH converges to zero, the transport function of SLoT model becomes equivalent to the s = 1 case in the K-S model.After the de-doping time exceeds the 7 min, the WH undergoes a sudden increase, supporting the deviation of the s = 1 fit of our experimental data.

Figure S7 . 14 Figure S8 . 15 Figure S9 . 16 Figure S10 .
Figure S7.Zabrodskii plot of the TDAE-treated PEDOT:PSS-TFSA films.The slope of the fit line equals to the value of −1/m, which gives the information of the dimension m of the VRH mechanism.

Figure S11 .
Figure S11.Electrical parameters of the TDAE-treated PEDOT:PSS-TFSA films extracted from Hall effect measurement.a) Temperature-dependent electrical mobility μH and b) carrier concentration n.

Table S1 .
Information on the diffraction peak parameters derived from GIWAXS data.The values of crystal coherence length were calculated using the Scherrer formula.