Ionic Liquid Designed for PEDOT:PSS Conductivity Enhancement

abundant elements to exhibit a high level of flexibility, electrical conductivity, and transparency in the visible ... as-prepared solution-cast PEDOT...
0 downloads 5 Views 3MB Size
Subscriber access provided by UNIV OF DURHAM

Ionic Liquid Designed for PEDOT:PSS Conductivity Enhancement Ambroise de Izarra, Seongjin Park, Jinhee Lee, Yves Lansac, and Yun Hee Jang J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.7b10306 • Publication Date (Web): 10 Apr 2018 Downloaded from http://pubs.acs.org on April 10, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Ionic Liquid Designed for PEDOT:PSS Conductivity Enhancement Ambroise de Izarra,a,b,1 Seongjin Park,a,1 Jinhee Lee,a Yves Lansac,b,c,* Yun Hee Janga,* a

Department of Energy Science and Engineering, DGIST, Daegu 42988, Korea GREMAN, UMR 7347, CNRS, Université François Rabelais, 37200 Tours, France c Laboratoire de Physique des Solides, CNRS, Université Paris-Sud, 91405 Orsay, France b

Supporting Information Placeholder ABSTRACT: Poly-3,4-ethylenedioxythiophene:polystyrenesulfonate (PEDOT:PSS) is a water-processable conducting polymer promising for the use in transparent flexible electrodes and thermoelectric devices, but its conductivity is not satisfactory. Its low conductivity is attributed to the formation of hydrophilic/insulating PSS outer layers encapsulating the conducting/hydrophobic p-doped PEDOT cores. Recently a significant conductivity enhancement has been achieved by adding ionic liquid (IL). It is believed that ion exchange between PEDOT:PSS and IL components helps PEDOT to decouple from PSS and to grow into large-scale conducting domains, but the exact mechanism is still under debate. Here we show through free energy calculations using density functional theory on a minimal model that the most efficient IL pairs are the least tightly bound ones with the lowest binding energies, which would lead to the most efficient ion exchange with PEDOT:PSS. This spontaneous ion exchange followed by the nanophase segregation between PEDOT and PSS with formation of a -stacked PEDOT aggregate decorated by IL anions is also supported by molecular dynamics performed on larger PEDOT:PSS models in solution. We also show that the most efficient IL anions would sustain the highest amount of charge carriers uniformly distributed along the PEDOT backbone to further enhance the conductivity providing that they remain in the PEDOT domain after the ion exchange. Hence our design principles is that the high-performance IL should induce not only an efficient ion exchange with PEDOT:PSS to improve the PEDOT morphology (to increase mobility) but also a uniform high-level p-doping of PEDOT (to enhance intrinsic conductivity). Based on these principles, a promising (electron-withdrawing, but bulky, soft, and hydrophobic) new IL pair is proposed. KEYWORDS: transparent flexible electrode, transparent conducting polymer, thermoelectric polymer, PEDOT:PSS, conductivity enhancement, ionic liquid, density functional theory, non-equilibrium Green’s function formalism, molecular dynamics simulation 1. Introduction. Transparent electrodes are an essential component of optoelectronic and photovoltaic devices. Since the most popular transparent electrode materials at this moment, indium tin oxides (ITO), have serious drawbacks (especially for commercialization of flexible, wearable, or printable devices) such as scarcity of indium, mechanical brittleness, and high-temperature fabrication.1,2 It would be desirable to develop an ITO alternative, which can be synthesized and fabricated at low temperature and at low cost from abundant elements to exhibit a high level of flexibility, electrical conductivity, and transparency in the visible range.1,2 Therefore, a transparent, water-processable, and thermally-stable conducting polymer, PEDOT:PSS (Figure 1),1-6 which is a composite of positively-charged p-doped PEDOT (poly-3,4-ethylenedioxythiophene) and negatively-charged PSS (poly-4-styrenesulfonate), has received a great deal of attention.7-13

Figure 1. (Left) PEDOT:PSS polymers treated with EMIM:X (X = Cl, ES, TCM, and. TCB) ionic liquids and (right) their morphology (schematic),14 mobility, and conductivity (measured).15 Shown together is the hypothetical HCCP anion. EMIM:HCCP is expected to be as efficient as EMIM:TCB in the morphology control and the mobility/conductivity enhancement of PEDOT:PSS. Adapted from Ref. 14 (ACS Applied Materials and Interfaces, 2017, 9, 819-826, Copyright 2017 American Chemical Society) and Ref. 15 (Advanced Materials, 2016, 28, 8625-8631, Copyright 2016 Wiley).

PEDOT:PSS has also been used as hole transport (injection or collection) layer between organic active layer and electrode in organic light-emitting diodes and photovoltaic cells12 and as promising materials for various applications such as thermoelectric devices,16,17 stretchable and biocompatible sensors to realize artificial electronic skins,18 and low-energy-consuming non-volatile electrochemical devices to mimic neural synapses.19,20 However, untreated as-prepared solution-cast PEDOT:PSS films show poor electrical transport properties (< 10 S cm–1 of conductivity dc and < 10–3 cm2 V–1 s–1 of mobility ),2,4,9,21,22 most likely because the excessive amount of insulating and hydrophilic PSS chains encapsulate the conducting and hydrophobic PEDOT cores and inhibit the formation of well-aligned PEDOT conducting networks (Figure 1).10,21,23 This is far lower than 104 S/cm of ITO2,6 and also than 103 (up to ~8000) S/cm of highly-ordered PEDOT:Cl and PEDOT:Tos (Tos = tosylate = p-toluenesulfonate = PTS ~ monomer unit of PSS)24,25 films produced by in-situ vapor-phase polymerization.26-28 A PEDOT:PTS film spin-casted from polar organic solvents has also been determined to be crystalline29 (contrary to amorphous PEDOT:PSS)30 and its structure, properties, and electronic structure have been studied extensively.24,25,31-35 Contrary to all-connected PSS encapsulating PEDOT in excess amount, small PTS anions would easily find positions around PEDOT chains to neutralize them with the minimum amount without intervening their crystalline assembly.24 Therefore, a great deal of effort has been made to achieve crystalline PEDOT by separating PSS from PEDOT:PSS with polar solvents,2,6,7,10,16,22,24 surfactants,8,36 acids,9,21 and inorganic/organic salts11,17,37 added to PEDOT:PSS solutions. Such dipoles and small ions would screen the interaction between PEDOT+ and PSS–, separate PSS– from PEDOT+, induce a coil-to-extended conformational change of PEDOT, and achieve a conducting network of extended and ordered PEDOT

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

phase.10,21,22,36 Indeed, the conductivity of PEDOT:PSS films has been improved by 2-3 orders of magnitude up to 103 S/cm. Along this line, a significant (up to 5000-fold) conductivity enhancement has been achieved by vigorously mixing PEDOT:PSS aqueous solutions with ionic liquids (ILs) and sonicating them in deionized water, acetone, and isopropyl alcohol (Figure 1).15 The most outstanding 5000-fold enhancement of conductivity (from 0.4 to 2103 S/cm) was achieved by PEDOT:PSS treated with ethyl-3methylimidazolium (EMIM) tetracyanoborate (TCB) IL (Figure 1), similarly to the improvement (from 0.68 to 2084 S/cm) achieved in a previous study with the same system.38 More modest 3500-, 2000-, and 900-fold enhancements (to 1405, 840, and 359 S/cm) were also achieved by PEDOT:PSS treated with EMIM:X ILs where X is tricyanomethanide (TCM), ethylene sulfate (ES), and chloride (Cl) anions, respectively (Figure 1).15 The proposed mechanism is again that ionic exchange between PEDOT+:PSS– and EMIM+:X– would separate PEDOT+ from PSS– to form a network of ordered PEDOT+:X–phases, which increases the mobility . This hypothesis is supported by x-ray photoelectron spectroscopy (which shows the PEDOT-PSS interaction monotonically decreasing along the series, TCB > TCM > ES > Cl), grazingincidence wide-angle x-ray scattering (which shows the -stacking distance monotonically decreasing along the series), electron spin resonance (which shows monotonically increasing inter-chain coupling), and transmission electron microscopy (which shows monotonically increasing amounts of nanofiber growth).15 However, the mechanism is not completely clear. There has been a claim that a high level of inter-chain ordering is not necessary for efficient charge transport of PEDOT because disordered PEDOT is as conductive as ordered one as long as a sufficient number of charge carriers are present.39,40 This claim implies that electron transport along the backbone of p-doped PEDOT, which would be less sensitive to inter-chain ordering, is as important as (or more important than) electron transport across -stacked backbones.29 Density functional theory (DFT) calculations on pristine and doped PEDOT crystals have also shown that the intra-chain transport is significantly faster than the inter-chain transport.25,31,33,34 In this case, another important factor to increase the conductivity would be the p-doping density of PEDOT, which would in turn depend on the power and the concentration of dopants as well as the interaction between the p-doped PEDOT and the anions staying near it to keep the charge neutrality. In fact, a similar behavior was also observed for our IL-treated PEDOT:PSS cases:15 the IL-dependence of the series of conductivity (; left graph at the bottom of Figure 1) does not show exactly the same pattern as the IL-dependence of the series of mobility (; directly related to the degree of ordering; right graph at the bottom of Figure 1). It appears that the conductivity of EMIM:TCB-treated PEDOT (red squares in the two graphs in Figure 1) is further improved by additional factors beyond the ordering, and that additional factor is most likely the number of charge carriers (i.e., the degree of p-doping) in PEDOT. Indeed, a non-monotonic variation of charge carrier density (n) as a function of IL anions X (1.70, 1.50, 1.88, and 2.54 for Cl, ES, TCM, and TCB, respectively; 1021) is observed from reflectance spectroscopy,15 while n is known to be independent of the content of PSS.41 Since a similar order-of-magnitude change in conductivity has been observed for pristine PEDOT coupled with various dopant anions,27 our interpretation of such findings is that a significant amount of X anions may remain in the PEDOT phase (as PEDOT:X as proposed in previous studies)38,42 even after the rigorous washing and induce additional pdoping of PEDOT.43 In fact, it is rather unlikely that positivecharged PEDOT fragments would stay together without anionic glues between them. The X anion can play the role of condensing

Page 2 of 9

agent (like positively-charged condensing agents of negativelycharged DNA)44 which holds the p-doped PEDOT chains together, and the sustainability of p-doping of such PEDOT:X phase (i.e., PEDOT-to-X electron transfer or vice versa) would play a significant role in altering the conductivity (    n) of EMIM:Xtreated PEDOT:PSS. Good X anions in EMIM:X IL would play the multiple role of (1) disrupting the PEDOT-PSS attraction and bringing PEDOT together (to increase ) and (2) increase (or sustain) the degree of p-doping in PEDOT (to increase n). To confirm such roles of IL at the molecular level and to develop guidelines for designing better IL, we herein define key properties relevant to these two aspects of conductivity-enhancing mechanism and calculate them on model compounds of IL-treated PEDOT:PSS. As a key parameter describing the spontaneity of the IL-triggered PEDOT-PSS phase separation, we calculate in Section 2.1 the free energy of ion exchange ∆∆GX between PEDOT:PSS and EMIM:X. A trimer of EDOT (tri-EDOT) is chosen as a minimal model of PEDOT according to a report24,45 that one unit of positive charge is carried over at least three EDOT units of p-doped PEDOT, while a PTS unit is used as a minimal model of PSS (Figure 2).

Figure 2. (a-b) PEDOT:PSS polymer (modeled by tri-EDOT:PTS), EMIM:X IL (X = PTS, Cl, ES, TCM, TCB, and newly-designed HCCP), and ion exchange between them; (c-d) optimized geometries and ion-pair binding free energies (black lines) of EMIM:X and PEDOT:X complex as well as free energies of ion exchanges between them (red lines) which are calculated using Eq. (1’) and the ion-pair binding free energies of PEDOT:PSS (235 kJ/mol) and EMIM:PSS (317 kJ/mol). Color code: H (black), B (green), C (grey), N (blue), O (red), S (yellow), and Cl (dark green)

In Section 2.2 we confirm that the ∆∆GX parameters obtained with this minimal model correlate with the efficient EMIM:X-triggered PEDOT-PSS nanophase segregation simulated in their mixed aqueous solutions. As the second key parameter describing the charge carrier density (i.e., degree of p-doping) and the charge transport property of the backbone of IL-treated PEDOT:PSS. i.e., PEDOT

2

ACS Paragon Plus Environment

Page 3 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

coupled with X after the ion exchange, we calculate the positive charge carried by PEDOT (Section 2.3) and the current-voltage (IV) curve (Section 2.4) in a periodic single-chain PEDOT:X model with six EDOT units per unit cell, which is required to describe bipolaronic states.26 We then propose a new IL which would improve the PEDOT:PSS conductivity as effectively as EMIM:TCB. 2. Results and Discussion 2.1. Ion exchange free energy: DFT. The free energy of ion exchange ∆∆GX0 between PEDOT:PSS (modeled by tri-EDOT:PTS) and EMIM:X (Figure 2) is calculated from the standard Gibbs free energy (∆G0) of each set of ion pairs before and after the exchange: GX0  G 0 ( PEDOT : X)  G 0 ( EMIM : PSS)

. (1)  G 0 (PEDOT : PSS)  G 0 (EMIM : X) A more negative value of ∆∆GX0 indicates more spontaneous ion exchange by IL. The standard Gibbs free energy ∆G0 of each ion pair (PEDOT:X and EMIM:X, where X is PSS, Cl, ES, TCM, and TCB) in the aqueous phase at 298 K is given by the sum of the total energy at 0 K (E0), the zero-point energy (ZPE) and the Gibbs free energy change from 0 to 298 K (G0298K): G 0  E0  ( ZPE)  G0298K . (2) We obtain E0 from a full geometry optimization performed at the B3LYP46-48/6-31++G(d,p) level of DFT with Jaguar v8.5 (Schrödinger, LLC, New York, NY, 2014)49 starting from the initial geometries selected out of 10,000- to 300,000-step Monte Carlo conformation searches50 carried out with the OPLS-AA force field (FF)51,52 implemented in MacroModel v9.1 (Schrödinger, New York, 2005). The final structures shown in Figure 2 are confirmed to correspond to equilibrium geometry using a normal mode analysis, which also gives ZPE and the vibrational contribution to the free energy contribution G0298K. The rotational and translational contributions to G0298K are estimated by the equipartition theorem and the ideal gas approximation.20,21 Such an approach has been shown suitable to describe PEDOT-complex conformations.53 We choose B3LYP despite concerns about its reliability in describing long-range charge transfer because our previous B3LYP calculations have reproduced surprisingly well the UV/visible absorption spectra of charge-transfer (donor-acceptor) compounds54-58 and also because the binding energies and the ion exchange energies calculated with B3LYP are within 4% of error (see Supporting Information) from those calculated with a long-range-corrected (or range-separated hybrid) LC-PBE functional59,60 after tuning the range-separation parameter  to 0.17 a.u.1 using Gaussian09.61 The final geometries, where the PTS and TCB anions are bound to tri-EDOT mostly from the side (Figure 2 and Supporting Information), capture the binding geometries observed by x-ray crystallography on PEDOT:PTS crystals25,29,31 and MD simulations on PEDOT:TCB clusters (Section 2.2 and Supporting Information). We have established15 a simple correlation between the conductivity-enhancing power of EMIM:X IL and its ion-pair binding energy ∆Eb which is defined as [E0,pair – E0,cation – E0,anion] and calculated as –378, –358, –296, and –276 kJ/mol when X is Cl, ES, TCM, and TCB, respectively, in agreement with previous stuides.62,63 When an IL (e.g. EMIM:TCB) has ∆Eb whose magnitude is much smaller than that of EMIM:PSS (–368 kJ/mol), the IL would readily dissociate and participate in ion exchange with PSS to disturb the PEDOT:PSS binding. This simple correlation with the conductivity enhancing power of IL is still valid with the ion-pair binding free energy (instead of energy) ∆∆Gb, which is defined as [∆G0pair – ∆G0cation – ∆G0anion] and calculated as –352, –306, –248, and –227 kJ/mol for EMIM:X (black line with square marks; Figure 2) and as –317 kJ/mol for EMIM:PSS. When combined with the definition of ∆∆Gb, the equation (1) is rewritten in terms of ∆∆Gb as follows:

GX0  Gb0 (PEDOT : X)  Gb0 (EMIM : PSS)  Gb0 (PEDOT : PSS)  Gb0 (EMIM : X)

. (1')

The remaining ∆∆Gb values are calculated as –258, –221, –200, and –180 kJ/mol for PEDOT:X (black line with triangle marks; Figure 2) and as –235 kJ/mol for PEDOT:PSS. Indeed, all the anions bind more strongly to EMIM+ than to PEDOT+ whose singleunit positive charge is spread over three monomer units, but the difference in the two binding energies varies quite strongly with anion: 82 (PSS), 94 (Cl), 85 (ES), 48 (TCM), and 47 (TCB) kcal/mol. The preferred binding to EMIM than to PEDOT is prominent for small and point-charge-like (and thus more hydrophilic) anions and less significant for bulkier and softer (and thus more hydrophobic) anions: TCB ~ TCM < PSS ~ ES < Cl. When PEDOT:PSS and EMIM:TCB (or TCM) are mixed and sonicated vigorously, new hydrophilic EMIM:PSS ion pairs would form spontaneously and disperse in solution, leaving hydrophobic PEDOT:TCB (or TCM) pairs behind. Indeed, such a favorable ion exchange is predicted from the negative value (–35 kJ/mol) of the free energy (∆∆GX; red line in Figure 2). On the other hand, PSS and ES anions, having sulfonate in common, would show similar binding characteristics toward EMIM and PEDOT, and indeed the free energy of ion exchange induced by EMIM:ES is calculated to be negligible (∆∆GX = 2 kJ/mol; Figure 2). The point-charge-like Cl anion would prefer charge-localized EMIM to charge-dispersed PEDOT even more than PSS does, and thus the ion exchange from EMIM:Cl and PEDOT:PSS to EMIM:PSS and PEDOT:Cl would not easily occur, as revealed by the positive value of free energy of ion exchange (∆∆GX = 12 kJ/mol; Figure 2). Based on such understanding, we propose a new design rule for anion X in EMIM:X IL: a bulky, soft, charge-dispersed, and hydrophobic (but non-planar; see below Section 2.3) anion with a minimal ion-pair binding (free) energy ∆Eb (or ∆∆Gb) of EMIM:X, which leads to a significantly negative free energy of ion exchange with PEDOT:PSS (∆∆GX). A hypothetical heptacyanocyclopentenide (HCCP; Figure 1) anion is shown to meet all the criteria including the lowest ion-pair binding free energy of EMIM:HCCP (∆∆Gb; Figure 2, black line with square mark) and the most negative ion-exchange free energy with PEDOT:PSS (∆∆GX = –38 kJ/mol; Figure 2, red line) among the series considered in this work. 2.2. Ion exchange in mixed solution: MD. To validate the ionexchange mechanism predicted by DFT calculations and to explore at a larger length-scale its consequences on the resulting morphology of PEDOT, a series of MD simulations are performed on periodic model systems representing mixed aqueous solutions. We first explore the minimal tri-EDOT:PTS model mixed with EMIM:TCB IL that has been reported to lead to the highest conductivity. We then probe the robustness of our findings by using longer 6-unit EDOT oligomers (6EDOT) and substituting the PTS monomer units by 16-unit SS oligomers (16SS) and investigating various concentrations of IL. Most of simulation details including FF are taken from a previous MD work of Franco-Gonzalez and Zozoulenko on PEDOT:PTS systems without IL.34 Parameters for bonded and van der Waals (vdW) non-bonded FF are taken from OPLSAA64 implemented in GROMACS 5.1.4,65,66 while atomic-charge parameters used for Coulombic non-bonded FF are taken from the electrostatic-potential-fitted (ESP) charges obtained from the DFT calculations on each isolated species. Water solvent molecules are described by the SPC/E model.67 Each system is prepared by inserting PEDOT and PSS (or PTS) at random positions and orientations in a computational box of about 6 nm on each side, which is sufficiently larger than each oligomer, and followed by solvation in water. The resulting system of a total of about 20,000 atoms is first equilibrated during 30-ns NVT run at 293 K using the modified

3

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Berendsen thermostat.68 Subsequent 30-ns NPT run is performed at 1 atm using the Parrinello-Rahman barostat69 to complete the equilibration and to analyze the behavior of the system without IL. The leap-frog integration scheme70 with a time step of 2 fs is used. Then, to investigate the effect of IL on the PEDOT morphology, water molecules are randomly chosen and replaced by the desired amount of EMIM and X, and then the system is equilibrated during 30-ns NVT and subsequent 30-ns NPT runs. Simulations on 12 pairs (0.1 M) of tri-EDOT:PTS in water (6696 water molecules) are shown in Figure 3. The last snapshot of the simulation without IL (box size 5.9 nm; density 1.0 kg/m3; Figure 3a, left) shows three small -stacked clusters composed of several units of tri-EDOT (blue) decorated by PTS (green), as also observed in the previous study of Franco-Gonzalez and Zozoulenko.34 The cluster analysis performed on the 60-ns run with a friends-offriends algorithm finding all the molecules within a maximum distance rc of 3.5 Å (and 4.0 Å which gives essentially the same results) from a given molecule confirms that the three clusters form after less than 10 ns and stay quite stable (or metastable or dynamically arrested) over the next 50 ns without IL (Figure 3b, left panel, black dots) and also indicate that each cluster is composed on average of 2 or 5 tri-EDOT units and 1-4 PTS units (Figure 3b, right, red and turquoise bars). This is consistent with the radial distribution function (RDF) between the C atoms in the tri-EDOT backbones, gCC(r) or RDF(CC), computed over the 30-ns NPT simulation (Figure 3c, top left, black curve), which shows only four sharp peaks at multiples of d- (3.7 Å) and a broad peak beyond that similarly to the report of the work of Franco-Gonzalez and Zozoulenko.34 This is also consistent with the coordination number (CN = integrated RDF) which is saturated at 3-4 tri-EDOT neighbors per tri-EDOT (Figure 3c, bottom left, black curve).

Page 4 of 9

After 12 or 24 pairs (0.1 or 0.2 M) of EMIM and TCB are added to this final state, the three small clusters quickly (< 20 ns) gather and form a single -stacked cluster composed of 12 tri-EDOT units, as shown in the final MD snapshot (Figure 3a, middle/right), the time evolution of the number of clusters (Figure 3b, left, red), and the CN(CC) curve (Figure 3c, bottom left, blue and red). A higher number of RDF(CC) peaks (Figure 3c, top left), i.e., six with 12 pairs (blue) and eight with 24 pairs (red), indicate that the -stacking order of the cluster increases with higher concentration of IL (24 pairs). The RDF(CS) and CN(CS) between the C atoms of triEDOT backbone and the S atom of PTS fall down and shift out (Figure 3c, blue), while the RDF(CB) and CN(CB) between the triEDOT backbone C atoms and the B atom of TCB grow up (Figure 3c, red). Indeed, the cluster analysis for the case of 24 IL pairs (Figure 3b, right) shows that the single cluster of 12 tri-EDOT units (blue) is surrounded by TCB anions mostly (~17; gold) and by less amounts of PTS anions (~4; purple) and EMIM cations (~8; black) [or by ~10 TCB, ~5 PTS, and ~2 EMIM units in the case of 12 IL pairs], confirming the higher affinity of TCB than PTS to PEDOT suggested by DFT in Section 2.1.

Figure 4. 60-ns MD simulations on aqueous solutions of larger models of PEDOT:PSS, 16 tri-EDOT units and a 16SS chain (Model 1) as well as 16 longer 6EDOT units and two 16SS chains (Model 2), mixed with 0-128 pairs of EMIM:TCB and 64 pairs of ES. (a) Cluster size (number of tri-EDOT units only) distribution in Model 1 with 0-64 EMIM:TCB/ES pairs (left, red/blue; probability scale [0, 0.55]) and cluster composition in Model 1 with 64 EMIM:TCB/ES pairs (right, bottom/top); (b-c) Final snapshots and RDFs of Model 1 with 64 EMIM:TCB/ES pairs (left/middle) and Model 2 with 64/128 EMIM:TCP pairs (right). Color code: tri-EDOT/6EDOT (blue), 16SS (green), EMIM (yellow), TCB/ES (red), and water (removed for clarity).

Figure 3. 60-ns MD simulation on aqueous solutions of 12 tri-EDOT:PTS pairs and subsequent 60-ns MD simulation after adding 12 or 24 EMIM:TCB pairs to it. (a) Final snapshot. Color code: tri-EDOT (blue), PTS (green), EMIM (yellow), TCB (red), and water (removed for clarity); (b) cluster analyses: time evolution and composition; as well as (c) RDFs and CNs: gCC(r) between tri-EDOT units, gCB(r) between tri-EDOT and TCB, and gCS(r) between tri-EDOT and PTS.

Similar but slightly different behavior is observed with more realistic models employing a single 16SS oligomer (assumed fully deprotonated; -16|e|) instead of multiple PTS units (Figure 4). MD simulations of 60 ns are performed on the aqueous solution of 16 tri-EDOT units and one 16SS chain (6585 water molecules; box size 5.9 nm; density 1.0 kg/m3) and then another 60-ns runs after mixing with 12, 24, 48, or 64 pairs (0.1-0.6 M) of EMIM:TCB. During the 60-ns runs, 12 and 24 pairs of EMIM:TCB are not successful in replacing 16SS from tri-EDOT:16SS and inducing triEDOT aggregation into a large one: their cluster size distribution and RDF(CS) are not far from those of pristine tri-EDOT:16SS (Figure 4a, left, red, three bottom panels; Figure 4c, left; blue vs.

4

ACS Paragon Plus Environment

Page 5 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

black). However, high concentrations (48-64 pairs) of EMIM:TCB finally induce aggregation of 15-16 tri-EDOT units into a large stacked cluster, as seen in the cluster size distribution (Figure 4a, left, red, top) as well as the final snapshot (Figure 4b, left) and the RDF(CC) curve (Figure 4c, left, red) of the 64-EMIM:TCB case. This tri-EDOT cluster is surrounded mostly by 30-60 (peaked at ~53) TCB anions (Figure 4a, right, gold bars), and the depletion of the 16SS sulfonate groups around it is significant according to the RDF(CS) curve (Figure 4c, middle, red). It is understandable in terms of entropy that segregation of all-connected 16SS chain away from tri-EDOT clusters into water requires higher number (48; Figure 4a-b) of EMIM:TCB (or significantly longer time) than required for detachment/dispersion of small PTS units (12; Figure 3). [Similar behavior is observed with even larger models composed of 16 6EDOT units and two 16SS chains when they are mixed with 64 or 128 pairs (0.6 or 1.2 M) of EMIM:TCB in water (Figure 4b-c, right).] On the other hand, when the same pristine tri-EDOT:16SS (16:1) solution is mixed with the same high concentration (64 pairs) of EMIM:ES instead of EMIM:TCB, none of such behavior (ionic exchange, segregation of 16SS, or aggregation of tri-EDOT units) is observed. On the contrary, this solution behaves just as the pristine solutions of tri-EDOT:PTS (Figure 3a, left) and tri-EDOT:16SS (Figure 4a, left, bottom; Figure 4c, left and middle, black): three or four small tri-EDOT clusters persist during the 60-ns run (according to the final snapshot; Figure 4b, middle), and each of these clusters is composed of no more than 10 (mostly 4-5) tri-EDOT units (according to the cluster composition analysis; Figure 4a, left, blue) and surrounded by a significant amount of 16SS sulfonate groups [according to the RDF(CS) curve; Figure 4c, middle, green]. The ES anions, which are more hydrophilic than TCB and even than PTS, prefer staying dispersed in water than replacing PSS and staying near tri-EDOT clusters, just as predicted from the DFT calculations in Section 2.1 above. 2.3. Electronic structure of ion-exchanged PEDOT:X. The PSS- anions in PEDOT:PSS and the X- anions remaining in PEDOT:X domains after ion exchange may play a simple role of accompanying oxidized PEDOT+ cations, keeping their p-doped states, and thus sustaining their high charge carrier density n. However, the interaction between PEDOT and X, which would vary (rather ionic or rather covalent, rather localized or rather dispersed, and so on) according to the characteristics and the electronic structure of the anion X, may redistribute electrons in the PEDOT:X complex, inducing a subtle variation in the p-doping state of PEDOT. Therefore, the X-induced variation in the degree of p-doping (and the charge carrier density n) of PEDOT:X is qualitatively estimated on a periodic single chain model of the same series of PEDOT:X as considered in Section 2.1 (except PEDOT:Cl which was found to be unrealistic; Figure 2). The periodic unit cell contains two repeating units of PEDOT:X (six EDOT units with two units of positive charges, 6EDOT2+, and two X– anions with one X positioned above the PEDOT chain and the other below it; Figure 5a), which is the minimum-size model to consider the singlet (zero-spin) bipolaronic states. Built from the DFT structure optimized in Section 2.1, this periodic system is further optimized using the PBE DFT functional71,72 (as in other studies on PEDOT:PTS,31,32 although its reliability in describing charge transfer needs to be verified), the D2 dispersion correction of Grimme,73 the double-zeta basis set, and the (113) MonkhorstPack k-point grid implemented in the Atomistix Toolkit (ATK) 2014.2 code (QuantumWise, Copenhagen, Denmark).74 At the optimized structure, the band structure (not shown) as well as the total and projected density of states (P)DOS are calculated at the

(1130) Monkhorst-Pack k-point grid (Figures 5b-d). The Mulliken population analysis is also performed to calculate the positive charge carried by PEDOT, that is, the degree of p-doping of PEDOT fine-tuned by anion X (Figure 5e).31

  Figure 5. (a) Top and side views of a periodic cell of a single-thread PEDOT:X model; (b) DOS of un-doped PEDOT; (c-d) PDOS of PEDOT in PEDOT:X where X is PSS, ES, TCM, TCB, and HCCP, which is displayed with respective to the Fermi level of each species [-4.04 (PSS), -4.06 (ES), -3.88 (TCM), -4.32 (TCB), and -4.46 (HCCP) eV, which is marked as a vertical line] or with respect to the vacuum level; (e) Mulliken charge of each PEDOT monomer unit (numbered as in Figure 5a) in PEDOT:X.

Comparison between the DOS of neutral PEDOT (Figure 5b) and the PDOS of PEDOT in PEDOT:X (Figure 5c) indicates that the neighboring anion X pushes a significant portion of the valence band of PEDOT above the Fermi level (EF), keeping the partial oxidation (that is, p-doping) and a metallic behavior of PEDOT.31 The degree of p-doping induced by the valence band shift is prominent with TCB but, surprisingly, not with TCM. Indeed, the Mulliken charge summed over each tri-EDOT repeating unit in PEDOT:X is 0.62 (PSS), 0.66 (ES), 0.63 (TCM), and 0.75 (TCB) in electron unit (Figure 5e), showing again the outstanding p-doping power of TCB. Hence, while PEDOT:TCM and PEDOT:TCB show similar degrees of phase separation from PSS (Section 2.1), the resultant PEDOT-abundant phase would be more p-doped in PEDOT:TCB than in PEDOT:TCM. A higher charge carrier density n and in turn a higher conductivity  would be exhibited by PEDOT:TCB than by PEDOT:TCM, which in fact agrees with the experimental findings (see the orange and red squares in the two graphs at the bottom of Figure 1).15 This behavior should be relevant to the shift of EF: –4.04 (PEDOT:PSS), –4.06 (PEDOT:ES), –3.88 (PEDOT:TCM), and –4.32 (PEDOT:TCB) eV (Figure 5d). The EF of PEDOT:TCB is the lowest among the series and the EF of PEDOT:TCM is the highest among the series. It is understandable that PEDOT:PSS and PEDOT:ES show not only similar binding characteristics (Section 2.1) but also similar electronic structure, i.e., similar EF, similar (P)DOS curves, and essentially the same amount of p-doping, because they have the sulfonate end group in common. However, it is surprising to see such a difference between PEDOT:TCM and PEDOT:TCB in electronic structure. We suppose that the covalent nature of the interaction between PEDOT and TCM in the -stacked PEDOT:TCM (Figure

5

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2) induces TCM-to-PEDOT electron back-donation (by overlap or hybridization between molecular orbitals) and reduces the positive charge (p-doping) carried by PEDOT (Figure 5e, green), failing to sustain the complete p-doping in PEDOT which is achieved in the rather ionic PEDOT+-TCB– bond in PEDOT:TCB (Figure 5e, blue). We would therefore avoid IL additives with planar aromatic anions for electronics application, no matter how efficient they are for phase separation with their low IL binding energy. Such aromatic IL additives would not dissolve easily in water in all the cases. We would not recommend IL additives containing small hardbase anions such as sulfonate, oxide, fluoride, and chloride, either. Not only they are inefficient in achieving the phase separation (with the positive free energy of ion exchange; Figure 2), their interaction with PEDOT is so localized and point-like that electrons in PEDOT would redistribute due to them so that the positive charge is localized around their binding sites, leading to fluctuation in charge carrier density along PEDOT, as exhibited by PEDOT:ES (orange) and PEDOT:PSS (red) in Figure 5e. Such fluctuation would provide additional charge scattering sources which disturb charge transport and lower conductivity of PEDOT:ES and PEDOT:Cl. Such fluctuation would also create a curvature in the PEDOT chain (as seen from the end-to-end distance per monomer, which is in the order PEDOT:PTS ~ PEDOT:ES < PEDOT:TCM < PEDOT:TCB = 13.2 Å; Figure 2) provides extra charge scattering sources. In fact, such a charge fluctuation could be interpreted as a slight polaron (spin-polarized) contribution persisting among more stable (as supported by our calculations; not shown) singlet (zero-spin) bipolaron states in the cases of PEDOT:PSS/PTS/ES, as suggested in a previous study (~7% polaron and ~93% bipolaron in PEDOT:PSS).26 This explains the experimental observation that the polaron peak detected from the electron spin resonance (ESR) spectra15 of pristine PEDOT:PSS decreases when treated with EMIM:X IL and the extent of the reduction is higher when treated with more effective IL such as EMIM:TCB/TCM. Such a decrease of the polaron contribution has also been ascribed to higher inter-chain coupling and closer inter-chain stacking in the PEDOT phases, as seen in the electron paramagnetic resonance (EPR) spectra26 where the polaron peak of PEDOT:PSS decreases with the treatment with polar organic solvent and then completely disappears in PEDOT:Tos crystals. However, our current results obtained with a single-chain model indicate that the variation in the polaron contribution could also be an intrinsic intra-chain phenomenon directly caused by a subtle variation in the PEDOT-X interaction, which would persist in amorphous domains of PEDOT:PSS films. Therefore, our single-chain model, which is obviously very simple, could be quite useful (at least more useful than a perfect crystal model) to describe the charge distribution and transport in solution-cased IL-treated PEDOT:PSS films. Hence, our design rule for efficient EMIM-based IL additives enhancing PEDOT:PSS conductivity is that they should have bulky, charge-dispersed, hydrophobic-but-water-soluble, and non-planar soft-base anions. Our newly-proposed EMIM:HCCP, whose HCCP anion is bulky, non-planar, polarizable, and highly electron-withdrawing with seven nitrile groups, would satisfy this rule. We predict that the EMIM:HCCP would be as efficient as EMIM:TCB in inducing not only an efficient nanophase segregation between PEDOT and PSS (owing to the weak binding between EMIM and bulky HCCP; ∆∆GX = –38 kJ/mol; Figure 2) but also a high degree of p-doping (0.76 electron units per tri-EDOT due to a significant downward EF shift to –4.46 eV; Figure 5) and a high level of charge carrier density over the whole PEDOT chain with an extended conformation (13.3 Å per monomer unit; Figure 2). 2.4. Electron transport of PEDOT:X. Qualitative evaluation of electron transport behavior of PEDOT:X (relative to PEDOT:PSS)

Page 6 of 9

is carried out by (1) building a two-probe model junction device with a supercell of the single-thread PEDOT:X, whose geometry was optimized in Section 4, (2) applying bias voltages V up to 2 V in a step of 0.2 V between two electrodes (represented by two ends of the thread itself; Figure 6), and (3) calculating the tunneling current I at each bias voltage V. The DFT-based non-equilibrium Green’s function (NEGF) formalism75 is used to calculate the transmission probability T(E,V) with a fully self-consistent treatment of the electrode-molecule-electrode interaction, and the tunneling current I is evaluated by integrating T(E,V) over E within a bias window [–V/2, V/2] with the Landauer-Büttiker formula:75 (3) I  G0  T ( E ,V ) [ f ( E   L )  f ( E   R )] dE , where G0 is the conductance quantum, f(E) represents the FermiDirac distribution, and L (R) is the chemical potential of the left (right) electrode. The junction model shown in Figure 6 as well as the NEGF approach may not capture the electron transport through the whole device accurately, ignoring inter-chain transport, hopping, scattering, or contact resistance at the molecular-electrode interface, but it would still describe the intrinsic tunneling transport through a PEDOT backbone influenced by dopants X inside the bulk phase. The same level of DFT calculations as in Section 2.1 are performed with (1150) k-point grids. Indeed the calculated I-V curves (Figure 6) demonstrate the enhanced metallic character of PEDOT:X (colored curves) compared to un-doped PEDOT (black curve) at the voltages considered, which has been expected from Figures 5b-c. Although the I-V curves are fluctuating most likely due to the narrow energy band near the Fermi level of the electrodes made of a single-thread PEDOT:X, we still see clearly that PEDOT:TCB (blue) and PEDOT:HCCP (magenta) would indeed exhibit a higher electron conductivity than PEDOT:TCM (green), PEDOT:ES (orange), and PEDOT:PSS (red).

Figure 4. I-V curves calculated on two-probe junction device models of single-thread PEDOT:X (colored curves) and un-doped PEDOT (black curve).

3. Summary. Our minimal-model DFT calculation supported by larger-scale MD simulation demonstrates that the role of EMIM:X IL in conductivity enhancement of PEDOT:PSS is to induce a partial decoupling of PEDOT from PSS by ion exchange and a high level of uniformly-distributed p-doping in PEDOT with exchanged

6

ACS Paragon Plus Environment

Page 7 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

anion X. The calculations suggest that a high-performance IL additives should have (1) low binding energy between EMIM and X to guarantee favorable ion exchange with PEDOT:PSS and (2) strong but soft p-doping power, that is, electron-withdrawing but polarizable anion X. A bulky, non-planar, polarizable, hydrophobic, but water-soluble anion with multiple electron-withdrawing groups and no hydrogen-bond-accepting/donating groups would be promising. Based on such design principles, a new IL additive with a hypothetical heptacyanocyclopentenide (HCCP) anion is proposed. We expect that PEDOT:PSS treated with EMIM:HCCP would exhibit improved conductivity and become more suitable to be used for transparent flexible electrodes or thermoelectric devices. The MD simulation, which was carried out mostly on EMIM:TCB with medium-size models in the present work for a proof-of-concept purpose, will be extended in our future work to consider various types of ILs and to take into account the interconnectivity and entanglement of the PEDOT:PSS polymer in its long-time-scale behavior. This would require larger models of PEDOT:PSS (beyond 6EDOT and 16SS of our current study) and advanced simulation schemes which can overcome dynamically-arrested states more efficiently such as parallel tempering, simulated annealing, simulated spin coating (shear force and solvent evaporation) and/or coarsegrained models.

ASSOCIATED CONTENT

(6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20)

Supporting Information (21)

DFT validation and FF details are available free of charge on the ACS Publications website.

(22)

AUTHOR INFORMATION

(23)

Corresponding Author

(24)

*E-mail: [email protected] *E-mail: [email protected]

(25) (26)

ORCID Yves Lansac: 0000-0002-3955-417X Yun Hee Jang: 0000-0002-6604-5813

Author Contributions 1These

(27)

authors contributed equally.

(28)

Notes The authors declare no competing financial interest.

ACKNOWLEDGMENT

(30)

This work was supported by the Basic Research Program of NRF (2016R1A2B4009037) and the Human Resources in Energy Technology Program of KETEP (20174030201590) of Korea. The computing time was supported by the KISTI Grand Challenge Program (KSC-2016-C3-0012).

(4) (5)

(31) (32) (33) (34)

REFERENCES (1) (2) (3)

(29)

Inganas, O. Nat. Photon. 2011, 5, 201. Na, S.-I.; Kim, S.-S.; Jo, J.; Kim, D.-Y. Adv. Mater. 2008, 20, 4061. Pei, Q.; Zuccarello, G.; Ahlskog, M.; Inganas, O. Polymer 1994, 35, 1347. Groenendaal, L. B.; Jonas, F.; Freitag, D.; Pielartzik, H.; Reynolds, J. R. Adv. Mater. 2000, 12, 481. Elschner, A.; Kirchmeyer, S.; Lovenich, W.; Merker, U.; Reuter, K. PEDOT: Principles and Applications of an Intrinsically Conductive Polymer; CRC Press: Boca Raton, FL, 2010.

(35) (36) (37) (38) (39) (40)

Crispin, X.; Jakobsson, F. L. E.; Crispin, A.; Grim, P. C. M.; Andersson, P.; Volodin, A.; van Haesendonck, C.; Van der Auweraer, M.; Salaneck, W. R.; Berggren, M. Chem. Mater. 2006, 18, 4354. Kim, Y. H.; Sachse, C.; Machala, M. L.; May, C.; Müller-Meskamp, L.; Leo, K. Adv. Funct. Mater. 2011, 21, 1076. Vosgueritchian, M.; Lipomi, D. J.; Bao, Z. Adv. Func. Mater. 2012, 22, 421. Xia, Y.; Sun, K.; Ouyang, J. Adv. Mater. 2012, 24, 2436. Palumbiny, C. M.; Liu, F.; Russell, T. P.; Hexemer, A.; Wang, C.; Müller-Buschbaum, P. Adv. Mater. 2015, 27, 3391. Yu, Z.; Xia, Y.; Du, D.; Ouyang, J. ACS Appl. Mater. Interfaces 2016, 8, 11629. Sun, K.; Zhang, S.; Li, P.; Xia, Y.; Zhang, X.; Du, D.; Isikgor, F. H.; Ouyang, J. J. Mater. Sci.: Mater. Electron. 2015, 26, 4438. Wen, Y.; Xu, J. J. Polym. Sci. Part A: Polym. Chem. 2017, 55, 1121. Teo, M. Y.; Kim, N.; Kee, S.; Kim, B. S.; Kim, G.; Hong, S.; Jung, S.; Lee, K. ACS Appl. Mater. Interfaces 2017, 9, 819. Kee, S.; Kim, N.; Kim, B. S.; Park, S.; Jang, Y. H.; Lee, S. H.; Kim, J.; Kim, J.; Kwon, S.; Lee, K. Adv. Mater. 2016, 28, 8625. Kim, G.-H.; Shao, L.; Zhang, K.; Pipe, K. P. Nat. Mater. 2013, 12, 719. Fan, Z.; Du, D.; Yu, Z.; Li, P.; Xia, Y.; Ouyang, J. ACS Appl. Mater. Interfaces 2016, 8, 23204. Roh, E.; Hwang, B.-U.; Kim, D.; Kim, B.-Y.; Lee, N.-E. ACS Nano 2015, 9, 6252. Gkoupidenis, P.; Schaefer, N.; Garlan, B.; Malliaras, G. G. Adv. Mater. 2015, 27, 7176. van de Burgt, Y.; Lubberman, E.; Fuller, E., J.; Keene, S., T.; Faria, G. C.; Agarwal, S.; Marinella, M. J.; Talin, A. A.; Salleo, A. Nat. Mater. 2017. Kim, N.; Kee, S.; Lee, S. H.; Lee, B. H.; Kahng, Y. H.; Jo, Y.-R.; Kim, B.-J.; Lee, K. Adv. Mater. 2014, 26, 2268. Ouyang, J.; Xu, Q.; Chu, C.-W.; Yang, Y.; Li, G.; Shinar, J. Polymer 2004, 45, 8443. Lang, U.; Müller, E.; Naujoks, N.; Dual, J. Adv. Func. Mater. 2009, 19, 1215. Zotti, G.; Zecchin, S.; Schiavon, G.; Louwet, F.; Groenendaal, L.; Crispin, X.; Osikowicz, W.; Salaneck, W.; Fahlman, M. Macromolecules 2003, 36, 3337. Kim, E.-G.; Bredas, J.-L. J. Am. Chem. Soc. 2008, 130, 16880. Bubnova, O.; Khan, Z. U.; Wang, H.; Braun, S.; Evans, D. R.; Fabretto, M.; Hojati-Talemi, P.; Dagnelund, D.; Arlin, J.-B.; Geerts, Y. H.; Desbief, S.; Breiby, D. W.; Andreasen, J. W.; Lazzaroni, R.; Chen, W. M.; Zozoulenko, I.; Fahlman, M.; Murphy, P. J.; Berggren, M.; Crispin, X. Nat. Mater. 2014, 13, 190. Cho, B.; Park, K. S.; Baek, J.; Oh, H. S.; Koo Lee, Y.-E.; Sung, M. M. Nano Lett. 2014, 14, 3321. Fabretto, M. V.; Evans, D. R.; Mueller, M.; Zuber, K.; Hojati-Talemi, P.; Short, R. D.; Wallace, G. G.; Murphy, P. J. Chem. Mater. 2012, 24, 3998. Aasmundtveit, K. E.; Samuelsen, E. J.; Pettersson, L. A. A.; Inganas, O.; Johansson, T.; Feidenhans, R. Synth. Met. 1999, 101, 561. Breiby, D. W.; Samuelsen, E. J.; Groenendaal, L.; Struth, B. J. Polym. Sci., Part B: Polym. Phys. 2003, 41, 945. Shi, W.; Zhao, T.; Xi, J.; Wang, D.; Shuai, Z. J. Am. Chem. Soc. 2015, 137, 12929. Zhang, B.; Wang, K.; Li, D.; Cui, X. RSC Adv. 2015, 5, 33885. Munoz, W. A.; Singh, S. K.; Franco-Gonzalez, J. F.; Linares, M.; Crispin, X.; Zozoulenko, I. V. Phys. Rev. B 2016, 94, 205202. Franco-Gonzalez, J. F.; Zozoulenko, I. V. J. Phys. Chem. B 2017, 121, 4299. Rudd, S.; Franco-Gonzalez, J. F.; Singh, S. K.; Khan, Z. U.; Crispin, X.; Andreasen, J. W.; Zozoulenko, I.; Evans, D. J. Polym. Sci., Part B: Polym. Phys. 2018, 56, 97. Palumbiny, C. M.; Schlipf, J.; Hexemer, A.; Wang, C.; MüllerBuschbaum, P. Adv. Electron. Mater. 2016, 1500307. Xia, Y.; Ouyang, J. Macromolecules 2009, 42, 4141. Badre, C.; Marquant, L.; Alsayed, A. M.; Hough, L. A. Adv. Func. Mater. 2012, 22, 2723. Armel, V.; Rivnay, J.; Malliaras, G.; Winther-Jensen, B. J. Am. Chem. Soc. 2013, 135. Mayevsky, D.; Gann, E.; Garvey, C. J.; McNeill, C. R.; WintherJensen, B. Phys. Chem. Chem. Phys. 2016, 18, 19397.

7

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 9

(41) Stocker, T.; Kohler, A.; Moos, R. J. Polym. Sci. B: Polym. Phys. 2012, 50, 976. (42) Döbbelin, M.; Marcilla, R.; Salsamendi, M.; Pozo-Gonzalo, C.; Carrasco, P. M.; Pomposo, J. A.; Mecerreyes, D. Chem. Mater. 2007, 19, 2147. (43) Brooke, R.; Fabretto, M.; Krasowska, M.; Talemi, P.; Pering, S.; Murphy, P. J.; Evans, D. J. Mater. Chem. C 2016, 4, 1550. (44) Lansac, Y.; Degrouard, J.; Renouard, M.; Toma, A. C.; Livolant, F.; Raspaud, E. Sci. Rep. 2016, 6, 21995. (45) Kahol, P. K.; Ho, J. C.; Chen, Y. Y.; Wang, C. R.; Neeleshwar, S.; Tsai, C. B.; Wessling, B. Synth. Met. 2005, 151, 65. (46) Becke, A. D. Phys. Rev. A 1988, 38, 3098. (47) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785. (48) Slater, J. C. The self-consistent field for molecules and solids; McGraw-Hill, 1974. (49) Greeley, B. H.; Russo, T. V.; Mainz, D. T.; Friesner, R. A.; Langlois, J. M.; Goddard III, W. A.; Donnelly, R. E.; Ringnalda, M. N. J. Chem. Phys. 1994, 101, 4028. (50) Chang, G.; Guida, W. C.; Still, W. C. J. Am. Chem. Soc. 1989, 111, 4379. (51) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. J. Am. Chem. Soc. 1996, 118, 11225. (52) Kaminski, G. A.; Friesner, R. A.; Tirado-Rives, J.; Jorgensen, W. L. J. Phys. Chem. B 2001, 105, 6474. (53) Gangopadhyay, R.; Das, B.; Molla, M. R. RSC Adv. 2014, 4, 43912. (54) Ku, J.; Lansac, Y.; Jang, Y. H. J. Phys. Chem. C 2011, 115, 21508. (55) Ku, J.; Song, S.; Park, S. H.; Lee, K.; Suh, H.; Lansac, Y.; Jang, Y. H. J. Phys. Chem. C 2015, 119, 14063. (56) Ku, J.; Gim, Y.; Lansac, Y.; Jang, Y. H. Phys. Chem. Chem. Phys. 2016, 18, 1017. (57) Gim, Y.; Kim, D.; Kyeong, M.; Byun, S.; Park, Y.; Kwon, S.; Kim, H.; Hong, S.; Lansac, Y.; Jang, Y. H. Phys. Chem. Chem. Phys. 2016, 18, 15054. (58) Abdo, N. I.; Ku, J.; El-Shehawy, A. A.; Min, J.-K.; El-Barbary, A. A.; Jang, Y. H.; Lee, J.-S. J. Mater. Chem. A 2013, 1, 10306. (59) Vydrov, O. A.; Heyd, J.; Krukau, A. V.; Scuseria, G. E. J. Chem. Phys. 2006, 125, 074106. (60) Vydrov, O. A.; Scuseria, G. E. J. Chem. Phys. 2006, 125, 234109. (61) Frisch, M. J., et al.; Revision A.02 ed.; Gaussian, Inc.: Wallingford, CT, 2016. (62) Zhang, Q.-G.; Wang, N.-N.; Yu, Z.-W. J. Phys. Chem. B 2010, 114, 4747. (63) Mao, J. X.; Lee, A. S.; Kitchin, J. R.; Nulwala, H. B.; Luebke, D. R.; Damodaran, K. J. Mol. Struct. 2013, 1038, 12. (64) Jorgensen, W. L.; Tirado-Rives, J. Proc. Natl. Acad. Sci. 2005, 102, 6665. (65) Berendsen, H. J. C.; van der Spoel, D.; van Drunen, R. Comput. Phys. Commun. 1995, 91, 43. (66) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435. (67) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P. J. Phys. Chem. 1987, 91, 6269. (68) Bussi, G.; Donadio, D.; Parrinello, M. J. Chem. Phys. 2007, 126, 014101. (69) Parrinello, M.; Rahman, A. J. Appl. Phys. 1981, 52, 7182. (70) Hockney, R. W.; Eastwood, J. W. Computer Simulations Using Particles; McGraw-Hill, New York, 1981. (71) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (72) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1997, 78, 1396. (73) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104. (74) José, M. S.; Emilio, A.; Julian, D. G.; Alberto, G.; Javier, J.; Pablo, O.; Daniel, S.-P. J. Phys. Condens. Matter 2002, 14, 2745. (75) Datta, S. Superlattice Microst. 2000, 28, 253.

8

ACS Paragon Plus Environment

Page 9 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

TOC Figures

9

ACS Paragon Plus Environment