15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Research Article www.advmat.de Strong Suppression of Thermal Conductivity in the Presence of Long Terminal Alkyl Chains in Low-Disorder Molecular Semiconductors Ekaterina Selezneva, Alexandre Vercouter, Guillaume Schweicher, Vincent Lemaur, Katharina Broch, Aleandro Antidormi, Kazuo Takimiya, Veaceslav Coropceanu, Jean-Luc Brédas, Claudio Melis,* Jérôme Cornil,* and Henning Sirringhaus* 1. Introduction While the charge transport properties of organic semiconductors have been extensively studied over the recent years, the field of organics-based thermo- Thermoelectric materials offer a simple electrics is still limited by a lack of experimental data on thermal transport solution for direct heat-to-electricity con- version from a variety of heat sources via and of understanding of the associated structure–property relationships. To the Seebeck effect. Worldwide, about 2/3 fill this gap, a comprehensive experimental and theoretical investigation of of primary energy is currently wasted as the lattice thermal conductivity in polycrystalline thin films of dinaphtho[2,3- heat.[1] Therefore, there exist great oppor- b:2′,3′-f ]thieno[3,2-b]thiophene (Cn-DNTT-Cn with n = 0, 8) semiconduc- tunities for enhancing the energy effi- tors is reported. Strikingly, thermal conductivity appears to be much more ciency of many power generation and isotropic than charge transport, which is confined to the 2D molecular layers. industrial processes. Current thermo- electric materials operate far from theo- A direct comparison between experimental measurements (3ω–Völklein retical efficiency limits. There are intense method) and theoretical estimations (approach-to-equilibrium molecular ongoing research efforts to improve effi- dynamics (AEMD) method) indicates that the in-plane thermal conductivity is ciency and enable wider applications in strongly reduced in the presence of the long terminal alkyl chains. This evolu- waste heat harvesting.[2–4] One class of tion can be rationalized by the strong localization of the intermolecular vibra- materials being explored for this purpose is that of organic semiconductors (OSCs). tional modes in C8-DNTT-C8 in comparison to unsubstituted DNTT cores, The efficiency of a thermoelectric mate- as evidenced by a vibrational mode analysis. Combined with the enhanced rial is determined by the dimensionless charge transport properties of alkylated DNTT systems, this opens the pos- figure of merit zT = S2σT/(κe+ κph), where sibility to decouple electron and phonon transport in these materials, which S [V K–1] denotes the Seebeck coefficient; provides great potential for enhancing the thermoelectric figure of merit ZT. σ [S m–1], the electrical conductivity; κe and κph [W m–1 K–1], the electronic and Dr. E. Selezneva, Dr. G. Schweicher, Prof. H. Sirringhaus Prof. K. Broch Optoelectronics Group Institut für Angewandte Physik Cavendish Laboratory Universität Tübingen University of Cambridge Auf der Morgenstelle 10, 72076 Tübingen, Germany JJ Thomson Avenue, Cambridge CB3 0HE, UK Dr. A. Antidormi E-mail: hs220@cam.ac.uk Catalan Institute of Nanoscience and Nanotechnology (ICN2) A. Vercouter, Dr. V. Lemaur, Dr. J. Cornil CSIC and BIST Laboratory for Chemistry of Novel Materials Campus UAB, Bellaterra, Barcelona 08193, Spain University of Mons Prof. K. Takimiya Mons 7000, Belgium Emergent Molecular Function Research Group E-mail: jerome.cornil@umons.ac.be RIKEN Center for Emergent Matter Science (CEMS) Wako, Saitama 351-0198, Japan The ORCID identification number(s) for the author(s) of this article Dr. V. Coropceanu, Prof. J.-L. Brédas can be found under https://doi.org/10.1002/adma.202008708. School of Chemistry and Biochemistry & Center for Organic Photonics © 2021 The Authors. Advanced Materials published by Wiley-VCH and Electronics (COPE) GmbH. This is an open access article under the terms of the Creative Georgia Institute of Technology Commons Attribution License, which permits use, distribution and Atlanta, GA 30332-0400, USA reproduction in any medium, provided the original work is properly Prof. J.-L. Brédas cited. Department of Chemistry and Biochemistry The University of Arizona DOI: 10.1002/adma.202008708 Tucson, AZ 85721-0088, USA Adv. Mater. 2021, 33, 2008708 2008708 (1 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de phononic components of the thermal conductivity, respec- to be in the range 0.1–0.8 W m–1 K–1. However, a detailed under- tively; and T [K], the absolute value of the average temperature standing of structure–property relationships remains elusive. between the cold and hot sides. For a good thermoelectric mate- One reason for the lack of studies of the thermal transport rial, it is also desirable to decouple electron and phonon trans- properties is that thermal conductivity is a challenging trans- port; the notion of an "electron crystal–phonon glass" origi- port coefficient to measure reliably. Thermal conductivity κ nally proposed by Slack,[5] in which electron mean-free paths is defined according to Fourier’s heat conduction equation: are long while phonon mean free paths are short, remains an ∆T Q = − κ , where Q [W m–2] is the heat flux passing through important concept, whose implementation has been attempted ∆l in different classes of inorganic materials with skutterudites[6] the sample; Δl [m], the sample length; and ΔT [K], the tem- and clathrates[7] being among the best-known examples. perature difference across the sample. Parasitic heat transfer Herringbone-stacked alkylated thienoacene-based molecular through radiative (infrared) heat exchange with the surround- materials have recently emerged as some of the best performing ings and losses due to thermal resistance at interfaces present OSCs, a result of p-type charge carrier mobilities that can reach in virtually every experimental system create high uncertainty over 10 cm2 V–1 s–1.[8–10] Among this family, dinaphtho[2,3-b:2′,3′-f] in the measured heat flux values.[37,38] Another unexpectedly thieno[3,2-b]thiophene (DNTT) derivatives have been investigated large uncertainty in thermal conductivity measurements comes in detail both experimentally and theoretically; the results point from measurements of sample dimensions, as has been dem- to a favorable 2D character of charge transport (i.e., within the onstrated by an international round-robin testing of bulk ther- layers).[11–18] Importantly, the origin of their high charge mobili- moelectric materials.[39] As a result, even for measurement ties has recently been attributed to their reduced dynamic dis- setups designed in accordance with well-established measure- order, coupled to an isotropic electronic coupling pattern within ment standards, the combined measurement uncertainty can the plane of charge transport. Indeed, alkylation has been dem- reach up to 20% for bulk materials. Reduced sample dimen- onstrated to: i) shift in-plane phonon modes to higher frequen- sions in the case of thin films worsen the situation and the cies, hence suppressing their impact on disorder; and ii) reduce experiments require extra care at every step to keep the overall the amplitude of the out-of-plane long-axis sliding motions (the uncertainty within acceptable limits. so-called killer-mode).[19–21] Combined to good environmental sta- bility and ease of processing through vapor and solution deposi- tion techniques,[22] these materials thus constitute ideal candidates for thermoelectric applications. However, a detailed investigation 2. Results and Discussion of their ability to transport heat has not been performed yet. To date, only few studies have been dedicated to the thermal In this work, we report a comprehensive experimental and transport properties of OSCs.[23] The thermal conductivity in theoretical investigation of the lattice thermal conductivity in organic materials is usually assumed to be low due to lattice polycrystalline thin films of dinaphtho[2,3-b:2′,3′-f ]thieno[3,2- disorder. More specifically, the limited thermal conductivity b]thiophene (Cn-DNTT-Cn with n = 0, 8) semiconductors (the in such systems is attributed to localization of lattice vibra- molecular structures are given in Figure 1a). Considering the tions described via the Einstein model of isolated atomic oscil- anisotropy of the material properties and the fact that DNTT lations[24] with heat being carried out through the lattice via molecules tend to have a preferential orientation with their random walk rather than through wavelike motions of collective long molecular axis nearly perpendicular to the substrate plane, oscillations.[25] The exact mechanism of phonon localization, it is important to distinguish between the in- and out-of-plane however, is not well understood. The Einstein-like behavior transport directions. Wang et al.[23] measured the thermal con- was observed in small single crystals of alpha-monoclinic sele- ductivity of DNTT thin films in the out-of-plane direction with nium[26] and, in spite of their microcrystallinity, in C60/C70.[27] the traditional 3ω-method.[40] However, in view of the potential More recently, very low values of thermal conductivity were also thermoelectric applications, the thermal transport in the in- found in layered WSe2 crystals – which was speculated to be plane direction of thin films of small-molecule semiconductors caused by phonon localization induced by random stacking of is more relevant since it aligns with the direction of fast elec- 2D crystalline thin sheets[28]—and in the fullerene derivatives trical transport in these materials. Here, we measured the in- PCBM and PCBNB[29,30]—which was explained, similarly to plane thermal conductivity of undoped DNTT and its alkylated C60/C70, by localization of vibrational modes of the rigid bucky- derivative C8-DNTT-C8 following our previously established ball molecules. The in-plane and out-of-plane thermal conduc- protocol that reduces measurement uncertainty.[41] The experi- tivities of various systems (pentacene, Alq3, C60, rubrene, TIPS- mental setup is schematically presented in Figure 1b. pentacene, CuPc)[31–36] with sample ranging from polycrystals Thermal conductivity measurements were performed in the to single crystals, have been measured macroscopically using in-plane direction according to the 3ω–Völklein method[42,43] techniques such as ac-calorimetry, the 3ω method or time implemented in a commercial Linseis Thin Film Analyzer domain thermoreflectance measurements and have been found (TFA).[44] More information about the method and its advan- tages and can be found in the Supporting Information. The samples were prepared through thermal evaporation Dr. C. Melis of a DNTT layer on top of the membrane of the measure- Dipartimento di Fisica ment chips using a masking shutter that allows the deposi- Università di Cagliari, Cittadella Universitaria Monserrato (Ca) 09042, Italy tion of four different thicknesses within the same conditions. E-mail: claudio.melis@dsf.unica.it Adequate masking insured specific deposition over the active Adv. Mater. 2021, 33, 2008708 2008708 (2 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 1. Molecular structures of the investigated materials and experimental details of the thermal conductivity measurements implementing the 3ω–Völklein method. a) Molecular structures of the investigated materials; i.e., DNTT (top) and C8-DNTT-C8 (bottom). b) Schematic description of the measurement setup. In a cross-sectional view (inset) the interface across which thermal resistance can contribute to measurement uncertainty is marked in red. c) Thermal resistance of DNTT and d) C8-DNTT-C8 as inverse function of their thickness. The intercepts with the y-axis are used to estimate the contribution from the thermal contact resistance at the interface between the film and the substrate; the corresponding error bars repre- sent the standard errors of the linear regression. area. Grazing-incidence wide-angle X-ray scattering (GIWAXS) arise in the present setup is that between the sample film and and atomic force microscopy (AFM) investigations allowed us the membrane (marked in red in Figure 1b (inset)). The con- to analyze the film microstructures and surface morphologies tribution from this interface can be estimated by plotting the as well as to determine their respective thicknesses. ∆T apparent thermal resistance, R th = [K W–1], with respect to The maximum thickness of the DNTT films was ≈240 nm Q and all films showed a polycrystalline structure with a preferred the sample thickness and approximating the intercept of the orientation where the long molecular axis is nearly perpendic- curve with the y-axis.[41] For all samples under investigation, ular to the substrate plane. In the case of C8-DNTT-C8 films, the total thermal resistance (sample + membrane) was clearly the growth rate was enhanced and the maximum thickness was separable from that of the empty membrane, indicating a clear ≈ 530 nm (for similar deposition conditions, see Experimental contribution of the sample to the thermal response (Figure S5, Section). The thinnest C8-DNTT-C8 film showed a preferred Supporting Information). orientation while thicker C8-DNTT-C8 films showed powder- The evolutions of Rth versus sample thickness are presented like distribution of crystallites (causing rings in the GIWAXS in Figure 1c,d. Note that since the direction of heat transport is data) and only a slightly preferred orientation of the crystallites in the plane of the sample film, the sample thickness defines (peaks on rings), pointing to an edge-on orientation as in the the cross-sectional area of heat transfer, not the sample length. case of DNTT (Figure S3, Supporting Information). The AFM The effective thermal resistance is thus plotted versus inverse topography for the C8-DNTT-C8 films also revealed larger fea- sample thickness. Linear extrapolation of the intercepts with tures and a higher surface roughness (Figure S4, Supporting the y-axis including the uncertainty of the linear regression Information). In spite of these differences, the average crystal- resulted in small, yet not negligible values. However, the clear lite size was found to be the same for all samples in the study, linear trends for both DNTT and C8-DNTT-C8, consistent with in the range of ≈7–10 nm (Table S1, Supporting Information). the scaling sample cross-section, indicated that the apparent As pointed out previously, parasitic heat losses at interfaces thermal resistance is dominated by the material property and with thermal contact resistances are among the major sources that the offset is likely caused by the uncertainties associated of uncertainties in thermal conductivity measurements. One with the repeatability of thermal conductivity measurements of the interfaces where thermal contact effects are likely to and film thickness determination. Adv. Mater. 2021, 33, 2008708 2008708 (3 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de complications due to orientational changes with increasing thickness, the thermal resistance as a function of inverse thick- ness exhibit a clear linear behavior across the entire thickness range (Figure 1c,d) for both molecules. This also suggests that orientational effects are not the dominant factor when com- paring the thermal conductivity extracted for the two mole- cules. The room temperature value of ≈0.05 W m–1 K–1 is com- parable to the ultralow thermal conductivities observed in a few systems, such as C60/C70,[27] layered WSe2[28] crystals, and the fullerene derivatives PCBM and PCBNB.[29,30] Our experimental results thus suggest that the in-plane thermal conductivity is significantly reduced by adding alkyl chains on the terminal aromatic rings. Indeed, such observa- tions are convincingly strengthened by the following theoret- ical estimations. Herein, we focused exclusively on the lattice thermal conductivity since the electronic contribution to the thermal transport is expected to be weak in neutral or slightly doped OSCs,[48] see the discussion in the Supporting Infor- mation; all our experimental data were collected on undoped Figure 2. Measured in-plane thermal conductivities of thin films of samples. Owing to the air stability of DNTT-based materials,[22] DNTT and C8-DNTT-C8 as a function of temperature. The error bars rep- we do not expect significant unintentional doping by environ- resent the combined uncertainty associated with standard deviation in mental contaminants (O2 and H2O). This is supported by our the repeatability of measurements due to variability in thermal contacts experience with FET devices made with the two molecules and other experimental factors and standard deviation associated with the thickness determination (including film non-uniformity and surface which generally exhibit very low OFF currents.[21] roughness). We estimated the lattice thermal conductivity in single crystals made of DNTT cores and their alkylated derivatives The temperature evolutions of the thermal conductivities of C8-DNTT-C8 via the "approach-to-equilibrium molecular DNTT and its alkylated derivative, C8-DNTT-C8, are presented dynamics" (AEMD) method.[49] In essence, AEMD computes in Figure 2. Due to high thermal resistance of the films under the lattice thermal conductivity of a material based on the investigation, we were able to obtain reliable data only from rapid decay time of a thermal gradient initially created along the effective sample area on top of the smaller membrane; as a specific crystal orientation. More precisely, this property is a result, it was not possible to correct for the radiative losses in deduced from the exponential fit of the time-decreasing tem- these experiments. Thus, the reported data present an apparent perature difference between the hot and cold region of the thermal conductivity that includes thermal conduction through simulation box with an appropriate solution of the 1D heat the material and conduction through infrared radiation; hence, ∂T ∂2T equation:[49] = D 2 . Therefore, the lattice thermal con- the values are overestimated. ∂t ∂x The thermal conductivity decreases with temperature, which ductivity, κ [W m–1 K–1], can be straightforwardly derived from is consistent with the trend found in crystalline materials.[45] the thermal diffusivity, D = κ/ρCp [m2 s–1], provided that the The decrease may appear statistically insignificant in case of mass density, ρ [kg m–3], and heat capacity, Cp [J kg–1 K–1], of C8-DNTT-C8; however, the trend is more evident when the error the system are well defined. Here, we relied on the Dulong– bars do not include the standard deviation associated with the Petit model,[50] which considers the specific heat capacity Cp to thickness determination (Figure S6, Supporting Information). be strictly equal to 3R. Further description of the method, its Since the film thickness is not expected to change significantly advantages as well as an example of a typical AEMD simulation as a function of temperature (substantial thermal expansion can be found in the Supporting Information. and material degradation typically occur at higher temperatures The inverse of the lattice thermal conductivity versus the (above 100 °C[46,47])), this standard deviation would cause an inverse of the length of the supercell along the three major offset in the mean thermal conductivity value without changing crystal axes of DNTT and C8-DNTT-C8 is shown in Figure 3. the overall trend. We note that in the case of C8-DNTT-C8 sam- We note that these two organic compounds exhibit a mono- ples, this trend was not found in the thicker films presenting clinic structure (with a = 6.187 Å, b = 7.662 Å, c = 16.210 Å, only a slightly preferred orientation of crystallites, compared α = 92.490° for DNTT and a = 5.987 Å, b = 7.861 Å, c = 34.066 to the thinnest film for which the preferential alignment was Å, β = 99.860° for C8-DNTT-C8; respectively[21]) and a her- clearly demonstrated (Figure S6, Supporting Information). The ringbone packing motif, which is essentially maintained by most striking result, however, is the extremely low thermal CH–π interactions in the ab plane, with the c-axis lying per- conductivity of the alkylated derivative C8-DNTT-C8 com- pendicular with respect to the substrate plane. A linear regres- pared to its nonalkylated counterpart. Since the C8-DNTT-C8 sion through the data gives lattice thermal conductivity values film with the preferred orientation of crystallites had similarly of 0.79, 0.73, and 1.40 W m–1 K–1 along directions a, b, and c low thermal conductivity, the thermal conductivity reduction for DNTT and 0.35, 0.31, and 1.14 W m–1 K–1 along the same cannot be correlated solely with the decrease in the prefer- directions for C8-DNTT-C8. In the latter case, our results are ential alignment of crystallites. We also note that despite the consistent with those of the theoretical work of Shi et al.,[51] Adv. Mater. 2021, 33, 2008708 2008708 (4 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 3. a,b) Inverse of the lattice thermal conductivity as a function of the inverse of the box length along directions a, b and c for DNTT (a) and C8-DNTT-C8 (b). Each inset represents the layered herringbone packing of DNTT and C8-DNTT-C8 in the ab plane. For the sake of clarity, the alkyl chains have been omitted in the inset of part (b). who have reported very similar thermal conductivity values on the thermal conductivity of C8-DNTT-C8 are very similar in the ab plane by using the NEMD method for the study of to those observed in the state-of-the-art fullerene derivatives dioctyl[1]benzothieno[3,2-b][1]-benzothiophene (C8-BTBT-C8). PCBM and PCBNB; such a low thermal conductivity could be The thermal conductivity ratios a/b, c/a, and c/b are 1.08, 1.77, of great interest for the development of efficient thermoelec- and 1.92 for DNTT and 1.13, 3.26, and 3.68 for its alkylated tric applications. We also draw attention to the fact that the derivative. Of notable interest is that the phonon mean free calculated in-plane thermal conductivities κin are 3.8 and 6.6 paths (MFPs), lbulk [Å], can be extracted from the analytical equa- times higher for DNTT and C8-DNTT-C8 when compared to tions associated with the extrapolation curves. Without dwelling the corresponding experimental measurements. The origin of on theoretical concepts that have already been discussed in these overestimations could be threefold: i) the predominance detail elsewhere,[52–54] we simply mention that such a deduction of the harmonic approximation in the expression of the poten- is made possible by combining the kinetic formulation of the tial energy (i.e., for bonds and angles) while anharmonic terms 1 would yet better account for phonon–phonon interactions; thermal conductivity (namely, κ = ρC pτ bulk 2 l bulk; where τbulk [ fs] 3 ii) the occurrence of scattering processes at grain boundaries is the bulk phonon relaxation time) with the decomposition of due to the polycrystalline nature of the organic thin films, as τbulk into separate terms due to the independence of the various confirmed by the GIWAXS experiments; and/or iii) the pres- scattering events arising in the system, as predicted by the Mat- ence of impurities that can strongly affect thermal transport by thiessen’s rule. Hence, the average phonon MFPs along direc- acting as phonon scattering centers. Another source of overes- tions a, b, and c are 129.0, 293.0, and 129.0 Å for DNTT and timation could arise from the absence of quantum corrections 15.0, 43.6, and 44.0 Å for C8-DNTT-C8, respectively. Unexpect- in our calculations; nevertheless, the Dulong–Petit model[50] edly, in contrast to the generally 2D character of charge trans- can be considered as valid since the MD simulations are con- port in molecular semiconductors with very weak electronic ducted at room temperature while the Debye temperature θD coupling between molecules lying in adjacent layers,[11–13,55,56] of many OSCs rarely exceeds 100 K.[58,59] It is worth noting that the computed thermal conductivity appears to be much more Wang et al.[23] reported an out-of-plane thermal conductivity isotropic, with the most efficient direction for heat transport κout = 0.45 ± 0.06 W m–1 K–1 for a DNTT thin film with a thick- actually corresponding to the interlayer axis. Since the 3ω–Völk- ness of 50 nm, as measured at room temperature by means lein method used experimentally on the polycrystalline samples of the differential 3ω method. By combining this value with provides an isotropically averaged value through the ab plane, the current experimental data, we obtain an anisotropy factor we have defined in turn an average thermal conductivity as κin = κ out exp /κ inexp ≈ 2.5, which is very similar to the corresponding theo- (κa + κb)/2 for ease of comparison with the experimental data. retical ratio κ out theo /κ intheo ≈ 1.9, even though it must be borne in On that basis, the calculations indicate that the thermal conduc- mind that the values of both in-plane and out-of-plane thermal tivity is reduced within the layers upon alkylation of the DNTT conductivities are overestimated by our theoretical calculations. core, which is fully consistent with the experimental results. The origin of such a strong suppression of thermal transport In addition, this structure–property relationship is consistent in the alkylated derivatives can be explained via the estimation of with a recent joint experimental and theoretical investigation of the participation ratio (PR) of the lattice vibrational modes.[60] As thermal transport properties of nonalkylated and alkylated BTBT described in the Experimental Section, this parameter is derived derivatives probed by thermal scanning microscopy along the from the diagonalization of the dynamical matrix and offers out-of-plane (interlayer) direction.[57] From a quantitative per- a quantitative estimate of the spatial extension of each vibra- spective, the effects of alkylation of DNTT lead to a decrease in tional mode. More specifically, it allows to distinguish between the calculated in-plane thermal conductivity by a factor of ≈2.3, extended modes (wavelike motions or collective oscillations), with this drop being experimentally even more pronounced characterized by a large PR value (>0.4), and localized modes (a factor of ≈ 4). As mentioned earlier, our experimental data (Einstein model of isolated atomic oscillations), which have a PR Adv. Mater. 2021, 33, 2008708 2008708 (5 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 4. a) Estimated participation ratio (PR) as a function of frequency for DNTT (red) and C8-DNTT-C8 (blue). b) Atomic displacement fields for two eigenmodes in a section of DNTT with thickness of 5 Å. Displacements (properly scaled) are shown as blue [red] vectors for localized [extended] modes with PR = 0.05 [0.35] respectively. close to zero. The calculated participation ratios in DNTT and of the atoms, leading to reduction of the spatial extension of the C8-DNTT-C8 are shown in Figure 4a as a function of frequency. vibrational modes, i.e., mode localization. This drop ultimately As can be seen, the DNTT participation ratios are found to be leads to remarkably low experimental thermal conductivities on systematically larger than those in C8-DNTT-C8 over the whole the order of 0.05 W m–1 K–1, comparable to the lowest values frequency window. This underscores the critical role played experimentally observed to date. Since, among the two sys- by long alkyl chains in strongly localizing the lattice modes in tems investigated here, C8-DNTT-C8 is known from field-effect C8-DNTT-C8 with respect to DNTT. The localized character of transistor (FET) studies to exhibit higher carrier mobilities, the modes is expected to hinder their capability of transmitting our results suggest the possibility of decoupling electron and heat, hence leading to a pronounced decrease in the total in- phonon transport in alkylated DNTT systems and realizing the plane thermal conductivity. A more intuitive understanding of so-called “phonon glass–electron crystal" concept, which is of the difference between extended and localized modes is gained high interest for enhancing the thermoelectric figure of merit by looking at their atomic displacements. To this aim, we show ZT. The combination of molecular design and accurate inves- in Figure 4b the displacement field of a high-frequency localized tigations of charge and thermal transport properties will open mode with PR = 0.05 and a low-frequency extended mode with new pathways for improved thermoelectric applications. PR = 0.35 in a section of DNTT with thickness of 5 Å. While in the extended mode almost all atoms participate in the motion, in the localized mode only a small subset of the atoms is signifi- cantly vibrating around the equilibrium positions. 4. Experimental Section Materials: DNTT was purchased from Sigma-Aldrich and used as received. C8-DNTT-C8 was synthesized according to previously described procedures and supplied by Nippon Kayaku.[15,61] 3. Conclusions The organic thin films were deposited by thermal evaporation (Angstrom Engineering, Inc.) at a rate of 0.5 Å s–1 under the pressure We have conducted a detailed experimental and theoretical of 10–7 Torr and substrate rotation of 25 rpm, using a masking shutter. study of the thermal transport properties in the crystalline Calculation of the Lattice Thermal Conductivity: The MD simulations structure of nonalkylated and alkylated DNTT derivatives. were performed with the Large-scale Atomic/Molecular Massively Thermal conductivity measurements were performed with Parallel Simulator (LAMMPS) package.[62] The systems are described the 3ω–Völklein technique following a protocol with reduced with a properly reparametrized version of the OPLS-AA force field whose measurement uncertainty. On the theoretical side, we have bonded and nonbonded parameters are implemented in the LAMMPS code via the LigParGen server.[63] The localized bond-charge corrected exploited the robustness and accuracy of the AEMD method to CM1A (1.14*CM1A-LBCC) charge model is employed.[64] We have verified estimate the phononic component of the heat diffusion along that this methodology generates thermal molecular motions around their crystal lattices. Our results emphasize the critical impact the crystal equilibrium geometry, without spurious drifts typically of establishing robust structure–property relationships in order observed with raw force fields. Moreover, simulations with a properly to design the phonon characteristics in organic semiconduc- tuned force field are able to depict reliably the low frequency modes tors and hence their heat transport properties. A direct com- of interest for thermal conductivity when compared to corresponding parison between experimental measurements and theoretical quantum-chemical calculations.[65] The DNTT and C8-DNTT-C8 unit cells are elongated along each direction of interest for analyzing the heat estimations highlights a noticeable drop of the in-plane thermal transport and are then replicated using periodic boundary conditions. conductivity upon addition of long alkyl side chains to the aro- These supercells are first optimized at constant lattice parameters before matic rings, which were found to disrupt the collective motion relaxing the box dimensions during a second optimization step. Next, Adv. Mater. 2021, 33, 2008708 2008708 (6 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de the systems are successively equilibrated in the NVT and then in the Supporting Information NPT ensemble, using a velocity-Verlet algorithm to solve the equation of motions and the Nosé–Hoover thermostat and barostat to monitor Supporting Information is available from the Wiley Online Library or the temperature and the pressure. Each simulation lasts for 1 ns under from the author. standard temperature and pressure conditions with a timestep of 1.0 fs, using 10 Å and 12 Å as van der Waals and electrostatic cutoff distances, respectively. In the AEMD approach, a controlled step-like temperature profile is applied by keeping fixed atomic positions in one half of the Acknowledgements simulation box while consecutively thermostating at = 362.5 K E.S., A.V., and G.S. contributed equally to this work. E.S., G.S., and H.S. and = 237.5 K the other half of the system during two short NVT gratefully acknowledge funding from the Centre for Advanced Materials simulations. An NVE simulation up to 3 ns is then needed to dissipate for Integrated Energy Systems (CAMIES, Grant No. EP/P007767/1). G.S. the initial ΔT = – = 125 K thermal gradient. Finally, a size- acknowledges postdoctoral fellowship support from The Leverhulme dependent lattice thermal conductivity is obtained by reinjecting the n =1 Trust (Early Career Fellowship supported by the Isaac Newton Trust). following fitting constant ∑ C ne −α n2Dt into the 1D heat equation.[49] Note A.V. is grateful to the Georgia Institute of Technology for partial support 5 that Cn and αn are coefficients depending on the size of the supercell, of his stay in the School of Chemistry and Biochemistry & Center for the integer n and the initially imposed temperature gradient. Organic Photonics and Electronics. V.C. and J.L.B. acknowledge support Estimation of the Participation Ratio: We studied the vibrational from the College of Science of the University of Arizona. The work in properties of both DNTT and C8-DNTT-C8 by setting up and diagonalizing the Laboratory for Chemistry of Novel Materials was supported by the the dynamical matrix of one snapshot of the samples, given by: Consortium des Équipements de Calcul Intensif (CÉCI), funded by the Fonds de la Recherche Scientifique de Belgique (F.R.S.-FNRS) under Grant No. 2.5020.11. J.C is a FNRS Research Director and A.V. is a −1 ∂Fiα Diα , jβ = (1) research follow from FNRS. K.B. thanks Prof. Schreiber (University of mi m j ∂r jβ Tübingen) for access to equipment.  In this equation, the Greek letters indicate the (x,y,z) Cartesian components while the Latin indices are used for labeling atoms. Here mi is the mass of the ith atom. Fiα is the force on the ith atom Conflict of Interest along direction α and rjβ an infinitesimal displacement of atom j along direction β. The authors declare no conflict of interest. The calculation of the first derivative in the above equation has been performed by finite difference with an atomic displacement of 5 × 10−4. The dynamical matrix has then been diagonalized by means of the SLEPc library[66] by obtaining the eigenvectors es and eigenvalues ω2s where Data Availability Statement s = 1, . . ., 3N counts eigenmodes. The data that support the findings of this study are openly available The participation ratio (PR) is finally estimated as:[67] in Apollo—The University of Cambridge Repository at https://doi. 2 org/10.17863/CAM.68874.  N 2  ∑ e i ,s  1  i =1  (2) PR = N N 4 Keywords ∑ e i ,s i =1  molecular dynamics, organic semiconductors, thermal conductivity, providing a normalized estimation of the subgroup of atoms involved thermoelectrics in the s-th vibrational mode. The spatial extension of such a subgroup is linked to the localized or extended nature of that mode: for extended Received: December 24, 2020 modes PR ≈ 1, while localized modes have a smaller ratio, down to the Revised: April 16, 2021 limit PR = 1/N for a mode completely localized on a single atom. It is Published online: August 3, 2021 worth noticing that, according to its very definition, a PR exactly equal to unity is obtained solely for vibrational modes in ideal crystalline systems, in which the atomic displacements are perfectly periodic within the sample. In turn, its value rapidly decreases if tiny inhomogeneities are [1] C. Forman, I. K. Muritala, R. Pardemann, B. Meyer, Renewable Sus- present in the atomistic coordinates or as a consequence of a possible tainable Energy Rev. 2016, 57, 1568. numerical uncertainty in the atomic displacements. In both cases, the [2] D. M. Rowe, Renewable Energy 1999, 16, 1251. extended character of the vibrational modes is preserved. Calculated [3] X. Liu, Y. D. Deng, S. Chen, W. S. Wang, Y. Xu, C. Q. Su, Case Stud. values of PR for extended modes in noncrystalline systems lie in the Therm. Eng. 2014, 2, 62. range [0.4–0.6]. [4] Y. Lyu, A. R. M. Siddique, S. H. Majid, M. Biglarbegian, Atomic Force Microscopy (AFM): The measurements were performed S. A. Gadsden, S. Mahmud, Energy Rep. 2019, 5, 822. with an MFP-3D AFM System (Asylum/Oxford Instruments) in AC [5] G. A. Slack, CRC Handbook of Thermoelectrics, CRC Press, Boca (noncontact) mode. The samples thickness was obtained by calculating an average step height in the topography scans taken at the sample Raton, FL, USA 1995. edges in the proximity of the four corners. [6] X. Shi, J. Yang, J. R. Salvador, M. Chi, J. Y. Cho, H. Wang, S. Bai, Grazing-Incidence Wide-Angle X-ray Scattering (GIWAXS): GIWAXS J. Yang, W. Zhang, L. Chen, J. Am. Chem. Soc. 2011, 133, 7837. measurements were performed using a Xeuss 2.0 SAXS/WAXS system [7] A. Saramat, G. Svensson, A. E. C. Palmqvist, C. Stiewe, E. Mueller, (Xenocs) with a Dectris Pilatus3R 300 K detector using a wavelength of D. Platzek, S. G. K. Williams, D. M. Rowe, J. D. Bryan, G. D. Stucky, 1.5406 Å and an angle of incidence of 0.2°. For the measurements, the J. Appl. Phys. 2006, 99, 023708. sample was placed in a vacuum chamber to reduce air scattering. [8] C. Mitsui, T. Okamoto, M. Yamagishi, J. Tsurumi, K. Yoshimoto, Thermal Conductivity Evaluation: Calculations were performed K. Nakahara, J. Soeda, Y. Hirose, H. Sato, A. Yamano, T. Uemura, according to Linseis et al.[44] from the raw data obtained on Linseis TFA. J. Takeya, Adv. Mater. 2014, 26, 4546. Adv. Mater. 2021, 33, 2008708 2008708 (7 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2021, 37, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202008708 by Readcube (Labtiva Inc.), Wiley Online Library on [15/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de [9] G. Schweicher, G. Garbay, R. Jouclas, F. Vibert, F. Devaux, [38] P. Diaz-Chao, A. Muniz-Piniella, E. Selezneva, A. Cuenat, Meas. Sci. Y. H. Geerts, Adv. Mater. 2020, 32, 1905909. Technol. 2016, 27, 085002. [10] S. Fratini, M. Nikolka, A. Salleo, G. Schweicher, H. Sirringhaus, Nat. [39] H. Wang, W. D. Porter, H. Böttner, J. König, L. Chen, S. Bai, Mater. 2020, 19, 491. T. M. Tritt, A. Mayolet, J. Senawiratne, C. Smith, F. Harris, P. Gilbert, [11] W. Xie, K. Willa, Y. Wu, R. Häusermann, K. Takimiya, B. Batlogg, J. W. Sharp, J. Lo, H. Kleinke, L. Kiss, J. Electron. Mater. 2013, 42, C. D. Frisbie, Adv. Mater. 2013, 25, 3478. 654. [12] S. Haas, Y. Takahashi, K. Takimiya, T. Hasegawa, Appl. Phys. Lett. [40] D. G. Cahill, R. O. Pohl, Phys. Rev. B 1987, 35, 4067. 2009, 95, 022111. [41] E. Selezneva, R. Di Pietro, X. Jiao, C. R. McNeill, H. Sirringhaus, [13] G. Schweicher, V. Lemaur, C. Niebel, C. Ruzié, Y. Diao, O. Goto, APL Mater. 2019, 7, 081118. W. Lee, Y. Kim, J. Arlin, J. Karpinska, A. R. Kennedy, S. R. Parkin, [42] F. Völklein, H. Reith, A. Meier, Phys. Status Solidi A 2013, 210, 106. Y. Olivier, S. C. B. Mannsfeld, J. Cornil, Y. H. Geerts, Z. Bao, Adv. [43] A. Sikora, H. Ftouni, J. Richard, C. Hébert, D. Eon, F. Omnès, Mater. 2015, 27, 3066. O. Bourgeois, Rev. Sci. Instrum. 2012, 83, 054902. [14] B. Peng, S. Huang, Z. Zhou, P. K. L. Chan, Adv. Funct. Mater. 2017, [44] V. Linseis, F. Völklein, H. Reith, K. Nielsch, P. Woias, Rev. Sci. 27, 1700999. Instrum. 2018, 89, 015110. [15] M. J. Kang, I. Doi, H. Mori, E. Miyazaki, K. Takimiya, M. Ikeda, [45] D. G. Cahill, R. O. Pohl, Ann. Rev. Phys. Chem. 1988, 39, 93. H. Kuwabara, Adv. Mater. 2011, 23, 1222. [46] K. Kuribara, H. Wang, N. Uchiyama, K. Fukuda, T. Yokota, [16] T. Uemura, K. Nakayama, Y. Hirose, J. Soeda, M. Uno, W. Li, U. Zschieschang, C. Jaye, D. Fischer, H. Klauk, T. Yamamoto, M. Yamagishi, Y. Okada, J. Takeya, Curr. Appl. Phys. 2012, 12, S87. K. Takimiya, M. Ikeda, H. Kuwabara, T. Sekitani, Y. Loo, T. Someya, [17] T. Uemura, C. Rolin, T. H. Ke, P. Fesenko, J. Genoe, P. Heremans, Nat. Commun. 2012, 3, 723. J. Takeya, Adv. Mater. 2016, 28, 151. [47] M. J. Kang, E. Miyazaki, I. Osaka, K. Takimiya, A. Nakao, ACS Appl. [18] C. Rolin, E. Kang, J. H. Lee, G. Borghs, P. Heremans, J. Genoe, Nat. Mater. Interfaces 2013, 5, 2331. Commun. 2017, 8, 14975. [48] D. Wang, W. Shi, J. Chen, J. Xi, Z. Shuai, Phys. Chem. Chem. Phys. [19] S. Illig, A. S. Eggeman, A. Troisi, L. Jiang, C. Warwick, M. Nikolka, 2012, 14, 16505. G. Schweicher, S. G. Yeates, Y. H. Geerts, J. E. Anthony, [49] C. Melis, R. Dettori, S. Vandermeulen, L. Colombo, Eur. Phys. J. B H. Sirringhaus, Nat. Commun. 2016, 7, 10736. 2014, 87, 96. [20] S. Fratini, S. Ciuchi, D. Mayou, G. T. de Laissardière, A. Troisi, Nat. [50] N. W. Ashcroft, N. D. Mermin, Solid State Physics, Harcourt College Mater. 2017, 16, 998. Publishers 1976. [21] G. Schweicher, G. D’Avino, M. T. Ruggiero, D. J. Harkin, K. Broch, [51] W. Shi, J. Chen, J. Xi, D. Wang, Z. Shuai, Chem. Mater. 2014, 26, D. Venkateshvaran, G. Liu, A. Richard, C. Ruzié, J. Armstrong, 2669. A. R. Kennedy, K. Shankland, K. Takimiya, Y. H. Geerts, J. A. Zeitler, [52] P. K. Schelling, S. R. Phillpot, P. Keblinski, Phys. Rev. B: Condens. S. Fratini, H. Sirringhaus, Adv. Mater. 2019, 31, 1902407. Matter Mater. Phys. 2002, 65, 144306. [22] K. Takimiya, S. Shinamura, I. Osaka, E. Miyazaki, Adv. Mater. 2011, [53] D. P. Sellan, E. S. Landry, J. E. Turney, A. J. H. McGaughey, 23, 4347. C. H. Amon, Phys. Rev. B 2010, 81, 214305. [23] X. Wang, K. D. Parrish, J. A. Malen, P. K. L. Chan, Nat. Sci. Rep. [54] D. Wang, L. Tang, M. Long, Z. Shuai, J. Phys. Chem. C 2011, 115, 2015, 5, 16095. 5940. [24] A. Einstein, Ann. Phys. 1911, 340, 898. [55] J. Cornil, J. P. Calbert, J. L. Brédas, J. Am. Chem. Soc. 2001, 123, 1250. [25] D. G. Cahill, S. K. Watson, R. O. Pohl, Phys. Rev. B 1992, 46, 6131. [56] Y. Tsutsui, G. Schweicher, B. Chattopadhyay, T. Sakurai, J. B. Arlin, [26] M. Meissner, A. Tausend, D. Wobig, Phys. Status Solidi A 1978, 49, C. Ruzié, A. Aliev, A. Ciesielski, S. Colella, A. R. Kennedy, V. Lemaur, 59. Y. Olivier, R. Hadji, L. Sanguinet, F. Castet, S. Osella, D. Dudenko, [27] J. R. Olson, K. A. Topp, R. O. Pohl, Science 1993, 259, 1145. D. Beljonne, J. Cornil, P. Samorì, S. Seki, Y. H. Geerts, Adv. Mater. [28] C. Chiritescu, D. G. Cahill, N. Nguyen, D. Johnson, A. Bodapati, 2016, 28, 7106. P. Keblinski, P. Zschack, Science 2006, 315, 351. [57] M. N. Gueye, A. Vercouter, R. Jouclas, D. Guérin, V. Lemaur, [29] J. C. Duda, P. E. Hopkins, Y. Shen, M. C. Gupta, Phys. Rev. Lett. G. Schweicher, S. Lenfant, A. Antidormi, Y. Geerts, C. Melis, 2013, 110, 015902. J. Cornil, D. Vuillaume, Nanoscale 2021, 13, 3800. [30] X. Wang, C. D. Liman, N. D. Treat, M. L. Chabinyc, D. G. Cahill, [58] P. Ramadoss, N. Buvaneswari, J. Chem. 2011, 8, 120260. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 88, 075310. [59] S. Fakruddin Babavali, T. S. Nori, C. Srinivasu, Mater. Today: Proc. [31] N. Kim, B. Domercq, S. Yoo, A. Christensen, B. Kippelen, 2019, 18, 2073. S. Graham, Appl. Phys. Lett. 2005, 87, 241908. [60] A. Cappai, A. Antidormi, A. Bosin, D. Narducci, L. Colombo, [32] R. C. Yu, N. Tea, M. B. Salamon, D. Lorents, R. Malhotra, Phys. Rev. C. Melis, Phys. Rev. Mater. 2020, 4, 035401. Lett. 1992, 68, 2050. [61] T. Yamamoto, K. Takimiya, J. Am. Chem. Soc. 2007, 129, 2224. [33] H. Zhang, J. W. Brill, J. Appl. Phys. 2013, 114, 043508. [62] S. Plimpton, J. Comput. Phys. 1995, 117, 1. [34] Y. Okada, M. Uno, Y. Nakazawa, K. Sasai, K. Matsukawa, [63] L. S. Dodda, I. de Cabeza de Vaca, J. Tirado-Rives, W. L. Jorgensen, M. Yoshimura, Y. Kitaoka, Y. Mori, J. Takeya, Phys. Rev. B 2011, 83, Nucleic Acids Res. 2017, 45, W331. 113305. [64] L. S. Dodda, J. Z. Vilseck, J. Tirado-Rives, W. L. Jorgensen, J. Phys. [35] Y. Yao, M. Shahi, M. M. Payne, J. E. Anthony, J. W. Brill, J. Mater. Chem. B 2017, 121, 3864. Chem. C 2016, 4, 8817. [65] N. G. Martinelli, Y. Olivier, S. Athanasopoulos, M.-C. R. Delgado, [36] D. Trefon-Radziejewska, J. Juszczyk, A. Fleming, N. Horny, K. R. Pigg, D. A. da Silva Filho, R. S. Sánchez-Carrera, E. Venuti, J. S. Antoniow, M. Chirtoc, A. Kazmierczak-Balata, J. Bodzenta, R. G. Della Valle, J.-L. Brédas, D. Beljonne, J. Cornil, ChemPhysChem Synth. Met. 2017, 232, 72. 2009, 10, 2265. [37] E. Selezneva, C. Stacey, P. Diaz-Chao, A. Muniz-Piniella, A. Cuenat, [66] V. Hernandez, J. E. Roman, V. Vidal, ACM Trans. Math. Software in Thermoelectric Materials and Devices, Energy and Environment 2005, 31, 351. Series, Vol. 17 (Eds: I. Nandhakumar, N. M. White, S. Beeby), Royal [67] A. Antidormi, X. Cartoixà, L. Colombo, Phys. Rev. Mater. 2018, 2, Society of Chemistry, Cambridge, UK 2017, pp. 109–132. 056001. Adv. Mater. 2021, 33, 2008708 2008708 (8 of 8) © 2021 The Authors. Advanced Materials published by Wiley-VCH GmbH