Fluorescence and Electroluminescence of J ‑ Aggregated Polythiophene Monolayers on Hexagonal Boron Nitride

: The photophysics of a semiconducting polymer is manipulated through molecular self-assembly on an insulating surface. Adsorption of polythiophene (PT) monolayers on hexagonal boron nitride (hBN) leads to a structurally induced planarization and a rebalancing of inter- and intrachain excitonic coupling. This conformational control results in a dominant 0 − 0 photoluminescence peak and a reduced Huang − Rhys factor, characteristic of J-type aggregates, and optical properties which are signi ﬁ cantly di ﬀ erent to both PT thin ﬁ lms and single polymer strands. Adsorption on hBN also provides a route to explore electroluminescence from PT monolayers though incorporation into hybrid van der Waals heterostructures whereby the polymer monolayer is embedded within a hBN tunnel diode. In these structures we observe up-converted singlet electroluminescence from the PT monolayer, with an excitation mechanism based upon inelastic electron scattering. We argue that surface adsorption provides a methodology for the study of fundamental optoelectronic properties of technologically relevant polymers. He helium ﬂ ow cryostat and held under a vacuum of approximately 10 − 6 mbar. Electrical measurements of completed P3DT devices were acquired using a 1612B Keithley Sourcemeter.

T he environment and interchain interactions of polymeric semiconductors are important determinants in the electronic and photophysical properties of their associated devices and thin films. 1−3 Semiconducting polymers offer significant advantages over their small-molecule counterparts 4,5 due to increased conjugation which enhances intrachain exciton and charge transport. 6,7 However, the mechanical flexibility inherent to polymer chains leads to a range of structural conformations which can both limit intrachain coupling and cause variations in the optical properties of single, isolated chains. 8−10 An investigation of the fundamental properties of polymer chains would ideally require a configuration in which both the conformation and the interaction between neighboring chains can be independently controlled, but in practice, this is very difficult to achieve. For example, in thin films of polythiophene (PT), one of the most extensively studied polymers, the polymer backbone is planarized in a lamellar structure, but the resulting optical properties are strongly influenced by interactions with neighboring molecules. 11,12 Consequently, the effects of planarization, which can control the degree of conjugation, and H-type aggregation, arising from cofacial polymer packing, are hard to disentangle. 12−14 Effects due to nearest neighbors can be eliminated in studies of single PT molecules in a frozen matrix, 15−17 but in this configuration polymers are in uncontrolled, nonplanar conformations in the absence of side-chain engineering approaches which can produce deterministic backbone configurations. 18,19 It is therefore highly desirable to identify a configuration in which the polymer conformation can be controlled, but decoupled from interchain interactions, thus allowing investigations of fundamental optoelectronic polymer properties. We show below that the adsorption of polymer monolayers on a planar insulating surface provides a route to the realization of this goal.
The optical properties of organic polymers can be greatly affected by H-and J-type aggregation, determined by the relative interplay of inter-and intrachain interactions. 13 Polymerization leads to conjugated arrays of fluorophores, where the excitonic coupling within the polymer chain arises due to both through-bond and electrostatic dipole−dipole interactions. 6,7,10 Intrachain coupling between subunits leads to J-type aggregation, 6 in which coherent coupling between chromophores shifts the exciton band to lower energy for the optically allowed transition at k = 0, where k is the exciton wave vector. This leads to a red shift in the emission maximum and an enhancement in photoluminescence intensity, 20,21 benefiting applications requiring bright, and spectrally sharp, luminescence. In contrast to intrachain interactions, interchain interactions within aggregates of organic polymers often lead to H-type aggregation, where coupling between neighboring molecules shifts the exciton energy at k = 0 to higher energy, leading to an indirect bandstructure and suppression of the direct 0−0 optical transition. 13, 22 It has recently been shown that H-and J-type aggregation, and therefore the optical properties, of small organic molecules can be controlled through the variation of their relative position and orientation within self-assembled domains which form on the surface of hexagonal boron nitride (hBN) and are stabilized, for example, through hydrogen-bonding interactions. 23−26 Molecular monolayers adsorbed on supporting flakes of hBN, an insulating 2D material, can also be integrated into hybrid van der Waals heterostructures, forming tunnel diodes, which enable the electroluminescence from the molecular layers to be explored, 25,27 resulting in various effects such as photon up-conversion and selective spin triplet excitation. 28−32 Polythiophene can also be adsorbed on the surface of hBN; the polymer backbone can be resolved within monolayer-thick islands using atomic force microscopy under ambient conditions allowing the molecular scale characterization of these self-assembled monolayers. 33 In light of these results, it is interesting to consider whether the order imparted by adsorption influences the optical properties of a polymer as previously reported for small molecules. 4,34 In this paper, we show that the self-assembly of polymeric monolayers on an insulating surface provides a route to an environment which is complementary to both thin films and single molecules frozen in a matrix and, moreover, allows characterization of conformation at the molecular scale. Specifically, excitonic coupling, and hence the luminescence, of semiconducting polymers can be explored without the need for π-stacking of molecules 22 or side chain engineering. 17,18 By depositing self-assembled monolayers of the PT derivative poly[3-decylthiophene-2,5-diyl] (P3DT) on hBN it is possible to form aggregates exhibiting the photophysical properties of Jtype aggregates, which reveal a much lower Huang−Rhys factor and more extended vibronic structure than has previously been reported for polythiophene. [15][16][17]35 It is also possible to integrate these polymeric monolayers into van der Waals tunnel diodes, allowing the electron-induced generation of excitons and electroluminescence from an embedded monolayer of polythiophene. We believe that this approach could be extended to a wide range of technologically relevant semiconducting polymers and provide further insight into the photophysical properties of these important materials.

RESULTS
Photoluminescence from Monolayer P3DT Aggregates on hBN. After deposition from solution onto precleaned hBN substrates (see the Methods), P3DT forms monolayer islands with a lamellar structure, as depicted in Figure 1a,b, due to the interdigitation of the decyl side groups attached to the polythiophene backbone. 33 As described in our previous work, atomic force microscopy (AFM) can be used under ambient conditions to resolve individual polymer chains and determine the morphology of self-assembled P3DT. 33 The AFM images in Figure 1c show P3DT islands of monolayer thickness with the polymer backbone parallel to the substrate. From the AFM images we determine a fractional surface coverage of 0.68, with no evidence of multilayers. We also observe a number of "kinks" in polymer chains at the edges of islands, whereby polymer chains are bent through 180°. The extent of straight polythiophene sections, those between either the end of a chain or a "kink", was determined to be 12.0 ± 4.6 nm. An average number of 7−8 such parallel sections were found per polythiophene island, separated by a distance of 1.99 ± 0.05 nm.
In order to explore the influence of conformation and morphology on the resonant interactions determining H-and J-type aggregation, the temperature dependence of the P3DT photoluminescence was measured as shown in Figure 2. At temperatures down to 6 K, the 0−0 peak remains dominant and the vibronic satellite peaks can be resolved up to the 0−3 peak. The 0−0 and associated vibronic peaks were red-shifted by, respectively, 43 and 53 meV between 300 and 6 K, with a reduction in the full width half-maximum of the 0−0 peak from 86 to 38 meV, see Figure 2, but minimal change in peak intensity, see the Supporting Information. The approximate linear dependence of red-shift on temperature (see Figure 2b) suggests increasing exciton diffusion to lower energy sites at low temperature. 17,38 The gradient of the red-shift of the 0−0 peak, 0.13 meV/K, is smaller than the value for isolated chains of polythiophene derivatives (0.2−0.5 meV/K), indicating that the local planarity of the polymer adsorbed on hBN is greater than that of single strands in a frozen matrix. 10,17,39 At low temperature, vibronic peaks in the photoluminescence spectra up to the 0−3 index were resolved; the energy separation of successive vibronic peaks, ΔE = 0.177 ± 0.001 eV at 6 K extracted by fitting to a Franck−Condon progression (fitted peaks are shown in SI), is close to the energy associated with the vinyl stretch of the polythiophene backbone. 40 The measured ratio of the intensities of the 0−0 peak and its 0−1 vibronic satellite peak (see Figure 2c) reveals a weak temperature dependence which contradicts the expected enhancement of the 0−0 peak relative to the 0−1 peak for Jtype aggregates, 6,22,41 suggesting a more complicated balance between inter-and intrachain interactions. The strength of interchain and intrachain coupling, respectively, J inter and J intra , can be estimated from the temperature dependence of the I 0−0 /I 0−1 peak ratio using the H/J-type aggregate model. 13 The peak ratio has a maximum value at (I 0−0 /I 0−1 ) max at a temperature T max , and according to this model, 4FJ inter = K B T max , where F is the Franck−Condon factor and I The parameter S 0 is the Huang−Rhys factor which characterizes the strength of vibrational coupling for thiophene units in the noninteracting limit (i.e., assuming that changes to the fluorescence line shape due to excitonic coupling are neglected). The Huang−Rhys factor of P3HT has been estimated in the literature to be in the range S 0 = 1−2 through a comparison of models for intrachain excitonic coupling with data for single chains. 12,13,15,36,42 Taking the Huang−Rhys factor measured by Clark et al. for a dilute P3HT solution 11 as a lower bound for S 0 , where S 0 = 1 (F = 0.37), we obtain J inter = 3 ± 1 meV and J intra = 88 ± 29 meV from experimental values (Figure 2c) for (I 0−0 /I 0−1 ) max = 5.8 and T max = 60 K for P3DT on hBN.
Our experimental values show an increase in intrachain and decrease in interchain interactions in comparison with P3HT nanofibers 36 for which J inter ∼ 9 meV and J intra ∼ 60 meV, while the interchain coupling is much less than the value J inter ∼ 30 ACS Nano www.acsnano.org Article meV reported for P3HT thin films. 11,12 In addition, our experimental value for J inter is close to the estimate, J inter = + 1 meV, calculated using the point dipole approximation (we assume a single point dipole interacts with a chain of neighboring dipoles with a chain separation (Figure 1) of 2 nm, a value 43 for the transition dipole moment of 4 D, and taking into account screening from the hBN substrate). 24 We also estimate the coherence number, N c , from the relationship I 0−0 /I 0−1 = N c /S 0 , 6,12,22 where we again take S 0 = 1 as a lower bound, giving N c ∼ 5 and a coherence length, L c ∼ 4 (L c = N c − 1), of thiophene units over which the exciton is delocalized, equivalent to ∼1.6 nm. This value is much less than the typical lengths, 12.0 ± 4.6 nm, of straight polymer sections (see Figure 1c) confirming that finite size effects are not expected to be significant.
The temperature dependence of the photoluminescence, discussed above, shows that the self-assembly of P3DT on hBN leads to a conformation which enhances intrachain coupling and suppresses interchain coupling and is consistent with our initial hypothesis that the sharp spectral features are due to planarization of the polymer backbone and the absence of face-to-face packing for the monolayer films we study. Our results show that adsorption of monolayers provides an environment which is complementary to thin films and single molecule studies.
Electroluminescence from P3DT Monolayers in van der Waals Heterostructures. It is also possible to incorporate P3DT monolayers on hBN, similar to those discussed above, into van der Waals heterostructures to produce a hybrid polymer/2D electrical device. We have adapted a device architecture, which was developed recently to form analogue structures with embedded small organic molecules, to form a heterostructure in which a P3DT monolayer is embedded between two few-layer hBN tunnel barriers and overlapping few-layer graphene (FLG) contacts (see Figure 3a). 27 This device structure is further encapsulated between two thicker (>10 nm) hBN flakes. Completed devices, as illustrated in Figure 3b, are constructed in such a way that, under an applied bias, current flows vertically between the fewlayer graphene (FLG) contacts by tunnelling across the hBN tunnel barriers, between which a P3DT partial (0.68 fractional coverage) monolayer is embedded. This type of device is fabricated by the sequential pick-up and mechanical transfer of flakes of graphene and hBN, including flakes on which the P3DT has been predeposited. The technique is described in Svatek et al. 27 with more details in SI.
The current−voltage dependence of a completed device consisting of hBN tunnel barriers with thicknesses of 1 ML (top) and 2 ML (bottom), respectively, was acquired at room temperature and is highly nonlinear, as expected for a tunnel diode. 27,44−46 As the bias, V, applied to the device was increased, electroluminescence was observed. The electroluminescence (EL) and photoluminescence (PL) spectra of the encapsulated P3DT show dominant peaks at E EL = 1.93 eV and E PL = 1.95 eV, respectively, which are broadened with respect to uncapped P3DT monolayers, with the full width half-maximum increasing from 68 to 140 meV between uncapped and capped room temperature photoluminescence measurements; this increase is likely due to variations in the local environment of the encapsulated P3DT within the heterostructures, for example, due to structural defects such as wrinkles and blisters which are common in van der Waals heterostructures. 47 As shown in Figure 3d, the vibronic peaks in the photoluminescence and electroluminescence of capped P3DT are less clearly defined than uncapped P3DT. In order to compare with our earlier analysis of vibronic structure, the I 0−1 /I 0−0 peak ratio was extracted from the P3DT photoluminescence, (I 0−1 /I 0−0 ) Capped-PL and electroluminescence spectra, (I 0−0 /I 0−1 ) Capped-EL , of capped P3DT. We find lower values for these quantities, (I 0−1 /I 0−0 ) Capped-PL = 0.05 and (I 0−1 /I 0−0 ) Capped-EL = 0.12, than for uncapped P3DT at room temperature, (I 0−1 /I 0−0 ) PL = 0.19. It is possible that effects such as pressure and interactions with two hBN interfaces within the heterostructures could lead to changes in the vibrational coupling of P3DT, 48 however the presence of inhomogeneities, as discussed above, complicate a comparison with homogeneous uncapped P3DT monolayers.
The observation of EL with spectral features which are similar to the PL of capped and uncapped P3DT indicates that charge transport between the two FLG electrodes generates excitons within the P3DT monolayer, which relax though the emission of photons resulting in the measured EL. The evolution of EL with increasing bias measured at 300 K (in Figure 4) shows no apparent change in the position of the EL peak with increasing voltage. From the anticipated band alignment of the highest occupied molecular orbital (E HOMO = 5 eV), the lowest unoccupied molecular orbital (E LUMO = 3 eV) of P3DT, and the work function of FLG (W FLG = 4.5 eV), we rule out direct injection and recombination of electrons and holes as a route to the formation of neutral excitons within the P3DT monolayer; this mechanism would require V EL ∼ 2| W FLG − E LUMO |/e ∼ 3 V, much larger than the measured onset voltage of EL, V EL = 1.6 V.
In fact, we observe EL centered at E EL = 1.93 eV even at the onset voltage, V EL = 1.6 V; i.e., the photon energy is greater than the energy gained by a single electron traversing the device, E EL > eV EL , a process known as photon up-conversion. ACS Nano www.acsnano.org Article As discussed in recent papers, 27,30,32 this indicates that molecules are excited into an intermediate state through inelastic scattering of a tunnelling electron, and then, through a further excitation, into the excited singlet state S 1 . It has been argued that the lifetime of the intermediate state must be longer than the average traversal time, τ T of electrons through each fluorophore. We estimate τ T ∼ 0.13 μs from the measured current density (0.37 pA nm −2 ) close to the threshold for EL, assuming an effective fluorophore area of 4 nm 2 . In common with recent studies of EL from organic molecules using scanning tunnelling luminescence, 27,30,32 and molecular/2D hybrid tunnel devices, 27 we identify the intermediate state as the spin triplet, T 1 , since other excited states, for example vibrationally excited states, are expected to have much shorter lifetimes. In this scenario, a molecule is excited into a triplet state, T 1 , via inelastic scattering of a tunnelling electron followed by a further excitation to an excited singlet, S 1 , state which relaxes through the emission of a photon. Note that for our observed onset voltage, we measure a photon upconversion, |E EL − eV EL |, of approximately 0.35 eV, corresponding to ∼14 k B T at room temperature, ruling out simple explanations based on thermally excited electrons. The triplet state of polythiophene is reported to have an energy of approximately ∼1.2 eV. 49 Interestingly, we do not observe EL from P3DT in this spectral range, although direct emission from the triplet state has been reported in related studies of some, but not all, small organic mole-cules. 27,29,30,32,50 The absence of triplet electroluminescence could be due to reduced intersystem crossing rates or more rapid triplet−triplet annihilation (TTA) for P3DT (relative to small molecules such as perylene diimide for which triplet emission is observed) resulting in an EL signal too low to detect using our experimental setup. The mechanism for the T 1 → S 1 transition may involve a second inelastic tunnelling process, 27,32 possibly via an initial transition to a higher lying state in the triplet manifold for which it has been shown that intersystem transfer may be more rapid. 51 Alternatively, this transition may involve a TTA process. 52 In our recent studies of triplet emission from a small molecule it was possible to rule out TTA as a mechanism, but in the absence of direct triplet emission in our devices we are unable to discriminate between these possible mechanisms and in addition note that Thomas et al. 53 have shown that the triplet population can be affected by J-type aggregation in P3HT further complicating a comparison with previous studies of electroluminescence from a molecular monolayer.
The efficiency of our devices ∼2 × 10 −7 photons/electron at 300 K is comparable to that observed in similar devices and STML experiments. 27,30,32,50 This low value is at least partially due to the fractional coverage of the P3DT layer (see Figure  1c) of 0.6−0.7; this coverage was chosen to reduce the occurrence of multilayers and aggregates, while the effective coverage of emissive sites is further reduced by inefficiencies in molecular packing and the area taken up by interdigitated decyl side chains, which do not contribute significantly to the frontier molecular orbitals or light-matter interactions of the polythiophene backbones.

CONCLUSION
We have shown that the sharp spectral features which we observe for self-assembled P3DT are directly due to the conformation of a polymer monolayer which result from adsorption on hBN which leads to a stabilization of a planar backbone and the absence of face-to-face packing. The adsorbed configuration provides an environment which is complementary to both thin films and single molecules in a frozen matrix. Specifically, adsorption on hBN facilitates the study of the intrinsic optical properties of planarized molecules, eliminating the complications of polythiophene in both thin films, in which planarization occurs due to the lamellar structure but is accompanied by strong interchain coupling leading to H-type aggregation, and as single molecules for which H-type aggregation is eliminated, but the molecules adopt an uncontrolled nonplanar conformation. By incorporating P3DT monolayers into hybrid van der Waals tunnel diodes, we also demonstrate electroluminescence from monolayer thick films of P3DT, where excitations are generated by inelastic scattering of electrons tunnelling across the hybrid hBN/P3DT/hBN junction. The associated observation of photon up-conversion in these devices is both highly relevant to the study of low voltage optoelectronics and provides a route to excite spin-triplet states in conducting polymers. We anticipate that this approach can be applied to a broad range of semiconducting polymers and copolymers offering the possibility to determine fundamental polymeric properties and optimize the bright and fast optical response from atomically thin organic layers. . EL of a polythiophene device was measured at 300 K for a series of applied voltages from 1.5 to 2.5 V in both forward and reverse bias. The EL intensity is shown plotted against applied voltage and photon energy with separate color scales (a). For a second device, also at 300 K, the total number of counts per second was extracted by integrating background subtracted spectra over the whole wavelength range. A logarithmic plot of the integrated counts versus voltage shows clear subthreshold emission, with the spectrum acquired at −1.6 V shown inset for reference.

ACS Nano
www.acsnano.org Article METHODS Sample Preparation. As in our earlier work, 24,33 hBN substrates were prepared by mechanical exfoliation of crystallites onto thermally oxidized 90 nm SiO 2 using the scotch tape method, leaving hBN flakes on the SiO 2 surface with typical lateral sizes and thicknesses of the order of 10 μm and 10 nm, respectively. hBN substrates were cleaned by annealing with a butane gas torch and cooled immediately prior to the deposition of P3DT chains from solution. Regiorandom P3DT (molecular weight 30000−100000) was purchased from American Dye Source (ADS 510) and was dissolved in toluene (≥99.9%, Sigma-Aldrich) to produce a solution of 2.5 μg mL −1 concentration. P3DT films on hBN were prepared by immersion in solution for 45 s and removal of excess solvent using a nitrogen gun.
Devices were fabricated using a modified version of transfer techniques discussed in the literature and described in our recent work. 27,54 Using a micromanipulation stage, polypropylene carbonate stamps were used to sequentially pick up a thick hBN flake, FLG, ultrathin hBN and a second ultrathin hBN layer onto which a P3DT had been predeposited. Devices were completed by the transfer of the hBN/FLG/hBN/P3DT/hBN stack onto a predeposited heterostructures of hBN and FLG (stamped from PDMS 55 ) on predeposited gold contacts on 300 nm SiO 2 .
Sample Characterization. AFM measurements of the polythiophene film morphology and van der Waals heterostructures at various stages during fabrication were carried out using both the Asylum Research Cypher S and MFP-3D instruments with NuNano Scout 70 probes.
Fluorescence spectroscopy and electroluminescence measurements were carried out using a Horiba MicOS optical spectrometer with a 50× objective (NA: 0.5) and a 405 nm delta diode excitation source with a pulse rate of 100 MHz, an average power of approximately 10 μW, and a spot size of approximately 2 μm. For all fluorescence measurements, samples were placed in an Oxford Instruments Microstat He helium flow cryostat and held under a vacuum of approximately 10 −6 mbar.
Electrical measurements of completed P3DT devices were acquired using a 1612B Keithley Sourcemeter.
Photoluminescence spectra of P3DT monolayers on hBN from 6 to 300 K, details of the fitting procedures, and an extended description of the device fabrication method (PDF)