Donor–acceptor stacking arrangements in bulk and thin-film high-mobility conjugated polymers characterized using molecular modelling and MAS and surface-enhanced solid-state NMR spectroscopy

DPP-DTT adopts a donor-on-acceptor stacking arrangement which is preserved in thin films.


DFT calculations on polymer fragments
Computational calculations on the structural fragment were performed using Gaussian 09.
Structures were generated using the GaussView package and fully optimized at the B3LYP level of theory using the 6-31G(d) basis set. For the NMR calculations, polymer fragments were constructed where the aliphatic chains were replaced with methyl groups to reduce the computational cost. Two optimisations were carried out; one with the thiophene groups oriented to make a weak hydrogen bond with the DPP moiety, and one with the thiophene groups oriented away from the weakly hydrogen bonded position ( Figure S1).  Table S1. Reference shieldings of 31.7 and 197 ppm were used for 1 H and 13 C, respectively, determined from a separate calculation on a tetramethyl silane molecule. We note that a high chemical shift of 9 ppm is predicted for H5 only when weakly hydrogen bonded (conformation 1). When H5 is not weakly hydrogen bonded (conformation 2), a shift of 6.7 ppm is predicted, similar to the other aromatic proton sites in the structure. Table S1. Calculated 1 H and 13 C chemical shifts for the two conformations of the polymer fragment shown in Figure S1. For CH, CH 2 and CH 3 groups, values are averaged across the sites in the fragment.  Figure S2. Figure S2. Periodic DFT-simulated 13 C NMR spectra for the type I and type II structures.

Transfer integral calculations at the DFT level
The transfer integral has been calculated within the dimer fragment approach as implemented in the Amsterdam Density Functional (ADF) package 5 for oligomers containing two repeating units. In this approach, the orbitals of the dimer are expressed as linear combination of molecular orbitals of the fragments that are obtained by solving the Kohn-Sham equations. Especially, the site energies, ε 1 and ε 2 , and the transfer integrals t 12 , are obtained by computing the following matrix elements: Where  i and  j correspond to the HOMO/LUMO orbitals of the isolated molecules (i.e. fragments).
The transfer integral t 12 has been evaluated at density functional theory (DFT) level using the B3LYP (Becke, three-parameter, Lee-Yang-Parr) hybrid functional 6 with a Double Zeta basis set.
However, due to the non-orthogonality of the fragment orbital basis set, the transfer integral value is not uniquely defined and depends on the definition of the energy of the origin. 7 The problem is solved by applying a Löwdin transformation to the initial electronic Hamiltonian resulting in the following expression of the transfer integral: Where the parameter S 12 represents the orbitals overlap.

MD Computational Details
The strategy that we have used to probe the conformational space of DPP-DTT polymer chains in the bulk consists in coupling molecular mechanics and molecular dynamics simulations on a unit cell containing one monomer unit that is replicated using periodic boundary conditions to mimic an infinite system. All molecular mechanics/dynamics calculations have been performed within the Materials Studio (MS) 6.0 package 8 using a force-field derived from the Dreiding force-field 9 in which three torsion potentials have been reparameterized against reference B3LYP/cc-pvtz calculations; i.e., the torsion between the DPP and thiophene units, between the thiophene and thienothiophene segments and the torsion corresponding to the rotation of a methylpropyl chain with respect to the DPP core (see Figure S3). The atomic charges have been obtained by fitting the electrostatic potential calculated at the mp2/cc-pvdz level on a DPP-DTT dimer. successively at 600K and 1000K; (iv) and finally, longer quenched dynamics (t = 500 ps) using, as starting points, the most stable structure of the last quenched dynamics in step iii, are performed at increasing temperature (300K, 600K, and 1000K) following the procedure developed in steps ii and iii. Figure S4 shows the integrated intensities of a deconvoluted fit to the 1 H MAS NMR spectrum in Figure 4b in the main text. The integrated intensities of the H5 and H6 resonances are related by a factor of two, indicating that essentially all thiophene groups are in the weakly hydrogen bonded orientation.  Figure S5 shows a 1H DQ-SQ NMR spectrum recorded for DPP-DTT bulk polymer using a BABA 10 recoupling time of one rotor period for DQ excitation and reconversion to favour short-range 1 H -1 H proximities. The H6 -H9 autocorrelation at ( SQ,  DQ ) = (6.75 ppm, 13.5 ppm) has a much lower relative intensity than in the spectrum recorded with a BABA recoupling time of two rotor periods ( Figure 5a in the main text), confirming that this correlation corresponds to the longer-range intermolecular proximity between H6 and H9 in adjacent polymer chains, as observed in the MDsimulated structures.

Preparation of Polymer Samples
For all experiments on the bulk polymer, DPP-DTT was used as received from Sigma Aldrich without further purification or treatment. To prepare drop-cast and spin-coated films, DPP-DTT polymer was dissolved in dichlorobenzene to a concentration of 5 mg/ml and heated at 80 °C for 5 hours followed by 100°C for 30 minutes. Drop-cast films were then prepared by dropping the solution from a pipette onto a thickness #0 (0.085 -0.13 mm) microscope cover slip and annealing at 100 °C for 1 hour. Spin-coated films were prepared by dropping the solution onto a thickness #0 microscope cover slip rotating at 500 rpm, before annealing at 100 °C for 1 hour.

DNP sample Preparations Details
For DNP experiments on the bulk DPP-DTT polymer, approximately 1 mg of material was chosen initially, in order to minimize the amount of the material that was contaminated by the polarizing agent. The small fragment of polymer was broken into smaller pieces by grinding the sample in liquid nitrogen using a pestle and mortar. However, it was still not possible to obtain a powder by this approach; instead the sample was reduced to inhomogeneous flakes with a maximum size of around 0.5 mm. The flakes were then impregnated with the polarizing solution and no swelling was observed upon addition of the solution. The impregnated sample was loaded into the centre of a 3.2 mm sapphire rotor with KBr powder to maximise the signal enhancement. Figure S8   The thickness of the spin-coated film was estimated to be 400 nm by carrying out AFM measurements on an area of the film with a scratch (see Figure S9a,b). Because the film was so thin, it was not possible to remove it from the glass cover slip. Instead, the cover slip was coarsely crushed in a mortar and pestle. SEM images of the fragments ( Figure S9c)   It was not possible to measure the 13 C CP polarization enhancement for the spin-coated film since no 13 C signal could be obtained in the absence of microwave irradiation. However, a value of ε 1H = 15 was measured on the 1 H solvent signal. This is less than the value of 45 measured for the dropcast film using protonated TCE solvent (Figure 8d), but the DNP enhancement still enables a 13 C CPMAS NMR spectrum to be recorded, which would otherwise be impossible for the spin-coated film. Indeed it is remarkable that a natural-abundance 13 C signal can be obtained from such a small amount of sample. Furthermore, since the sample remained coated on the glass slide during the NMR experiment, DNP polarisation transfer was only possible from one side. Further work needs to be carried out to determine how far the polarisation penetrates into coated film samples and how much of the sample is therefore being observed. 13,14 Figure S10 shows a DNP-enhanced 13    1D 1 H direct excitation experiments were acquired with a rotor-synchronized spin echo sequence in order to suppress the background signal of the probe. π/2 and π pulses of 2.5 μs and 5.0 μs (100 kHz) were used respectively. The echodelays (τ) were set to one rotor period. Conventional cross-polarization (CP) experiments were used for the acquisition of the 1D 13 C CPMAS spectra at a MAS rate of 11 kHz. The CP step was achieved with a proton radio-frequency (RF) field of 75.5

DNP-MAS
kHz. 13 C chemical shifts were referenced to the TCE solvent resonance at 74 ppm. Other experimental details are summarized in Table S3. The 2D 1 H-13 C HETCOR spectra were recorded with a 3.