Impact of hydrophilic side chains on the thin film transistor performance of a benzothieno–benzothiophene derivative

Side-chain engineering in molecular semiconductors provides a versatile toolbox for precisely manipulating the material's processability, crystallographic properties, as well as electronic and optoelectronic characteristics. This study explores the impact of integrating hydrophilic side chains, specifically oligoethylene glycol (OEG) units, into the molecular structure of the small molecule semiconductor, 2,7-bis(2(2-methoxy ethoxy)ethoxy) benzo[b]benzo[4,5] thieno[2,3-d] thiophene (OEG-BTBT). The investigation includes a comprehensive analysis of thin film morphology and crystallographic properties, along with the optimization of deposition parameters for improving the device performance. Despite the anticipated benefits, such as enhanced processability, our investigation into OEG-BTBT-based organic field-effect transistors (OFETs) reveals suboptimal performance marked by a low effective charge carrier mobility, a low on/off ratio, and a high threshold voltage. The study unveils bias stress effects and device degradation attributed to the high ionization energy of OEG-BTBT alongside the hydrophilic nature of the ethylene–glycol moieties, which lead to charge trapping at the dielectric interface. Our findings underscore the need for a meticulous balance between electronic properties and chemical functionalities in molecular semiconductors to achieve stable and efficient performance in organic electronic devices.


OEG-BTBT thin film transistor device architecture
Supporting Information

X-Ray reflectivity measurements ODTS self-assembled monolayer
X-ray reflectivity data of the used substrates: 300 nm thermally oxidized silicon (grey curve) and ODTStreated thermally oxidized silicon (black curve).Fitting of the experimental data (red curves) reveal a mass density of 2.3 g cm -3 and a surface roughness of 0.37 nm for the untreated substrate.For the ODTS-treated substrate a mass density was found to be between 1.3 g cm -3 and 1.4 g cm -3 , and a film thickness of 1.6 nm with a surface roughness of 0.3 nm was determined.

Energy cutoff and k-mesh convergence
For the calculation of the Raman spectrum with DFT, the energy cutoff and k-mesh were chosen such that the total energy per atom is converged to below 0.5 meV.To that end, we performed calculations of the total energy per atom with energy cutoffs of 400 eV, 500 eV, 600 eV, 700 eV, 800 eV, 900 eV, 1000 eV, 1100 eV and 1200 eV at a 1×2×2 k-mesh.The results are shown in Figure S5.Additionally, to determine the necessary k-point sampling of the first Brillouin zone for obtaining the charge densities, calculations with 1×2×2, 2×5×5, 3×7×7 and 4×10×9 k-meshes were performed at a cutoff energy of 400 eV.The resulting energies are presented in Figure S6.These data show that the energy per atom is converged to below 0.5 meV for an energy cutoff of 900 eV and a k-mesh of 1×2×2.Supporting Information

Raman spectrum for the fully optimized cell
In addition to calculating the Raman spectrum for the experimental unit cell of OEG-BTBT, we also fully relaxed the lattice parameters.This yielded the following lengths and angles between the unit cell vectors: 18.423 Å, 7.587 Å, 8.253 Å, 90°, 100.419°, and 90°.Calculating the Raman spectrum for that relaxed unit cell resulted in a clearly worse agreement with the experimental spectrum, as shown in Figure S7.We attribute this to thermal expansion effects neglected in the DFT-relaxed cell, which corresponds to a T→0K structure.The resulting too small unit cell causes a shift of all Raman peaks to too high wavenumbers.Supporting Information

Anharmonicities
Schweicher et al. found strong anharmonicities in C8-BTBT-C8 and C8-DNTT-C8, which prompted us to investigate this matter also for OEG-BTBT. [1]We analyzed possible anharmonicities of the lowfrequency modes in the same fashion as we did in the SI of reference 2: [2] We displaced the structure along the mass-weighted eigenvectors of an eigenmode.The magnitude of this displacement is given by the displacement distances .It was chosen to range from -3 to +3 Å with 60 steps.A single point  calculation was performed for each displacement distance with the data shown as blue dots in . They are compared to the shapes of the harmonic approximation of the PES derived from the harmonic frequencies of the eigenmodes , whereby the composite index denotes the band index and the reciprocal lattice vector.These frequencies were obtained from the lattice dynamics calculations using phonopy as described in the main text. [3]The energies as a function of the displacement distance are then given by where is the effective mass of the mode .This equation corresponds to a term in the Hamiltonian

𝜇 𝜆 𝜆
of the quantum harmonic oscillator equation, when writing it as a superposition of independent onedimensional harmonic oscillators.The effective mass was calculated as described in the SI of reference [4] with the eigenvector and the mass of each atom .The sum runs over all atoms . is shown as a solid orange line in Figure S8.Its minimum is aligned with the energy at the equilibrium structure and defined as the zero of the energy axis.In the following, we focus on bands 4 to 7 at the Γ-point, which are the lowest optical bands and at the same time at  the lowest bands with non-zero energy.
As can be seen in Figure S8 for modes 4 and 5, the explicitly calculated curves lie close to the   () harmonic energies.For mode 6, the displaced structure at Q=-0.1 has a lower energy than the Q=0 structure by 0.035 meV.It appears that the explicitly calculated energy curve is shifted with respect to the curve for the harmonic calculation, which is similar to the observations in the SI of reference 2 for an -quinacridone crystal. [2]We attribute this small shift to the relaxation procedure, which for the chosen convergence criterion apparently yields an optimized structure that does not exactly correspond to the minimum with respect to Mode 6.We note in passing, that in both cases (OEG-BTBT and quinacridone), the mode corresponds to a rotation of the molecule.Most importantly, this shift in the origin of the displacement axis has only a negligible impact on the curvature of the potential and, thus, the eigenfrequency.As reference energy, we consider 25 meV, which is k B T at room temperature and in a classical picture equals the average total energy of a phonon mode.For all modes discussed so far, there are hardly any deviations between the explicitly calculated energy of the displaced structure and the energy obtained from the harmonic oscillator model at this reference energy (safe the rigid shift of the potentials for mode 6).The situation is different for mode 7 (a bending motion of the side chains): At very small displacements the orange line and the blue data points coincide, but already at displacements associated with an energy of 25 meV there are significant differences in the displacements suggesting non-negligible anharmonicities, e.g., at room temperatures.Still, this has no impact on the Raman spectra, as of the considered modes only modes 4 (at 20.45 cm -1 ) and 6 (at 31.58cm -1 ) are Raman active, whereas mode 5 (at 28.2 cm -1 ) and mode 7 (at 32.24 cm -1 ) are not.
Figure S8: Energy (relative to the energy of the equilibrium structure) versus eigenmode displacement Q for modes 4 to 7. The blue dots show the explicitly calculated energies, while the orange lines show the parabolic potential energy surfaces derived from eigenfrequencies within the harmonic approximation.The dashed black line indicates the thermal energy at room temperature (25 meV).We present here calculations for the experimental unit cell and note in passing that equivalent results have been obtained for the DFT-relaxed unit cell.

Figure
Figure S4: X-ray reflectivity measurements and corresponding fits of n-octadecyltrichlorosilane (ODTS) self-assembled monolayer deposited on SiO 2 substrate.

Figure S5 :
Figure S5: Convergence test of the energy per atom in terms of energy cutoff.The shown total energies per atom at the equilibrium geometry are relative to the energy at an energy cutoff of 1200 eV.The convergence criterion of 0.5 meV per atom is shown as a black dashed line.

Figure S6 :
Figure S6: Convergence test of the energy per atom in terms of k-mesh.The shown total energies per atom at the equilibrium geometry are relative to the energy with a k-mesh of 4×10×9.The convergence criterion of 0.5 meV per atom is shown as a black dashed line.

Figure S7 :
Figure S7: Experimental (orange) and simulated Raman spectra of OEG-BTBT.The calculations were performed for the experimental (blue) and fully DFT relaxed (purple) unit cells.For the simulated spectra, in the upper panel (downwards-pointing spectra), Raman-active modes are indicated by bars scaled with their Raman intensities.The lower panel illustrates the frequencies of all modes (including Raman-active and Raman-inactive ones).