Pristine Poly(para-phenylene): Relating Semiconducting Behavior to Kinetics of Precursor Conversion

We investigated unsubstituted poly(para-phenylene) (PPP), a long-desired prototype of a conjugated polymer semiconductor. PPP was accessed via thermal aromatization of a precursor polymer bearing kinked, solubility-inducing dimethoxycyclohexadienylene moieties. IR spectroscopy and Vis ellipsometry studies revealed that the rate of conversion of the precursor to PPP increases with temperature and decreases with film density, indicating a process with high activation volume. The obtained PPP films were analyzed in thin-film transistors to gain insights into the interplay between the degree of conversion and the resulting p-type semiconducting properties. The semiconducting behavior of PPP was further unambiguously proven through IR and transistor measurements of molybdenum trioxide p-doped films.

. a) Absorption and emission spectra of the precursor molecule in thin films. b) Absorption and emission spectra of the converted PPP in thin films. c) Tauc plot used to calculate the band gap of both materials. α denotes the absorption coefficient ν the photon frequency and h the plank constant. Both the here used Tauc plot and the more commonly used extrapolation of the absorption onset yield identical results.
Absorption measurements were done using a Jasco V-660 spectrophotometer while emission spectra were recorded with a Jasco FP-6500 spectrofluorometer.

S-3
Ultraviolet photoelectron spectroscopy | He1 line and secondary electron cutoff of a converted PPP film Figure S2. a)/b) Helium-1 spectra of a PPP precursor spincoated on a silicon substrate and cured for 120 min at 300°C. c) Secondary electron cut-off (SEC) of the same sample. The SEC has been measured with an applied bias of V=-3V by which the spectrum was shifted.

S-4
Comments on transistor fabrication and precursor coating techniques.
Initially several different device layouts were tested, yielding devices with unsatisfactory performance. Parylene-c was used as gate dielectric in a top gate architecture with gold as a contact material (as a bottom or top contact device). Furthermore the commercial dielectric material Cytop was tested in a bottom-contact top gate architecture (again with gold as contact material). We attribute the lack of sufficient device performance of these devices to the fact that both of these two dielectric materials do not possess a sufficiently high leak resistance in comparison to the necessary high voltages (breakthrough of the dielectric before device operation). We found no improvement for the mobility through a reduction of the channel length (minimum L=2.5 µm was investigated) in the devices presented in the main paper.
Coating techniques unsuccessfully investigated for the fabrication of ordered precursor layers:  Controlled evaporation of the precursor solution (a technique close to dip-coating). 1  Off-center spin coating.
 Conversion of P1 in melts of biphenyl, p-terphenyl, imidazole and pyrene.

S-5
Evolution of the transfer characteristics for repeated measurements Figure S3. a) Evolution of the transfer characteristic for repeated measurements with increasing drain voltage in a logarithmic plot (top) and in a square root plot (bottom). A clear decrease in turn-on voltage can be seen. Device was cured for 60 min at 300 °C (L=5 µm).   Figure S4. a) DFT calculated relative transmission spectrum of a PPP precursor layer on silicon. The used molecular structure is given in the inset. b) Experimental relative transmission spectrum of a 72 nm thick PPP precursor layer on silicon (black) and the corresponding SCOUT fit (red). The strongest experimental absorption bands in the fingerprint range marked with dashed lines are assigned to calculated vibrational modes (see Table SIR1) c) Modelled dielectric function (real part: green; imaginary part: orange) obtained from the fit in b) consisting of 47 Brendel oscillators and constant dielectric background of 2.71.   Figure S5. a) DFT calculated relative transmission spectrum of a PPP layer on silicon. The used molecular structure is given in the inset. b) Experimental relative transmission spectrum of a 55 nm thick PPP layer on silicon (black) and the corresponding SCOUT fit (red). The strongest experimental absorption bands in the fingerprint range marked with dashed lines are assigned to calculated vibrational modes (see Table SIR2) c) Modelled dielectric function (real part: green; imaginary part: orange) obtained from the fit in b) consisting of 43 Brendel oscillators and constant dielectric background of 3.07. Fitting procedure | IR spectra of intermediate annealing steps The IR spectra for intermediate annealing steps were fitted by using a layer-stackmodel in SCOUT, a software for spectral modelling. 2 The used layer-stack is shown in Figure S7 and consists of the substrate (silicon), a precursor layer and a PPP layer. The dielectric models used for the precursor and the PPP are shown in Figure  S4 and S5.  Figure S6. a) DFT calculated relative transmission spectrum of a PPP precursor layer on silicon. b) Experimental relative transmission spectra of 43 nm PPP precursor on silicon as cast and for varying annealing times at 300°C with corresponding fits (black). The fitted relative volumes of PPP for each annealing time are given in corresponding colors together with the used layer-stack-model on the right. c) DFT calculated relative transmission spectrum of a PPP layer on silicon.

S-9
Scanning electron microscopy | pictures of precursor layers with varying layer thickness  Figure S20).
AFM Measurements were carried out on a Bruker Nanoscope MultiMode VIII in the ScanAsyst PeakForce mode. The resulting images were treated with the software Gwyddion for row alignment, scar correction and plane leveling. 5*5 µm 2 scan Area 2*2 µm 2 scan Area Figure S9. Root mean square roughness (RMS) of precursors with varying film thickness calculated via AFM images. Two sets of AFM images of different scan size were recorded at different positions. Due to the overall small RMS its calculation is very sensitive to small defects, rendering the trend much less clear then the obvious decrease of voids seen from Figure S8. Defects have been mask to reduce this effect further.

Conversion monitoring | IR spectroscopy and UV-Vis ellipsometry
To clarify if the UV-Vis ellipsometric measurements, which are much faster and more facile compared to IR spectroscopy, can be used to determine the degree of precursor conversion, the normalized change in layer thickness Δdnorm. and refractive index Δnnorm. are plotted in Figure S9 and S10 against the fraction of PPP determined by IR spectroscopy. The clear linear dependence for both quantities reveals a strong linear correlation of the change in layer thickness as well as refractive index and the degree of conversion into PPP. This means the conversion process can easily be monitored using UV-Vis ellipsometry in combination with a simple Cauchy layer approach for data analysis.

S-15
Angular dependent IR Measurements Figure S13. a) Experimental relative transmission spectra of a 45 nm thick PPP precursor layer on silicon measured near normal (7°, black) and oblique (70°, red) incidence of light. b) Experimental relative transmission spectra of a 34 nm thick PPP layer on silicon (thin film from S12 a) heated for 1h at 300°C in N2) measured near normal (7°, black) and oblique (70°, red) incidence of light. Angular dependent IR measurements do not indicate a preferred orientation of the molecules neither in as processed nor in tempered films. Output curve of doped/undoped devices | Figure S14. Exemplary normalized output characteristic for a doped (solid) and undoped (symbols) device. The arrows mark a decrease in current in undoped devices. Sample fabricated from a 3,5mg/ml solution, annealed for 60min, L=10 µm. S-17 Device stability | Transfer curve of the same device 1.5years apart Figure S15. Transfer characteristic of the same transistor measured the first time on the 17. Jan 2017 (solid line). A repeated measurement on the 7. June 2018 resulted in near identical result. Sample fabricated from a 3,5mg/ml solution, annealed for 60min, L=10 µm. Note the strong hysteresis on the first sweep. This behavior can be attributed to the filling of interfacial traps at the Si/SiO2 interface.
Most devices measured 1.5years apart showed similar device operation with the exception of an increase of the change of device failure in the form of shorts. The mechanism behind this increased chance of device failure remains unclear.

S-18
Grazing-Incidence Wide-Angle X-ray Scattering (GIWAXS) Figure S16. GIWAXS of thin films of a) a pristine P1 film and b) a fully converted film (300h, 2h). Both films investigated show no signs of order.
GIWAXS experiments were performed by means of a solid anode X-ray tube (Siemens Kristallofl ex X-ray source, copper anode X-ray tube operated at 35 kV and 40 mA), Osmic confocal MaxFlux optics, X-ray beam with pinhole collimation, and a MAR345 image plate detector. The beam size was 0.5 × 0.5 mm, and samples were irradiated just below the critical angle for total reflection with respect to the incoming X-ray beam (≈0.10°). Intensity / a.u.
pristine P1 Figure S17. 1D-XRD of thin films of a) a pristine PPP film and b) a fully converted film (300h, 2h). Both films of the pristine material aswell as fully converted PPP show no signs of order.
1D XRD measurements were recorded using a Siemens D-500 powder diffractometer (Cu-Ka: 0.1541 nm) with a scan rate of 0.1/20 s. 2D-WAXS measurements were performed using a custom setup consisting of the Siemens Kristalloflex X-ray source (copper anode X-ray tube, operated at 35 kV/20 mA).

S-20
Atomic Force Microscopy (AFM) | Comparison of annealing length at 300°C Figure S18. AFM of thin films of the precursor annealed for different times at 300°C. I off V th Figure S19. Sqrt(Id) vs gate voltage plot of the doped device from Figure 7. A line was fitted to the linear part of the sqrt(Id) plot. Taking the current at 0 V gate voltage to be the off current allows now the calculation of the threshold voltage for the accumulation channel, reading around 70V. At voltages higher than Vth the current from the accumulation conduction follows the Shockley equation while at voltage below Vth it reads Ioff