Fabrication of Amorphous Silicon–Carbon Hybrid Films Using Single-Source Precursors

The aim of this study was the preparation of different amorphous silicon–carbon hybrid thin-layer materials according to the liquid phase deposition (LPD) process using single-source precursors. In our study, 2-methyl-2-silyltrisilane (methylisotetrasilane; 2), 1,1,1-trimethyl-2,2-disilyltrisilane (trimethylsilylisotetrasilane; 3), 2-phenyl-2-silyltrisilane (phenylisotetrasilane; 4), and 1,1,2,2,4,4,5,5-octamethyl-3,3,6,6-tetrasilylcyclohexasilane (cyclohexasilane; 5) were utilized as precursor materials and compared with the parent compound 2,2-disilyltrisilane (neopentasilane; 1). Compounds 2–5 were successfully oligomerized at λ = 365 nm with catalytic amounts of the neopentasilane oligomer (NPO). These oligomeric mixtures (NPO and 6–9) were used for the preparation of thin-layer materials. Optimum solution and spin coating conditions were investigated, and amorphous silicon–carbon films were obtained. All thin-layer materials were characterized via UV/vis spectroscopy, light microscopy, spectroscopic ellipsometry, XPS, SEM, and SEM/EDX. Our results show that the carbon content and especially the bandgap can be easily tuned using these single-source precursors via LPD.


Thin Layer Materials Deposition and Solution Parameters
Thin layer materials were spin coated on glass substrates (25 x 25 mm).Prior to the spin coating, the substrates were cleaned with deionized water, acetone and sonication in isopropanol (2 h).For all compounds different solutions were made.The optimized conditions can be seen in Table 1.The spin coating speed for depositing the layers was varied from 3000 rpm to 9000 rpm at 10-20 s.Further spin coating parameters and their optimization are depicted in Table 2.The successfully spin coated materials were then deposited at 500 °C to achieve amorphous Si/Si-C layers.Optical Properties UV/Vis-spectroscopy was recorded with an UV/Vis-spectrometer Lambda 35 from Perkin Elmer.The settings can be seen in Table 3.The absorption spectra were measured in thin films on glass substrates after spin coating and deposition at 500 °C.
Table S4.Settings for absorption measurements of thin layer materials.

Start wavelength [nm] 800
End wavelength [nm] 300 igure S31.Ellipsometric parameter  and  of thin layer MeISO.The dotted green and blue lines show the measured values, the red curves show the fitted calculated data.The model used for the calculation consisted of a pole site in the UV outside the measured spectral range, a DC offset, an (asymmetric) Cody-Lorentz-as well as a much smaller (symmetric) Gaussian broadened-oscillator in the UV-VIS-NIR range and a Drude term for the long-wavelength absorption in the IR.

TMSISO -Generated and Experimental
igure S32.Ellipsometric parameter  and  of thin layer TMSISO.The dotted green and blue lines show the measured values, the red curves show the fitted calculated data.The model used for the calculation consisted of a pole site in the UV outside the measured spectral range, a DC offset, an (asymmetric) Cody-Lorentz-as well as a much smaller (symmetric) Gaussian broadened-oscillator in the UV-VIS-NIR range and a Drude term for the long-wavelength absorption in the IR.

NPO -Generated and Experimental
igure S33.Ellipsometric parameter  and  of thin layer NPO.The dotted green and blue lines show the measured values, the red curves show the fitted calculated data.The model used for the calculation consisted of a pole site in the UV outside the measured spectral range, a DC offset, an (asymmetric) Cody-Lorentz-as well as a much smaller (symmetric) Gaussian broadened-oscillator in the UV-VIS-NIR range.Here, no Drude term was necessary to describe the IR part of the absorption.

PhISO -Generated and Experimental
igure S34.Ellipsometric parameter  and  of thin layer PhISO.The dotted green and blue lines show the measured values, the red curves show the fitted calculated data.The model used for the calculation consisted of a pole site in the UV outside the measured spectral range and a DC offset.No absorption could be found for this layer.It is to say that the very small thickness of this layer additionally to general restrictions of ellipsometric sensitivity to small absorption sets a strong limit for absorption determination.
Elemental analysis and further images of the surface layer were obtained by Freskida Goni via SEM/EDX.SEM micrographs and EDX spectra were collected using Tescan VEGA 3 SEM (Oxford Instruments plc, Abingdon, United Kingdom) with tungsten source filament working at 20 kV.For SEM imagines, a resolution of 5 µm and a working distance of around 15 mm was used, and the EDX spectra were collect4ed at a 0-10 keV scale.Prior the analysis, the samples were sputter-coated with gold.Table 5 summarizes the obtained data in aspect of elemental composition of the different layers.Small amounts of different elements were omitted.The amount of oxygen was omitted as well, to be able to compare the silicon and carbon amount respectively Additionally, images of the surface area of the different layer materials were obtained.The high carbon values are assigned to carbonaceous compounds such as CO 2 etc., which were adsorbed to the surfaces during storage under atmospheric conditions.The adsorption of small molecules from the atmosphere is a well-known phenomenon in surface analysis.The origin of the carbon signal becomes obvious by considering the NPO sample, on which 19.5 at% C were detected although the thin layer investigated consisted of a neopentasilane oligomer.

Computational Methods
The optimization of conformations for 1-5 was performed with the composite DFT method PBEh-3c. [1]In this method, the electronic PBE0 energy is corrected by Grimme's D3BJ dispersion correction, and the geometrical counterpoise energy (gCP) is added for correction of the basis set superposition error.The basis set def2-SVP is applied in this method.The geometries were checked to be minima at the potential energy surface by calculation of the harmonic frequencies.The calculations were performed in the solvent n-hexane using the conductor-like polarizable continuum model (CPCM) for solvation. [2]Timedependent DFT (TD-DFT) was applied for the computation of 15 vertical excitations, and the UV/Vis spectrum was then simulated from these data by Gaussian broadening of the peaks with a half width at half height of 10 nm using the program gabedit.Molecular orbitals were generated by the program Avogadro, [3] and the UV spectrum was simulated by the program gabedit. [4]The program ORCA 5.0.3 was used for all DFT calculations. [5]mulated UV/Vis Spectra and Molecular Orbitals The simulated UV/Vis spectra of compounds 1-5 (Figure S41) are systematically blue-shifted compared to the experimental spectrum, which is a known behavior of the PBEh-3c functional.Nevertheless, general features of the computed spectra compare well with the experiment, and the character of the bands can be explained by interpreting the molecular orbitals (see Table S8 and Figure S40).In general, the S1 band is very weak with the exception of compounds 2 (R=Me) and 5 (Ring structure).The high-intensity band as well as the S1 bands are interpreted in the following text.
Compounds 1 (R=SiH 3 ) and 3 (R=SiMe 3 ) provide similar spectra, although the very weak S1 band is red-shifted in compound 3 relative to 1 by ca. 10 nm, which is rationalized by the SiMe 3 instead of SiH 3 group.The same effect can be seen when methyl is introduced in compound 2 (R=Me), where the main effect is also a higher intensity of the bands.The ring compound 5 increases this behavior, and the high-intensity of the HOMO-LUMO band also from a σσ* transition of the ring system.Additionally, this band has approximately the same wavelength as in compound 2. Compound 4 (R=Ph) shows a different characteristic, as both S1 and S2 bands consist of linear combination of the frontier orbitals, where the HOMO-LUMO excitation is the second S2 band with high intensity.Because the respective orbitals are localized at the phenyl substituent only, the expected high intensity of a normal * transition occurs in compound 4 at the intense S2 band, which is strongly shifted to the red relative to compound 2.
Table S8.Most important simulated PBEh-3c UV vertical excitations: wavelength (in nm), oscillator strength (italic, in brackets) and respective orbital contributions with squared LCAO coefficients c 2 (in brackets).The first S1 excitation as well as the most intense band is tabulated.H denotes HOMO, L denotes LUMO.

Figure S27 .Figure S28 .
Figure S27.Microscopic image with 200x magnification of the same deposited layer of 15 wt% 9 (left: homogenous; right inhomogenous parts of the thin layer material)

Figure S29 .
Figure S29.Tauc plot for determining the band gap of a thin film for 7.

Figure S30 .
Figure S30.Tauc plot for determining the band gap of a thin film for 8.
Figure 27-31 show the surfaces of the different thin layers.

Figure S41 .
Figure S41.Simulated UV spectra of compounds 1-5 (from top to bottom).To simulate the spectrum, 20 vertical excitations were calculated so that the bands fall off at the edge of the low wavelengths.

Table S3 .
Optimized spin coating parameters of the different solutions.

Table S5 .
Summary of the silicon, carbon and oxygen amount of different thin layer.

Table S6 .
Surface Composition of the thin layers determined by XPS measurements ...single value, detected on only one measurement spots on the sample *

Table S7 .
Elemental Analysis of the oligomers performed after photolysis.