The use of a-SiC:H thin-films alloys in p-i-n (a-SiC:H(p)/a-Si:H(i)/a-Si:H(n)) photovoltaic components has widely attracted the research interest, especially towards the direction of optimizing the optoelectronic attributes of a-SiC:H that aim to the exploitation of a wider range of the solar spectrum. A major effort towards this direction concerns the increase of the carbon concentration in the a-SiC:H thin-films alloys, which causes a consequent increase of their optical energy-band gap up to a critical ceiling value, while afterwards the energy-band gap decreases. However, the increase of carbon concentration induces to the alloys structural disordering that is related to subsequent downgrade of their optoelectronic attributes. Therefore, another direction for the photovoltaic performance optimization regards the control of the hydrogen concentration to the a-SiC:H thin-films alloys composition, which results to better structures in comparison with the carbon-control techniques. Aiming at promoting the multidisciplinary demand of modern material physics, this paper broadens the conceptualization of complex network analysis of time-series by applying the visibility graph algorithm (NVG) to a DC activation-energy instead of time-series data, to study the performance of a-SiC:H thin-films alloys, as a function of temperature and hydrogen flow. Laboratory experiments are conducted to measure the dependence of structural, optical, and electrical properties of a-SiC:H thin films, which are deposited by using the RF-sputtering process on the substrate temperature, for different hydrogen flow rates, so that to gain the optimum material quality. The available data of the study regard measurements of the DC activation-energy (Ea) of a-SiC:H thin-films alloys, which were extracted from four distinct temperature flows referring to 0, 9, 14, and 20sccm (standard cubic centimeters per minute), each configuring a temperature curve calibrated at 30, 100, 120, 140, 160, 180, 220, 250, 270, 290, 300, and 320oC.
The methodological framework is composed by a multilevel analysis consisting of three steps; the first is based on statistical inference analysis, the second on pattern recognition, and the third on network analysis. The statistical inference analysis revealed that at zero hydrogen-flow levels, the dc activation-energy is statistically indifferent to the temperature, whereas, for non-zero hydrogen-flows temperature affects the semiconductor’s structure. Also, the application of non-zero hydrogen-flows can statistically change the levels of DC activation-energy, but variations amongst DC activation-energy levels cannot be considered as statistically significant. In a pattern recognition approach, the supply of hydrogen-flow to the semiconductor body led to better semiconductor structures at lower temperatures, where a zone of 17-20sccm appeared with the better activation energy levels. The network analysis of the visibility graphs revealed a rich-club configuration at a temperature range [120,180]oC and three distinct DC activation-energy states, which correspond to different structural (semiconductor) behaviors of the a-SiC:H thin-films alloys. The overall analysis provided insights of dealing with multivariate structural analysis, within the context of insufficient information.