Node-of-Influence Network Analysis for Targeted Condition Sequencing in Plasma Chemical Reaction Networks

Abstract

It was demonstrated that a plasma chemical reaction system can be represented as a directed bipartite variable-relationship (VR) graph to accurately represent node influence, and that targeted node-influence analysis of this graph can determine improved condition sequencing for a target outcome. A novel graph algorithm (OCARINA) was adapted for VR-graphs to give a measure of the net-influence of incremental increase (NIII) of one variable on another at any depth in the graph. Additionally, two conventional node-influence measures, the Eigenvector Centrality Index (ECI) and Katz Centrality Index (KCI), were also trialled on the VR-graph. The electron energy (ε) node influence was evaluated on a “baseline” continuous sequence of 10 ns 1 eV pulses in a 0D chemical-kinetic simulation using ECI, KCI and OCARINA NIII at three depths. KCI appeared to give meaningful values for ε influence in the whole graph but not on specific nodes, ECI gave no meaningful results. OCARINA (O3 targeted) suggested each successive ε pulse had diminishing influence on O3 formation, though analysis differed for each NIII depth. O3 concentrations in simulations with different numbers of ε pulses decreased with each additional pulse, correlating with the OCARINA analysis. O3, NO and O Species production in simulations of two ε pulses with one or both pulses incrementally changed by 10% from the baseline also largely agreed with the OCARINA results for each species on a baseline simulation of two consecutive 1 eV electron energy pulses. Additionally, it was found the NIII at antecedent depths corresponded to effects in subsequent phases in simulations.