Multi-layer patent citation networks: mapping technological trajectories and assessing lock-in risks through main path analysis

In the context of globalised innovation, patent citation networks are essential for tracing technology diffusion and evolution. This paper constructs a multi-layer citation network based on patent families and proposes a novel main path analysis (MPA) approach to explore semiconductor development. A cross-layer weighting method is introduced to account for citations across patent offices, with weights adjusted by citation frequency to better reflect global knowledge flow. Using data from China, the US, Europe, and Japan, the study uncovers hierarchical dependencies in innovation diffusion. Main-path nodes concentrate in IPC group H01L21 (semiconductor manufacturing), with Japan and the US leading in citations and patent counts. Same-layer citations dominate (mostly above 70%), showing strong domestic path dependence and weak global diffusion. Stable technological breadth, slower geographic diversification, and the dominance of CNIPA, USPTO, EPO, and JPO highlight risks of lock-in, underscoring the need for cross-regional integration and process diversification.