Organizational Deep Learning Capability and Manufacturing Quality Intelligence in Integrated Circuit Manufacturing
Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Deep learning is increasingly used in integrated circuit manufacturing, but its contribution to quality management depends on how organizations turn analytical outputs into usable information. This study examines the associations among deep learning application capability (DLAC), quality inspection effectiveness (QIE), quality risk identification capability (QRIC), and manufacturing quality performance (MQP). Drawing on organizational information processing theory and a dynamic capability perspective, we propose that DLAC is associated with MQP through the generation of reliable inspection information and its interpretation as quality risk knowledge. A cross-sectional survey of professionals in integrated circuit and related semiconductor manufacturing in mainland China yielded 384 valid responses. Confirmatory factor analysis supported the four-construct measurement model. Covariance-based structural equation modeling showed positive associations between DLAC and QIE (β = 0.54, p < 0.001), DLAC and QRIC (β = 0.25, p < 0.001), QIE and QRIC (β = 0.45, p < 0.001), and QRIC and MQP (β = 0.47, p < 0.001). The sequential indirect association from DLAC to MQP through QIE and QRIC was 0.114 (95% bootstrap confidence interval, 0.073--0.164). These findings suggest that organizational use of deep learning is associated with perceived manufacturing quality performance through two connected information-processing capabilities. The cross-sectional, self-reported data do not establish causal effects or changes in objectively measured production quality.
Source: Research Square
Author: RongHao Zhang
Institution: Beijing University of Posts and Telecommunications
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7225886576
