Database-driven divergence in carbon footprint estimates for Chinese refined copper supply chains

Refined copper is critical to the global energy transition, yet carbon footprint estimates for Chinese primary production vary because background databases represent concentrate supply, electricity systems, and process conditions differently. Using 2025 operational data from two Chinese smelters, this study developed cradle-to-gate inventories for Ausmelt-based and double-flash production systems and evaluated them across five software-database-modeling configurations. Across these configurations, the observed Ausmelt case yielded 5.9-7.1 t CO2e/t Cu, whereas the double-flash case yielded 4.6-6.3 t CO2e/t Cu. Because feed composition and operating conditions were not normalized, this contrast represents the sampled plant systems rather than an intrinsic technology effect. Upstream concentrate production was the main source of inter-configuration divergence. A controlled parameter-transfer check reproduced the SimaPro result in OpenLCA within 0.008%, indicating transfer fidelity rather than native cross-software equivalence; the estimated OpenLCA-ecoinvent 3.11 result was 1.31% lower than the reconstructed ecoinvent 3.10 case. Paired Monte Carlo analysis confirmed the case-level ordering and showed that inter-configuration dispersion exceeded within-configuration parameter uncertainty. After consistently rebasing the electricity-factor trajectory, combined process-efficiency improvement and grid decarbonization reduced modeled electricity-related emissions by approximately 19% in 2030 and 96% in 2060. Oxygen-production efficiency provided additional reductions, although long-term mitigation remained dominated by grid decarbonization. These findings identify low-carbon concentrate supply, efficient air separation, grid decarbonization, and regularly updated localized databases as priorities for comparable refined-copper carbon footprints.