Rational design of carbon-based metal-free electrocatalyst architectures for selective oxygen reduction reaction: From structure–performance correlations to predictive strategies

The escalating energy crisis and environmental pollution necessitate the development of clean energy technologies and remediation strategies, wherein the oxygen reduction reaction (ORR) plays a dual role. It is central to the efficiency of fuel cells and metal–air batteries and enables the green electrosynthesis of hydrogen peroxide (H₂O₂). However, the intrinsic competition between these pathways limits practical applications: The four-electron (4e⁻) pathway requires suppression of H₂O₂ byproduct, while the two-electron (2e⁻) pathway demands avoidance of O–O bond cleavage for high H₂O₂ selectivity. Carbon-based metal-free electrocatalysts (CMFECs) have emerged as promising candidates for regulating ORR selectivity due to their low cost, structural tunability, and excellent stability. This review focuses on the rational design of these catalysts through multiscale structural engineering, systematically discussing three primary strategies: heteroatom doping, defect engineering, and microenvironment modulation of active sites. We first elucidate the mechanistic origins of ORR activity and selectivity, introducing key theoretical descriptors based on the adsorption free energies of intermediates *OH and *OOH (ΔG_*OH and ΔG_*OOH). Subsequently, we provide an in-depth analysis of how specific structural motifs—such as pyridinic nitrogen, pentagonal defects, and hierarchical pores—modulate the local electronic structure and mass transport to govern reaction pathways. By integrating theoretical calculations with in-situ characterization, we aim to establish clear structure–performance relationships, shifting catalyst design from empirical trial-and-error toward rational paradigms. Finally, we highlight the frontier applications of these catalysts in energy storage, H₂O₂ production, and in-situ environmental remediation. Current challenges and future directions, including advanced characterization, multiscale simulations, and machine-learning-assisted design, are also discussed.