The escalating global CO₂ emissions from industrial processes and power generation demand urgent development of low-carbon technologies. Electrochemical CO₂ reduction (CO₂RR) offers a sustainable pathway to convert CO₂ into valuable fuels and chemicals, but its practical application is hindered by the complexity of real-world feedstocks, particularly flue gas [1]. Flue gas, a mixture of CO₂ (4–15%), nitrogen, oxygen (1–15%), and contaminants, presents significant challenges for CO₂RR due to its low CO₂ concentration, which limits solubility and mass transport, and the presence of O₂, which competes with CO₂RR via the oxygen reduction reaction (ORR), drastically reducing Faradaic efficiency even at trace levels [2].

Current strategies to mitigate these issues include hydrated ionomer coatings and polymers of intrinsic microporosity (PIMs), which selectively hinder O₂ transport while promoting CO₂ reduction [3]. However, these approaches face limitations in long-term stability and precise control over the local reaction microenvironment. Direct reactive capture (DRC) methods, such as amine scrubbing and carbonate/bicarbonate electrolytes, eliminate the need for energy-intensive CO₂ purification but suffer from poor selectivity and hydrogen evolution, impacting efficiency [4].

Non-aqueous systems, particularly those using aprotic or non-nucleophilic solvents, emerge as a promising alternative [5]. Binary solvent electrolytes, combining an aprotic solvent with a weak proton donor (e.g., water or ethanol), enable fine-tuning of the local microenvironment, enhancing CO₂ solubility and suppressing hydrogen evolution reaction (HER). These systems maintain selective CO₂ reduction pathways while providing controlled proton availability, addressing the dual challenges of low CO₂ concentration and O₂ interference.

This study introduces a binary solvent-engineered strategy for CO₂ electroreduction directly from O₂-containing flue gas. By systematically tuning the proton donor environment in acetonitrile-based electrolytes, we establish a linear correlation between HER Faradaic efficiency and hydrogen-bond donating (HBD) ability, while ORR selectivity decreases exponentially with reduced HBD strength. Theoretical calculations confirm that HBD strength moderates HER activation but strongly suppresses ORR, aligning with experimental observations. Proton-free environments, such as the dimethyl sulfide/acetonitrile (DMS/ACN) system, effectively suppress both HER and ORR due to disrupted hydrogen-bond networks. Under moderate pressures, DMS/ACN achieves near-quantitative Faradaic efficiencies for CO production even at 1% CO₂ and up to 15% O₂, with operational stability exceeding 100 hours and an energy consumption of 30.7 GJ ton⁻¹ CO. Coupled with a high-efficiency triple-junction solar cell, the system attains a solar-to-fuel conversion efficiency of ~5.5%, comparable to aqueous systems using pure CO_2r. This approach not only advances CO₂RR toward industrial relevance but also provides new mechanistic insights into selective electrocatalysis under realistic conditions.