The direct ethanol fuel cell (DEFC) represents one of the most exciting future clean energy solutions in modern research, because ethanol can be sustainably produced from biomass, is relatively nontoxic and, most importantly, has a high energy density. [1][2][3][4][5][6][7][8] The exceptional energy density is due to the transfer of 12 electrons from ethanol during complete electrochemical oxidation (as opposed to six electrons from methanol or two from hydrogen). The practicality of such a device is contingent on its ability to selectively catalyze the total oxidation of ethanol to CO 2 . [9,10] However, the CO 2 selectivity in the current ethanol fuel cells is very low, and the main products are acetic acid (resulting in the transfer of only four electrons) and acetaldehyde (only two electrons) in most systems reported. [11][12][13] Herein, we address the origin of low CO 2 selectivity in the DEFC, arguably the most important question to be answered in the field, using first-principles calculations.The pioneering work on DEFCs can be traced back to the 1950s, [14] but it was not until later that the selectivity was comprehensively investigated with IR spectroscopy showing CO 2 to be a minor product.[15] Behm and co-workers [1] performed a thorough investigation on the selectivity under a wide range of conditions and employing a wide range of morphologies, and they convincingly showed that platinum catalysts exhibit selectivity towards CO 2 in the region of 0.5-7.5 %, which is far short of the selectivity needed for economic implementation of the technology. This problem has proven difficult to surmount empirically. Recent work has made significant progress in terms of activity and selectivity, [2] but further improvements are required. Theoretical studies have made considerable advances [2,[16][17][18] and identified the platinum monoatomic step as the most likely site for total ethanol oxidation and concluded that the close-packed surfaces are unsuitable. [17] Despite the extensive experimental and theoretical work that has been carried out, the inhibiting factors in CO 2 formation remain unclear. There are good reasons for this: 1) the catalytic reactions occur on solid surfaces in the presence of a solvent, resulting in a system that is complex to understand at the molecular level; and 2) electrocatalysts operate at an applied potential (i.e. bearing charge), leading to more complications. Hence, it is extremely difficult to characterize the molecular-level surface processes by using experimental techniques, and it is also a huge computational challenge to realistically model the system. Without a clear understanding of the issue, strategies to overcome the problem remain limited to trial and error.Although the selectivity problem has been identified, a fundamental understanding of the low CO 2 selectivity observed is still missing. In fact, the low selectivity of CO 2 in ethanol fuel cells is in contrast to the general consensus in chemistry: CO 2 is significantly more stable than the major products acetic...