The conversion of intermittent renewable energy resources in the form of chemical bond, such as hydrogen production from electrochemical water splitting, is a promising way to satisfy the future global energy demand and address the environmental issues. The oxygen evolution reaction (OER) is the bottleneck among the overall reaction due to its endergonic thermodynamics and complicated 4H+/4e- transfer process. Despite extensive efforts have been made to develop cheap, efficient and robust heterogeneous electrocatalysts for OER, their heterogeneous nature makes the deep investigation of the active sites and reaction mechanisms challenging. Benefiting from high atomic efficiency, excellent intrinsic activity, and well-defined active motifs, the atomically dispersed catalysts provides an opportunity to get insight of the catalysts. The Chapter 2 demonstrates a Co single-atom precatalyst immobilized on N-doped carbon support (Co-N-C) can be transformed to a Co-Fe double-atom catalyst (Co-Fe-N-C) via electrochemical activation in KOH containing Fe impurities. The Co-Fe-N-C exhibited a turnover frequency among one of the highest values of the state-of-the-art OER catalysts. Electrochemical, microscopic, and spectroscopic data, including atomic-resolution electron microscopy and operando X-ray absorption spectroscopy (XAS), reveal an atomically dispersed dimeric Co-Fe moiety as the active site of the Co-Fe-N-C. On the basis of Co-Fe-N-C, the Chapter 3 further describes a general synthesis of Co, Fe, and Ni-containing double-atom catalysts from their single-atom precursors via in situ electrochemical transformation. All these double-atom catalysts have molecule-like bimetallic active sites, which resembles the possible key active centers of bimetallic heterogeneous catalysts. In Chapter 4, a systematic mechanistic insight of the double-atom catalysts and the catalytic process were provided, by combining both electrokinetics analysis and operando XAS results. These mechanisms follow a similar O-O bond forming step and all exhibit bimetallic cooperation, while each mechanisms diverge in the catalytic site and the source of OH- for O-O bond formation as well as the order of proton and electron transfer. The works from Chapter 2-4 suggested that the double-atom catalysts not only provide an attractive molecular platform for fundamental studies of heterogeneous OER electrocatalysts, but also establish a bridge between homogeneous and heterogeneous OER catalysis. The traditional OER catalysts typically occurs via a mechanism involving four consecutive proton-coupled electron transfer (PCET) steps, which has a performance limit imposed by the scaling relationship of various oxygen intermediates. The Chapter 5 describes using operando Raman spectroscopy and electrokinetic analysis to study two active model OER catalysts, FeOOH-NiOOH composite and NiFe layered double hydroxide (LDH). The data that FeOOH-NiOOH operates by a bifunctional mechanism where the rate-determining O-O bond forming step is the OH- attack on a Fe=O coupled with a hydrogen atom transfer to a nearby NiIII-O site, whereas NiFe LDH operates by a conventional mechanism of four consecutive PCET steps. The bifunctional mechanism, hitherto only supported by theoretical computations, has the potential to circumvent the scaling relationship limit of conventional OER mechanism. Thus, this work provides the first experimental evidences to validate such novel mechanisms.