Understanding Advanced Transition Metal‐Based Two Electron Oxygen Reduction Electrocatalysts from the Perspective of Phase Engineering

Non‐noble transition metal (TM)‐based compounds have recently become a focal point of extensive research interest as electrocatalysts for the two electron oxygen reduction (2e− ORR) process. To efficiently drive this reaction, these TM‐based electrocatalysts must bear unique physiochemical properties, which are strongly dependent on their phase structures. Consequently, adopting engineering strategies toward the phase structure has emerged as a cutting‐edge scientific pursuit, crucial for achieving high activity, selectivity, and stability in the electrocatalytic process. This comprehensive review addresses the intricate field of phase engineering applied to non‐noble TM‐based compounds for 2e− ORR. First, the connotation of phase engineering and fundamental concepts related to oxygen reduction kinetics and thermodynamics are succinctly elucidated. Subsequently, the focus shifts to a detailed discussion of various phase engineering approaches, including elemental doping, defect creation, heterostructure construction, coordination tuning, crystalline design, and polymorphic transformation to boost or revive the 2e− ORR performance (selectivity, activity, and stability) of TM‐based catalysts, accompanied by an insightful exploration of the phase‐performance correlation. Finally, the review proposes fresh perspectives on the current challenges and opportunities in this burgeoning field, together with several critical research directions for the future development of non‐noble TM‐based electrocatalysts.


Introduction
−7] Its pivotal role as a disinfectant against SARS-CoV-2 during the COVID-19 pandemic, as acknowledged by the World Health Organization, highlights its significance. [8]Moreover, H 2 O 2 is also a clean fuel due to its high energy density and can be effectively used as both fuel at the anode and oxidant at the cathode in a H 2 O 2 fuel cell. [9,10]The global market of H 2 O 2 is poised to reach a substantial worth of US$ 4.1 billion by 2028, with a predicted compound annual growth rate of 3.8% from 2023−2028. [1]−13] An alternative method involves on-site production of H 2 O 2 through the direct mixing of gaseous H 2 and O 2 over noble metal-based catalysts, however, this poses safety challenges, necessitating additional security measures to prevent H 2 explosions. [11,14]−16] Nevertheless, ORR exhibits a thermodynamical preference to follow a four electron (4e − ) pathway, leading to the undesired product H 2 O. [17] Both 2e − ORR and 4e − half reactions of ORR play pivotal roles in electrocatalysis.While the 4e − ORR is known for its higher theoretical energy efficiency due to the involvement of more electrons in the reaction, it is important to note that the product attained from the 2e − ORR (i.e., H 2 O 2 ), holds distinct advantages over that of the 4e − ORR, which yields H 2 O. Furthermore, the integration of 2e − ORR with hydrogen oxidation in fuel cells presents promising economic prospects, facilitating the simultaneous green synthesis of H 2 O 2 alongside electricity generation. [13,15]This integration holds potential for efficient distributed generation and energy storage systems.During periods of electricity scarcity, 2e − ORR in the fuel cell reactor can generate H 2 O 2 and power at the same time.Conversely, excess electricity can catalyze the conversion of H 2 O 2 into H 2 and O 2 for energy storage. [15]Therefore, 2e − ORR electrocatalysis emerges as the preferred choice, not only due to the superior value of its products but also in line with the concept of the electrochemical H 2 O 2 cycle.
−20] As abovementioned, a couple of pioneering works have identified noble metals or alloys (e.g., Pt-Hg [21] and Pd-Hg) [22] as highly effective 2e − ORR electrocatalysts for the electrosynthesis of H 2 O 2 with both satisfactory activity and selectivity, but the prohibitive cost and scarcity of noble metals hinder their practical application. [23]Within this context, carbon nanomaterials with low cost have been extensively investigated as an alternative to noble metal-based electrocatalysts and great progress has been made in recent years. [3,24,25]Nevertheless, the vast majority of documented carbon nanomaterials exhibit optimal performance exclusively in alkaline environments.−28] Non-noble transition metal (TM, e.g., Fe, Co, Ni, Mn, Cu)-based compounds (e.g., oxides, chalcogenides, phosphides, and single atoms) feature several appealing merits including earth abundance, low cost, benign environmental compatibility as well as tunable composition and structure, showing great potential as advanced 2e − ORR electrocatalysts. [14,29]In this context, the paradigm shift toward investigating TM-based compounds as advanced 2e − ORR electrocatalysts has begun in the quest for sustainable and efficient H 2 O 2 production.
−36] In this regard, a considerable amount of exploratory work has recently begun to focus on modifying the phase structure of TM-based compounds for efficient 2e − ORR.Several strategies for phase structure modification such as elemental doping, defect creation, heterostructure construction, coordination tuning, crystalline design, and polymorphic transformation, have been proposed and demonstrated to be effective in improving the 2e − ORR performance of TM-based compounds.The outlined strategies are collectively referred to as phase engineering, by which the catalytic sites, mechanisms, and behaviors are capable of being contrapuntally regulated at the atomic level.−44] To this end, a timely summary of this encouraging work in this field is of great significance for guiding the development of improved electrocatalysts for 2e − ORR.While several reviews in recent years have explored reaction mechanisms, electrocatalyst categories, and reaction electrolyzers for 2e − ORR have been covered in recent years, [11,12,14,23,26,29,45−54] there remains a notable gap-a systematic review dedicated to elucidating TM-based 2e − ORR electrocatalysts from the perspective of phase engineering.In view of this, our contribution offers a comprehensive review of phase engineering strategies for TM-based electrocatalysts and delves into details of elemental doping, defect creation, heterostructure construction, coordination tuning, crystalline design, and polymorphic transformation (Figure 1).Moreover, this review provides an introduction to the fundamental principles of electrocatalytic 2e − ORR and explicates the scientific significance of phase engineering, establishing the dependency of performance (selectivity, activity, stability) on phase.Additionally, it emphasizes the pressing challenges and outlines open research questions in the future development of TM-based electrocatalysts.This contribution aims to serve as a comprehensive resource, shedding light on the cutting-edge field of phase-engineered TM-based electrocatalysts for 2e − ORR.

Fundamentals of 2e− ORR
Electrocatalytic ORR involves the electrochemical reduction of O 2 molecules, resulting in the production of H 2 O 2 and H 2 O through pathways involving four and two electrons, respectively. [48]pecifically, in 2e − ORR electrocatalysis, the reaction mechanisms are dependent on the pH values of the electrolyte, as listed below: [29,46,48,52,55]  In acid media: In details: * +O 2 + ( H + +e − ) → HOO * (1a) In basic media with a pH value > 11.6, the product is in the form of HO 2 − : In details: * +O 2 + H 2 O + e − → HOO * + OH − (2a) HOO * +e − → HO 2 − + * (2b) In the above reactions, * is denoted as the active sites on the catalysts for ORR.Moreover, it is important to mention that for electrocatalytic 2e − ORR, some undesirable competing or side reactions probably also take place, which severely decrease the H 2 O 2 production ability: i) 4e − ORR to produce H 2 O in acid media: In basic media: ii) Further electroreduction of the H 2 O 2 product from 2e − ORR in acid media: In basic media: iii) Spontaneous disproportionation of the H 2 O 2 product from 2e − ORR in acid media: 2H 2 O 2 → 2H 2 O+O 2 (7)   In basic media: From the above reaction mechanisms, it is clear that 2e − ORR tends to generate HO 2 − in the basic electrolytes, which could be interpreted based on the previously proposed electron transfer mechanisms within the varied double layer structures at the electrode (catalyst)/electrolyte interface.It is important to note that while the concentration of both acid and basic electrolytes used for ORR is usually ≥0.1 m, the focus remains on the Helmholtz layer, neglecting influences from the diffusion layer.The operational potentials for ORR are typically positive against the potential of zero charges.Within these situations, in the acid solutions, solvent water dipoles, chemisorbed O 2 (denoted as *O 2 ), as well as specifically adsorbed OH − from the dissociation of H 2 O (denoted as *OH − ) constructs the inner-Helmholtz plane (IHP), while the solvated O 2 molecules and electrolyte anions, as well as hydronium ions (H 3 O + ) occupy the outer-Helmholtz plane (OHP).When using a TM-based catalyst for ORR, the first electron transfers from the active TM site to the *O 2 , and the adsorbed superoxide species (denoted as *O 2 •-) form.Subsequently, the protonation of the *O 2 •-takes place after the proton of the H 3 O + migrates from OHP to the IHP, yielding the key ORR intermediate (*HOO).Due to the high mobility of protons in acidic media, this step and the following protonation are not rate limiting.Consequently, the formation of the *HOO intermediate is rapid, hindering its timely desorption.As a result, the H 2 O tends to be the final product of ORR for those TM-based catalysts.This corresponds to the term of inner-sphere electron transfer mechanism.In the case of basic electrolytes, the double layer structure at the TM-based catalyst/electrolyte interface changes.The H 2 O molecules serve as both solvent and proton source required for ORR.The IHP in alkaline electrolytes consists of solvent water dipoles, *O 2 , and *OH − (i.e., the adsorbed electrolyte anions), while the well-solvated alkali metal ions and solvated O 2 molecules populate the OHP.The abovementioned inner-sphere electron transfer could still occur, except that the H + for the protonation of the *O 2 •-mainly comes from the neighboring co-adsorbed water molecules.However, the strong alkaline media features a slow proton transfer rate from water, and this possibly impedes follow-up protonation, thereby hampering the complete four electron/proton transfer required for H 2 O formation during ORR.Moreover, it's noteworthy that an outer-sphere electron transfer likely predominates for TM-based catalysts in this environment.Specifically, solvated molecular O 2 in the OHP could interact with surface *OH − through a hydrogen bond with very low energies (< 35 kJ mol -1 ), facilitating thermodynamically feasible electron transfer from TM sites to the solvated O 2 in the OHP.Hence, the solvated O 2 •-in the OHP forms, which might then undergo desolvation, followed by being adsorbed on the catalyst surface and protonated, finally forming the adsorbed HO 2 -anion.Generally, in alkaline aqueous electrolytes, the TM is rapidly covered by OH − forming a passivated TM layer that favors the desorption of HO 2 -as the eventual product. [46,56]Herein one should note that although hydrogen bonds could also form between the *OH − in IHP and the solvated O 2 in OHP in the acid media, their interaction energy may not facilitate an outersphere electron transfer. [56]Nevertheless, it is essential to highlight that the current research on pH-dependent ORR activity, selectivity, and stability for TM-based catalysts is relatively scarce, suggesting that exceptions are likely to exist, which deserve more attention.
Based on the above discussion, the key to stabilizing the efficient production of H 2 O 2 is to suppress the undesirable competing 4e − ORR and disproportionation reactions.Notably, there are already mature methods in the industry to relieve the disproportionation reactions, such as the fleet stabilization of products through the addition of silicates or magnesium salts. [57]Hence, the pivotal point to elevate the efficiency of 2e − ORR is to maintain its satisfactory activity, and simultaneously inhibit or even eliminate the occurrence of the competing 4e − ORR, while this is highly decided by the property of electrocatalysts.Note that in order to design an ideal ORR catalyst with exceptional capability to electrosynthesize H 2 O 2 , different relevant influencing factors should be taken into account collectively.

The Evaluation Protocols and Criteria on the 2e− ORR Performance
Unlike the well-established protocols for 4e − ORR, there has been no consensus among the broader scientific community regarding specific protocols proposed by a particular institution or agency for the exclusive evaluation of 2e − ORR performance.However, a couple of previous reports have put forward reliable protocols. [46,51,58]Drawing upon these suggested criteria and insights gleaned from our own experimental experiences, we especially recapitulate a series of important points as follows.
The assessments on 2e − ORR electrocatalysis should take both rotating ring-disk electrode (RRDE) and bulk electrolysis (H-cell or flow cell) measurements into account.The former is facile, repeatable, and time-saving, suitable for the preliminary screening of the catalyst performances.When used for the ORR test, the disk current (I disk , from ORR into H 2 O 2 or H 2 O) and ring current (I ring , from the oxidation of in situ evolved H 2 O 2 ) are obtained.In most cases, by comparing the onset potential among the probed catalysts, the activity can be identified, where a more positive onset potential indicates superior activity. [46]Moreover, a highly active catalyst should exhibit an onset potential quite close to the theoretical equilibrium potential.In most studies, the onset potential is defined as either the potential corresponding to 5% of the diffusion-limited current, [59,60] or a current density of 0.1 mA cm −2 . [61,62]dditionally, the reaction kinetics is also an important parameter to evaluate the catalyst activity, and a lower Tafel slope means a more favorable kinetics.It can be visualized by the Tafel slope derived from the kinetic current density (J k ), that is: [63] where E, b, and c correspond to applied voltage, Tafel slope value, and constant, respectively.Besides, the J k can be obtained from the Koutecky-Levich diffusion equation: [64] 1 At the same time: [64] 1 where J m and J d represent measured current density and limiting diffusion current density, respectively.Besides, n, F, C 0 , D O2 , v, and w corresponds to electron transfer number, Faraday constant, saturated O 2 concentration in the electrolyte, O 2 diffusion coefficient in the electrolyte, kinetic viscosity of the electrolyte, and rotation rate of RRDE electrode, respectively.Furthermore, the number of active sites and the intrinsic activity are other two aspects that would determine the presented activity of the probed samples, which can be reflected by their electrochemically active surface area (ECSA) and turnover frequency (TOF), respectively. [46,51]The former is proportional to the double-layer capacitance (C dl ) that can be obtained by typical cyclic voltammetry (CV) within the non-Faradaic potential region.If the provided value of C dl (ECSA) increases (decreases), it can be concluded the surface active site number accordingly rises (drops). [4,65]However, the quantification of the exact surface ORR-active sites poses a considerable challenge, primarily due to the complexity of identifying active sites for the targeted catalyst surface, especially those accessible during the ORR electrochemistry. [51]Recently, for a couple of metal-based catalysts in atomic site or nanoparticle forms, the use of specific poisoning molecules (e.g., SCN − ) with a strong selective affinity toward those metal centers has been used as a promising approach.This method indeed calculates the surface active site number by estimating the amount of such molecules attached to the active metal sites. [51,66,67]However, such methods are only effective when the poisoning molecules render all the blocked sites inactive.Accordingly, determining the exact number of active sites remains elusive for the majority of other types of 2e − ORR electrocatalysts.Furthermore, as the amount of active sites is an inevitable parameter to calculate the TOF value, the quantification of poisoning molecules is often inapplicable for most catalysts, leading to challenges in accurately determining their TOF.In such cases, the alternative approach to assessing the intrinsic activity involves normalizing the kinetic current of 2e − ORR against the C dl value.In particular, the as-obtained specific current (j s ) with a higher value at a fixed potential indicates a better intrinsic activity. [25,65]n addition to the activity, selectivity is another important aspect of performance evaluation on the 2e − ORR electrocatalysts.The selectivity of H 2 O 2 can be obtained from the equation below: where N represents the collection efficiency, which reflects the fraction of product from the disk electrode to one detected on the ring electrode.The N of RRDE can be determined by the fast redox reaction of [Fe(CN) 6 ] 3−/4− in the inert atmosphere-saturated electrolyte. [51]Based on the relevant works reported so far, overall, a H 2 O 2 selectivity of over at least 90% within a wide potential window could be regarded as a criterion for judging "selective" catalysts.Meanwhile, corresponding to the H 2 O 2 selectivity, the electron transfer number (n) can be also calculated as: Apparently, the closer the value of n is 2, the higher the selectivity of ORR toward producing H 2 O 2 .Herein, it is important to note some potential pitfalls during RRDE measurements: [51,68] i) The N of RRDE is dependent on the loading mass of the probed catalysts that are deposited on the disk electrode, and usually quite different from the value offered by the manufacturer.Therefore, it is recommended to detect the exact value of N of RRDE before each ORR measurement; ii) The loading of the deposited catalysts can impact the measurement results.Excessive amounts of catalysts deposited with a thick layer can lead to the trapping of products, resulting in an underestimation of the evolved H 2 O 2 by the ring disk.Meanwhile, the confined H 2 O 2 on the catalytic sites may undergo further reduction.Moreover, a thick catalyst layer normally shows a rough surface, on which turbulence might occur when the electrolyte flows induced by the rotating RRDE, consequently leading to a decrease of N. Conversely, too little catalyst deposition would induce the contribution from the substrate to the presented ORR performances, as the glassy carbon of RRDE has been verified to be inherently active for 2e − ORR.In this regard, achieving proper loading is extremely significant, and the deposited layer should be as thin and smooth as possible while covering the entire substrate surface.
Finally, the stability of the targeted catalysts can also be assessed through RRDE tests.Typically, the chronoamperometry (CA) is adopted to yield the I ring -time and I disk -time curves, providing insights into how long the 2e − ORR activity and selectivity of the probed catalysts can be sustained.Besides, the stability assessment can be also supplemented by accelerated degradation testing (ADT). [4]n scaled-up scenarios and even (quasi-)practical situations, the evaluations based on bulk electrolysis tests are indispensable.Briefly, at a fixed ORR potential, CA is conducted in a H-type cell or flow cell, where the catalysts are deposited on the cathode.The accumulated H 2 O 2 in the electrolyte is consequently quantified, and the corresponding production rate and Faraday efficiency (FE) are accordingly calculated.Of note, i) FE is also considered as a descriptor for the selectivity of the targeted catalysts, while they are generally inconsistent with the H 2 O 2 selectivity measured by RRDE.However, FE normally reflects the real selectivity of 2e − ORR more accurately; [58] ii) The currently populated quantitative techniques for the H 2 O 2 in the electrolytes include redox titration, ultraviolet-visible (UV-vis) spectrophotometry, and colorimetric test strips, respectively.Among them, the former two are highly recommended, and the latter provides only semi-quantitative results due to its unsatisfactory accuracy. [69]Finally, from an industrial perspective, for the electrosynthesis of H 2 O 2 by 2e − ORR, based on the latest developments of bulk electrolysis in flow cells and other thoughtfully designed reactors (e.g., solid electrolyte-configuring reactor), stable generation of current densities over 100 mA cm −2 for hundreds of hours with a high H 2 O 2 FE of over 90% could be considered as a significant criterion. [2,19,65,70,71]igure 2. a) The calculated volcano plot for the 2e − (blue) and 4e − (red) ORR electrocatalysis, based on plotting U T as a function of ΔG HO* (lower horizontal axis) and ΔG HOO* (upper horizontal axis).Reproduced with permission. [21]Copyright 2013, Springer Nature Limited.b) Scaling relationships for the binding free energy of HOO*/O* with that of HO* based on the (111) surface of different metals.Reproduced with permission. [72]Copyright 2018, American Chemical Society.c) Illustration of kinetically elevating the selectivity of 2e -ORR by inhibiting either thermally cleaving based on the two neighboring active sites, or electrochemically reductive elimination.Reproduced with permission. [29]Copyright 2022, American Chemical Society.d) ORR activity trend plotted as a function of binding energies of both the O* and HO*.Reproduced with permission. [31]Copyright 2004, American Chemical Society.

Key Factors Determining the 2e− ORR Performance
The 2e − ORR performance of the electrocatalysts can be evaluated from the selectivity, activity, and stability.Achieving optimal performance in targeted 2e − ORR electrocatalysts requires strategic modifications across four key factors, i.e., the thermodynamics and kinetics of active sites toward reaction intermediates, the exposure, and accessibility of the active sites, the bulk and surface charge transfer capability, as well as the mass and ion transport ability.

Thermodynamics and Kinetics of Active Sites
According to the reaction equations (1−2), the HOO* is the sole intermediate involved in the 2e − ORR process.Therefore, elucidating the thermodynamic and kinetic properties of the active sites toward the HOO* is of high significance during ORR.Utilizing the density functional theory (DFT) calculations, Siahrostami et al. plotted the volcano relationship for electrocatalytic 2e − (blue) and 4e − (red) ORR based on the thermodynamic limit-ing potential (U T ) as a function of binding free energy of HO* (ΔG HO* , lower horizontal axis) and HOO* (ΔG HOO* , upper horizontal axis) (Figure 2a). [21]Herein note that ΔG HO* has a liner relationship with ΔG HOO* in ORR: ΔG HOO* = ΔG HO* + 3.2 eV (Figure 2b). [72]As is shown in Figure 2a, an activity-ideal catalyst appears at the peak of the calculated volcano diagram for the 2e − ORR, where its ΔG HOO* is 4.2 ± 0.2 eV and the equilibrium potential for 2e − ORR ( O2/H2O2 ) is correspondingly 0 V (the dotted line), which means that the best activity can be achieved with negligible overpotentials.In addition to the activity, the selectivity of ORR can also be evaluated by the thermodynamic adsorption free energy of O* and HO* (note these two still correlate in a scaling relation: ∆G O* = 2∆G HO* , [47,72] Figure 2b), as whether the O−O bond would dissociate directly determines the sequent ORR routes.Moving left from the volcano peak of 2e − ORR in Figure 2a, a stronger HO* binding is presented, and concurrently a more positive U T of 4e − route can be observed, which means ORR is preferable to form H 2 O instead of H 2 O 2 as the main product.Interestingly, on the right side, starting from the volcano peak of 2e − ORR where the binding of HO* (also for HOO*) gradually weakens, the volcano curve of both 2e − and 4e − ORR overlap with the same U T , indicating the activity deterioration for both of these two pathways.Moving further to the right, the even weaker adsorption of HO* (also for O*) enables the preservation of O−O of HOO*, therefore the catalyst (for example, Au in Figure 2a) exhibits an enhanced selectivity but decreased activity (weaker HOO* adsorption) toward the 2e − pathway. [21,29]In fact, when considering ORR selectivity, apart from the thermodynamic factors, the kinetical concerns should also be taken into  2c, upper right. [29]Note that the cleavage of O−O bond can also be achieved by the electrochemically reductive elimination through the proton-coupled electron transfer (PCET) process.Using the CoS 2 structure as an illustrative example, the absence of the active-site ensembles results in only one O of HOO* being well adsorbed by the active atom (Co).Consequently, the PCET to the O bonded by the active atoms is apparently more favorable than to the distant O, hence favoring the formation of H 2 O 2 (Figure 2c, bottom right). [29]he above knowledge generalizes how the thermodynamics and kinetics of active sites influence the presented ORR activity and selectivity.When directing our focus to the active sites themselves, it becomes evident that the inherent properties of TM sites for ORR cannot rival those of benchmark noble metal sites (particularly Pt).Specifically, the aspect of activity reflected by the transfer kinetics of proton/electron to the stably surfaceadsorbed oxygen-containing intermediates is intrinsically inferior to that of noble metals such as Pt and Ag. [31,73]Based on a microkinetic model, Nørskov et al. explicitly evaluated and plotted the reaction rate of oxygen reduction for various pure metals as a function of their own oxygen binding energies, where Pt displays the best activity and lowest overpotential for ORR (Figure 2d). [31]esides, when considering the ORR route, Pt and Ag also demonstrate the best 2e − ORR selectivity among various pure metals, as unveiled by the scaling relationships between the ΔG HOO* and ΔG O* on their (111) surfaces plotted by Ulissi and co-workers.The trend indicates that the weaker the ΔG O* , the better the selectivity exhibited. [30]Last but not least, from the aspect of stability, pure TM atoms are prone to dissolution under the operating potentials of the ORR process when compared with noble metals. [21]herefore, in order to achieve excellent 2e − ORR activity and selectivity in TM-based catalysts, rationally optimizing the intrinsic properties of the involved TM atoms is of tremendous importance.First of all, the ∆G HOO* of the TM sites should be optimized to be close to 4.22 (U = 0 V; or 3.52 eV, U = 0.7 V) as much as possible, which can be realized through the electronic structure regulation. [30,41]Secondly, deliberately prolonging the distance between the adjoining active TM−TM atoms to surpass that of O−O of HOO* is also considered as an effective approach. [29,37]This could lead to the kinetically increased acti-vation barrier of O−O cleavage and transfer barrier of protoncoupled electron toward the distant free O site of HOO*.The approach suppresses O−O dissociation by avoiding thermal break and electrochemical reductive elimination, eventually promoting the selectivity of H 2 O 2 electrosynthesis.Herein, it is necessary to point out that the premises to satisfy the above two points is that the other neighboring component atoms possess thermodynamically destabilized/unfavorable binding ability toward O* and/or HO* (so-called "inert atoms"), which can be enabled by finely tuning the local coordination environments and the electronic state of the involved binding sites. [4,21]The characteristic feature of TM-based catalysts (including compounds and atomic-site catalysts), wherein ensembles of active adsorption sites are typically absent, inherently positions them to meet these requirements. [29,74]n addition, the Pourbaix diagrams of different elements should be utilized in screening appropriate TMs and other component elements for constructing stable catalysts under various applied electrolyte pH and potentials. [75]These diagrams could provide insights into corrosion resistance across different ORR scenarios.Additionally, when considering well-defined TM-based compounds, their respective bulk phase Pourbaix diagrams, available in the Materials Project database, [76] serve as valuable references for 2e − ORR. [48,77]Compared to those based on the bulk phase, surface Pourbaix diagrams offer a more direct explanation and prediction of the surface stability of a catalyst.−79] In principle, by combining the Pourbaix diagrams of individual involved elements with the bulk phase and surface structure ones, the stability of most fabricated TM-based catalysts for ORR can be predicted, regulated, and improved.It is crucial to highlight here that besides the corrosion/dissolution of atoms of the TM-based phases, the "surface reconstruction" from the pristine phase to the thermodynamically more stable one (not water soluble) might also occur under ORR conditions, [79] as experimentally identified. [80]−84] In the case of 2e − ORR, due to the current lack of deep insights into the relationship between "surface reconstruction" of the TM-based "precatalysts" and the maintenance of the presented catalytic efficiency, whether to promote or avoid the occurrence of surface reconstruction deserves more attention.

Exposure and Accessibility of the Active Sites
During the electrocatalytic 2e − ORR, the utilization of surface and bulk active sites of a catalyst directly correlates with its presented activity and stability, i.e., accelerating the catalysis at a high reaction rate under low overpotential (against the theoretical equilibrium potential) and maintaining such a working efficiency in the long term, requiring the sufficient exposure and high accessibility of the active sites toward electrolytes (reactants and intermediates). [25,31,84,85]This is strongly dependent on the surface properties (e.g., morphology and porosity), bulk structure, and size of the electrocatalysts.−92] Lastly, although this kind of investigation in the field of electrochemical 2e − ORR is rather limited, the observations on TM-based catalysts serving for other oxygen electrocatalysis processes are very common, we thus can still deduce that iv) the crystallographic features of these catalysts, including crystallinity, lattice parameters, and solubility under a certain applied bias and pH value of aqueous electrolyte, would greatly determine whether they can exhibit better utilization and durability of active sites during 2e − ORR. [4,28,29]orrespondingly, screening and developing TM-based catalysts with proper crystallinity, the well-exposed domain of crystal unit, and good corrosion resistance of the active sites are greatly desired. [4,29,36,93,94]

Bulk and Surface Charge Transfer Capability
The charge transfer ability of the catalyst is another crucial factor controlling the catalytic performance. [95,96]An accelerated charge transfer contributes to delivering a larger current density during catalysis, [97] resulting in improved activity.Remarkably, in the case of the electrochemical process of ORR, the charge transfer kinetics of a catalyst is in nonlinear correlation with its selectivity of a 2e − route. [39]Although some discoveries from Sun and co-workers demonstrate that elevating the electron transfer to a certain extent is beneficial to the 2e − ORR selectivity, [98,99] a similar claim of faster charge transfer enhancing the 4e − ORR pathway can also be found in the literature. [100,101]This contradictory point can be probably accounted for from two aspects.First, regarding an ORR process, the preference for the 2e − /4e − route of an electrocatalyst can be collectively influenced by different factors, e.g., intrinsic adsorption/desorption abilities of the different catalytic sites and the number of accessible catalytic sites. [95]Of great significance, albeit occasionally overlooked is the consideration of charge transfer ability at both bulk and surface regions.Given that the interface reactions between the electrode (catalyst) surface and electrolyte dominate the overall heterogeneous electrocatalysis, [102] the charge transfer on the catalyst surface is quite important.In comparison to its bulk counterpart, this surface charge transfer property more directly impacts the resulting ORR selectivity.However, to date, a comprehensive understanding of these associated understandings remains elusive.
104] On the contrary, the interfacial charge transfer kinetics was seldom explored and explicated, possibly due to the lack of specialized and direct characterization methods.To this end, the transient photo-induced voltage (TPV) characterizations developed in our group based on a stimuli-response principle can serve as an effective solution, which can directly detect and analyze the charge transfer at the catalyst surface, resulting from the charge separation after photoexcitation.On the basis of this point, the interfacial charge transfer ability during the heterogeneous electrocatalysis process in aqueous electrolytes can be assessed. [90]Taking advantage of this technique, our recent contributions focusing on the surface charge transfer manifested that appropriately slowing down electron transfer at the interface between the electrocatalyst surface and electrolyte can substantially enhance the 2e − route selectivity, [4,8,41] This deliberate slowing down of electron transfer prolongs the residence time of electrons within the catalyst surface, preventing rapid and extensive binding reactions with O-containing absorbates. [105]As a result, the competing 4e − pathway can be greatly inhibited.
In summary, we advocate the following principles for designing TM-based catalysts for 2e − ORR from the perspective of charge transfer ability: i) increasing the electron transfer (especially within the bulk of the targeted catalysts) to a certain extent to achieve a higher activity; ii) giving more importance to the surface charge transfer kinetics, which should be moderately decelerated to realize a preferable 2e − pathway.

Mass and Ion Transport Ability
An outstanding ability of mass and ion transport, especially that sufficient O 2 reactants are capable of being dissolved and diffusing toward the catalyst's surface, coupled with the swift and timely evacuation of H 2 O 2 (or HO 2 − in the alkaline media) products from the vicinity of the catalyst has an enormously positive effect on the resulting activity, selectivity, productivity, and stability of the 2e − ORR electrocatalysis. [52,53,106]On one hand, the dissolution and diffusion capability of the O 2 reactant toward the active surface (sites) of the electrodes (catalysts) will determine the reaction kinetics of the sequent ORR process.However, due to the intrinsic solubility and diffusion limitation of O 2 in aqueous solutions, [107] enhancing the transport of the O 2 reactant is essential for most ORR reactions, directly impacting catalytic activity.On the other hand, the rapid removal of H 2 O 2 (HO 2 − ) products helps prevent the accumulation of evolved products and the blockage of active surface sites, thereby enhancing overall activity.Furthermore, it effectively inhibits further electroreduction and disproportionation reaction of 2e − ORR products, as well as the catalyst poisoning and corrosion caused by the evolved H 2 O 2 (HO 2 − ) bonded on the catalyst surface, [14,52] contributing significantly to the activity, productivity, selectivity, and stability of H 2 O 2 synthesis from ORR.Thus, for a TM-based catalyst, expediting the diffusion of O 2 reactants and the removal of H 2 O 2 (HO 2 − ) is extremely significant for exceptional 2e − ORR capability.This can be achieved through feasible protocols, including i) the catalyst optimization [108−110] and ii) electrochemical cell design, [23,111] according to the previous explorations.In the aspect of the catalyst optimization, three points are suggested to be specially considered: i) increasing the porosity and ii) hydrophobicity, as well as iii) reasonably adjusting the charge density (local electric field) of the catalyst surface.On the one hand, the increment of the surface porosity or building a (macro/meso)porous skeleton for a catalyst is highly recognized to enhance the mass transport ability. [106]Tuning the surface of the catalyst to be hydrophobic is also conducive to the O 2 gas transfer. [112,113]This is due to the fact that a hydrophilic surface will naturally result in the strong affinity of the water molecules in the aqueous electrolytes, shielding the access to the active surface sites and the detachment of the evolved products, thus exasperating the reaction kinetics and the eventual performance. [83,114]While current strategies predominantly focus on improving carbon-based catalysts for 2e − ORR, [106,112,115−118] they are also in principle applicable to TMbased catalysts.Beyond these two commonly employed regulatory methods, it is worthwhile to emphasize another underrated approach to improving the mass transport and reaction kinetics, i.e., altering the charge density (local electric field) of the TM-based catalyst surface for 2e − ORR.Our recent works carefully compared various TM-based compounds with different surface situations, revealing that an optimum surface charge density (local electric field) can distinctly control the concentration (coverage) of charged ions surrounding the catalyst surface during the ORR process, and more importantly, taking advantage of such modulations can effectively and instantly drive the products of 2e − ORR catalyzed by TM-based compounds away from the catalyst surface, thereby preventing a decline in working efficiency. [8,16,42]n addition, it is believed that the rational design of the electrochemical reactor cells is also able to effectively boost the mass and ion transport capability during the ORR process.Herein, we have to point out that the widely utilized reactors for 2e − ORR electrocatalysis, especially at the laboratory level, include rotating RRDE and H-cell setups, which are easily manipulated to electrochemically and chemically examine the H 2 O 2 evolution.Nevertheless, during electrochemistry, the presented catalytic performances significantly suffer from the poor solubility and diffusion of the O 2 reactant, as well as the high local concentration of H 2 O 2 (HO 2 − ) product in the immediate vicinity of the electrode surface.To enhance the mass transport ability and reach the goal of industrial application, depositing the TM-based catalysts on the gas diffusion electrodes (GDEs) equipped in a suitable electrochemical flow cell with continuous flow and sufficient supply of O 2 is an effective and practical approach. [4,23,44,119]In detail, the catalyst inks are typically coated on a hydrophobic support, such as a carbon cloth treated with polytetrafluoroethylene (PTFE). [23,120]One side of this support coated with carbon and hydrophobic polymers serves as the contact face to O 2 gas, through which the constant O 2 flow can more smoothly diffuse into the GDEs.Meanwhile, the catalyst side of this support contacts the electrolyte, where the products are formed during the ORR process, and the continuous circulation of reactants and products by such flow cells can in principle suppress the accumulation of H 2 O 2 (HO 2 − ) in the local region surrounding the catalyst (electrode) surface. [23,120]Re-markably, to realize accelerated mass transport, higher working efficiency, and lower energy consumption, some electrode/cell parameters still need to be reasonably and delicately designed, especially catalyst loading and deposition method, GED engineering (such as wettability and porosity), aeration rate, oxygen utilization efficiency, etc. [29,107,120−125] So far, in the cases focusing on developing GDEs in flow cells, the utilized catalysts are still mostly confined to the carbon-based ones, yet these reports can still inspire and guide the researchers to develop the appropriate GDEs integrated with TM-based catalysts for 2e − ORR electrocatalysis.
Consequently, according to the above elaborations on the four most essential influencing factors determining the resulting 2e − ORR performance, the corresponding four protocols to boost the H 2 O 2 electrosynthesis ability of TM-based ORR catalysts have been systematically illustrated in Figure 3, including optimizing the thermodynamics and kinetics toward intermediates, promoting the active site exposure and accessibility, adjusting the bulk and surface charge transfer, as well as elevating the mass and ion transport.

Phase Engineering and its Application in Electrocatalysis
Phase engineering has become a prevalent strategy for tailoring TM-based catalysts to suit various electrocatalytic applications, with improvements stemming from the four aspects discussed in the preceding section.However, a precise definition of phase engineering within the context of 2e − ORR is yet to be established.Herein, we aim to present a succinct concept of phase engineering, that is, elaborately controlling and altering the presence, type, amount, arrangement period and mode, local coordination, and electronic state of atoms within the defined TMbased catalysts.This approach enables profound, effective, thorough, and even precise modifications and optimizations on their physicochemical, crystallographic, and geometric properties at the atomic level. [126,127]−130] Moreover, changing the arrangement period and mode of atoms can occur through altering crystallinity, and/or polymorphic transformation of the TMbased 2e − electrocatalysts. [4,8,77]Furthermore, the local configurations and geometries for the atom arrangements can be specially altered by the coordination tuning of the important active atoms in the TM-based materials. [131,132]Notably, the electronic structure of atoms in the modified catalysts also accordingly changed unavoidably, influenced by nearly all of the above-phase engineering approaches. [133]Herein, the occurrence of phase engineering is typically accompanied by the involvement of a large degree or even all atoms in the targeted catalysts, hence resulting in the penetrating changes of entire bulk phase structures with various modified features. [126]Another point worth noting is given the fact that the growth of nanomaterials is strikingly affected by their surface characteristics and kinetics, the formation of unconventional phases that are thermodynamically unfavorable, or the distinctive heterostructured phases could be achieved, which can bring about distinct atomic arrangements, surface active sites, and electronic structures.−137] On the basis of the above descriptions, phase composition, and structure should be regulated at the atomic scale for TMbased electrocatalysts to optimize the 2e − ORR activity, selectivity, stability, or even productivity.Accordingly, six phase engineering strategies naturally emerge as follows: i) elemental doping, ii) defect creation, iii) heterostructure construction, iv) coordination tuning, v) crystalline design, and vi) polymorphic transformation.Figure 4 also illustrates the derivative ideas of phase engineering strategy and some of their typical forms based on TM-based 2e − ORR electrocatalysts (e.g., multiple atoms doping, ligand grafting, phase transition from 2T to 3R structure, sulfides heterophase, expanding layer spacing or triggering cation vacancy, and inducing disordered local structure).−144] It is also not an exception for the majority of the reported TM-based materials serving for the electrosynthesis of H 2 O 2 by 2e − ORR.Notably, the performances facilitated by phase engineering for 2e − ORR are generally achieved through one or a combination of the four fundamental aspects toward TM-based catalysts:, i.e., promoting the exposure and accessibility of the active sites, optimizing the thermodynamics and kinetics of active sites toward reaction intermediates, enhancing the bulk and surface charge transfer capability, and/or boosting the mass and ion transport ability.The associated details will be exhaustively elaborated in the following section.

Elemental Doping
Elemental doping can be roughly defined as the introduction of foreign atoms/groups at a low concentration into the host lattice.
The doping of foreign atoms into the metal compounds typically occurs in two forms, i.e., substitutional doping or interstitial doping, mostly depending on the size of the dopant atoms.Substitutional doping involves the replacement of host atoms, occupying their lattice sites, while interstitial doping places dopants within the interstitial spaces of the targeted lattice. [145]In both doping scenarios, the local atomic presence, types, and arrangements of the catalysts undergo significant variations leading to lattice distortion/strain, and alterations in electron density, resulting in notable changes in the physicochemical and crystallographic properties of the materials. [142,146,147]ased on the above considerations, the doping of foreign elements into appropriate TM-based catalysts has proven to be a viable strategy for enhancing catalytic performance in the field of 2e − ORR electrocatalysis.Specifically, i) the facile heteroatom doping could endow the TM-based metal catalysts with a more porous surface morphology with substantially enhanced specific surface area, thereby enlarging the surface exposed active sites, as well as accelerating the mass/ion transport. [88,148,149]Taking the study conducted by Cai and co-workers as an example, the introduction of Mn dopants into CuS (denoted as Mn-CuS-x, where x values increasing from 1 to 5 meant the increment of the Mn proportion) enabled by a simple one-pot hydrothermal method not only doubled or tripled the Brunauer-Emmett-Teller (BET) surface area of the pristine CuS, but also indicated the formation of the porous surface with larger specific surface area, by which a much improved 2e − ORR activity was achieved. [88]ii) Foreign elemental doping induces a noticeable transformation in the electronic structure of the pristine TM-based catalysts, resulting in a narrowed band gap.This alteration proves advantageous for enhancing their electrical conductivity, [98,150] which has been well illustrated by comparing the density of states (DOS, Figure 5a) for the TiO 2 with and without foreign Mn atom doping in the work from Sun et al. [150] iii) The electron density of the active atoms is usually altered, so that they can serve as more appropriate adsorption sites, and display an optimized charge interaction with the reaction intermediates. [98,148,150,151]For example, Fei and co-workers developed a single atom catalyst prototype in which the atomic Co sites were coordinated with N atoms and dispersed on the carbon support (CoN 4 -C).Interestingly, a series of advanced synchrotron-based spectroscopic techniques, including X-ray absorption near-edge structure (XANES), as well as Fourier transformed extended X-ray absorption fine structure (FT-EXAFS) and the associated Wavelet transform EXAFS (WT-EXAFS) spectra, revealed that the doping the foreign P atoms into the carbon matrix (CoN 4 -PC) could withdraw more electrons from Co atoms leading to an elongated Co−N bond (Figure 5b-d).Hence, the P doping-induced tensile strain toward the Co-N coordination (Figure 5e), potentially created electron-deficient Co centers, serving as more suitable active sites for the 2e − ORR. [152]However, it's important to note that contradictory conclusions have been drawn in different studies, where active metal atoms with both higher and lower oxidation states, influenced by the charge interaction with foreign dopants, were reported to enhance the 2e − ORR process. [98,148,150,153]Therefore, many factors, such as the type, position, and concentration of dopants, as well as the doping manner would collectively contribute to the eventual catalytic performances.Besides, another important point worth mentioning is that the dopant atoms with inherent catalytic capability, exhibiting more appropriate electron density than the host atoms, can act as the real active sites.For instance, in the case of Mn-substituted CuS, Mn dopants donated electrons to the surrounding atoms, optimizing the binding with the intermedi-ates HOO* (Figure 5f,g). [88]It's worth emphasizing that adjusting the electron density of the active sites alters their electronic interactions with oxygen-containing intermediates through electrostatic forces.For example, the combination of the X-ray photoelectron spectroscopy (XPS) with XANES and EXAFS spectra confirmed the substitution of S 2− with lower electronegativity to the O 2− of NiP 4 Mo 6 -based polyoxometalates (POMs), suppressing the interaction between the anionic atoms of such a POM with the proton (positive charge) of the ORR intermediate HOO*, thereby inhibiting the tendency of the O−O bond cleavage. [128]Last but most extensively studied, iv) from a thermodynamic point of view, the modulated electronic structures induced by the elemental doping could enable the host materials to achieve an optimized d-band center and desirable binding energies toward the oxygen-containing intermediates.Of note, the d-band center can be employed as an effective descriptor to demonstrate the binding strength toward the O 2 and oxygencontaining reaction intermediates during ORR, and generally, a d-band center closer to the Fermi level represents a stronger adsorption ability. [148,153]  ) and Mn-doped TiO 2 (Mn-TiO 2 ).Reproduced with permission. [150]Copyright 2021, American Chemical Society.b) The Co K-edge XANES spectra and c) k 2 -weighted FT-EXAFS spectra of CoN 4 -C and CoN 4 -PC, as well as references (Co foil and CoO).The inset of b): the fitted oxidation state of Co from the associated XANES spectra.d) The k 2 -weighted Co K-edge WT-EXAFS spectra of CoN 4 -C and CoN 4 -PC, as well as the reference Co foil.e) Schematic of the tensile strain toward the Co-N coordination induced by P doping.Reproduced with permission. [152]opyright 2023, Elsevier.f) The charge density difference of Mn-doped CuS (Mn-CuS), where the cyan and yellow colors represent the charge depletion and accumulation in the space, respectively.g) The OOH* adsorption energies and the associated charge states of the active sites for CuS and Mn-CuS.Reproduced with permission. [88]Copyright 2023, Elsevier.h) Free energy diagram for 2e − ORR electrocatalysis at different possible catalytic sites of (101) Cu-TiO 2 , including Cu doping-induced lower-valence Ti 4c 3+ , Ti 5c 3+ , and their synergy (Ti 4c-5c 3+ ), as well as Cu dopant, and Ti equivalent to the one within the pristine TiO 2 .i) The corresponding ΔG O* at these sites (Ti 4c 3+ , Ti 5c 3+ , Ti 4c-5c 3+ , Cu, and Ti).Reproduced with permission. [98]Copyright 2022, Tsinghua University Press and Springer-Verlag GmbH Germany.
reaction intermediates enhancing ORR activity. [148]This trend was also observed in experiments with titanium-doped metal sulfides, where the ORR activity increased with the d-band center shifting toward the Fermi level. [154]More significantly, the elemental doping could especially promote the non-linear relationship between the adsorption strengths toward HOO* and O*, [98,150] a critical factor for achieving both outstanding activity and selectivity in 2e − ORR electrocatalysts (vide supra).For example, as the adsorption free energy calculation results revealed (Figure 5h,i), a moderate value of ΔG HOO* was obtained on the lower-valence Ti 3+ atoms (specifically, cooperative Ti 4c 3+ and Ti 5c 3+ sites (Ti 4c-5c 3+ )) emerged from the pristine TiO 2 after Cu doping (Cu-TiO 2 ), which was very close to the ideal values (3.52 eV, U = 0.7 V).Concurrently, the largest ΔG O* value can be found for Ti 4c-5c 3+ , indicative of its weak adsorption ability toward the terminal O of HOO* and favorable conditions for 2e − ORR electrocatalysis. [98]he cases of elemental doping into TM-based catalysts to enhance the electrosynthesis of H 2 O 2 from 2e − ORR have been summarized in Table 1.Notably, the doping of foreign cations and anions can be easily achieved through straightforward hydrothermal, annealing, or refluxing methods.These studies consistently demonstrate excellent performance improvements resulting from this phase engineering strategy.However, it's Refs.
Adv. Mater.2024, 36, 2400140 crucial to highlight that the scope of these investigations has been confined to alkaline media, and the conditions in lower-pH electrolytes remain unclear.

Defect Creation
The presence of defects within the heterogeneous catalyst is believed to remarkably disrupt the periodic arrangement of atoms within the crystals, which leads to variations in their physicochemical and crystallographic properties.Accordingly, through rationally creating defects with various types, amounts, and locations, the resulting electrocatalytic capacities can be precisely altered. [162,163]−166] Markedly, over the years, the great roles of defect creation in ameliorating the ability of TMbased compounds to electrogenerate H 2 O 2 through ORR have been also progressively discovered, and the associated characteristic cases have been summarized in Table 2.−173] It is worth highlighting that the creation of oxygen vacancies within these compounds can be readily accomplished through simple and costeffective methods, primarily involving hydro/solvothermal techniques, heat treatments, acid etching, or their combination.Consequently, the incorporation of oxygen vacancies holds promise as an effective and universally applicable strategy for enhancing the catalytic performances of pristine TM-based oxides in the 2e − ORR.On the other hand, elucidating the mechanism through which the emergence of oxygen vacancies enhances 2e − ORR capability is pivotal, which significantly promotes the adsorption of O 2 reactants and their protonation into the key intermediate HOO*so that the activity of 2e − ORR is tremendously enhanced.8][169][170]172] This was clearly demonstrated by the work of Zou et al., who developed a novel {001} facets-exposed -Fe 2 O 3 catalyst with rich oxygen vacancies, which showed exceptional H 2 O 2 production ability in wide-pH electrolytes. DFT alculation results on thermodynamics revealed that the existing O vacancies facilitated the adsorption of O 2 with a more negative binding energy.Meanwhile, the O−O bond was elongated to be beneficial to the subsequent protonation into HOO* (Figure 6a).More interestingly, apart from the enhancement of the catalytic activity, the selectivity was also elevated as the O 2 molecules favored adsorption in an"end-on" configuration to occupy oxygen vacancies on the catalyst surface (Figure 6a).This strategic configuration substantially prevented the cleavage of the O−O bond.[172] Following this work, Wen and co-workers further used the in situ X-ray absorption spectroscopy (XAS) technique, experimentally tracking the filling of oxygen vacancies within amorphous copper hydroxide supported on graphene oxide (A-Cu(OH) 2 /GO) by the oxygencontaining intermediates during ORR process.As is reflected by the in situ XANES spectra (Figure 6b), the oxidation state of Cu significantly increases when the applied potential reaches levels conducive to extensive ORR.This phenomenon was attributed to the adsorption of oxygen-containing intermediates by Cu sites and the withdrawal of the electrons from these Cu atoms.Upon reducing the potential back to the ORR-inactivated region, the oxidation state of Cu accordingly dropped, indicating the high reversibility and reusability of the active Cu centers enabled by the oxygen defects.[89] Yet, it's essential to note that controversy persists regarding the real active sites for oxygen-deficient TM-based oxides for 2e − ORR.For the O defects-enriched TiO 2-x ORR catalyst, Sun et al. combined the in situ Raman techniques with DFT calculations, proposing that the oxygen vacancies steadily located in the near surface areas of TiO 2-x , rather than those surface ones which were easily filled by the oxygen-containing intermediates, more possibly acted as the real active sites for the evolution of H 2 O 2 .[99] Besides the role of oxygen vacancies in facilitating the 2e − ORR, another noteworthy function is the profound tuning of the electron density of neighboring TM atoms within the host oxides, and consequently, tuning them to be more active for the proceeding of 2e − ORR.[150,173] Akin to the situations in other fields of electrocatalysis, oxygen vacancy could lead to escalated electric conductivity of the investigated TM-based oxides, collaboratively promoting their abilities for the production of H 2 O 2 from ORR. [167,171,172] The versatile role of oxygen vacancy in facilitating 2e − ORR naturally inspires the community to consider whether other types of anionic vacancy can also positively affect the ability of TM-based compounds to electrosynthesize H 2 O 2 through ORR.In this regard, we intentionally developed nickel diselenide with abundant selenium vacancies (NiSe 2 -V Se ) for alkaline 2e − ORR, which demonstrated impressive catalytic activity and selectivity.[41] The intriguing and abnormal point of such Se vacancies is that they were negatively charge-polarized, which was verified by the Rietveld refinement of the powder X-ray diffraction (XRD) patterns of the Se-deficient NiSe 2 .NiSe 2 -V Se was determined to own an expanded lattice volume than that of the pristine NiSe 2 .Conversely, the generation of normal vacancy typically leads to the shrinkage of the lattice.Remarkably, we further employed the DFT simulations, calculating the relationship of the lattice volume of NiSe 2 with the charge injected into the Se vacancy, which revealed that the more negative charge-polarization of Se vacancy (trapping more negative charges) will accordingly linearly expand the lattice volume of NiSe 2 .Based on this point, a series of merits were induced to enhance the 2e − ORR capability.First of all, the proper lattice expansion can stabilize the crystal structure of NiSe 2 , which was verified by the Halder-Wagner method of crystallographic analysis.Second, through the TPV characterizations, it was found that the surface charge-polarized Se vacancies induced the electron-trapping effect, thereby slowing up the surface charge transfer kinetics during the ORR process.Consequently, the electrons tended to reside at the surface of NiSe 2 -V Se for a longer period, thus, the rapid and sufficient binding reaction of the surface active sites with the O-containing species was inhibited.This was conducive to enabling the ORR to proceed in a 2e − pathway.Third, similar to the function of O   Reproduced with permission.[172] Copyright 2020, Wiley-VCH.b) In situ XANES spectra (Cu K-edge) of A-Cu(OH) 2 /GO catalyst recorded under the applied potential shifting from 1.0 to 0.5, and finally back to 1.0 V. Inset: the magnified absorption edges.Reproduced with permission.[89] Copyright 2021, Chinese Chemical Society.The differential charge density distribution between the adsorbed *OOH intermediate and catalyst slabs, i.e., c) NiSe 2 and d) NiSe 2 -V Se .Note that the grey and green sphere corresponds to the Ni and Se atoms, respectively.Also, the positive and negative charge is represented by the red and blue color isosurface, respectively. Reprouced with permission.[41] Copyright 2022, The Authors.Reproduced with permission.[44] Copyright 2021, Wiley-VCH.
vacancies, the Se ones managed to optimize the electronic structure of the pristine NiSe 2 and its free adsorption energy toward the pivotal intermediate HOO*, giving rise to enhanced reaction thermodynamics. [41]This can also be verified through the differential charge density distributions between the catalyst slab and the adsorbed HOO* (Figure 6c,d), in which more obvious charge localization induced by the negative-charge polarized Se can be found in the case of NiSe 2 -V Se , suggesting a strengthened binding of HOO* by the active Ni sites.Based on the above three points, inducing the different anionic vacancies within the TMbased compounds beyond the oxygen ones is highly promising to elevate 2e − ORR performances.However, despite the potential benefits, there is a notable scarcity of investigations in this specific domain.
It is worth mentioning that similar to the effects of anionic vacancies, manipulating the cationic ones within the TM-based compounds has also been confirmed to be a feasible strategy to strengthen the 2e − ORR capability. [129,174]Briefly, such improvements can be generalized from two aspects, i.e., the optimizations toward geometric and electronic structures of the studied materials.On the one hand, it is conceivable that the absence of a certain number of TM atoms within the host lattice will enlarge the distance between the surrounding TM sites.This, in turn, can hinder the cleavage of the O−O bond for the intermediate HOO* adsorbed on those TM sites and enable the oppression of 4e − ORR tendency. [129] 6f).This observation clearly suggests that 4e − ORR dominated in the unmodified Ni 2 P where the abundant Ni vacancies were absent. [174]On the other hand, the electronic structure of the host TM-based compounds can be finely tuned by the formation of TM vacancies.This enables the optimization of the binding of reaction intermediates by the TM vacancy-enriched TM-based compounds, exerting a profound influence on their catalytic activity. [174]Using the study by Zhao et al. as an example, it was observed that in comparison to pristine Ni 2 P, which exhibited excessively strong adsorption of HOO*, the presence of Ni vacancies evidently weakened the adsorption of HOO* Specifically, in the ORR activity volcano plot (Figure 6g), Ni 2−x P-V Ni is located at the point (ΔG HOO* = 4.21 eV and U L = 0.69 V) closer to the peak, as compared to the pristine Ni 2 P. Ideally, when a catalyst is located at the volcano peak (ΔG HOO* = 4.22 eV and U L = 0.7 V), its activity for 2e − ORR is maximized with negligible overpotentials.Consequently, the 2e − ORR activity of Ni 2−x P-V Ni significantly surpasses that of pristine Ni 2 P, which has been also verified by the RRDE tests.In addition, the DFT also disclosed that the ΔG HOO* values vary with increasing Ni vacancy concentration, but only within a specific range.Hence, it could be concluded that during the ORR process, reducing the TM amount in a TM-based catalyst can probably alter its electronic structure, as well as the adsorption strength with the key intermediate, such as HOO*.Upon reaching an optimum reduction degree (an optimum TM vacancy concentration), its 2e − ORR activity may be improved to the maximum. [174]n addition to the lattice vacancy, the crystal defects caused by the lattice distortion/strain can also significantly contribute to the improvement of the 2e − ORR ability of TM-based catalysts. [44,88,175]Examples include the presence of zigzag edges [176,177] and interlayer defects [178]   pathway during the ORR process. [175]In fact, similar improvement effects on binding HOO* intermediate can also be attained by inducing the interlayer defects in TMDs.Gao et al. discovered that narrowing the interlayer spacing of CoSe 2 (as illustrated by the corresponding HRTEM and FFT results in Figure 6h) can notably optimize its adsorption toward HOO* during 2e − ORR, and they experimentally confirm this phenomenon besides the DFT simulations.In detail, as the operando XAS spectra demonstrated, the reversible oxidation state of Co and coordination number of Co-O can be found for sc-CoSe 2 (the modified CoSe 2 with a narrowed interlayer gap), accompanied by the initiating, proceeding, and ending of the ORR (Figure 6i,j).Whereas, in the case of the pristine CoSe 2 with large interlayer spacing accommodating diethylenetriamine (DETA) molecules, no visible oxidation state changes of Co were reflected by the operando XANES spectra during an ORR operational cycle.Meanwhile, after liberating the coordinated DETA, the Co-O coordination number of this sample remained stable with the proceeding and ending of ORR (Figure 6i,j).This supported the fact that the unmodified pristine CoSe 2 exhibited an over-strong HOO* adsorption strength and thus a poorer 2e − ORR ability.At last, it is necessary to mention that the altered electronic structure originating from diminishing the interlayer distance endowed the sc-CoSe 2 with an improved electric conductivity, which further contributed to elevating the 2e − ORR capability, as revealed by the ultraviolet photoelectron spectroscopy (UPS) findings and DFT calculations. [44]ased on the above overview, defect engineering aimed at boosting 2e − ORR performance of TM-based catalysts primarily focuses on introducing lattice vacancies, particularly oxygen vacancies.However, oxygen vacancies are more commonly observed in TM-based oxides, which often fall short in exhibiting competitive 2e − ORR performance in low-pH environments compared to precious metal-based catalysts.As a solution, more research attention should be directed toward 2e − acid ORRpromising TM-based chalcogenides, particularly those 2D TMbased dichalcogenides (TMDs) known for their abundant and flexible defect types. [176]

Heterostructure Construction
Coupling other components can directly modify the atomic arrangements of the targeted TM-based catalysts, especially those atoms located near the interface regions.−182] Recently, the construction of heterostructures has emerged as a promising strategy to enhance the performance of TM-based materials 2e − ORR electrocatalysis, as demonstrated in Table 3.Interestingly, it can be easily found that the majority of such reported works adopted carbonaceous materials, such as MXenes, graphenes, nitrogen-doped carbon, and carbon black species, to couple the targeted TM-based compounds ranging from nanoparticles to downsized atomic sites. [108,130,183,184]−187] These can be well demonstrated by the increased BET surface areas [188] and ECSA [189] after the introduction of heterophasic carbon.It is also worth mentioning that the mass transport ability which directly determines the prospects for industrial applications can be especially promoted by constructing the 3D open porous skeleton after assembling the TM-based catalysts with carbon species. [108,190]Another interesting point to note is that the performance of an electrocatalyst in 2e − ORR is influenced by its pore sizes.Mesopores are typically preferred as they facilitate mass transport and product release during 2e − ORR.[193] This point can be well exemplified by the heterostructured catalyst, i.e., MnO nanoparticles/Mn-N x moieties that were incorporated into the N-doped carbons (Mn-O/N@NCs), developed by Byeon and co-workers. [191]The RRDE results demonstrated that the Mn-O/N@NCs sample with moderate pore size on its carbon exhibited the best activity and selectivity for 2e − ORR.
ii) Coupling other components typically results in the altering of the lattice and electronic structures of the pristine TM-based catalysts, optimizing their reaction thermodynamics and kinetics for 2e − ORR. [130,184,189]iii) Because of the superior intrinsic conductivity of most carbon-based materials, the overall combined system endows with a promoted electron transfer ability during the ORR process. [130,189,194]iv) The TM-based electrocatalysts supported/encapsulated by the carbonaceous materials ensure the homogenous distribution of such catalysts, effectively addressing issues such as disordered stacking or severe aggregation of the active species, thus avoiding burying of active sites. [183,184,189]urthermore, the confinement provided by carbon supports stabilizes dispersed TM-based components, mitigating corrosion by electrolytes during electrochemistry and suppressing the 2e − ORR performance deterioration. [108,188,189]These two points are verified by the comparison experiments between core-shell heterostructured TM-based compound/carbonaceous material and carbon-free bare TM-based compound, such as CoSe 2 nanoparticles into nitrogen-doped carbon nanotubes (CoSe 2 @NCNTs) and bare CoSe 2 , [130] as well as CoTe nanoparticles in the nitrogendoped carbon (CoTe@NC) and CoTe. [189]In both cases, the heterostructured catalyst exhibited decreased charge transfer resistance (R ct ) compared with the associated single Co-based compounds, as elucidated by the electrochemical impedance spectroscopy (EIS), indicating the unequivocal role of carbonaceous materials in promoting electron transfer.Moreover, a similar trend was observed for the C dl measurements, confirming that the introduction of a carbon shell induces a homogenous distribution of active TM-based compounds, thereby exposing more active sites.Besides, through inductively coupled plasma-mass spectrometry (ICP-MS) analysis on the electrolytes used for longterm stability tests, a significantly higher Co leaching rate was observed for CoTe compared to CoTe@NC, proving that the carbon shell serves as a protective layer mitigating electrolyte corrosion.
As a consequence, both CoSe 2 @NCNTs and CoTe@NC exhibited significantly improved activity, selectivity, and stability of electrocatalytic 2e − ORR compared with CoSe 2 and CoTe, respectively.On the other hand, it is noteworthy that TM-based compounds, such as CoTe in CoTe@NC, primarily act as the real active species for the progression of ORR.Herein it is necessary to point out that probably based on the above points iii) and iv), most TM-based catalysts coupled with carbonaceous materials can display promising electrochemical activity for 2e − ORR under lower-pH environments, i.e., in the neutral and acid media (as is listed in Table 3), This characteristic adds a layer of interest considering cost and industrialization compared to performance in alkaline solutions.Typically, the electrocatalysis reactions in the neutral media might be exceedingly limited by its poor conductivity, while the acid electrolytes usually lead to the leaching of active components of TM-based catalysts during the ORR process. [29,84]In addition, v) in many cases of TM-based catalysts/carbon composites, the functional groups or dopants attached to the carbon-based phase could serve as additional active sites beyond the ones from TM-based catalysts themselves, [130,185,195] thus further boosting the 2e − ORR electrocatalysis.Nevertheless, despite the notable progress achieved, the productivity of electrosynthesizing H 2 O 2 remains below satisfactory levels when contemplating the practical application scenarios. [108]This is probably because when producing H 2 O 2    and d) the free-standing CoN 4 /VG gas-diffusion cathode (which was subject to the hydrophobic pre-treatment by being exposed into to 0.1% PTFE solution and then heated under an argon atmosphere).Reproduced with permission. [108]opyright 2022, Royal Society of Chemistry.e) Illustration of O 2 adsorbed on the active Zn atoms at the interfaces of ZnO@ZnO 2 , where red and gray spheres correspond to O and Zn atoms, respectively.f) The pulse voltammetry protocol section containing a reductive and oxidative pulse together with the associated current responses.g) Capacitance against different applied potentials derived from the pulse voltammetry of ZnO and ZnO@ZnO 2 in the O 2 and Ar-saturated electrolytes.Reproduced with permission. [196]Copyright 2023, Royal Society of Chemistry.
through bulk electrolysis, those catalysts in the form of powders are always deposited on the collectors (e.g., carbon paper) with the assistance of polymeric binders (especially Nafion), which easily gives rise to the issues of the increased impedance, blocked active sites, insufficient deposited active species, and weak catalyst adhesion. [88,108]To this end, combining the substantial advantages caused by constructing TM-based catalysts/carbonaceous materials heterostructures, depositing such hybrid composites on the conductive supports without any polymeric binder is highly promising to realize the massive H 2 O 2 generation in acid electrolytes with higher economic values. [108]For example, Lu and co-workers specially coupled the vertically aligned graphene (VG) nanosheets with the hydrophobic carbon fiber paper through a binder-free plasma-enhanced chemical vapor deposition, followed by anchoring tetrahedrally coordinated Co single (CoN 4 /VG).When directly serving as a free-standing gasdiffusion cathode (after the hydrophobic treatments) of a flow cell, in the linear sweep voltammetry (LSV) tests (Figure 7a), a high current of -240 mA can be rapidly attained at a low cell voltage (−1.8 V) in the O 2 -saturated 0.1 m HClO 4 .This is signifi-cantly superior to that of the same cell equipped with the counterpart cathode prepared through the conventional drop casting strategy (CoN@CNTs GDE).Besides, the EIS tests uncovered that the self-supported CoN 4 /VG exhibited much decreased R ct and mass transfer resistance (R ms ).Based on these, the productivity for H 2 O 2 electrosynthesis of the flow cell with CoN 4 /VG gas-diffusion cathode was also examined.A satisfactory H 2 O 2 yield rate of 4 mol g −1 h −1 was achieved at −1.8 V cell voltage, realizing the concurrently exceptional activity (high working current) and massive productivity (Figure 7a,b).The above enhancement in catalytic capability was due to the alleviated particle aggregation, increased exposed active sites, accelerated charge transfer, and promoted mass transport brought by the 3D open porous free-standing electrode architecture (Figure 7c,d).Furthermore, constructing heterostructures between the active Co species and carbon materials not only enhances the aforementioned points but also prevents the leaching of active Co atoms during bulk electrolysis in acidic media (where negligible Co amount could be found) thus ensuring the stability of catalytic performance. [108]owever, challenges persist with hybrid TM-based catalyst/carbonaceous materials employed for electrogeneration of H 2 O 2 , particularly the daunting task of determining the real active structure and the operation active sites responsible for the catalysis process.Except for the TM-based catalysts themselves, the defects, functional groups, and dopants of the carbonaceous components can also directly govern the 2e − ORR electrocatalysis. [36]In view of this, very recently, the carbon-free TM-based heterostructures have been also developed as 2e − ORR electrocatalysts. [103,119,196,197]For example, Huang et al. explicitly demonstrated that the unsaturated coordinated Ni created by building heterostructured Ni-based compounds offered active adsorption sites for binding O 2 molecules and thereby displayed an outstanding 2e − ORR ability in 0.1 m KOH. [197]Encouraging progress has been also gained in the area of more practically desirable neutral 2e − ORR, which can be exemplified by the core-shell heterostructured ZnO@ZnO 2 catalyst developed by Kang et al. [196] Through the electrochemical tests, the authors have substantiated that the ZnO@ZnO 2 exhibits significantly enhanced 2e − ORR performances than those of its individual components, which can be ascribed to the formation of the catalytically active Zn sites at the heterostructure interface (Figure 7e).Especially, DFT calculations uncovered that compared to bare ZnO and ZnO 2 , the ZnO@ZnO 2 with active interfacial Zn atoms gained an optimized ΔG HOO* close to the ideal value, as well as the largest ΔG O* , suggesting its best activity and selectivity for 2e − ORR, respectively.Notably, as the kinetic process of the bias-driven electrons, that are transferred to the surface active sites of the targeted catalysts and then consumed by participating in the electrocatalysis (reacting with the O 2 reactants), is typically difficult to be directly detected, the pulse voltage-induced current (PVC) technique was specially employed for ZnO@ZnO 2 and ZnO under ORR electrochemistry.Figure 7f illustrates the PVC protocol sections between anodic (0.75 V vs RHE) and cathodic (0.29 V vs RHE) potentials without iR correction, as well as the associated current responses.Through the integration of the current responses against the cathodic voltage pulses and the subtraction of oxygen reduction background, the charges stored in the catalyst under different potentials were quantified, yielding the corresponding capacitance (including double layer capacitance and pseudocapacitance) values (Figure 7g).Compared to the tests under Ar atmosphere, in the O 2 -saturated electrolytes, the capacitance of ZnO@ZnO 2 and ZnO were both larger at 0.6 V versus RHE, as the surface adsorption of O 2 molecules led to an additional dielectric.Subsequently, unlike Ar-saturated electrolytes where minimal changes were observed (due to the absence of ORR), the capacitance of ZnO@ZnO 2 and ZnO both noticeably increased starting from 0.48 and 0.5 V versus RHE (close to their associated onset potential of ORR), respectively, corresponding to the involvement of adsorbed O 2 molecules in oxygen reduction.Finally, with the shift of the applied potential to the diffusioncontrolled ORR region, the stored electrons tended to be balanced with the consumed ones.Thus, this PVC technique offers a new approach to directly visualizing the kinetic behavior of electrons at the electrocatalyst/electrolyte interface from the initiation and progression of ORR.Moreover, it also elucidates the transition of adsorbed O 2 molecules from dielectric to reactants during the ORR process. [196]Unfortunately, the studies on carbon-free TM-based heterostructures for 2e − ORR remain scarce.

Coordination Tuning
In order to accomplish both exceptional activity and selectivity for a 2e − ORR catalyst, optimizing the electronic structures of the active sites to obtain a favorable interaction with the reaction intermediates is of great significance.Simultaneously, the desirable ensemble effects of the isolated active sites need to be implemented by tailoring their atomic configurations and geometries.Thereby, the binding energies of the key intermediates HOO* could be adjusted to approach the optimum value, and the adsorption of O* can be remarkably diminished. [21,74,204]In view of this aspect, the rational coordination tuning toward the catalytic sites within the TM-based catalysts that range from the bulk to the nano and even atomic scale could be capable of directly realizing both of the electronic regulations and ensemble effects, eventually boosting the catalytic performance for the electrosynthesis of H 2 O 2 .
In recent years, extensive investigations have been conducted on TM-based catalysts with coordination modifications, with a particular focus on those featuring atomic-site structures (as summarized in Table 4).These catalysts are distinguished by atomically dispersed active sites that are isolated with significant interdistances.−207] In addition, with the vast developments of the theoretic simulations, as well as (in situ) XAS and aberration-corrected TEM techniques, more opportunities have been offered to determine the exact atomic structures of such catalysts and their change during and after ORR, as well as their active sites (structures) and reaction mechanism.Beyond the inherent features of bearing isolated active TM sites, the local coordination environments of the TM motifs can be finely altered by modifying either the TM centers or the surrounding atoms, hence presenting an enhanced 2e − ORR catalytic performance. [14,207]Specifically, the former modifications normally comprise the changes in the type and amounts of the TM centers. [204,208]On one hand, in the pioneering work of Strasser et al., [6] a series of TM-based single atoms catalysts (M-N-C, that is, Mn, Fe, Co, Ni, and Cu atomic centers dispersed on the nitrogen-doped carbon matrix via the coordination between metal centers and N atoms) were systematically examined.Among them, the Co-N-C catalyst with Co-N 4 active motif exhibited optimum H 2 O 2 evolution ability in the acid electrolyte.This catalyst demonstrated superior selectivity and activity for 2e − ORR, along with the least H 2 O 2 reduction activity, likely attributed to its optimal binding with oxygen-containing intermediates.Conversely, other M-N-C counterparts tend to favor a 4e − ORR process under identical operational environments.Following this work, Liu et al. further experimentally proved the superiority of Co-based single-atom catalysts for acidic 2e − ORR among the above five TM-based catalysts (N-coordinated TM motifs dispersed on carbon support) (Figure 8b).Of note, they plotted the relationship of the binding energies of ORR intermediates      All potential versus RHE; In addition, the stability, production rate, and FE included herein are based on the bulk electrolysis measurements.
(*O, *OH, and *OOH) with the d-bond center and valence electron numbers of the TM within the TM-based single atom catalysts.As shown in Figure 8c, from Mn to Cu, the number of valence electrons of the TM atoms rose with the energy drop of the d-band center relative to the Fermi level, accompanied by diminished adsorption toward those intermediates.This clearly explained that the optimal binding energies of a Co-based single atom catalyst for an active and selective 2e − ORR can be correlated with the moderate state of its d-bond center (Figure 8d). [204]ased on the above two encouraging works, tuning the type of TM centers to substantially modify the electronic interactions between the targeted atomic-site (especially the single atom ones) catalysts and the ORR intermediates were adapted, leading to very different ORR selectivity and activity.For example, contrary to the Co-based one, the Fe single atom catalyst showed very good 4e − ORR activity and selectivity. [204,209]On the other hand, it is clear that the activity and selectivity of the Co-N-C catalyst demand further improvement to satisfy the requirements of efficient and robust generation of H 2 O 2 under practical working scenarios with high current density.To this end, introducing another or even more metal centers coupled with the single CoN x motif has been confirmed as a feasible solution. [208,210]For instance, Du and colleagues connected another N-coordinated In motif with CoN x one by the bridging O atoms (denoted as CoIn-N-C), as directly visioned by the aberration-corrected high-angle annular dark field scanning transmission electron microscopy (AC-HAADF-STEM) together with the associated Z-contrast analysis.Moreover, the XANES and EXAFS spectra further helped deduce the local coordination environment of CoIn-N-C, where InN 4 and CoN 4 units shared the two bridging O atoms. [208]The most interesting point for this work is that differing from the conventional dual atom catalysts consisting of two TM centers favoring a 4e − ORR pathway, [211−213] the in situ surface-enhanced Raman scattering spectroscopy (SERS), isotope experiments, and DFT calculations revealed that during the acid ORR process, the redox-inactive but O-affinitive In atoms adsorbed OH − from H 2 O molecules.Based on this, the 3d orbital structures of the real active Co atoms were subsequently altered, thereby more thermodynamically favorable reduction of O 2 into H 2 O 2 , even in a more practical three-phase flow cell at a high current density of 100 mA cm −2 . [208]Remarkably, beyond the inherent redox chemistry and oxophilicity of the metals themselves, creating different metal-centered motifs within the carbon matrix introduces more complex coordination environments for metal-N and metal-metal modules.This complexity leads to variations in the binding energies of ORR intermediates, consequently shifting the ORR pathways.This stands in contrast to the situations of most dual or multi-atom catalysts based on the TM centers, where the 4e − transfer process dominates.This can be well illustrated by the trimetallic-atom catalyst containing distinctive Co 2 TMN 8 configuration (each Co was coordinated by 2.5 N atoms, while TM referred to Mn, Fe, Co, Ni, and Cu), [210] which all showed a trend toward 2e − ORR in acid media.When discussing coordination modifications on non-metallic atoms surrounding the atomic metal centers for 2e − ORR electrocatalysis, these modifications can be classified into tuning in the first coordination sphere (CS) as well as the second or farther ones (Figure 8e). [151,205,206,214,215]A well-organized and systematic work by Qiao et al. provides an illustrative example. [205]In their study, they compared the Co coordinated by a two N and two O motif embedded into carbon support (denoted as CoNOC), with counterparts where the central Co was coordinated by four N (denoted as CoNC) and four O (denoted as CoOC) to showcase the impact of coordination tuning in the first CS.The in situ attenuated total reflectance surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) coupled with the DFT simulations signified that replacing the surrounding N atom using O atom can significantly reduce the binding strength of *OOH by the Co center.Interestingly, the increase in coordinated O atoms weakened *OOH adsorption on Co atoms, promoting the migration of the active site.In this case, the O-adjacent C in the second CS was enabled to be responsible for adsorbing/desorbing *OOH.Furthermore, Bader charge analysis revealed that introducing epoxy groups in the second CS of CoNOC further optimized the binding of *OOH by facilitating electron withdrawal from Co atoms (electronic regulation).Simultaneously, this induced that the adsorbed *OOH inclined toward Co, stabilizing the O−O bond of *OOH (spatial confinement). [205]n fact, exerting the composition tuning of the first CS, including altering the type and amounts of the non-metallic atoms, the abovementioned phenomenon has been also widely observed in other TM-based atomic-site catalysts for 2e − ORR beyond the Cobased ones, [19,100,206,209,216−219] i.e., the interaction with the ORR intermediates was finely modified to favor a 2e − ORR process, probably resulting from the charge polarization between the nonmetallic atoms in the first CS and the central TM atoms.The corresponding representative cases include the Fe-C-O motif where O replaces the N atoms within the typical 4e − ORR-favorable Fe-C-N type coordination, [209] the unique terdentate configuration of W 1 N 1 O 2 motif where the central W atom was coordinated with two O atoms and one N atom, [110] as well as the super-coordinated N 4 Ni 1 O 2 unit consisting the single Ni atom with surrounding four N and two O atoms. [19]Moreover, it is worthwhile to point out that apart from the compositional adjustments, very recent works uncovered that the proper tuning from the aspect of coordination position of the first CS is also able to boost the 2e − ORR capability of the TM-based atomic-site electrocatalysts. [207,220]Shi et al. compared two types of Co-based single atom catalysts, i.e., Co 1 −N 2 and Co 1 −N x motifs anchored at the edge and in-plane of the carbon supports, respectively, and found the former displayed a much improved 2e − ORR performance.The in situ XAFS and attenuated total reflectance surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) together with the DFT calculations elucidated that compared with the rigid and stable basal Co 1 −N x units.The investigation revealed that the edge-hosted Co 1 −N 2 can be flexibly self-adapted into higher-valence Co 1 −N 2oxo with peripheral oxygen functional groups during the ORR process (Figure 8f,g).This dynamic transformation optimized the electronic state of Co atoms, resulting in a desirable *OOH adsorption (enhancing activity) and increased activation barrier for the O−O bond dissociation of this intermediate (enhancing selectivity). [220]esides, as demonstrated in the aforementioned study by Qiao et al., [205] adjusting the coordination in the second or even more distant CS of TM-based atomic-site catalysts by incorporating the oxygen-containing group has been also considered as an effective approach to elevating the electrosynthesis of H 2 O 2 from ORR.This has been widely confirmed by other reports com-bining computational modeling with experimental analysis, emphasizing the practicality of this approach. [151,214,215]For example, Han and Kim et al. developed a distinctive Co-based singleatom catalyst consisting of high-coordinated Co-N 5 moieties surrounded by epoxy groups (Co-N 5 -O-C).This catalyst exhibited exceptional 2e -ORR activity and selectivity, surpassing its epoxyabsent counterpart (Co-N5-C) across wide-pH electrolytes.Impressively, in the alkaline media, the flow cell containing the Co-N 5 -O-C catalyst showed an excellent H 2 O 2 production rate (11.3 mol g -1 h -1 ) and FE (82.5%) at an industrial-relevant current density of 200 mA cm -2 , again outperforming the cell with Co-N 5 -C.In addition to the experimental findings, the computational simulations disclosed that the electron-withdrawing epoxy groups effectively modulate the electronic state of the Co center thereby facilitating the 2e − ORR route thermodynamically and kinetically favorable, thus contributing to significant enhancements in catalytic performance.Besides, this work highlights that modifying the amounts of epoxy oxygen in the second or farther CS can also influence the presented 2e − ORR performances. [151]owever, owing to the complex surface properties and the diverse forms oxygen could exist, determining the exact kind of oxygen functional group contributing to the electrogeneration of H 2 O 2 is of great significance.In this context, Lu et al. specially prepared different common types of surface oxygen functional groups in the second or farther CS of the single atom catalysts composed of CoN 4 motifs and carbon nanotube substrates.Based on their experimental findings, they determined that the enhancements in 2e − ORR were mainly attributed to the synergies between CoN 4 units and the adjoining epoxy groups.Notably, epoxy groups played a more pivotal role in promoting H 2 O 2 production than that of other oxygen functional groups, such as ketonic, hydroxyl, and carboxyl groups. [214]According to these examples, one might conclude that the amounts and types of the incorporated species in the second or farther CS would be both influencing factors on the resulting 2e − ORR abilities of the TMbased single atom catalysts.When realizing an optimum tuning, the electron structure of the active center can be effectively modified, then the ORR can be directed to prefer the 2e − pathway thermodynamically and/or kinetically.Interestingly, in addition to the oxygen-containing groups, doping other species, e.g., phosphorus and sulfur atoms, into the second or farther CS can also electronically modify the metal centers, thereby boosting the 2e − ORR ability of the targeted single atom catalysts, [152,221] including the one based on Co center. [152]lthough coordination tuning toward the TM-based atomicsite catalysts, particularly the single atom ones, has gained impressive advances, experimentally capturing and clarifying the real active structures (motifs) and the catalytic sites of this type of catalysts during the ORR process are still tricky.This challenge arises from the surface complexity including the dynamic reconstruction of metal-centered motifs and oxygenation of carbon species under electrochemical conditions. [207,214,220]Consequently, recent focus has also shifted toward modifying the atomic coordination of the well-defined TM-based nanocompounds with simper surface situations. [4,74,187,197,222]−227] This leads to effective i) geometric effect, ii)  [19] Copyright 2022, Wiley-VCH.b) Faradic efficiency of H 2 O 2 production from ORR for TM-based single atom catalysts (Mn-NC, Ni-NC, Cu-NC, Fe-NC, and Co-NC), as well as reference sample nitrogen-doped carbon (NC).c) Relationship of the binding energies of ORR intermediates with the d-bond center and valence electron numbers of the TM within such five TM-based single atom catalysts, as well as the associated d) plot of the limiting potential (U) as a function of ΔG *OH .Reproduced with permission. [204]Copyright 2019, Elsevier.e) Schematic of a typical single atom catalyst with highlighted center metal, as well as the first and second CSs.Reproduced with permission. [205]Copyright 2021, American Chemical Society.f) Co K-edge XANES spectra and g) k 2 -weighted FT EXAFS spectra in R-space for edge-hosted Co 1 −N 2 (denoted as CoNOC) under ex-situ and open circuit potential (OCP), as well as the in situ states including at onset potential (0.65 V), at 0.4 V, and after reaction (AR).Inset of f): the differential spectra obtained after subtracting the ex-situ XANES spectrum signal from the other four ones.Reproduced with permission. [220]Copyright 2023, Wiley-VCH.electronic modification, as well as iii) optimization of the interfacial electron transfer kinetics, eventually boosting the resulting 2e − ORR capability.Specifically, first of all, the distinctive geometry featuring isolated binding sites to destabilize the adsorption of O* binding can maintain the O−O bond of OOH*, improving the selectivity of converting O 2 to H 2 O 2 . [4,74]Exemplified by a series of TM-based selenides, i.e., CoSe, NiSe, and Cu 2-x Se, [74] these compounds consist of central TM metals coordinated by Se atoms, eliminating the hollow sites constructed by successive three or more metal atoms (as illustrated on the reference sample, pure metal Pt) (Figure 9a).Note that such hollow sites are the typical O* adsorption sites on the surface of pure metal catalysts.Isolating the active TM atoms with the weak O-affinity Se atoms endows O* intermediates to preferentially adsorb on the top of TM sites, resulting in weaker O* adsorption and disrupting the linear scaling relationship between ΔG OOH* and ΔG O* (Figure 9a,b).Following this work, the report from Shi et al. further proved that the geometric effect of the isolated active sites can be maximized by coordinating different inactive atoms.Among the developed nickel chalcogenides (NiO, NiS 2 , and NiSe 2 ), the Se 2− anion, owing to its larger size, effectively separates active Ni atoms, increasing the distance from For comparison, the linear scaling relationship between ΔG OOH* and ΔG O* (O* adsorbed on hollow (top) sites) for FCC metals was plotted using the orange (purple) solid lines.Reproduced with permission. [74]Copyright 2021, American Chemical Society.c) The calculated PDOS of Nb within different TM-based columbites.Reproduced with permission. [225]Copyright 2021, Wiley-VCH.TPV spectra and the associated decay times for d) Ni 3 B and Ni 2 B, as well as e) NiB, NiB 2 , and NiB 3 .Reproduced with permission. [4]Copyright 2022, The Authors.f) Crystal structures with the corresponding lattice parameters of c-CoSe 2 and c-NiSe  100) at the standard equilibrium potential of 2e − ORR.In addition, the yellow shaded regions represent the potential window in which the clean catalyst surfaces free of adsorbates had lower free energy than that of the catalyst surfaces adsorbing O* and/or OH*.Also, the inset scheme of atomic structures illustrates that one O* and one OH* were adsorbed on their preferential binding sites of the cell surfaces, consisting of two TM and four Se sites.Green, magenta, orange, red, and white spheres represent Ni, Co, Se, O, and H atoms, respectively.Reproduced with permission. [38]Copyright 2022, The Authors.
neighboring Ni atoms and enhancing the selectivity of H 2 O 2 evolution from ORR electrocatalysis. [224]Second, the electronic regulations can directly alter the interactions of the catalytic sites with the O-containing intermediates, by which an improved activity and selectivity for H 2 O 2 electrosynthesis can be attained. [225,227]or instance, Liu et al. reported that among the binary MOFs comprising active Ni centers coordinated with different metal atoms, the Zn atoms especially induced the electron-deficient Ni atoms with a higher oxidation state, thereby optimizing its adsorption toward the OOH* intermediates. [227]Using the projected density of states (PDOS) calculations (Figure 9c), Chen and co-workers also revealed that different TM atoms (i.e., Mn, Fe, Co, Ni, and Cu) can shift the d-band center of active Nb atoms within the TM-based columbites (TMNb 2 O 6 ) nanocrystals, directly affecting the ORR activity and selectivity. [225] 9e, in which their signal intensity even dropped to values less than zero over time, indicating a pronounced electron-trap effect on their surfaces.Analyzing the decay time and trends of the TPV spectra of these five samples, overall, the NiB 2 exhibited the slowest interfacial transfer kinetics.Therefore, NiB 2 achieved the most limited electron feeding to the catalyst surface, inhibiting the rapid and sufficient reaction with O 2 reactants and OOH* intermediates.As a result, the 4e − ORR selectivity was lowered to the maximum extent. [4]Finally, it is imperative to emphasize that the stability of H 2 O 2 electrosynthesis, a determining factor for practical applications, has long been neglected.Beyond the above three points mentioned earlier focusing on improvements in catalytic activity and selectivity, the role of compositional adjustment in promoting the durability of 2e − ORR elucidated by the brilliant work from Jin et al. deserves special consideration. [38]They computed and compared the surface stability under aqueous electrolytes for the two cubic pyritetype selenides where Se was coordinated with Ni (c-NiSe 2 ) and Co (c-CoSe 2 ), respectively (crystallizing in the same space group Pa-3, details in Figure 9f).Based on the change of the Gibbs free energy corresponding to the adsorption of O* and/or OH* on the most thermodynamic stable catalyst surface equilibrated with water, they established the relationship between the applied potential and the electrolyte pH. Figure 9g illustrates that the Se sites of c-CoSe 2 preferentially adsorbed O*, while the Se of c-NiSe 2 had a significantly lower O* affinity. [38]The shift of the preferential O* adsorption site from Se to Ni evidently contributed to stabilizing the c-NiSe 2 structure, as the decomposition of pyrite-type structures would be initiated primarily by the oxidation of dichalcogenide anions, followed by the leaching of the TM cations. [29,38]lso, in both of the probed selenides, Ni exhibited a weaker binding toward OH* than that of Co, which enabled the c-NiSe 2 to remain an adsorbate-free surface over a wider potential range (Figure 9g). [38]The above surface adsorbate analysis based on the DFT simulations can be well corroborated by solid experimental evidence.In particular, during RRDE stability tests, both TM and Se in c-NiSe 2 exhibited slower leaching rates compared to those in c-CoSe 2 .This data concludes that tuning the coordinated TM can substantially enhance the intrinsic resistance of TM-based compounds to surface oxidation and corrosion during operational 2e − ORR in aqueous environments.This conclusion holds true not only for altering coordinated TM atoms but also for changing the chalcogen atoms coordinated with fixed TM atoms for acid 2e − ORR, as demonstrated in another interesting study by Jin and his co-worker demonstrated (c-CoS 2 vs c-CoSe 2 ). [77]part from implementing coordination tuning by the composition adjustments, building coordinatively unsaturated or lowcoordinated active TM centers within the associated TM-based nanocompounds can also elevate the resulting 2e − ORR ability.This is predominantly because, it significantly increases the number of electrocatalytically active sites for the adsorption of O 2 molecules and tunes the electronic structure of the active TM sites, optimizing their ability to bind O-containing intermediates. [187,197,222]A notable example of this is the catalyst developed by Luo and colleagues, where the catalysts of Cu nanowires-supported CoS x were developed.The predicted DOS calculations elucidated that the lower the coordination number of S atoms against the Co center, the more upshift of the dband energy level for CoS x , implying the decreased filling of the antibonding Co-3d orbitals, by which the adsorption toward the O 2p orbitals was accordingly enhanced.The optimum catalyst (one Co atom was coordinated with four S atoms) exhibited moderate Co-O bonding, thus possessing the best 2e − ORR performance. [153]It's essential to highlight that ORR-driven structural reconstructions are likely to occur in nanocompounds with coordinatively unsaturated or low-coordinated central TM atoms, and the extent of reconstruction can be influenced by the pH values of the electrolyte. [153,197,222]Despite the reconstructed TM-based (oxy)hydroxides having been claimed to be beneficial to oxygen adsorption and ORR stability, [197,222] the associated underlying mechanisms and other concerns, such as whether the reconstruction would continuously proceed with the occurrence of ORR, as well as how deep the reconstruction degree is favored for 2e − ORR, still remain unanswered.

Crystalline Design
The crystallinity of the TM-based compounds serves as a crucial indicator of the long-range ordered degree for those atoms that are periodically arranged within their cells, which is essential to determine their presented crystallographic, surface morphological, and electronic structural properties, such as lattice parameters, surface porosity, and electric conductivity. [142,230,231]anipulating the crystallinity of these compounds can significantly impact their electrocatalytic capabilities in various applications. [142,230,232,233]Especially, among most of these reported works, TM-based compounds the more amorphous or lower-crystallinity, lacking the long-range order, tend to perform better than those of their crystalline counterparts, as they are characteristic of the interesting local symmetries and unique electronic structures, as well as more exposed active defect sites (such as the element vacancies or unsaturated coordinated TM centers). [142,233,234]In view of these points, one can naturally envision that the rational crystalline design holds substantial promise for enhancing 2e − ORR ability of the TM-based compounds.Numerous studies, as summarized in Table 5, highlight that reducing crystallinity remarkably enhances the 2e − ORR ability.Herein, we have outlined three key aspects of reducing crystallinity to improve the H 2 O 2 evolution performances.First, by reducing crystallinity, more electrochemically active sites for O 2 adsorption and protonation become exposed on the catalyst surfaces, [88,149,234] which is particularly beneficial to improve the catalytic activity. [4]For example, in our previous work, when comparing the amorphous Ni 3 B with its crystalline counterparts, the former revealed a remarkably larger value of double-layer capacitance (C dl ) during the ORR process, meaning the amorphous structure possessed more electrochemically active sites for the electroreduction of O 2 . [4]The phenomenon that the amounts of exposed active sites represented by the C dl values increased with the reduction of the crystallinity of the TM-based compounds can also be observed for a series of 2D Ni-based MOFs electrocatalysts, Ni 3 (HITP) 2 (HITP means 2,3,6,7,10,11-hexaiminotriphenylene). [235]Nevertheless, we have to acknowledge that the enhancement in catalytic activity induced by low crystallinity has also been widely reported on those lowcrystallinity/amorphous TM-based ORR catalysts that favor a 4e − pathway. [230,232,236]This indicated that the selectivity of the ORR electrocatalysis for TM-based compounds could be synergistically influenced by many other factors, especially when revisiting the functional relationship of the U T with ΔG HO* /ΔG HOO* (as depicted in Figure 2a).Consequently, depending on the characteristics of the target compounds, the effects on the ORR selectivity should be always taken into consideration when creating more catalytically active sites by reducing their crystallinity.Second, the electronic structures of TM-based compounds can be significantly tuned through crystalline design, and unsurprisingly, the reaction energetics from both thermodynamic and kinetic aspects are correspondingly regulated, elevating both activity and selectivity of 2e − ORR. [89,153,234]This is probably ascribed to the emergence of the unique active coordination motifs after crystallinity reduction or amorphization, which are inclined to exhibit unique coordination symmetries and unusual electronic states of the TM atoms. [89,153,234]Taking an interesting work from Hong and co-workers as an example, they successfully fabricated amorphous and crystalline NiO nanosheets (named a-NiO and c-NiO NSs, respectively).Using a combination of hard and soft XAS analysis, they revealed that the rich unsaturated coordinated Ni 2+ centers in short-range ordered a-NiO NSs caused the symmetry change of the ligand field compared with that of long-range ordered c-NiO NSs.Especially, from the simulations of Ni L-edge XANES spectra for these two samples (Figure 10a), the Ni 2+ cations of c-NiO NSs merely occupied the NiO 6 octahedral (O h ) sites, yet for the a-NiO NSs, they exhibited a prime filling in the NiO 5 pyramidal (C 4V ) sites.Based on this, the corresponding electronic configuration of central Ni 2+ in the amorphous sample tremendously differed from the crystalline one, as upon extracting the apical O from a NiO 6 octahedron, its e g orbitals could break degeneracy caused by the orbital polarization.On the contrary, confined by the atomic spatial position, the orbital energy levels of t 2g orbitals were relatively less changed (Figure 10b), which can be further confirmed by the deformation   respectively.Reproduced with permission. [234]Copyright 2022, The Authors.The adsorption configurations of HOO* on e) the crystalline NiO x and f) the amorphous NiO x -C.The black, red, brown, and pink spheres corresponded to Ni, O, C, and H atoms, respectively.Reproduced with permission. [187]opyright 2022, American Chemical Society.
charge density and PDOS calculations (Figure 10c,d).Note that the calculation results revealed that for the e g comprised by the dz 2 and dx 2 -y 2 orbitals, which are located in the z orientation, their electron occupied states changed significantly in NiO 5 pyramids compared with those in NiO 6 octahedrons.In detail, the non-occupied orbital level of dz 2 experienced a substantial drop to ≈2 eV, coinciding with a simultaneous rise in the occupied orbital level around the Fermi level.Therefore, this suggests that the shift of electrons from the occupied orbital level to the nonoccupied one could be more easily driven by an external potential.The situations in the dx 2 -y 2 orbitals were in analogy to those in the dz 2 ones (Figure 10c,d).Note that through adjusting the occupying states of the dz 2 and dx 2 -y 2 orbitals, the targeted TM will accordingly show an altered binding ability toward the oxygen-containing intermediates.In the context of this study, the HOO* adsorption was strengthened for a-NiO NSs with NiO 5 pyramidally local structures, resulting in an ameliorated 2e − ORR activity. [234]Third, diminishing the crystallinity could lead to a 2e − ORR-favorable intermediate (HOO*, the only key intermediate for 2e − ORR) adsorption configuration, promoting selectivity for H 2 O 2 production. [187,234] 10e).Instead, in the case of the amorphous catalyst, HOO* was adsorbed in an end-on manner, with its distant O being free and situated far from the adsorption sites (Figure 10f).Benefiting from this configuration, the distance O−O bond was merely increased, which was highly beneficial to the preservation of O−O bond and facilitating selective 2e − ORR. [187]owever, the reduction in crystallinity or amorphization typically inevitably introduces complexity to the initial well-defined TM-based compounds, leading to uncertainties in both surface and bulk structures and properties.Moreover, rationally establishing the calculation models for those low-crystallinity or amorphous compounds is a formidable challenge, even though the theoretic simulations are considered as a powerful tool to interpret the origins of the superior activity and/or selectivity of the 2e − ORR electrocatalysts.To this end, based on the coordination information gained from the EXAFS findings, utilizing the variable-cell DFT-based molecular dynamics (vc-DFTMD) to build, optimize, and finally determine the models for amorphous TM-based compounds is probably a promising solution. [89]Moreover, considering that the amorphous TM-based compounds exhibit short-range ordered arrangements of atoms, creating cluster models represents a reasonable option.Note that in order to estimate the rationality of the as-built cluster models, it is necessary to calculate the binding energy per atom (deducing the required energy to break atomic bonds in the clusters). [4]Overall, implementing the phase engineering toward TM-based 2e − ORR catalysts via their crystalline design still remains a relatively underexplored field, deserving increased attention.

Polymorphic Transformation
Polymorphic transformation refers to a single material crystallizing in different structures, all sharing similar chemical compositions, but exhibiting distinct atomic arrangements and electronic structures. [167,237,238]Such different polymorphs exhibit varying catalytic performances in different electrocatalysis fields, including water splitting, [165,239] N 2 reduction, [240] O 2 reduction (4e − pathway), [241] and so on.Specifically, in the field of ORR, the early pioneering studies have already disclosed that the different spatial arrangements of active atoms and varying electronic structures within TM-based compounds of the same composition and stoichiometry induce differences in O 2 adsorption and activation, thus significantly impacting the reaction pathway of the following ORR pathways. [242]Following this, recently, a couple of intriguing reports focusing on the developing diverse polymorphs of the TM-based catalysts for highly active, selective, and stable 2e − ORR electrocatalysis have been reported, and an insightful understanding of the underlying mechanisms have been obtained (Table 6).Among them, the most extensively investigated category is TMD catalysts, probably due to their distinct 2D crystal structures.They typically consist of a TM plane enclosed by two chalcogen planes which are enlarged into infinite layers and bonded to each other through van der Waals force.By the redistribution of their atom stackings, TMDs can easily exist in the forms of rich polymorphs with varied inherent properties. [165,243]For example, the layered MoS 2 tends to crystallize in the two polymorphs, the octahedral 1T and trigonal 2H phases according to the arrangement of its S atoms.These two phases can transform between each other through the gliding of the intralayer atomic planes, and yielding completely different electronic structures.Particularly, the 2H-MoS 2 possesses a larger bandgap of 1.9 eV and a semiconductive feature, while the bandgap of 1T-MoS 2 is smaller, and thus displays metallic properties. [243]According to this point, Perivoliotis et al. intentionally introduced the metal-like 1T-MoS 2 into the TM (Fe, Mn, and Co)-based porphyrin/molybdenum disulfide nanocomposites to elevate the overall electric conductivity, and correspondingly, achieving a superior 2e − ORR performance. [103,200]In addition to the improvement of the ability of the electron transfer caused by the polymorphic engineering of TMDs, Sun et al. further found that the carbonaceous material coupling-induced polymorphic transformation of cubic c-CoSe 2 into its orthorhombic phase (o-CoSe 2 ).This transformation can concurrently lead to the reduction of the size of the nanoparticles, which facilitates the dispersive distribution of the catalyst with more exposed active sites, therefore contributing to enhancing the resulting 2e − ORR capability. [130]Apart from the above two reports related to the catalytic activity, Jin et al. especially studied the influences of different polymorphs of TMDs on the selectivity of 2e − ORR.ORR selectivity can be ensured. [77]esides TMDs, very recently, tuning the polymorphs of TMbased oxides was also demonstrated to be feasible to boost their abilities of H 2 O 2 electrogeneration from ORR. [8,237] In our previous work, by increasing evenly distributed metal components, a high-entropy perovskite oxide ceramic Pb(NiWMnNbZrTi) 1/6 O 3 was developed from the Pb(ZrTi) 1/2 O 3 prototype.The Rietveld refinement of the powder XRD patterns unveils that the pure perovskite oxide phases can be identified for both Pb(NiWMnNbZrTi) 1/6 O 3 and Pb(ZrTi) 1/2 O 3 , while they crystallize in the cubic structure (Pm 3m) and the lower-symmetry tetragonal structure (P4mm), respectively (Figure 11a,b).Note that such polymorphic transformation induced the strong release of lattice strain and thus an amelioration of the structural stability, as verified through the Williamson-Hall method in Figure 11c.In addition, the TPV findings imply that, compared to the tetragonal Pb(ZrTi) 1/2 O 3 , the cubic Pb(NiWMnNbZrTi) 1/6 O 3 exhibited a conspicuous electron accumulation effect on its surface, which therefore significantly suppressed the rapid and sufficient binding reaction between the surface active sites and Ocontaining intermediates.Benefiting from this, the 4e -pathway was inhibited, and conversely a 2e -pathway of ORR was more preferred for Pb(NiWMnNbZrTi) 1/6 O 3. Furthermore, for the first time, the finite elemental analysis (FEA) simulations, which can reflect the physicochemical characteristics at a mesoscopic scale, were adopted in the field of 2e -ORR electrocatalysis and modeled the charge distributions on the surfaces of the tetragonal Pb(ZrTi) 1/2 O 3 and cubic Pb(NiWMnNbZrTi) 1/6 O 3 .The results illustrate that the former catalyst possessed a more positively charged surface than that of the latter one, through which, the Pb(NiWMnNbZrTi) 1/6 O 3 embraced a weakened binding toward the OOH -intermediate, conducive to the selective 2e -ORR. [8]part from the enhancement of the structural stability and surface electron migration kinetics, the porosity and pore size of TM-based oxides can be optimized through polymorphic transformation, beneficial to the mass transport during 2e -ORR. [237]oreover, similar to TMDs, thermodynamical and kinetical improvements in the reaction energies during the ORR process can also be achieved by tuning the polymorphs of TM-based oxides. [237]ast but certainly not least, it is worthwhile to point out that among the currently reported 2e -ORR TM-based catalysts that are involved in polymorphic transformation, intentionally stabilizing a compound in its metastable phase is of high applicability and significant potential.This is because the construction of a metastable phase could bring the modified compound an unequilibrated surface with high reactivity, featuring interesting atomic arrangements and electronic structures, therefore leading to different inherent catalytic properties compared with its "stable counterpart phase". [135,158]Our group first discovered that the formation of metastable hexagonal SnO 2 (h-SnO 2 ) with a space group of P6 3 /mmc synthesized through microwaveassisted mechanochemical-thermal method can serve as a highly active and selective 2e -ORR electrocatalyst in the neutral media.Strikingly, it outperforms the conventional Rutile-SnO 2 phase with a space group of P4 2 /mnm (136), which can be attributed to the optimized free binding energy of HOO* and higher O* formation energy from the dissociation of HOO* based on the h-SnO 2 . [135]Inspired by this study, the utilization of the metastable phase of the TM-based compounds was also proposed and realized for neutral 2e -ORR electrocatalysis.Shao et al. developed a unique metastable Ni, whose atoms were stacked in a hexagonal close-packed (hcp) manner, distinctly differing from more commonly reported stable face-centered-cubic (fcc) Ni (Figure 11d).More importantly, addressing at the challenges faced by most metastable materials, which tend to revert to a stable state under external energy, the authors ingeniously doped P into hcp Ni (P-hcp Ni, illustrated in Figure 11d).This doping strategy elevated the relative energy barrier for the phase transformation, addressing the unsatisfactory phase stability under ORR operational environments.Notably, P-hcp Ni retained its hexagonal close-packed feature of metastable Ni even after exposure to strong external activation, as evidenced by the spherical AC-HAADF-STEM pattern (Figure 11e).The enhanced phase stability of P-hcp Ni, compared to the non-doped counterpart, was further evident under strong external activation (high-temperature calcination under H 2 atmosphere) (Figure 11f).Also, the incorporation of P not only improved the electronic structure of the pristine hcp Ni but also fine-tuned its adsorption strengths of the oxygen-containing intermediates.Consequently, P-hcp Ni exhibited an exceptional H 2 O 2 electrosynthesis ability, yielding an outstanding H 2 O 2 production rate of 4917.2 mmol g −1 h −1 in a flow cell using pure water as the electrolyte with a long duration of 140 h. [158]From the above two important studies, it becomes evident that building the metastable polymorph of TM-based compounds holds great promise for obtaining an efficient, selective, and durable 2e -ORR electrocatalysis.However, the related investigations remain relatively scarce, demanding further research attention and exploration.Reproduced with permission. [8]Copyright 2022, The Authors.d) Illustrations of crystal structures of stable fcc Ni (ABCABC stacking), as well as metastable hcp Ni and metastable P-hcp Ni (both ABABAB stacking).The different blue spheres all represent Ni atoms located at different layers while the pink spheres represent P atoms.e) The spherical AC-HAADF-STEM pattern of P-hcp Ni with a six-fold symmetry.f) The illustration of P-doping to stabilize the metastable hcp Ni.Reproduced with permission. [158]opyright 2023, Wiley-VCH.

Conclusions, Challenges, and Perspectives
In this comprehensive review, we delved into the recent advancements in the growing field of phase engineering of non-noble TM-based electrocatalysts for efficient 2e − ORR.Our focus has been particularly directed toward elucidating the intricate relationships between the phase characteristics (structure and composition) and the corresponding electrocatalytic performance (selectivity, activity, stability).The critical role played by phase engi-neering strategies in shaping these relationships has been thoroughly explored, shedding light on their significance within the broader system.This collective knowledge is expected to pave the way for the development of highly efficient TM-based electrocatalysts with optimized 2e − ORR performance.The progress observed in phase engineering has already started to unlock new possibilities for the practical application of TM-based 2e − ORR electrocatalysts.So far, the topic of phase engineering-driven TMbased electrocatalysts to boost 2e − ORR performance is still in its infancy.Far more theoretical modeling toward the identification of descriptors is needed to shed new light on the pivotal role of phase engineering with respect to the predictive and systematic design of such materials.We believe that our elucidation of underlying influences of various phase engineering strategies on the TM-based electrocatalysts can serve as a guideline for the research community, advancing the development of electrosynthesis of H 2 O 2 in an effective, applicable, and sustainable manner.Furthermore, to better enrich the connotation of phase engineering in the field of electrosynthesis of H 2 O 2 , we have outlined the primary experimental and technical challenges and opportunities in the following (Figure 12).

Enhancing Catalytic Properties Through Carbon Hybridization:
To enhance the catalytic properties of the TM-based electrocatalysts, many carbon materials were introduced and hybridized with them to improve the conductivity and dispersity.However, the surface of carbon materials may be easily modified or functionalized with oxygen-containing functional groups during phase engineering or electrochemical processes, resulting in its high intrinsic catalytic ability toward 2e − ORR. [207,214]Thus, clarifying which subunit in the hybridization system contributes to which aspect of the ORR performance (selectivity, activity, and stability) is highly desired.−250] Nonetheless, most of the reported efficient carbon-based catalysts are still confined under alkaline working environments. [26,247]Meanwhile, inspiring progress has been also achieved in determining the dominant active sites of such carbon-based catalysts for 2e − ORR, beneficial for gaining a more in-depth understanding of the associated reaction mechanism. [26,36,244]However, as mentioned above, the easy-to-oxygenate nature with dynamic structural transformations could take place for carbon-based catalysts, [207,246] inducing highly complex surface situations.Therefore, identifying the real active sites during ORR electrochemistry is of utmost importance for carbon-based catalysts in the near future.Such insights can pave the way for more rational designs of these catalysts, leading to enhanced 2e − ORR performances.Alkaline and Acidic/Neutral 2e -ORR Challenges: Most TM-based catalysts can deliver a satisfactory alkaline 2e -ORR capability after phase engineering.However, H 2 O 2 cannot stably exist in the alkaline media, as it tends to convert into HO 2 − .Compared to the alkaline electrosynthesis of H 2 O 2 , producing H 2 O 2 in neutral and acid media presents an attractive avenue as H 2 O 2 in neutral and acid media can be stabilized and converted into hydroxyl radical ( • OH), respectively.The former scenario holds significant promise, offering more opportunities for the direct utilization of earth-abundant pure water as the electrolyte, thereby presenting substantial economic prospects. [158]Conversely, the latter scenario, involving the generation of hydroxyl radicals, holds an inherent benefit for onsite water disinfection and tandem environmental remediation.Besides, acid 2e -ORR can be performed in the proton-exchange membrane cells. [29,251]However, most TM-based catalysts suffer from large ohmic loss and slow reaction kinetics in neutral environments, delivering poor performance.Moreover, these catalysts often exhibit poor 2e -ORR performance in acidic solutions due to corrosion, as well as an inherent preference for the formation of H 2 O (elaborated in section: 2.1.Fundamentals of 2e − ORR).In this regard, employing more innovative and robust phase engineering approaches emerges as a promising and potentially transformative solution to tackle this challenge.These insights aim to provide a roadmap for researchers, offering a holistic perspective that addresses both the current hurdles and potential avenues for discovery in this evolving field of ORR.Scaling up 2e -ORR Performance for Practical Application: Up to now, the majority of the reported TM-based catalysts were operated only at the laboratory level, delivering low current densities for ORR into H 2 O 2 .Achieving sustained current density higher than 100 mA cm -2 , and maintaining high Faradaic efficiency during prolonged electrolysis is crucial for practical application. [19,65]To achieve this goal, in addition to innovative electrochemical reactor designs, such as GDE and membrane, urgent attention is required to optimize the activity, selectivity, and durability of deposited TM-based catalysts through phase engineering.Structural Dynamics During ORR: The dynamic variations in oxidation state and local coordination of designed TM species during ORR, evidently differing from those of the pre-and post-ORR, are attributed to the high redox activity of TM species during electrochemical processes. [44,89]Thus, identifying the structural evolution of TM-based catalysts before, during, and after ORR necessitates simultaneous detection using multiple in situ spectroscopy techniques (e.g., XAS, Raman coupled with FTIR), but it remains a formidable challenge.Dependence of Phase Component and Structure: Although Pourbaix diagrams indicate changes in the chemical state of most TM species over a wide pH range during ORR, [29,252] some reports claim stability in the phase structure and components of TMbased compounds post electrochemical ORR.The results seem contradictory.This apparent contradiction may be attributed to near-surface reconstruction, not easily identified by conventional characterization methods.Therefore, the dependence of phase components and structure on pH condition, temperature, and applied potential against ORR deserves to be further explored and clarified by more investigations.Meanwhile, in addition to the TM-based electrocatalysts, other non-noble metalbased compounds such as rare earth-based ones, are an intriguing area for 2e − ORR electrocatalysts.

Role of DFT Calculations in Understanding TM-Based Catalysts:
DFT calculations are widely adopted to provide more insights into the electronic structures, conductivity, adsorption free energy diagram of intermediates, and reaction kinetics energy barrier during ORR of those TM-based catalysts under various phase engineering strategies.Accordingly, establishing a reasonable calculation model based on the well-understood structure is quite important and necessary, and correspondingly, many factors should be taken into consideration, especially the real active structures/sites during the dynamic electrochemical ORR process, as well as the structural stability after long-term reactions.

Figure 1 .
Figure 1.Phase engineering strategies for the construction of advanced TM-based electrocatalysts for 2e − ORR.
account, as demonstrated by Jin et al. employing the well-defined CoS 2 model.Briefly, in their study, they observed the adjoining two binding Co sites of CoS 2 are separated by disulfide anions, revealing an interatomic distance of 3.941 Å, much longer than that of O−O bond of HOO* (Figure 2c, left).In this regard, to thermally break the O−O bond by the two closest Co sites, the elongation of O−O bond and the lattice distortion of CoS 2 with shortened Co−Co distance are essential prerequisites, which will accordingly lead to a high activation barrier.Consequently, this process incurs a high activation barrier, making the retention of the O−O bond of HOO* during ORR more likely, favoring the 2e − reaction pathway (Figure

Figure 3 .
Figure 3.The proposed four design protocols for improving the 2e − ORR performance of TM-based electrocatalysts.

Figure 4 .
Figure 4.The derivative ideas of phase engineering strategy and some of their typical forms based on TM-based 2e − ORR electrocatalysts.
DFT calculations by Liu et al. demonstrated that proper B atom doping into Ni nanoparticles modified the DOS of Ni-3d orbital, shifting its d-band center closer to the Fermi level, which facilitated the adsorption of O 2 and

Figure 5 .
Figure5.a) The DOS of pristine TiO 2 (p-TiO 2 ) and Mn-doped TiO 2 (Mn-TiO 2 ).Reproduced with permission.[150]Copyright 2021, American Chemical Society.b) The Co K-edge XANES spectra and c) k 2 -weighted FT-EXAFS spectra of CoN 4 -C and CoN 4 -PC, as well as references (Co foil and CoO).The inset of b): the fitted oxidation state of Co from the associated XANES spectra.d) The k 2 -weighted Co K-edge WT-EXAFS spectra of CoN 4 -C and CoN 4 -PC, as well as the reference Co foil.e) Schematic of the tensile strain toward the Co-N coordination induced by P doping.Reproduced with permission.[152]Copyright 2023, Elsevier.f) The charge density difference of Mn-doped CuS (Mn-CuS), where the cyan and yellow colors represent the charge depletion and accumulation in the space, respectively.g) The OOH* adsorption energies and the associated charge states of the active sites for CuS and Mn-CuS.Reproduced with permission.[88]Copyright 2023, Elsevier.h) Free energy diagram for 2e − ORR electrocatalysis at different possible catalytic sites of (101) Cu-TiO 2 , including Cu doping-induced lower-valence Ti 4c 3+ , Ti 5c 3+ , and their synergy (Ti 4c-5c 3+ ), as well as Cu dopant, and Ti equivalent to the one within the pristine TiO 2 .i) The corresponding ΔG O* at these sites (Ti 4c 3+ , Ti 5c 3+ , Ti 4c-5c 3+ , Cu, and Ti).Reproduced with permission.[98]Copyright 2022, Tsinghua University Press and Springer-Verlag GmbH Germany.
All potential versus RHE; aCoupled with 2e − water oxidation reaction (WOR); In addition, the stability, production rate, and FE included herein are based on the bulk electrolysis measurements.

Figure 6 .
Figure 6.a) Illustration of O 2 adsorbed on the surfaces of {001} facets-exposed -Fe 2 O 3 with and without oxygen vacancy (O v ), as well as the associated O−O bond lengths in both cases.Reproduced with permission.[172]Copyright 2020, Wiley-VCH.b) In situ XANES spectra (Cu K-edge) of A-Cu(OH) 2 /GO catalyst recorded under the applied potential shifting from 1.0 to 0.5, and finally back to 1.0 V. Inset: the magnified absorption edges.Reproduced with permission.[89]Copyright 2021, Chinese Chemical Society.The differential charge density distribution between the adsorbed *OOH intermediate and catalyst slabs, i.e., c) NiSe 2 and d) NiSe 2 -V Se .Note that the grey and green sphere corresponds to the Ni and Se atoms, respectively.Also, the positive and negative charge is represented by the red and blue color isosurface, respectively.Reproduced with permission.[41]Copyright 2022, The Authors.e) Operando SR-FTIR spectra of Ni 2−x P-V Ni recorded under the applied potential shifting from 0.8 to 0.2 V. f) Operando SR-FTIR spectra of Ni 2−x P-V Ni and pristine Ni 2 P recorded at 0.8 and 0.2 V, respectively.g) The calculated volcano plot for 2e − ORR electrocatalysis, plotted based on the U T as a function of ΔG HOO* for Ni 2−x P-V Ni and pristine Ni 2 P, as well as other reference models.Reproduced with permission.[174]Copyright 2022, Wiley-VCH.h) The high-resolution transmission electron microscopy (HRTEM) images (Scale bar: 1 nm) along the direction of the thickness of the pristine CoSe 2 (upper left) and the modified sc-CoSe 2 (lower left), as well as the corresponding fast Fourier transform (FFT) patterns of the pristine CoSe 2 (upper right) and the modified sc-CoSe 2 (lower right).i) Operando XANES spectra of Co K-edge for the pristine CoSe 2 (up) and the modified sc-CoSe 2 (down) at open-circuit voltage (OCV), applied potentials from 0.6 to 0.0 V, and back to OCV (after ORR).j) The coordination numbers of Co-O/N and Co-Se fitted from the operando EXAFS spectra corresponding to the XANES spectra in i).Note that the coordination numbers of Co-Se for sc-CoSe 2 always kept unchanged during an operando cycle, signifying its robust Co-Se coordination environment.Reproduced with permission.[44]Copyright 2021, Wiley-VCH.
Figure 6.a) Illustration of O 2 adsorbed on the surfaces of {001} facets-exposed -Fe 2 O 3 with and without oxygen vacancy (O v ), as well as the associated O−O bond lengths in both cases.Reproduced with permission.[172]Copyright 2020, Wiley-VCH.b) In situ XANES spectra (Cu K-edge) of A-Cu(OH) 2 /GO catalyst recorded under the applied potential shifting from 1.0 to 0.5, and finally back to 1.0 V. Inset: the magnified absorption edges.Reproduced with permission.[89]Copyright 2021, Chinese Chemical Society.The differential charge density distribution between the adsorbed *OOH intermediate and catalyst slabs, i.e., c) NiSe 2 and d) NiSe 2 -V Se .Note that the grey and green sphere corresponds to the Ni and Se atoms, respectively.Also, the positive and negative charge is represented by the red and blue color isosurface, respectively.Reproduced with permission.[41]Copyright 2022, The Authors.e) Operando SR-FTIR spectra of Ni 2−x P-V Ni recorded under the applied potential shifting from 0.8 to 0.2 V. f) Operando SR-FTIR spectra of Ni 2−x P-V Ni and pristine Ni 2 P recorded at 0.8 and 0.2 V, respectively.g) The calculated volcano plot for 2e − ORR electrocatalysis, plotted based on the U T as a function of ΔG HOO* for Ni 2−x P-V Ni and pristine Ni 2 P, as well as other reference models.Reproduced with permission.[174]Copyright 2022, Wiley-VCH.h) The high-resolution transmission electron microscopy (HRTEM) images (Scale bar: 1 nm) along the direction of the thickness of the pristine CoSe 2 (upper left) and the modified sc-CoSe 2 (lower left), as well as the corresponding fast Fourier transform (FFT) patterns of the pristine CoSe 2 (upper right) and the modified sc-CoSe 2 (lower right).i) Operando XANES spectra of Co K-edge for the pristine CoSe 2 (up) and the modified sc-CoSe 2 (down) at open-circuit voltage (OCV), applied potentials from 0.6 to 0.0 V, and back to OCV (after ORR).j) The coordination numbers of Co-O/N and Co-Se fitted from the operando EXAFS spectra corresponding to the XANES spectra in i).Note that the coordination numbers of Co-Se for sc-CoSe 2 always kept unchanged during an operando cycle, signifying its robust Co-Se coordination environment.Reproduced with permission.[44]Copyright 2021, Wiley-VCH.
Zhao et al. modeled and compared the Ni vacancy-enriched nickel phosphide (Ni 2−x P-V Ni ) and the pristine Ni 2 P, illustrating that a Ni atom was ≈3.72 Å far from the nearest Ni atom within Ni 2−x P-V Ni , while the distance between two Ni atoms for the pristine Ni 2 P was ≈2.61 Å.Such a geometric feature of Ni 2−x P-V Ni resulted in HOO* being adsorbed by active Ni atoms in an "end-on" manner.As previously mentioned, this arrangement increased the activation barriers for the scission of the O−O bond in HOO*, thereby impeding the occurrence of bond breakage.More significantly, the advantage brought by the geometric feature of Ni 2−x P-V Ni can be further experimentally confirmed through operando synchrotron Fourier transform infrared (SR-FTIR) spectra.As the cathodic potential shifted, two bands assigned to the presence of Ni-O and HOO* became prominent and intensified for Ni 2−x P-V Ni (Figure 6e), indicating increased adsorption and stabilization of HOO* intermediates on the active Ni sites.However, under similar electrochemical conditions, the band associated with HOO* was not clearly visible in pristine Ni 2 P.Moreover, the hydroxyl radicals arising from the dissociation of the O−O bond of HOO* was evident in the spectra of the pristine Ni 2 P (at 0.2 V where ORR extensively proceeded), while such a peak did not appear in Ni 2−x P-V Ni at the same potential (Figure of 2D TM-based dichalcogenides (TMDs).Li et al. exfoliated the commercially available bulk MoTe 2 through ultrasonication in the N-methylpyrrolidone, obtaining few-layered hexagonal 2H MoTe 2 nanoflakes featured with exposed zigzag edges accommodating rich unsaturated Mo/Te bonds, which performed exceptional 2e − ORR in acid media (a maximum H 2 O 2 selectivity as high as ≈93% in 0.5 m H 2 SO 4 ).DFT calculations indicated that compared to the active sites located at the basal plane where the O−O bond of the adsorbed HOO* easily dissociated (over-large ΔG HOO* ), those catalytic atoms at the exposed zigzag edges of 2H MoTe 2 possessed much optimized free binding energy of the HOO* intermediate (reduced ΔG HOO* ), directly giving rise to the elevation of the catalytic activity.Conversely, such catalytic sites at the zigzag edges exhibited poor O * affinity, retaining the O−O bond of HOO* and thus concurrently ensuring the preference to a 2e Photochemical metal organic deposition

Figure 7 .
Figure 7. a) LSV curves, and b) production rate under different cell voltages of the flow cell with CoN 4 /VG gas-diffusion cathode in the 0.1 m HClO 4 media.Schematics of c) the conventional GDE made by drop casting,and d) the free-standing CoN 4 /VG gas-diffusion cathode (which was subject to the hydrophobic pre-treatment by being exposed into to 0.1% PTFE solution and then heated under an argon atmosphere).Reproduced with permission.[108]Copyright 2022, Royal Society of Chemistry.e) Illustration of O 2 adsorbed on the active Zn atoms at the interfaces of ZnO@ZnO 2 , where red and gray spheres correspond to O and Zn atoms, respectively.f) The pulse voltammetry protocol section containing a reductive and oxidative pulse together with the associated current responses.g) Capacitance against different applied potentials derived from the pulse voltammetry of ZnO and ZnO@ZnO 2 in the O 2 and Ar-saturated electrolytes.Reproduced with permission.[196]Copyright 2023, Royal Society of Chemistry.
Photochemical metal organic deposition

Figure 8 .
Figure 8. a) Synthetic scheme of the representative TM-based atomic-site electrocatalyst for 2e − ORR.Reproduced with permission.[19]Copyright 2022, Wiley-VCH.b) Faradic efficiency of H 2 O 2 production from ORR for TM-based single atom catalysts (Mn-NC, Ni-NC, Cu-NC, Fe-NC, and Co-NC), as well as reference sample nitrogen-doped carbon (NC).c) Relationship of the binding energies of ORR intermediates with the d-bond center and valence electron numbers of the TM within such five TM-based single atom catalysts, as well as the associated d) plot of the limiting potential (U) as a function of ΔG *OH .Reproduced with permission.[204]Copyright 2019, Elsevier.e) Schematic of a typical single atom catalyst with highlighted center metal, as well as the first and second CSs.Reproduced with permission.[205]Copyright 2021, American Chemical Society.f) Co K-edge XANES spectra and g) k 2 -weighted FT EXAFS spectra in R-space for edge-hosted Co 1 −N 2 (denoted as CoNOC) under ex-situ and open circuit potential (OCP), as well as the in situ states including at onset potential (0.65 V), at 0.4 V, and after reaction (AR).Inset of f): the differential spectra obtained after subtracting the ex-situ XANES spectrum signal from the other four ones.Reproduced with permission.[220]Copyright 2023, Wiley-VCH.

Figure 9 .
Figure 9. a) Illustration for O* adsorption on the optimized structures of (i) CoSe, (ii) NiSe, and (iii) Cu 2-x Se, as well as the reference pure metal, FCC Pt. b) ΔG OOH* and ΔG O* of CoSe, NiSe, and Cu 2-x Se based on their various facets, as well as the benchmark samples PtHg 4 (yellow star marker) and Pd 2 Hg 5 (orange star marker).Note that the optimized structures of CoSe, NiSe, and Cu 2-x Se were still labeled by (i), (ii), and (iii), respectively.For comparison, the linear scaling relationship between ΔG OOH* and ΔG O* (O* adsorbed on hollow (top) sites) for FCC metals was plotted using the orange (purple) solid lines.Reproduced with permission.[74]Copyright 2021, American Chemical Society.c) The calculated PDOS of Nb within different TM-based columbites.Reproduced with permission.[225]Copyright 2021, Wiley-VCH.TPV spectra and the associated decay times for d) Ni 3 B and Ni 2 B, as well as e) NiB, NiB 2 , and NiB 3 .Reproduced with permission.[4]Copyright 2022, The Authors.f) Crystal structures with the corresponding lattice parameters of c-CoSe 2 and c-NiSe 2 .g) Different O* and/or OH* coverages (top) and the associated Gibbs free energy adsorbed by c-NiSe 2 and c-CoSe 2 (100) surfaces equilibrated with water (bottom).Note that the bottom insets display the calculated binding free energy of O* and OH* on the preferential binding sites of c-NiSe 2 and c-CoSe 2 (100) at the standard equilibrium potential of 2e − ORR.In addition, the yellow shaded regions represent the potential window in which the clean catalyst surfaces free of adsorbates had lower free energy than that of the catalyst surfaces adsorbing O* and/or OH*.Also, the inset scheme of atomic structures illustrates that one O* and one OH* were adsorbed on their preferential binding sites of the cell surfaces, consisting of two TM and four Se sites.Green, magenta, orange, red, and white spheres represent Ni, Co, Se, O, and H atoms, respectively.Reproduced with permission.[38]Copyright 2022, The Authors.
Figure 9. a) Illustration for O* adsorption on the optimized structures of (i) CoSe, (ii) NiSe, and (iii) Cu 2-x Se, as well as the reference pure metal, FCC Pt. b) ΔG OOH* and ΔG O* of CoSe, NiSe, and Cu 2-x Se based on their various facets, as well as the benchmark samples PtHg 4 (yellow star marker) and Pd 2 Hg 5 (orange star marker).Note that the optimized structures of CoSe, NiSe, and Cu 2-x Se were still labeled by (i), (ii), and (iii), respectively.For comparison, the linear scaling relationship between ΔG OOH* and ΔG O* (O* adsorbed on hollow (top) sites) for FCC metals was plotted using the orange (purple) solid lines.Reproduced with permission.[74]Copyright 2021, American Chemical Society.c) The calculated PDOS of Nb within different TM-based columbites.Reproduced with permission.[225]Copyright 2021, Wiley-VCH.TPV spectra and the associated decay times for d) Ni 3 B and Ni 2 B, as well as e) NiB, NiB 2 , and NiB 3 .Reproduced with permission.[4]Copyright 2022, The Authors.f) Crystal structures with the corresponding lattice parameters of c-CoSe 2 and c-NiSe 2 .g) Different O* and/or OH* coverages (top) and the associated Gibbs free energy adsorbed by c-NiSe 2 and c-CoSe 2 (100) surfaces equilibrated with water (bottom).Note that the bottom insets display the calculated binding free energy of O* and OH* on the preferential binding sites of c-NiSe 2 and c-CoSe 2 (100) at the standard equilibrium potential of 2e − ORR.In addition, the yellow shaded regions represent the potential window in which the clean catalyst surfaces free of adsorbates had lower free energy than that of the catalyst surfaces adsorbing O* and/or OH*.Also, the inset scheme of atomic structures illustrates that one O* and one OH* were adsorbed on their preferential binding sites of the cell surfaces, consisting of two TM and four Se sites.Green, magenta, orange, red, and white spheres represent Ni, Co, Se, O, and H atoms, respectively.Reproduced with permission.[38]Copyright 2022, The Authors.
Third, taking advantage of the advanced TPV technique, for the first time, our recent work on a series of nickel borides disclosed that varying the composition ratio (Ni/B) can control the electron transfer at the surface of TM-based nanocompounds.Especially, as depicted in the TPV curves Ni 3 B and Ni 2 B (Figure 9d) showed decreasing spectral intensity over extended measurement times, with Ni 2 B exhibiting a longer decay time than Ni 3 B, indicating slower interfacial charge transfer for Ni 2 B. Notably, compared with Ni 3 B and Ni 2 B, completely different TPV decay curves of NiB, NiB 2 , and NiB 3 can be found in Figure Photochemical metal organic deposition

Figure 10 .
Figure 10.a) The Ni L-edge XANES spectra tother with the associated simulations for the amorphous a-NiO NSs and crystalline c-NiO NSs.b) The illustration of 3d orbital configurations of Ni 2+ within the NiO 6 octahedron and NiO 5 pyramid, respectively.c) The deformation change density of Ni 2+ within the NiO 6 octahedron (left) and NiO 5 pyramid (right), respectively.d) Deformation charge density and PDOS for a-NiO (upper) and c-NiO (below),respectively. Reproduced with permission.[234]Copyright 2022, The Authors.The adsorption configurations of HOO* on e) the crystalline NiO x and f) the amorphous NiO x -C.The black, red, brown, and pink spheres corresponded to Ni, O, C, and H atoms, respectively.Reproduced with permission.[187]Copyright 2022, American Chemical Society.
For instance, Wu et al. specifically compared the HOO* adsorption configurations on the crystalline NiO x and amorphous NiO x -C.They found that in the former scenario, HOO* was adsorbed by the crystalline catalyst in a side-on way, and the O−O bond of HOO* was elongated to ≈1.40 Å from 1.25 Å, thereby facilitating the occurrence of bond breakage and a 4e − ORR tendency (Figure All potential versus RHE; In addition, the stability, production rate, and FE included herein are based on the bulk electrolysis measurements.Adv.Mater.2024, 36, 2400140 As demonstrated by the theoretical calculations on the reaction energetics between c-CoSe 2 and o-CoSe 2 , despite the former exhibiting more thermodynamically active for 2e − ORR, the latter polymorph was more kinetically favored.In detail, because the selectivity of 2e − ORR depends on the maintenance for O-O bond of the adsorbed key intermediate HOO*, two possible scission pathways were taken into consideration.The first possibility is that the O-O bond could be broken via the reductive elimination (proton-coupled electron transfer toward the distant O of HOO*) into the adsorbed O* and H 2 O by-products, was excluded as the very strong interaction merely took place between the proximal O of HOO* and the binding sites of both polymorphs.In the second scenario for both catalysts, the O-O bond could undergo dissociation into O* (stabilized on Se atoms) and HO* (stabilized on Co atoms) through the ensemble effect of neighboring Co atoms.The dissociation barriers served as a descriptor directly reflecting the selectivity of the ORR pathway, with those of o-CoSe 2 slightly surpassing those of c-CoSe 2 , implying the intrinsically more selective 2e − ORR capability for o-CoSe 2 .Furthermore, the surface Pourbaix diagrams obtained from the DFT and computational hydrogen electrode (CHE) calculation results imply that the binding strengths of HO* by the preferential Co sites of both polymorphs were different.The o-CoSe 2 featuring a larger interatomic distance of Co-Co possessed an increased HO* binding strength than that of c-CoSe 2 , and possibly a higher HO* coverage.This, in turn, could decrease the number of Co sites involved in cleaving the O-O bond of HOO*, by which the 2e

Figure 11 .
Figure 11.The Rietveld refinement of powder XRD patterns for a) Pb(NiWMnNbZrTi) 1/6 O 3 and b) Pb(ZrTi) 1/2 O 3 .c) Lattice strain within the pristine Pb(ZrTi) 1/2 O 3 and Pb(NiWMnNbZrTi) 1/6 O 3 based on the Williamson-Hall method.Reproduced with permission.[8]Copyright 2022, The Authors.d) Illustrations of crystal structures of stable fcc Ni (ABCABC stacking), as well as metastable hcp Ni and metastable P-hcp Ni (both ABABAB stacking).The different blue spheres all represent Ni atoms located at different layers while the pink spheres represent P atoms.e) The spherical AC-HAADF-STEM pattern of P-hcp Ni with a six-fold symmetry.f) The illustration of P-doping to stabilize the metastable hcp Ni.Reproduced with permission.[158]Copyright 2023, Wiley-VCH.

Figure 12 .
Figure 12.The generally existing challenges in the field of TM-based electrocatalysts for 2e -ORR, and correspondingly, the proposed perspective against each of those issues.

Table 3 .
Recently reports on heterostructure engineering enabled TM-based catalysts showing enhanced 2e −

Table 4 .
Recently reports on coordination tuning enabled TM-based catalysts demonstrating enhanced 2e − ORR electrocatalysis.

Table 5 .
Recently reports on crystallinity-design enabled TM-based catalysts exhibiting enhanced 2e −

Table 6 .
Recently reports on polymorphic transformation enabling TM-based catalysts to behave enhanced 2e −