Recent advances in active sites identification and new M−N−C catalysts development towards ORR

The M–N–C catalysts are considered potential alternative to Pt-based catalysts for the oxygen reduction reaction (ORR) due to its low cost and promising electrocatalytic performance. However, the catalysts are yet to become truly applicable in terms of activity and stability, and addressing such issues necessitate for indepth understanding in the structure performance relationship, which is remain elusive to date. Herein, we summarize our research progress achieved on M–N–C catalysts in recent years. Firstly, we successfully synthesized atomically dispersed Fe–N–C catalysts and conducted a detailed in-situ spectroscopy study, where the high spin D2 states of FeN4 is found to be an active species. Subsequently, in order to address the catalyst utilization and the overall activity of the catalysts, we carried out studies in increasing the active site density through regulating the microstructure of the catalysts. Finally and most importantly, in order to address the intrinsic activity of the catalysts, we carried work in developing new active centers of the M–N–C catalysts, where the new single or dual center catalysts were developed. Some of these centers are able to increase the stability of the catalysts, where the Fenton reaction is largely alleviated, resulting in both enhanced catalytic activity and stability. We hope that as the research continues, commercially available high performance and high stability M–N–C catalysts may eventually be realized.


Introduction
With the extensive use of fossil fuels, an increasing amount of NOx, SOx [1] and CO 2 exhaust gases are produced, which brings unprecedented environmental crisis globally.Electrochemical energy devices, such as fuel cells, are promising alternatives to address these issues by interacting with the renewable energies (wind, solar, etc.).Fuel cell is an energy conversion technology that directly converts the chemical energy in the fuel (hydrogen, methanol, formic acid) into electrical energy through an electrochemical process without direct combustion.At present, the proton exchange membrane fuel cell (PEMFC) is on the verge of large scale commercialization, mainly targeting on fuel cell vehicles [2].
oxidation reaction (HOR), the ORR reaction at the cathode results in much higher over potential (>300 mV) due to its slow kinetic process [4], significantly reduces the actual overall efficiency of the fuel cell.
The ORR is an intricate reaction involving multiple electron transfer steps, and the reaction mechanism involves multiple elementary reactions and multiple intermediates.The oxygen reduction model proposed by Wroblowa is the currently recognized basic model [5] (figure 1), it suggests that there are two ORR paths, one is a direct four electronic pathway to yield water, the other is a 2e + 2e indirect pathway with H 2 O 2 as a reaction intermediate.The ideal oxygen reduction reaction (ORR) is a direct four electron reaction, because H 2 O 2 can cause catalyst degradation and electrode material corrosion, especially in the aid of Fenton reaction.What happens in real reactions is usually a mixture of the two paths, so it necessary to develop a catalyst of high four electronic reaction selectivity.
In terms of ORR performance, it has been found that the activity is closely related to the adsorption energy of the oxygen intermediate at the active site.Therefore, developing the active center with proper adsorption energy for oxygen-containing intermediates is key to achieve high catalysis efficiency.By density functional theory (DFT) calculation, the ORR activity is plotted against the oxygen binding energy ∆E 0 to get a volcanic curve, with Pt sitting nearest to the top of the apex [6].Therefore, Pt based catalysts are the best performed ORR materials, but its high price and scarcity largely impedes the commercialization of the PEMFC.Therefore, intensive work have been studied to develop catalysts with lower Pt loading [7] or even without platinum, i.e. the non-platinum catalysts.
In this short review, we are focusing on our recent progresses on the pyrolysis metal-nitrogen-carbon catalysts (M-N-C), which are considered the most prospective Pt alternatives.The history of this type of catalysts can be date back to 1964, when Co phthalocyanine was first discovered considerable ORR activity in alkaline solution [8].Later, it was found that pyrolysis can greatly improve its activity and stability in acidic solutions [9].The early M-N-C catalyst is directly synthesized using carbon black as the precursor, lacking precise control in the morphology and structure, and it is difficult to conduct detailed analysis of its active sites.Slowly, with continuous optimization of the catalyst, a series of atomically dispersed catalysts have been gradually developed [10].Today, most of the M-N-C catalysts are synthesized in such a way that most metal species are dispersed in single atom states, making them actually a family member of single atom catalysts (SACs).For instance, the metal organic framework materials are widely used as the precursors to ensure the single atom dispersion and the high active site density in the catalysts [11].At present, researchers often use zeolitic imidazolate framework material ZIF-8 as the host to synthesis catalyst.This material has the following three advantages [12]: (a) ZIF-8 contains Zn center and organic ligands that rich in N, so that N can be evenly distributed in the C framework and spontaneously bind to the metal either through early stage chelation or by pyrolysis.(b) It has excellent pore structure: a large number of micropores can accommodate more active sites.(c) A higher porous structure can be obtained by high-temperature pyrolysis cause of evaporation of its Zn metal center, and can prevent the aggregation of metal particles.
Here in, we summarize our recent research results on M-N-C catalysts based on ZIF-8.We first conducted a detailed in-situ analysis of the active site and discovered its dynamic changes during the reaction.Subsequently, in order to address the catalyst utilization and the overall activity of the catalysts, we carried out studies in developing the active site density through regulating the microstructure of the catalysts.Finally, we carried work in developing new active centers of the M-N-C catalysts, where the new single or dual center catalysts were developed.Some of these centers are able to increase the stability of the catalysts, where the Fenton reaction is largely alleviated, resulting in both enhanced catalytic activity and stability.[16].Reprinted with permission from [16].Copyright (2018) American Chemical Society.

Active sites identification
A comprehensive cognition of active sites for M-N-C catalyst is required before one can develop a suitable catalyst that is truly viable in PEMFC.In 2012, Wen et al synthesized a nitrogen-rich core-shell structure catalyst with iron-carbon nanorods as the core and graphitic carbon as the shell [13], and considered carbon-coated metals as active sites.In 2015, Zitolo and his coworkers used ZIF-8 as a precursor and synthesized an atomically dispersed Fe-N-C catalyst [14].Through Mössbauer spectroscopy, two different coordinations of D1 and D2 were observed, combined with EXAFS data fitting and DFT calculation, it is considered that the FeN 4 C 12 planar structure with one or two axially adsorbed O is the active site.Many studies on the identification and hypothesis of active sites have also been published [15].
Based on the early understanding of active sites, our group synthesized atomically dispersed Fe-N-C catalysts through the pore confinement strategy.The onset ORR potential of 0.92 V is equivalent to Pt/C, which is better than most non-noble metal catalysts [16].Subsequently, an in situ spectroscopic analysis was carried out to gain a further insight into its active sites.Firstly, two coordination environments of D1 and D2 were found in the 57Fe Mo ¨ssbauer spectroscopy (figure 2(a)).EXAFS results show that it contains Fe-N(O) bonds similar to FePc, and the hardly observed Fe-Fe scattering further verifies its atomic dispersion (figure 2(b)).Further fitting results show that the four-coordinated Fe-N structure and the one-coordinated Fe-O structure, and the real active site is considered to be the five-coordination Ox-Fe-N 4 structure.Fe L-edge XANES Fe mainly exists in the Fe-N-C catalyst in trivalent form (figure 2(c)), and the absence of the square planar fingerprint peak at 7117 eV implies a non-planar Fe-N 4 structure.Furthermore, we found that the XANES curve shifted forward as the potential increased (figures 3(a)-(c)) through ∆µ techniques, indicating the occurrence of Fe 2+ /Fe 3+ redox reaction during ORR.After DFT calculation and data fitting, it is suggested that the active site undergoes a dynamic redox process that changes with the potential.According to the equation O x -Fe 3+ -N 4 + e -↔HO-Fe 2+ -N 4 , when the potential is higher than the Fe( 2+/3+ ) redox potential, Fe 2+ is oxidized to Fe 3+ , and then reduced back to Fe 2+ as the potential drops.The reaction proceeds to improve the catalytic efficiency of oxygen reduction, and suggested that the real active site is FeN 4 C 8 with axially bonded oxygen atoms (figure 3(d)).
Through the above research, we have deepened our understanding of the active site and reaction mechanism, which will help us in designing new catalysts.

Microstructural control
In addition to the identification of active sites, the microscopic changes in the formation process of catalysts are also crucial.The influences of different metal sources, synthetic procedure and pyrolysis atmosphere on the morphology and structure of catalysts are discussed below, which is helpful for us to regulate the microstructure of catalysts.
We studied the effects of different types of Fe source (FeCl 3 , Fe(NO 3 ) 3 , Fe(acac) 3 ) on the final structure and performance of Fe-N-C catalyst [17].It was found that although FeCl 3 and Fe(NO 3 ) 3 could improve the graphitization degree of the precursor, Fe 3 C nanoparticles were produced and affected the performance of the catalyst.Among all the precursors, Fe(acac) 3 could ensure the atomic dispersion of iron and obtained the best catalytic performance when the content of Fe was 1.47 wt %.It is related to the molecular size and  degree of hydrolysis of Fe(acac) 3 , and appropriate molecular size and lower degree of hydrolysis are conducive to its dispersion in the precursor of ZIF-8, so as to obtain higher catalytic activity.
Different synthesis method and treatment could great influence the structure and properties of the catalysts.For the ZIF-8 based M-N-C ORR catalysts, although the traditional one-pot method is helpful for the uniform atom dispersion of the metal, it will lead to the waste of metal salts and the difficulty of further increasing the metal content.Therefore, we developed a 'second pyrolysis' method (figure 4), using the pyrolyzed carbonized ZIF-8 as the host and obtain a highly dispersed single-atom catalyst, which performance has been greatly improved compared to the one-pot method [18].In the subsequent analysis and comparison, we found that for the one-pot method, the metal doping efficiency is very low, which is mainly due to the hydrolysis of metal salts caused by trace water brought by Zn(NO 3 ) 2 * 6H 2 O, leading to the formation of inactive nanoparticles.In addition, excessive Zn will occupy the voids of ZIF-8 to further hinder the incorporation of metal, which leads to a large amount of waste of metal salt.In the subsequent carbonization process, the ZIF framework will be carbonized and contracted, losing its restriction on the metal, thereby resulting in the formation of a small number of particles.The 'second pyrolysis' method solves these problems.The carbonized ZIF-8 will not cause the hydrolysis of the metal salt, which improves the utilization rate of the metal.During the pyrolysis process, the carbon backbone is well retained and the metal agglomeration is inhibited.This leads to a higher degree of dispersion.In short, we have obtained a higher loading and more dispersed MNC catalyst through the 'second pyrolysis' method, which has advanced the half-wave potential by 19 mV, and improved the catalyst in the synthesis method.
The ZIF-8 is a microporous cage structure, the inner size of the cage is 11.6 Å, and the opening diameter is 3.4 Å.This structure can make the cage only accommodate one Fe(acac) 3 , thereby avoiding iron particles during the pyrolysis process.The host-guest method is a universally effective method for preparing single-atom catalysts.However, although micropores are essential for the formation of Fe SAC, the micropores themselves are not conducive to the transfer of substances during the reaction.The ultramicropores in the catalyst have been proven to hinder the mass transfer of O 2 and H 2 O, resulting in limited catalytic performance [19].By adjusting the pyrolysis atmosphere, we can effectively adjust the distribution of pore structure of catalyst materials.Thus we developed surface hydrogen etching method which can effectively adjust the density of active site and the pore structure distribution of the catalyst by adjusting the hydrogen concentration [20].We purposed a new idea to improve the performance of the catalyst.During pyrolysis, hydrogen will preferentially etch unstable carbon, thereby resulting in an increased N content with more edge N sites exposed, which is conducive to anchoring free Fe ions to form FeN 4 sites.In addition, hydrogen etches blocked carbon chips and enlarges the pore size, and facilitates the electron and mass transfer.The proportion of mesopores increases steadily with the increase of hydrogen concentration (figures 5(a)-(e)), and shows the best oxygen reduction activity at a concentration of 10%.If the hydrogen concentration continues to increase, the H 2 will cause excessive etching of the carbon material and damage the catalyst structure, thus resulting in decreased specific surface area and catalyst performance.

High activity catalysts
Improving the intrinsic activity of the M-N-C catalysts is highly challenging and has plagued the application of this material.According to the Sabatier theory, adjusting the adsorption behavior of active sites on oxygen-containing intermediates can effectively improve the ORR activity through to different ways: (a) adjusting the energy level of the active center by tuning the chelation environment of the metal center; (b) tailoring the electronic structure of the metal center by changing the number of the metal atom in the catalysis center, i.e. creating chelation structures with more than one metal atoms in the center.
Our group brought up a resorcinol-formaldehyde (RF) resin to anchors free Fe atoms, and add zinc to prevent iron accumulation and introduce oxygen to the active site [21].In this way, we successfully obtained a monoatomic dispersed Fe-N-C catalyst, after electrochemical tests, we found that its half-wave potential reaches 0.83 V which is significantly higher than the traditional Fe-N-C catalysts.Through further spectroscopy analysis and DFT calculation, we found a Fe-O-Fe structure (figure 6(d)).This stable Fe-O-Fe structure can weaken the binding strength of the Fe center and the ORR intermediate, and causes higher ORR activity (figures 6(b) and (c)).As a result, we realized the energy level regulation by controllable introduction of O into the active site and further improved the catalytic activity.
The state of the metal in the catalysts largely depends on the loading of the metal and its degree of dispersion in the catalyst.The uniform dispersion in the precursor helps to control the structure.Therefore, we use a ZIF mixture with a Zn and Co bimetallic center to control the dispersion of the metal, where Zn can effectively isolate Co to maintain atomic dispersion [22].Thus we have achieved an atomically dispersed Co catalyst.We discovered two atomic sites of Co-Co with a distance of 2.1-2.2Å through high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM), and discovered a new shortened Co-Co path (2.12 Å) by XAS, thus we first discovered the dual-core Co 2 N x structure (figures 7(a)-(e)).It further confirms that the structure is Co 2 N 5 containing axial OH by DFT calculation.The calculations show that for the traditional single-atom site, oxygen molecules tend to be adsorbed laterally or at the end on a single atomic site, while at the diatomic site, O 2 tends to be adsorbed on the diatomic metal by the bridge cis.Due to the relatively easy break of the weakened O-O bond, the bridge cis adsorption is beneficial to the 4e-ORR pathway and has better oxygen reduction activity.The RDE test found that its half-wave point reached 0.79 V in acidic medium, and the Co 2 N 5 site has a very high mass activity that is 12 times that of CoN 4 .The discovery of this dual-nuclear metal center opens a new path to develop new M-N-C catalysts.
Through the discovery of binuclear metal catalysts and the understanding of active sites, we also guided practice from theory and successfully synthesized an efficient Fe-Co diatomic catalyst [23], and confirmed  our calculation results from further analysis.Firstly, because it is difficult to directly measure the intermediate adsorption energy experimentally, it is hard to directly adjust the electronic energy level of the active center.However, it is found that Fe(III)/Fe(II) redox potential (Eredox) is an effective indicator of oxygen adsorption energy [24], which greatly alleviates this dilemma.In particular, high redox potential corresponds to weak Fe-O binding strength, and low redox potential corresponds to strong Fe-O binding strength.Through a series of theoretical calculations, it is found that the structure of FeCoN 5 can spontaneously adsorb OH in water at 0 V.Under the electric potential, the adsorption is stronger, forming a stable FeCoN 5 -OH structure, which the charge density of its central atom is lower than FeN 4 , and it is more difficult to lose electrons, making it have higher Eredox, thus achieving better catalytic activity.Based on the above calculations, we have successfully synthesized Fe-Co binuclear catalysts, realized the strategy of customizing the electronic structure and optimizing the geometric configuration.OH is firmly anchored on the FeCoN 5 at a wide potential window range.The Fe-Co binuclear active site is identified by XAS, the structure is the same as the previous DFT calculation and fitting results, realizing the process of guiding practice from theory (figures 8(a)-(c)).In the further RDE test, the FeCoN5-OH site had shown unprecedented ORR activity in acid medium, which 20 times than the mononuclear Fe-N 4 site, and its onset potential and half-wave potential reached 1.02 V and 0.86 V, respectively.
In this part, we discussed several ways to improve the activity of the catalyst, mainly focus on the electronic state regulation of the active center.We believe that with the further understanding of active sites and theoretical calculations, ORR catalysts with higher activity can also be developed.

High stability catalysts
Although Fe-based catalysts have the highest non-noble metal ORR activity, the stability problem caused by Fenton reaction is difficult to solve.Specifically, Fe can react with H 2 O 2 to produce reactive oxygen species  (ROS) [25].This kind of strong oxidizing oxygen free radicals will cause the degradation of the catalyst and the corrosion of the carbon substrate, which will seriously reduce the reaction activity [26].The solution to this problem is to reduce the yield of H 2 O 2 or develop a new low Fenton reaction stable catalyst.
The macrocyclic compound Cr phthalocyanine is considered to have no oxygen reduction activity due to its overly strong binding energy to oxygen.However, our group successfully prepared Cr SAC using the host-guest strategy and showed excellent activity in acidic solutions [27].Under the analysis and fitting of EXAFS, it was found that its active site is a flat CrN 4 structure without axial oxygen.Interestingly, we found a very low Tafel slope (37 mV) through the LSV curve, indicating that it is a 2 + 2 electronic reaction, which has been further confirmed by the same RRDE and peroxide reduction reaction experiments.Since it is an indirect four-electron process, the catalyst will produce a large amount of H 2 O 2 , but the content of H 2 O 2 is not a decisive factor in the degradation of the catalyst.Under the accelerated aging test (ADT), after 20 000 cycles, the half-wave potential is only reduced by 15 mV, which is significantly lower than the 31 mV of Fe-N-C.We believe that its extremely low Fenton reaction activity is the reason for its superior stability, and the production of ROS has been analyzed by UV absorption spectroscopy with 2,20-azinobis (3-ethylbenzthiazoline-6-sulfonate) (ABTS).Experiments show that the absorbance value of CrNC is only 4.3% of the Fe counterpart, which clearly proves the inhibitory effect on ROS and achieves a substantial improvement in the stability of the catalyst (figures 9(a)-(d)).The discovery of CrN 4 catalyst provides a new path for solving the stability problem of non-noble metal ORR catalysts.
We also synthesized a new type of Ru single-atom catalyst by using the host-guest strategy [28], which exhibits high activity and a lower Fenton reaction to maintain long-term stability.Through XAS fitting results, it was found that the active site is a RuN 4 structure containing axial O.The axial O can reduce the charge density of the Ru center, in order to reduce the adsorption of oxygen-containing intermediates and obtain higher oxygen reduction activity.After testing with RRDE, it was found that the onset potential and half-wave potential reached 0.92 V and 0.821 V in acidic medium, respectively, which are significantly higher than the Ru nanoparticles.In addition, the single-atom Ru catalyst showed great stability, with only a negative shift of 17 mV after 20 000 cycles at the half-wave potential (figures 10(a)-(i)).The excellent activity and stability in the actual fuel cell device further confirms the practical application capability of the Ru-N-C catalyst.
In this section, we discussed the solution to improve the stability of the catalyst, mainly starting from the Fenton reaction to develop a new stable catalyst, which provides the possibility for the large-scale commercial use of M-N-C catalysts in fuel cells.

Summary and outlook
The development of efficient non-precious metal catalysts is a highly promising yet challenging task.Fortunately, the discovery of M-N-C catalysts makes this route possible.The morphological structure of the catalyst can be improved by using ZIF as the precursor, but further exploration still requires a deep understanding of the active site and the help of theoretical calculations.In this regard, we summarized the work of our group in recent years, mainly focusing on the exploration of active sites, and proposed a new dynamic active site reaction mechanism.We also adjusted the microstructure of the catalyst by changing the metal precursor and inventing the second pyrolysis method and hydrogen etching method to further improve the activity of catalysts.Finally on the basis of the above work, new type of catalysts with high activity and stability were further developed.
For the following research, we believe that the exploration of real active sites should be further promoted, thereby guiding us to synthesize new catalysts with optimized electronic structure and adsorption feature.Meanwhile, in order to promote the practical application, it is necessary to study how to inhibit the Fenton reaction of Fe-based catalysts and develop new and efficient low Fenton reaction catalysts.In addition, effectively preventing the metal leaching from the active site is also a way to improve stability, which is worthy of in-depth study.In the actual PEMFC, because it is very different from the three-electrode system, we think it is necessary to systematically study the activity and stable expression of M-N-C catalyst, and develop more in-situ cell technology.To achieve the optimization of the catalyst from PEMFC perspectives.We believe that through continuous research, M-N-C catalysts will replace Pt to achieve large-scale commercialization of fuel cells.