Thermodynamic and kinetic considerations of nitrogen carriers for chemical looping ammonia synthesis

Ammonia (NH3) is a promising clean energy carrier, provided that its production is driven by renewable energy rather than fossil fuel-based Haber–Bosch (H–B) process. Chemical looping ammonia synthesis (denoted as CLAS) can intervene in the ubiquitous scaling relations in catalytic ammonia synthesis by separately feeding reactants to a nitrogen carrier to achieve atmospheric operation, which provides an alternative synthetic route to the H-B process. The key of CLAS is to develop efficient N carrier materials with suitable thermodynamic and kinetic properties. Metal nitrides and metal imides are two kinds of N carrier materials for the CLAS process, and H2 and H2O are commonly used as the hydrogen sources of NH3. Here, we first analyze the thermodynamic properties of the reactions of various metal nitrides and imides with water or hydrogen to produce NH3, N2 fixation on metals or metal hydrides, and the regeneration of metals from metal oxides, respectively. The thermodynamic calculation results display the reduction of main group metal hydroxide, early transition metal oxides, and rare earth metal oxides to the corresponding metallic state or hydrides, the nitridation of late transition metals to the corresponding nitrides, are the thermodynamic limiting steps for the metal nitride carriers. The metal imides, such as lithium imide and barium imide, have the relatively proper thermodynamics for two-step chemical looping reactions, however, their performance is limited by the thermodynamics of hydrogenation reaction. Moreover, for the thermodynamically unfavorable steps in the CLAS, we propose potential electrochemical processes to run the loop, such as molten salt electrolytic cell and solid electrolyte electrolytic cell. Finally, we put forward some strategies, such as controllable synthesis of N carriers and adding efficient catalysts, to improve the kinetics of chemical looping reactions.

The chemical properties of N carriers have great influences on their performances of CLAS. Nitrogen in MNs usually has a formal oxidation state of − 3 as a nitride ion (N 3− ), and their chemical properties vary greatly with the valence electrons of the metal [18,[38][39][40]. According to the difference of valence electrons, MNs can be classified as the nitrides of the s-block elements (s-MNs), nitrides of the p-block elements (p-MNs), transition metal nitrides (TMNs), and nitrides of the rare earth elements (RENs), etc. On the other hand, metal imides mainly exist in alkali and alkaline earth metal compounds, rare earth metal compounds, and ternary alkali (or alkaline earth)-transition metal compounds, etc. s-MNHs are of importance for applications in the field of hydrogen storage [41][42][43]. The reaction thermodynamics of MNs and MNHs reacting with H 2 O or H 2 to produce NH 3 as well as the regeneration of MNs and MNHs are the important criteria for their potential applications in the CLAS [27,30,34,44]. In the Sect. 3, we summarized and analyzed the thermodynamics of these reactions according to the type of N carriers and hope to help guiding for the selection of N carriers and the design of the new CLAS processes.
Some sub-reactions in CLAS processes are limited by their reaction thermodynamics, which results in difficulty in closing the chemical loops. For example, the reduction of Al 2 O 3 or Cr 2 O 3 to the corresponding metal acquires extremely high temperature (above 1000 °C) [27,44]. It is well-known that extrinsic stimuli (photo, electricity, plasma, and so on) could change the chemical equilibrium of a reaction and thus make the reaction occur under milder conditions [45,46]. In the industry, the production of metallic Li, Na, K, Al, etc. are carried out under electrolytic conditions because the reduction of their oxides to metals is difficult under normal thermal conditions [47]. Inspired by this, we propose several electrochemical CLAS processes in this perspective. On the other hand, the kinetics of sub-reactions are also crucial to the NH 3 production performances in a CLAS process. For example, the hydrolysis of AlN to produce NH 3 is thermodynamically feasible but is limited by the slow kinetics [48]. In the Sect. 5, we will also discuss some strategies to improve the kinetic properties of N carriers.

The reaction thermodynamics of metal nitrides and imides
The N carriers mainly include metal nitrides and imides. In a CLAS process, ammonia can be produced by hydrolysis or hydrogenation of N-rich nitrides or imides accompanied by the formation of oxides, hydroxides, N-poor nitrides, metals or hydrides. To complete the chemical loop, the N-rich nitrides and imides need to be regenerated. In this section, we summarize and analyze the reaction thermodynamics of ammonia production from metal nitrides, including main group metals and transition metals, and metal imides, as well as the regeneration reaction thermodynamics of metal nitrides and imides.

The reaction thermodynamics of main group metal nitrides
The main group metals include s-and p-block elements. There are a large number of main group metal nitrides. Here, we select Li, Be, Mg, Ca, Sr, Ba, and Al as representatives to analyze their reaction thermodynamics. All of the reaction equations about the main group metal nitrides-mediated CLAS are listed in Table S1, and the thermodynamic calculation results are shown in Fig. 2. Figure 2a shows the reaction thermodynamics of s-MNs and p-MNs hydrolysis. It is seen that the Gibbs free energies of the hydrolysis reactions for all s-MNs are negative below 800 °C, suggesting these reactions spontaneously occur from the viewpoint of thermodynamics. The hydrolysis of alkali and alkaline earth metal nitrides to generate NH 3 and corresponding alkali and alkaline earth metal hydroxides (s-MOHs) can take place at low temperatures. The hydrolysis of aluminum nitride (AlN) requires high-temperature conditions due to large kinetic barriers, although it is thermodynamically favorable in the temperature range in Fig. 2a. The product of AlN hydrolysis is aluminum oxide (Al 2 O 3 ) rather than aluminum hydroxide (Al(OH) 3 ) at high temperatures [27]. Figure 2b shows the reaction thermodynamics of the hydrogenation of s-MNs and p-MNs to alkali and alkaline earth metal hydrides (s-MHs) and metallic aluminum (Al). We could obtain the following conclusions: (i). The order of thermodynamic feasibility is Li 3 N ≈ Ba 3 N 2 > Sr 3 N 2 > Ca 3 N 2 > Mg 3 N 2 > AlN > Be 3 N 2 ; (ii). The thermodynamics of these reactions are more favorable at lower temperatures; (iii). The hydrogenation of Li 3 N, Ba 3 N 2 , Sr 3 N 2 , and Ca 3 N 2 to corresponding hydrides can spontaneously occur below 484 °C, 505 °C, 372 °C, and 289 °C, respectively (at the equilibrium constant (K) = 1).
There are generally two approaches to regenerating s-MNs from s-MOHs, i. e., (i) AOHs decompose to form metals, which are subject to nitrogen fixation to return to s-MNs; (ii) the hydrogenation of s-MOHs generates s-MHs, and the s-MHs then fix nitrogen to generate s-MNs. However, the N 2 fixation on s-MHs usually generates s-MNHs rather than s-MNs at relatively low-temperature conditions, which had been experimentally observed. We only discuss the former pathway  Table S1, in which the standard formation enthalpy and standard entropy data of the reactants and products are taken from ref. [121,122] here, and the N 2 fixation on s-MHs will be discussed in the next section. For Al 2 O 3 , we consider the following pathway to regenerate the N carrier: hydrogenation of Al 2 O 3 to Al (AlH 3 is unstable), and Al then fixes nitrogen to generate AlN. Figure 2c shows the reaction thermodynamics of s-MOHs decomposition to alkali and alkaline earth metal as well as the hydrogenation of Al 2 O 3 to generate Al. It is seen that all of these reactions are thermodynamically unfavorable in the shown temperature range. These highly endothermic reactions could be carried out using concentrated solar energy as the source of heat energy. Alternatively, the introduction of extrinsic stimuli such as electricity, photon, etc. is another promising and green method to make a thermodynamically unfavorable reaction occur under lower temperatures [32,52], which will be discussed in the next section. Figure 2d shows the reaction thermodynamics of N 2 fixation on alkali and alkaline earth metals as well as Al to generate s-MNs and AlN. It is seen that the reactions of N 2 fixation on these metals spontaneously occur from the viewpoint of thermodynamics, and the lower temperature is more favorable for these reactions.
In summary, the H 2 O-CLAS processes over s-MNs should overcome the thermodynamic barriers of the decomposition of s-MOHs to generate alkali and alkaline earth metals. Moreover, in the H 2 -CLAS over s-MNs, the N carrier will change from nitrides to imides. For the AlN-mediated H 2 O-CLAS, it is limited by the step of Al 2 O 3 reduction to Al. Furthermore, the hydrogenation of AlN to produce NH 3 is the thermodynamically determining step for AlN-mediated H 2 -CLAS.

The reaction thermodynamics of transition metal nitrides and rare earth metal nitrides
Transition metals (TMs) and rare earth metals (REs) contain d-or f-block valence electrons with versatile physical and chemical properties, and their nitrides are abundant [53][54][55][56]. In this section, we will discuss several transition metal nitrides (TMNs) including Ti, V, Cr, Mn, Fe, Co, and Zn as well as two rare earth metals nitrides (LaN and YN) (RENs). These nitrides could react with H 2 O to produce NH 3 and corresponding oxides (TMOs or REOs) [44,49,57]. In addition, they could also react with H 2 to produce NH 3 and generate the N-poor carriers, i.e., metals, N-poor TMNs [30], transition metal hydrides (TMHs) [58], or rare earth metal hydrides (REHs) [59].
To achieve an H 2 O-CLAS over TMNs and RENs, TMOs or REOs should be regenerated to the corresponding nitrides. TMOs and REOs should be first reduced to corresponding metals or metal hydrides. For the oxides of V, Cr, Mn, Fe, Co, and Zn, they could be reduced to the corresponding metals by H 2 . For the elements of La, Y, and Ti, their hydrides are usually more stable than their metallic states in the H 2 atmosphere, so we assume that the oxides of these three metals would be reduced to corresponding metal hydrides by H 2 [58,59]. Then, these metals and metal hydrides could fix N 2 to form corresponding metal nitrides so as to close the H 2 O-CLAS loop. On the other hand, to achieve an H 2 -CLAS over TMNs and RENs, the N-poor carriers should be nitridized to metal nitrides. In this section, we calculate the reaction thermodynamics of TMNs and RENs reacting with H 2 O and H 2 to produce NH 3 , reduction of TMOs and REOs to form the N-poor carriers, and the N 2 fixation on these N-poor carriers to generate TMNs and RENs. All of the reaction equations about TMNs and RENs-mediated CLAS are listed in Table S2, and the thermodynamic calculation results are shown in Fig. 3. Figure 3a shows the reaction thermodynamics of the nitrides reacting with H 2 O to produce NH 3 and corresponding oxides. It can be seen that all of these reactions could occur in the shown temperature range and low T is more favorable from the viewpoint of thermodynamics. The order of thermodynamic feasibility is YN > LaN ≈ Cr 2 N > Mn 4 N > Zn 3 N 2 > Mn 5 N 2 > TiN > VN > Fe 4 N > Co 3 N. Figure. 3b shows the reaction thermodynamics of the hydrogenations of these nitrides to produce NH 3 and generate corresponding N-poor carriers. It can be seen that the hydrogenations of Zn 3 N 2 , Fe 4 N, and Co 3 N are thermodynamically feasible below 400 °C for Fe and the whole shown temperature range for Zn and Co (at K = 1). Moreover, the hydrogenations of N-rich CrN and Mn 5 N 2 to N-poor Cr 2 N and Mn 4 N could occur under high H 2 pressure and low NH 3 concentration conditions and the other reactions are more thermodynamically difficult considering the value of Gibbs free energy. Figure. 3c shows the reaction thermodynamics of the regeneration of these nitrides from their oxides. The results show that the reduction of the late transition metal oxides (Fe 2 O 3 and CoO) to corresponding metals is thermodynamically feasible, but the reduction of the other metal oxides is thermodynamically unfavorable. Figure. 3d shows the reaction thermodynamics of N 2 fixation on TMs, N-poor TMs, and metal hydrides to generate corresponding metal nitrides. It can be seen that the nitridation of the TMs (V, Cr, and Mn), TMHs (TiH 2 ), N-poor TMNs (Cr 2 N, Mn 4 N), and REHs (LaH 2 , YH 3 ) is thermodynamically feasible. However, the nitridation of Fe, Co, and Zn is thermodynamically unfavorable. The synthesis of Fe 4 N, Co 3 N, and Zn 3 N 2 is usually via the nitridation reaction of metal or oxide powders in an NH 3 atmosphere [60][61][62].
The above thermodynamic calculation results indicate that the reduction of metal oxides of early transition metal (Ti, V, Cr, and Mn) and rare-earth metal (Y and La) and the N 2 fixation on late transition metals (Fe, Co, and Zn) are the 1 3 thermodynamic limiting steps in the TMNs and RENs-mediated H 2 O-CLAS, respectively. In addition, the hydrogenations of early TMNs and RENs and N 2 fixation on late transition metals are the thermodynamic limiting steps in the TMNs and RENs-mediated H 2 -CLAS.

The reaction thermodynamics of metal imides
Metal imides include s-MNHs [41,63,64], rare earth metals binary imides (RENHs) [65], ternary mixed cation imides [42,66], ternary mixed anion imides [67,68], etc. Due to the lack of thermodynamic data for most metal imides, we only discuss a portion of binary s-MNHs for application in the CLAS. s-MNHs can react with H 2 or H 2 O to produce NH 3 and generate s-MHs or s-MOHs, respectively. The decomposition of s-MOHs to metals, which is thermodynamically unfavorable, is discussed in the Sect. 3.1 [31]. Here, we discuss the s-MNHs-mediated H 2 -CLAS, which contains two steps: (i) hydrogenation of s-MNHs to produce NH 3 and generate s-MHs, and (ii) N 2 fixation on s-MH to generate s-MNH accompanied by H 2 release [34]. All of the reaction equations about ANHs-mediated CLAS are listed in Table S3, and the thermodynamic calculation results are shown in Fig. 4. Figure 4a shows the reaction thermodynamics of s-MNHs hydrogenation to produce NH 3 and generate s-MHs. The results suggest that there is a low equilibrium concentration for the reaction of ANHs hydrogenation due to the positive Fig. 3 The reaction chemistries of TMNs and RENs for CLAS. a. The Gibbs free energy of TMNs and RENs hydrolysis to produce ammonia as a function of temperature. b. The Gibbs free energy of TMNs and RENs hydrogenation to produce ammonia as a function of temperature. c. The Gibbs free energy of TMOs and REOs hydrogenation to generate metals as a function of temperature. d. The Gibbs free energy of metals (or metal hydrides) to generate corresponding nitrides as a function of temperature. The Gibbs free energy is calculated according to the reaction equations listed in Table S2 and the thermodynamic data of the reactants and products are taken from ref. [121,122] Gibbs free energy and the low temperature is more thermodynamically favorable. Figure 4b shows the reaction thermodynamics of s-MHs nitridation to generate s-MNHs and H 2 . The results show that the nitridation reactions of various s-MHs are thermodynamically feasible. Considering the thermodynamics of these two steps, the order of thermodynamic feasibility of s-MNHs in the H 2 -CLAS is BaNH ≈ Li 2 NH > CaNH > MgNH [34]. It should be mentioned that the CaNH and MgNH could decompose or (de)hydrogenated to Ca 2 NH (or Ca 3 N 2 ) [69][70][71] and Mg 3 N 2 [71,72], respectively, at the certain conditions, which would form new loops for CLAS.
Compared to metal nitrides-mediated H 2 O-CLAS and H 2 -CLAS, the metal imides-mediated H 2 -CLAS processes possess higher ammonia production rates and can be operated under milder reaction conditions as the literature reported [34]. However, the imides N carriers were less explored, and lack of thermodynamic data for these materials up to now. There is huge space to explore new imides materials, such as rare earth metals imides, and ternary imides, for application in the CLAS [73].

The thermodynamic improvement for the N carriers
An ideal N carrier material should possess suitable thermodynamic properties for each sub-reaction in the CLAS process. However, few materials can meet this requirement. For one thing, the development of new nitrides or imides materials should be encouraged. For another thing, the following strategies can be considered to improve the thermodynamic properties of N carriers. (i). Doping. The dopants can modify the chemical composition of N carriers and the bond strength of M-N, which would influence the thermodynamics of materials. Hargreaves et al. studied the role of lithium dopant in manganese nitride by combining theoretical and in-situ experimental investigations [74]. They found that the lithium dopant modifies the bulk and surface structure of manganese nitride and enhances hydrogen activation. Liu et al. studied the transition metal heteroatoms (Cr, Fe, Co, Ni, Mo) doped Mn 4 N and Mn 2 N lattices by DFT calculations [75]. They found that the dopant influences the covalency of metal-N bonding. (2). Compositing. The chemical reactions taking place in composite N carrier systems could be different from those in the single N carrier, which could form a new reaction pathway. Guo and Cao et al. developed multi-functional composite nitrogen carriers Li 2 NH-Mn 2 N and BaNH-Mn 2 N for CLAS [51]. In these composite nitrogen carrier systems, the reaction of nitride N carrier with hydride to form the N-poor carrier (and imide), are more thermodynamically favorable than those reactions of nitride N carrier with hydrogen to form the N-poor carrier (and NH 3 ), and NH 3 production could follow the below reactions: (i) AH + Mn 2 N → ANH + Mn 4 N + H 2 ; (ii)ANH + H 2 → AH + NH 3 . (3). Forming high-entropy compounds. High-entropy compounds have unique structures with the coexistence of antisite disordering and crystal periodicity, Fig. 4 The reaction chemistries of alkali and alkaline earth metal imides for CLAS. a. The Gibbs free energy of metal imides hydrogenation to produce ammonia as a function of temperature. b. The Gibbs free energy of metal hydrides nitridation to regeneration metal imides as a function of temperature. The solid entropies are not considered in the thermodynamic calculations due to the lack of entropies data for metal imides. The Gibbs free energy is calculated according to the reaction equations listed in Table S2, and the thermodynamic data of the reactants and products are taken from ref. [121,122] and the entropy of solids is not considered in the calculation of Gibbs free energy which provide opportunities for various applications [76][77][78]. Some high-entropy oxides have been applied in the chemical looping process, which provided outstanding performance for thermochemical water splitting [79,80]. We believe that there are potential promising chances to explore high-entropy nitrides or imides materials for CLAS [81,82]. It should be mentioned that these strategies not only improve the thermodynamic properties of N carriers but also could enhance the kinetics performance of CLAS.
With the development of computational theories and the improvement of calculation speed, theoretical calculation and artificial intelligence (AI) have become effective means to assist material design [83][84][85][86]. They can be used to assist in the screening and design of N carriers with suitable thermodynamic properties. Michalsky et al. used the method of computational screening to search for suitable perovskite redox materials for H 2 O-CLAS [87]. Musgrave and Holder et al. used the method of high-throughput equilibrium analysis to screen 1,148 metal nitride/metal oxide pairs for CLAS based on calculated Gibbs energies of these materials acquired from the Materials Project (MP) database [88]. Yang et al. investigated 1699 bicationic inorganic redox pairs for CLAS by high-throughput screening based on MP database [89]. It can be expected that theoretical calculation and AI will play a more important role in CLAS in the future.

Electricity-driven CLAS
Based on the above thermodynamic analysis, it can be seen that the performances of s-MNs, early TMNs, and RENs in the H 2 O-CLAS processes are limited by the conversion of s-MOHs to metal [31], the reduction of TMOs to TMs (or TMHs), and the reduction of REOs to REHs, respectively [44,57]. The performances of late TMNs in the H 2 O-CLAS processes are limited by the nitridation of TMs to generate TMNs [90]. Electrification provides a pathway to break the thermal equilibrium limit of a reaction, which can be used to address the difficult step in the chemical looping process [31]. Such as water splitting to produce hydrogen and reduction of carbon dioxide to produce chemicals are the extremely thermodynamically unfavorable reactions, however, they can be carried out efficiently under electrically driven conditions [91,92]. In addition, building electrochemical processes can contribute to achieving the storage and utilization of renewable energy and reducing carbon emissions [93,94]. Here, we list several electrochemical processes to achieve the CLAS process based on various N carriers. Figure 5a shows the theoretical voltage of some chemical sub-reactions at 25 °C and 400 °C. Figure 5b and c show the schematic diagrams of some electrolytic cells.

Electrolysis of s-MOHs to produce metal
As above mentioned, the industrial production of alkali and alkaline earth metals is using the method of electrolysis of their chlorides. Nørskov and Jaramillo et al. had proposed the electrolysis of LiOH to produce Li and proposed a Li 3 N-LiOH-Li-mediated H 2 O-CLAS [31]. In their work, the controlled molten salt mixture LiOH-LiCl was used as electrolyze to electrolysis of LiOH based on a molten salt electrolytic cell at cell potential minimum of 3.0 V and 450 °C. Similar to this process, the electrolysis of s-MOHs to produce corresponding metals can be used to construct s-MNsmediated H 2 O-CLAS processes. Figure 5a shows the thermodynamically electrolytic voltage of s-MOHs to metals at 25 °C and 400 °C. The values of all s-MOHs are in-between 2.5 V to 3.5 V. Alkali and alkaline earth metals can be produced in the molten salt system as shown in Fig. 5b. Besides, some Li (or Na, K, Mg, Al, etc.) battery devices could be used to produce corresponding metals [95,96]. In future, the low-temperature molten salt or ion liquid could be used as electrolyte to reduce the reaction temperature, and efficient electrode material should be developed to reduce operating voltage.

Electro-reduction of metal oxide to metal
The thermal reduction of early TMOs and REOs to metals or metal hydrides is thermodynamically unfavorable from the above thermodynamic analysis (Fig. 3c). However, the theoretical electrolytic voltages of various metal oxides are below 3.0 V (Fig. 5a). The order of voltage value for various metal oxides is which decreases with the increase in temperature. Moreover, the theoretical electrolytic voltage of Zn is only 0.2 V at 400 °C. Fray et al. found a solid metal oxide can be directly reduced to metal based on a molten salt electrolytic cell [97]. Subsequently, this sample method was widely used to produce various metals or alloys [98]. Besides, solid oxide oxygen-ion-conductors were used as electrolyte to produce various metals, such as Mg, Al, Ti, Ca, Fe, Cu, Ta, Cr, Ce. This process has advantages such as simplified design, lower cost, lower energy use, and zero emissions [99].
Recently, the technologies of solid oxide fuel cell (SOFC) and solid oxide electrolysis cell (SOEC) have made great progress in the development of electrolytic devices, electrolyte materials, electrocatalysts, and fabrication technology [100][101][102][103][104][105]. Here, we put forward to reduce metal oxide to produce metal by a SOEC device. Figure 5c shows the schematic diagram of a SOEC for the electrolysis of MOs. It will be important to develop a low-temperature O ion conductor and efficient electrocatalyst and optimize the electrode-electrolyte interface to achieve this process.

Kinetics improvement for CLAS
Beyond the thermodynamic property of N carriers, the kinetic property of N carriers is another key issue for the CLAS processes. Plenty of research have been performed to investigate the kinetics of chemical looping reactions. The following strategies have been generally used and demonstrated to be feasible for the kinetic improvement (Fig. 6).

Particle size control
In general, reducing the particle size could increase the reactivity of the materials, which have been used to improve the performance of O carrier materials [106]. Cwiertny and Grassian et al. investigated the particle size effects of α-Fe 2 O 3 nanoparticles as oxygen carriers for chemical looping methane combustion. They found that decreasing particle size of α-Fe 2 O 3 (from 350 to 3 nm) could increase the duration of CH 4 was completely converted to CO 2 due to smaller particles with higher surface area-to-volume ratios. Reduction and reoxidation cycling experiments showed that the efficiency of large particle sample (350 nm) reduced fast in the first three cycle and remained stable subsequently. However, the efficiency of small particle sample (50 nm and 3 nm) was continuous declination [107]. For the CLAS, our previous work showed that the NH 3 production rate was significantly enhanced when BaNH was dispersed on a large specific surface area Al 2 O 3 support, which could increase the number of reactive sites [34].  [109]. On the other hand, it had been found that N 2 activation is sensitive to the structure of active sites, for example, B5 site for Ru-based and C7 site for Fe-based in catalytic ammonia synthesis, and size-control could adjust the surface structure of catalysts and influence the number of active sites [110,111].

Architecture optimization
The architecture of N carriers and support could affect the transport and diffusion of reactants and products. It has been reported that the core-shell architecture of O carriers and support could increase the activity and selectivity of a chemical looping process [112,113]. For example, Li et al. synthesized a Fe 2 O 3 @La 0.8 Sr 0.2 FeO 3-δ (LSF) core-shell redox catalyst used in CH 4 partial oxidation reaction [113]. In this structure, iron oxide core serves as the primary source of lattice-oxygen and LSF shell provides an active surface and facilitates O 2− and electron conductions. Besides, the inert LSF shell could enhance the selectivity of syngas by reducing the reactivity of lattice oxygen to prevent complete oxidation of CH 4 , and improve stability of the primary oxide Fe 2 O 3 by preventing particles from aggregating. AlN is one of the most studied N carriers for CLAS. It was found that the pore structure and specific surface area of AlN have a great effect on its ammonia production performance. Zhang et al. [114] synthesized a supported Al-based nitrogen carrier (AlN/Al 2 O 3 ) with a mesoporous structure by the carbothermal reduction method. They found that the excess Al 2 O 3 led to the significant improvement in N-carrier's porosity and thus ammonia production reactivity. With the technological development of the controllable synthesis of materials, the rational design and precise control of new structures could potentially improve the performance of N carriers.

Surface modification
A chemical looping reaction is composed of the following several steps, i.e., the adsorption and activation of reactants on the surface, the diffusion of ions in the bulk phase, and the desorption of products. The surface properties of N carriers are important for their reactivity. The hydrolysis of AlN is thermodynamically feasible at low temperatures, but suffers from a large kinetic energy barrier. With the aid of DFT calculations, Bartel et al. proposed that the hydrolysis of AlN, including water adsorption, hydroxyl-mediated proton diffusion to form NH 3 , and NH 3 desorption, is enabled by the diffusion of protons across the AlN surface by a hydroxyl-mediated Grotthuss mechanism [115]. This work suggested that the surface properties can have a great influence on the reactivity of N carriers, which could be helpful to guide material design.. Moreover, a lot of research found that the surface modifications of O carriers could improve their activity and selectivity [116,117]. Li et al. [116] employed alkali metal (Li, Na, and/or K)-promoted La x Sr 2-x FeO 4−δ as redox catalysts for chemical looping oxidative dehydrogenation of ethane. They found that the surface layer of alkali metal oxide on the surface of catalysts increased the resistance for O 2diffusion from the bulk and its subsequent evolution into electrophilic oxygen species on the surface, so the nonselective oxidation of ethane is inhibited. In another work, they found that promoting mixed oxides with a small amount of Rh can lower the onset temperature of methane partial oxidation by as much as 300 °C [117]. It was suggested that Rh on the surface could enhance methane C-H bond activation, which leads to higher concentrations of surface CH x species for facile oxygen extraction from the surface. These surface modification methods in the design of O carriers may be helpful for the design of N carrier materials.

Catalysis promotion
One of the most common means to accelerate the rate of a reaction is via adding catalysts. Many works have shown that catalysts could enhance the performance of CLAS. Wu [74]. As mentioned before, doping also changes the thermodynamic property of N carriers. Moreover, our recent works have shown that TM and TMN catalysts could catalyze the N 2 fixation on AH and hydrogenation of ANH in the ANH-mediated H 2 -CLAS [34,51]. Except for the above strategies, external-field assistance, composition optimization, etc. could also be used to improve the kinetics. As mentioned in Sect. 4, an electricity-driven approach provides an alternative route to make those thermodynamically unfavorable reactions occur at low theoretical potentials. However, the kinetic behaviors in this process such as gas/ solid phase mass transfer, charge transfer, and gas-solid interfacial reaction etc. should be also considered. The abovementioned strategies for kinetics improvement can also be used in the electrochemical reactions. For example, we can control the size and architecture of N carrier materials as well as catalysts to enhance the gas/ions transfer in the pore and bulk, respectively. The surface modification could influence the charge transfer and surface reaction processes. Besides, active electrode materials are demanded to efficiently catalyze the electrode reactions and reduce the overpotentials. More studies are needed on this aspect to fulfill the promising electro-driven CLAS process.

Conclusion
Chemical looping ammonia synthesis is a promising research field due to the advantages in breaking scaling relations and avoiding competitive adsorption. The development of efficient N carriers is vital for this process. This paper summarized the reaction thermodynamics of H 2 -CLAS and H 2 O-CLAS based on N carriers of main group metal and transition metal nitrides, and metal imides. The thermodynamically limited steps, for example, the reduction of metal hydroxides/oxides to metals or metal hydrides, the hydrogenation of metal nitrides/imides, could be improved by the electricity-driven method. Moreover, the strategies for improving the chemical looping reaction kinetics were summarized, in which the controllable synthesis of N carriers and the development of efficient catalysts could be important research directions. In future, high throughput synthesis and artificial intelligence may be helpful for the screening of N carriers, and the engineering design of the reactor and operation process should be developed for practical application.