Trace to the Source: Self‐Tuning of MOF Photocatalysts

Benefitting from ultra‐high porosity, large specific surface area (SSA), and rich active sites, metal–organic frameworks (MOFs) have emerged as a star material in energy and environment related applications, especially in photocatalysis. In comparison with conventional photocatalysts, MOFs can be customized in terms of electronic structure, photo‐responsiveness, and morphological dimensions by rational design, which means that MOFs eliminate the need for other species to achieve enhanced photocatalytic performance, such as metals, compounds, and polymers, which may introduce extra costs, be time‐consuming, be multi‐phase, or have an unclear mechanism. Most previous review works related to MOF photocatalysis have mixed self‐tuning and assisted‐tuning, which often gives an incomplete understanding. In this work, the self‐tunning of MOFs by analyzing component (as clusters, ligands, and inclusions), defect (defect engineering, as linkers, clusters, and oxygen vacancies (OVs), and crystal (as facets, dimensions, and amorphization) is summarized. Then, outstanding examples of these strategies applied to water redox, CO2/N2 reduction, organic conversion, and environmental treatment are discussed. Finally, the opportunities and challenges faced in the self‐tuning of MOFs are analyzed from different perspectives. This paper is a comprehensive and systematic review on the self‐tuning of MOFs to enhance photocatalytic activity.


Introduction
Almost 80% of the global energy supply currently depends on the combustion of fossil fuels, which brings a series of severe of improvement strategies for MOF photocatalysts from both self-tuning and assisted-tuning perspectives. Self-tuning is to explore the influence of the physicochemical properties, defect structure, pore, and internal environment, crystal shape/facet/ size of the metal center/organic ligand on the photocatalytic properties of MOF from the elemental composition and structure. On the other hand, the assisted-tuning strategy obtains performance enhancement with the help of other active substances, which include both MOF and non-MOF-type catalysts. Among them, it should be noted that inclusion is an awkward object for classification, which we will explain in Section 3.3. Such a classification derives from the special connection between the structure and properties of MOFs, as the tunability and stability of the porous coordination network (PCN) structure can withstand linear modulation within certain limits. [26] It allows MOFs to obtain structural control and performance enhancement by self-tuning and desired changes with the help of external materials. Further, we will elaborate on the specific comparison between self-tuning and assisted-tuning strategies in Section 2.
Regarding the assisted-tuning strategy, almost all works are directed to heterojunction-type composite photocatalysts. [4,27] These works achieve a comprehensive performance enhancement with the help of a good MOF template and active guest material. However, it was evident that the MOF substrate plays an important role, and the self-tuning strategy is essential to achieve this goal. [28][29][30] Second, we found that most reviews on photocatalysis by MOFs present a mixture of these two strategies, which seems to have created a "mess." [22,24,[31][32][33][34][35][36] Although there have been relevant summary works such as ligand strategy, [37][38][39][40] defect regulation, [41][42][43] design of opening metal sites, [44] hybrid metal centers, [45][46][47][48][49] and crystal engineering, [50][51][52][53] these works only present the self-tuning strategy of MOFs from one aspect, and the incompleteness and obsolescence of these works have gradually emerged as time passes and research progresses. Thus, a comprehensive and targeted review on the selftuning of MOFs in photocatalysis still needs to be completed, and our present work hopes to fill this gap.  [19] Copyright 2019, Zhengzhou University.

Self-Tuning versus Assisted-Tuning
In general, the optimal modification of materials can be analyzed from the following two perspectives: On the one hand, the flexibility and tunability of the material itself will allow a certain degree of modest adjustment of the material itself to meet the higher performance requirements, called self-tuning. On the other hand, the compounding of other suitable materials with the parent material physically or chemically will also obtain an effect of 1+1>2, called assisted-tuning. [54][55][56][57][58][59] In addition, a novel carbon material obtained by MOF derivation is out of the scope of our review. [60] Of course, these two perspectives have their advantages and disadvantages. This section will briefly analyze the differences and connections between self-tuning and assisted-tuning from the photocatalysis of MOFs (Figure 2).

Component Engineering
Compared [4] to composites, MOFs have the advantage of homogeneity, and their tunable objects include metal centers, organic ligands, and channel molecules. This figurative modification strategy often allows self-tuning in the direction we expect and makes it easier to obtain explicit structural characterization and reliable mechanistic analysis. The composite material is characterized by "multiple" and "hybrid," and the process of compounding various materials by physical or chemical means not only requires at least one additional step in the synthesis procedure but also brings more headaches in characterization and analysis. Our recognition of complex materials is still preliminary due to the level of cognition and characterization techniques, and new theories and explanations are emerging. However, a unified explanation still needs to be provided. Thus, the self-tuning strategy has two significant advantages over the assisted-tuning strategy in compositional engineering. First, thanks to the development of coordination chemistry, MOFs have a uniform and defined composition, which makes our regulatory targets more specific. Second, the characterization of MOFs after self-tuning does not increase the difficulty due to their simple composition and well-defined structure.

Defects Engineering
For [61] MOF catalysts, the metal center is the active site for most catalytic reactions, and the "photosensitive antenna" property of the ligand is essential for photocatalytic reactions. Defective MOFs will allow more active sites to be exposed to promote the reaction. Combined with the ultra-high SSA and porosity, the chance of reactant collisions will increase dramatically. Thus, defect chemistry allows MOFs to reasonably regulate the active site, energy band structure, pore structure/environment, and SSA within a certain range. To this extent, defect engineering is a practical tuning strategy. While assisted-tuning strategies usually rely on the photosensitivity and catalytic properties of exterior materials, especially for composite MOFs photocatalysts, extra attention should be paid to the energy band matching principle to obtain effective performance enhancement. It undoubtedly adds a new challenge. Therefore, not only does the assisted-tuning strategy increase the time-consuming and labor-intensive process, but it also fails to obtain the expected results.

Crystal Engineering
In [62] crystallography, the thermodynamically and kinetically stable dominant facets formed under natural conditions are often lousy activity facets, and obtaining crystals with highly active facet exposures with the help of different strategies is one of the leading tuning goals of crystal engineering strategies. The self-tuning strategy is not only to gain highly active facets but should be optimal in terms of crystal shape, size, and long-range ordered structure. In addition, MOF substrates with advantageous facets, crystalline shapes, sizes, and long-range ordered structures can be more effectively compounded well with other materials through high-energy facets. It is an effective tool for expanding MOFs to MOFs composite materials.
In addition, other disadvantages of the assisted-tuning strategy include the wide dispersion in terms of composition, distribution, morphology, and properties. Such dispersion seriously affects the reliability of the composite, while the "pure" crystallization obtained by the self-tuning strategy often leads

Component Engineering of MOFs
It is well known that the compositions of MOFs consist of inorganic metal core clusters, organic linkers, and inclusion molecules that form 1, 2, or 3D extended coordination networks. Structure determines properties, while properties reveal structure. Therefore, modification or replacement of metal centers and organic ligands and encapsulation of active substances are regarded as three basic ideas for tunable material composition of MOFs, which also opens paths for structural complexity, functional diversity, and performance enhancement of MOFs. In this section, we will explore the recent development of these modification strategies in modifying or replacing metal clusters and ligands of MOFs and encapsulating active inclusions, respectively. From the relevant results in recent years, we systematically summarize and compare the excellent cases of composition modulation engineering of MOFs to optimize the photocatalytic performance of MOFs. In order to clearly understand the sub-point basis of the content of this section, the modifications, [63] substitutions, [64] and encapsulations [65,66] involved in the following are explained here in a targeted manner. By modification, we mean the addition of additional modifiers such as functional metal atoms/complexes, [67][68][69][70][71] or organic groups [72][73][74] to the original structure of MOFs (clusters or ligands). By substitution, we mean that other metal atoms or ligands replace part/all of the metal clusters [75][76][77] or organic linkers [78][79][80] on the parent MOFs to form multi-metal/ligand-MOFs. In contrast, species encapsulation refers to the artificial introduction of other active substances to occupy the MOFs pore cavities to obtain "Inclusions@MOFs (Inc.@MOFs)" type composites. [81] We believe that clarifying the relationship between them is necessary to understand MOFs' compositional tuning engineering and enhance their photocatalytic performance.

Clusters Engineering
The transformation strategy for metal clusters includes substitution and modification. Metal ions play an important role as active sites in catalytic reactions, which makes cluster engineering more cautious. In particular, it is easy to gain negative effects at the cost of losing active sites.

Metal Substitution Strategy
The metal substitution strategy refers to substituting metal atoms on metal clusters in the parent MOFs with other metal atoms. The most typical example is the substitution of Ti IV for Zr IV . [82] This strategy for constructing photoactivity MOFs can be divided into two types: in situ synthesis and post-modification. For the former, the in situ synthesis strategy is achieved by directly adding different metal atoms with similar charge number, ionic radius, and coordination number into the reaction systems. In contrast, the post-synthetic modification (PSM) is achieved by adding the synthesized MOFs into the solutions containing the metal atoms different from the added MOFs. Naturally, both methods have their advantages and disadvantages. In comparison with in situ synthesis method, PSM method can only achieve partial substitution of metal ions in most cases. [83,84] Moreover, the structure of MOFs is prone to collapse during the PSM process. However, the PSM method is simple and easy, and the obtained modified MOFs are isostructural with the parent MOFs in most cases. On the contrary, it is hard to obtain pure MOFs through in situ synthesis method. Through regulating the metal centers of the MOFs, not only the reactive sites can be tuned, but also their energy band structure will also be regulated, thus enhancing the photocatalytic activity. Due to the high stability and excellent photocatalytic activities, Zr IV , Ti IV , and Fe III -based MOFs are often selected as the models to carry out the metal substitution.
M/Zr-MOF: Due to its excellent stability and photocatalytic performance, research on the substitution of metal ions in Zr-MOF continues to be hot. The most common substitution metal ions include Ti, Ce, Hf, and Ni. Regarding Ti IV →Zr IV , computational studies show that different degrees of Ti IV substitution will primarily affect the electronic structure of the polymetallic MOFs, which explains why the photoactivity of Ti-MIL-125 is superior to that of Zr-UiO-66. For example, Musho's group obtained Ti x Zr 6-x -UiO-NH 2 with different degrees of substitution by post-synthesis exchange (PSE) methods, combined with theoretical calculations to simulate the effect of partial/complete substitution of Ti IV within the Zr-UiO-66 lattice on the energy band structure of MOFs (Figure 3a). [85] The experimentally synthesized Ti 5 Zr-UiO-66-NH 2 exhibits the smallest band gap of 2.60 eV, much lower than that of Zr 6 -UiO-66 at 3.76 eV. Such a bandgap modulation is attributed to the HOMO shift caused by bonding through TiOC. In the theoretical prediction, even fully substituted Ti 6 -UiO-66 with a band gap as low as 1.62 eV can be obtained. However, total replacement is hugely challenging for PSM and difficult to achieve thermodynamically and kinetically. Therefore, scholars are still actively exploring how to enhance PSE technology and improve the exchange environment to obtain the maximum exchange degree of multi-metal MOFs. For example, The PSM of multi-metal MOFs aided by microwave-assisted technology was reported by Tu et al. (Figure 3b) [86] This strategy significantly reduced the cation exchange time and greatly enhanced the degree of Ti IV →Zr IV exchange, obtaining Ti/Zr-UiO-66 with an exchange rate of more than 50% in just a few hours. We mentioned earlier that in situ synthesis has many advantages over PSM. In the report of Yuan et al., nearly ideal photoactive inorganic building units ([Ti 8 Zr 2 O 12 (COO) 16 ] clusters) were obtained by pre-mixing two different metal precursors (ZrCl 4 and Ti(O i Pr) 4 ) with a one-pot method, prior to the formal synthesis of MOF. [87] As shown in Figure 3c, the [Ti 8 Zr 2 O 12 (COO) 16 ] cluster consists of a vertebral structure with two Zr IV at the top and bottom and eight Ti IV at the center. Four µ 2 16 ] clusters exhibited high porosity and chemical stability with a photocatalytic hydrogen production rate of 594 µmol g −1 h −1 . Apart from Ti ions, the Ce or Hf ions can also replace the Zr ions in Zr 6 -clusters and form the Ce/Zr-MOFs. One of the fascinating examples was reported by Luo's group (Figure 3d,e). [88] The redox activity of Ce ions was introduced into Zr-UiO-66 through an in situ doping strategy to alleviate the limited redox ability of Zr ions and the difficulty of electron transfer from ligands to clusters. It was found that each Zr 6 -O cluster can only isomorphically replace one Ce atom in the secondary building unit (SBU) of UiO-66, that is, the formation of the CeZr 5 -cluster. When the doped Ce content exceeds 20%, it will decrease the photo-nitrogen fixation activity of Ce/Zr-UiO-66. The metal substitution strategy was an effective strategy for changing the energy band structure and forming electron traps to prevent the electron-hole pairs recombination, which should follow the principle that the doped metals be more readily reducible than the host metal to bring about a more pronounced change in band gap and band edge position for the MOFs (Figure 3f-i). [75] For example, the effect of different metal atom doping on the solar-catalyzed hydrogen production from Zr-UiO-66 was explored by Yu's group (Figure 3j-m). [89] Three transition metals (Co, Fe, and Ni) with valence electronic structure of 3d 6-8 4s 2 were used to obtain M/Zr-UiO-66-NH 2 nanoparticles (NPs) with heterometallic SBU by a one-pot in situ synthesis method. The optimized 0.5(Co/Zr)-UiO-66-NH 2 , 0.5(Ni/Zr)-UiO-66-NH 2 , and 0.5(Fe/Zr)-UiO-66-NH 2 exhibited about 35.22-, 13.84-, and 1.58-times higher hydrogen precipitation capacity than Zr-UiO-66-NH 2 , respectively. This dramatic increase in catalytic efficiency can be attributed to the rapid transfer of photogenerated charges from the ligand to Co (Fe or Ni) to Zr (ligand-metal-metal charge transfer (LMMCT)). The heterometallic doping of Zr-MOF, while maintaining its high stability and porosity, facilitates the coordination of the electronic structure, changes the active redox site, and promotes charge transfer in metal-to-metal excitation, thus improving photocatalytic activity.
M/Ti-MOF: In the same way as Ti IV →Zr IV -MOF, the doping of Ti-MOF with heterometallic ions (e.g., Zr, Ca, Fe, and Ni) can also have a good effect on the electronic structure, active sites, and optical properties of Ti-MOF. One of the most typical examples is the replacement of Ti IV ions in MIL-125-NH 2 by Zr IV ions, which will be compared to the case of Ti IV →Zr IV -UiO-66. For example, Belver's group reported the in situ synthesis of Zr IV partially replacing Ti IV to obtain a series of Zr x Ti-MOF (x = 15, 30, 60, 80, denoting the molar ratio of Ti: Zr) (Figure 4a-c). [90] The introduction of Zr IV provides a new energy level below the CB and above the VB, which can participate in the charge transfer process as a new state of electron motion and reduce charge recombination. In particular, although the PL spectra showed that Zr 6 Ti-MOF had the lowest carrier complexation rate, Zr 1 Ti 5 -MOF outperformed it in photocatalytic tests regarding photoactivity. It could be attributed to the poor framework stability of Ti-MIL-125 and the increased stability of the framework caused by doping a large amount of Zr IV . At the same time, in the case of Ti→Zr-MOF, Zr-MOF was able to withstand a significant degree of cation exchange and maintain the skeleton stability. According to the SBU analysis of two different types of MOFs (MIL-125 and UiO-66), the nodal structures and coordination modes of TiO 6 -clusters (MIL-125) and ZrO 8 -clusters (UiO-66) are different, which will determine the non-isomerism, degree of doping and framework stability of the multi-metal MOFs (Figure 4d). [75] In addition to Zr ions, other metal ion exchange is crucial to the structural and property effects of Ti-MOF, such as Ni, Fe, and Ca. In the in situ synthesis strategy, the composition and structure of the metal clusters in the mixed metal MOFs prepared by the one-pot method are closely related to the metal salt feeding ratio. Zhang's group achieved the gradient regulation of the target metal-oxygen cluster fraction by carefully tuning the molar ratio of Ni II /Ti IV experimentally (Figure 4j). [91] The Ni-oxo clusters in Ni-BDC direct evidence of the effective involvement of hetero metallic nodes in the excited states of charge separation was revealed for the first time (Figure 4g-i). [92] They prepared a Fe-doped Fe/ Ti-MIL-125-NH 2 with a target SBU of FeTi 7 -clusters (i.e., a Ti in each Ti 8 -cluster is occupied by Fe). The Fe K-edge X-ray absorp-tion spectroscopy (XAS) confirmed its metal-oxygen node coordination. At the same time, time-resolved optical and X-ray transient absorption (XTA) studies revealed its role as an electron trap facilitating long-lived photogenerated charge separation in the framework. It is noteworthy that the XTA measurements  [90] Copyright 2019, Elsevier B.V. d) Structures of metal nodes in MIL-125 and UiO-66 architectures. Adapted with permission. [75] Copyright 2018, ACS. e) Ti IV 2 Ca II 2 (µ 3 -O) 2 (H 2 O) 4 (CO 2 ) 8 tetramers acting as SBU and internal structure of the sodalite-type octahedral cages of MUV-10(Ca); f) photocatalytic generation of hydrogen after 24 h of irradiation. (e, f) Adapted with permission. [93] Copyright 2018, John Wiley & Sons, Ltd. g) Fe K-Edge XANES of NH 2 -MIL-125(Ti, Fe); h) proposed structure of NH 2 -MIL-125(Ti, Fe); i) Fe K-Edge XTA spectra and corresponding difference spectra for NH 2 -MIL-125(Ti, Fe). Inset: Kinetics monitored at 7.123 keV. (g-i) Adapted with permission. [92] Copyright 2020, RSC. j) Photocatalytic performance of as-prepared samples. Adapted with permission. [91] Copyright 2021, ACS.
show a sustained electronic reduction of the Fe-sites to the microsecond time range. The characterization results indicate that the Fe-doped Ti-MIL-125-NH 2 can effectively participate in the exciting charge separation state. Earlier, the effects of metal doping could only be studied with the help of theoretical calculations and simulations, and no substantial evidence was captured. For example, Javier et al. obtained Ca/Ti-MUV-10 (MUV = material from the University of Valencia) with micron dimensions and the general structural formula [Ti IV 3 Ca II 3 (µ 3 -O) 2 (btc) 4 (H 2 O) 6 ]·solvent through one-pot method (Figure 4e,f). [93] The SBU of Ca/Ti-MUV-10 consists of three hexacoordinated Ti IV and three hexacoordinated Ca II centers, eight µ 2 -O bridges, and two µ 3 -O bridges and H 2 O molecules to complement the coordination sites, forming an eight-sided cage-like structure. Based on the photoactivity performance of the Ti-oxo cluster itself, the enthalpies and free energies of Ca II exchange of several transition metals were calculated using the density functional theory (DFT) to verify that this strategy is favorable for photoactivity. The results indicate that Mn II with a high spin d 5 conformation is thermodynamically more favorable. Experimentally, Mn/Ti-MUV-10 did show better hydrogen production performance than Ca/Ti-MUV-10. Compared with Zr-MOF, the cation exchange of Ti-MOF is more complicated. First, the formation conditions of Ti-O clusters are harsh, which is the direct reason for the relatively late development of Ti-MOF. Second, the stability of Ti-MOF is poor, as evidenced by the framework collapse and performance degradation caused by the exchange. In particular, we did not find any relevant cases of the post-synthetic cation exchange of photocatalyst Ti-MOF.
M/Fe-MOF: Fe-MOFs are one of the new low-cost, environmentally friendly, and attractive materials. Increasing the complexity of metal clusters in Fe-MOF by post-synthesis or in situ doping is an effective strategy to modulate the electronic structure and catalytic properties. For example, they used Fe-PCN-250 (Fe 3 (µ 3 -O)(ABTC)) 1.5, ABTC = 3,3",5,5"-azobenzene tetracarboxylate) as a model for the one-pot synthesis of a series of Fe 2 M(Fe, Ni, Co, Mn)-PCN-250 with homocrystalline mixed metals for use as a Photo-Fenton catalyst for the degradation of methylene blue (MB) by Zhou's team. [94] According to the binding energy (BE) analysis of Fe 2p 3/2 by XPS: Fe 2 Mn (710.9 eV) < Fe 2 Co (711.1 eV) < Fe 2 Fe (711.2 eV) < Fe 2 Ni (712.0 eV), following the order of metal electronegativity: Mn (1.55) < Co (1.88) < Ni (1.91) (Figure 5a-d). A lower BE will promote a rapid transition between high/low oxidation states of the metal elements and can be an excellent photocatalyst. In addition, Wu and Chen et al. successfully used the same strategy  [95] Copyright 2021, Elsevier B.V. i) Material preparation process of two mechanisms for the formation of Fe II -CUS: the departure of anionic ligands and the reduction caused by DMF; j) XRD patterns of mixed-valence MIL-53(Fe) samples and the simulated patterns of MIL-53, MIL-53-1, and MIL-53-2; k) photodegradation of RhB with mixed-valence MIL-53(Fe) in visible light irradiation; l) band gap structure of mixed-valence MIL-53(Fe). (i-l) Adapted with permission. [101] Copyright 2019, ACS.
to develop bimetallic L-MIL-53 (Fe, Mn/Cu) and Al/Fe-MIL-53 photocatalysts with low crystallinity, respectively. [95,96] Unlike the in situ synthesis, the PSM strategy for metal clusters, Fe-MOFs as Photo-Fenton catalysts prefer to construct mixed valence (Fe II/III ) metal centers to obtain easily controllable active variable valence Lewis acid sites (Figure 5e-h). For example, in Fe-MIL-100, [97] Fe-MIL-88, [98][99] and Fe-MIL-101, [100] Fe II/III -CUC with mixed valence were obtained by post-oxidation under vacuum or inert atmosphere. In the report by Chen et al., the formation of Fe II -CUC can be attributed to two aspects: the de-liganding of anionic ligands (F − and OH − ) and the DMF-induced reduction. [101] The 1D mixed-valence Fe II/IIIoxo chain influences the energy band structure of Fe-MIL-53, especially controlling the position of the CB, thanks to the charge transfer of Fe II →O→Fe III in the Fe-oxo clusters (Figure 5i-l).
In addition to the above common metal-based MOFs, hetero-metal substitution strategies have also been investigated in other metal-based MOFs, such as Cu→Co-MOF, [102] Ni→Zn-MOF, [103] Cu→Zn-MOF, [104] and Na→Cu-MOF. [105] In conclusion, metal substitution is one of the promising strategies for modifying MOF photocatalysts. However, mastering the kinetic or thermodynamic codes is the only way to extend this strategy to other types of MOFs, whether the in situ doping or PSM.

Clusters Modification Strategy
Due to the presence of unsaturated coordinated metal sites in MOFs, many groups or molecules are coordinated at the end of metal clusters in addition to organic ligands, such as -OH, solvent molecules, and H 2 O. The delocalization of these terminal groups brings an ideal platform for modifying MOFs, which can implant photosensitive functional groups, metals, or molecules through metal or oxygen coordination, thus obtaining MOFs with enhanced photocatalytic activity. However, the modification strategy of the cluster will undoubtedly lose some of the Lewis acid sites but enhance the LMCT process. Therefore, there must be a balance between the enhanced LMCT and the loss of active metal sites, which determines the optimal level of metal modification. Benefitting from the excellent stability and the presence of unsaturated coordinated metal sites in MOFs, some polymetallic clusters are often selected as the models to complete metal site modification. In this part, we will highlight the Zr IV -, Ti IV -, and Fe III -based MOFs, which have usually been studied in photocatalysis. X→Zr-Oxo: The PSM strategy of metal clusters, also known as solvent-assisted ligand doping (SALI), was first proposed by Farha et al. The modification of metal clusters includes functional molecules and metals and their oxides/complexes. [106] For the former, Demel's group used diphenyl phosphate (DPPA) to replace the terminal hydroxyl group on the Zr-oxo clusters (PCN-222) to obtain hydrophobic and stable MOFs (Figure 6a-c). [107] The muscular coordination of Zr IV by DPPA and the modification of the environment around the ligand porphyrin significantly enhanced the photoactivation of O 2 → 1 O 2 . It distinguishes DPPA grafting into MOF via ligand amino groups instead of SBU. [108] In addition, unlike the PSM, selecting suitable metal precursors will result in situ-modified MOFs. For example, Sun et al. used the slow hydrolysis of Cp 2 ZrCl 2 (Cp = cyclopentadienyl) to construct trinuclear ligand-bridged metal-organic polyhedral (MOP, {[Cp 3 Zr 3 O(OH) 3 ] 4 (BDC) 6 } 4+ ). [109] According to the single-crystal X-ray diffraction analysis, an isolated cationic zirconocene coordination tetrahedron has a V 4 E 6 topology (V = vertex, E = edge), bridging with (NH 2 )H 2 BDC to form a classical tetrahedral cage. In the metal cluster of MOP, the Zr IV site is occupied by the Cp group, which solves the problem that MOP is challenging to apply in aqueous systems (Figure 6d-f). For the latter, the excellent coordination ability makes metal on metal a viable means (Figure 6g,h). [110,111] For example, a representative new Fe-UiO-66 that mounts FeO x to Zr-oxo clusters for photoinitiating and driving the activation of recalcitrant CH bonds was reported by Jiang's team. [68] The strategy is to post-ligand graft FeO x onto Zr-oxo clusters by microwaveassisted techniques, forming a FeO x →Zr-oxo charge transport path (metal-metal charge transfer (MMCT)) that can be excited under visible light (Figure 6i,j). In contrast, the original LMCT process is challenging to occur under visible light irradiation. In addition, the same strategy appeared in the study of Gao et al. with the introduction of an antagonistic electron ligand (Ferrocene, Fc) to modify the Zr-oxo clusters (UiO-66). [112] The introduction of Fc enabled the dual channel transfer of photogenerated electrons (dual LMCT channels) and enhanced photocatalytic efficiency (Figure 6k,l).
X/Ti-Oxo: The PSM of Ti-SBU has been reported in a previous case. [113] Interestingly, Wang and colleagues found that the combination of methanol and MIL-125 exhibited superior hydrogen precipitation activity than methanol and MIL-125-NH 2 under broad UV irradiation. [114] Such a phenomenon is attributed to the fact that the methanol molecule can act as a hole scavenger to coordinate with the Ti-oxo clusters and promote the separation of photogenerated carriers. In contrast, the holes on the amino group in MIL-125-NH 2 tend to become charge complex centers, which is detrimental to photocatalytic performance. While the modification of metal atoms tends to obtain a more prominent contribution, loading multiple Cu I onto Ti 8 (µ 2 -O) 8 (µ 2 -OH) 4 (MIL-125) resulted in high-density and homogeneous active sites throughout the MOFs. [71] The substantial deprotonation property of LiCH 2 SiMe 3 allows the strategy to be extended to different MOFs carriers, such as hcp-UiO-66 (Zr), MIL-53 (Al), DGIST-1 (Ti), ZSTU-1 (Ti), and MIL-125-NH 2 (Ti) (Figure 7a-c). In another study, a series of NiII-modified Ti-oxo by in situ synthesis of NH2-MIL-125-Ni0.5-1.5%/Ti (Figure 7d,e). [115] According to the DFT computational simulations, Ni II with high electron affinity can modulate the electron distribution of Ti-oxo clusters, further adjusting the band gap and band edge positions of MOFs and enhancing the LMCT process of charge transfer (Figure 7f-i).
X/Fe-Oxo: Integrating PDI (N, N'-di(propionic acid)-perylene-3,4,9,10-tetracarboxylic diimide) into the skeleton of MOFs to obtain PDI@MOFs materials with high stability and excellent optical properties is one of the effective means to enhance the photoresponse. Thus in the report by Li's group, a PDI supramolecular nanofiber with a terminal carboxyl group for mimicking the coordination modification of Fe-SBU of MIL-53 (Fe) by organic ligands to obtain an enhanced photocatalytic Z-type heterojunction composite. [116] Due to the good covalent grafting of PDI with Fe-oxo clusters, the interfacial contact effect of the two materials is significantly enhanced, which promotes the rapid transfer of photoelectrons through the phase interface boundary. According to the analysis, two materials have charge flow directions under light: LMCT and internal electric field (IEF). On the one hand, the optical jump of MIL-53(Fe) is attributed to the LMCT process. On the other hand, the polarization of the perylene nucleus and the terminal carboxyl group in the center of the PDI molecule occurs, and the photogenerated holes are transferred to the terminal carboxyl group by IEF (Figure 8a-c). In contrast, the conjugation of the perylene nucleus will cause the aggregation of the photogenerated charges. As a result, holes with strong oxidizing power accumulate on MIL-53(Fe), while electrons with strong reducing power accumulate on PDI, achieving effective separation of photogenerated electronhole pairs.
In addition, other cluster modification strategies for MOFs are being developed, such as Cr-oxo clusters modified by diamino-functionalized ionic liquids (DAIL) (Cr-MIL-101) (Figure 8d-f) [117] and Zn-oxo clusters modified by L-or D-pyrrolidin-2-ylimidazole (PYI) (Zn-MOF) (Figure 8g,h), [72] both of which are used for enhanced photocatalytic applications. In summary, the exchange and modification strategies of metal clusters are essential to promote the development of the photocatalytic field of MOFs. Although it is still in its early stages, and we cannot be familiar with the preparation and prediction of the results, cluster engineering strategies are necessarily a viable way out in the face of a large, flexible, and promising new class of materials like MOFs.

Ligands Engineering
As one of the pre-integrated MOFs components, organic linkers usually play three crucial roles in MOFs: First, as the main structure supporting the skeleton, and the size and shape of the ligands determine the MOFs topology, dimensionality, and pore size. [118,119] Second, organic ligands usually play an essential role as "photoreceptor antennas", which determine the photoresponse range of the MOFs and are responsible for the generation of photoexcited electrons that participate in the LMCT or MLCT process. [120][121][122] Finally, the linkers substituent group usually determines the MOFs' pore environment. [123,124] The ligand engineering strategies for MOFs generally include modification and substitution. The former can be divided into pre-modification and post-modification according to the order of introduction of the modified groups. The latter can be classified into hybrid linker and ligand exchange strategies according to the synthesis procedure.

Ligands Modifications
Pre-Modification: Briefly, the pre-modification strategy refers to the covalent grafting of simple groups (such as -NH 2 , -OH, -SH, -Br, -Cl, -F, -NO 2 , -CH 3 ) to the original ligand prior to the formal synthesis procedure of MOFs. This modification strategy is simple and easy to implement and can lead to a significant improvement, or even change, in MOFs performance. Thus, pre-modification was also one of the earliest ligand modification strategies to be developed. [8] In MIL-125, it was found that amino-functionalized MIL-125-NH 2 (NH 2 -BDC) exhibited superior photocatalytic performance than MIL-125(H 2 BDC). [125] This photoactivity difference not only depends on the amination-expanded photoresponsive region, [126] but it has been shown that in MIL-125-NH 2 , the photoexcited electrons are located on the Ti-oxo clusters, and the photogenerated holes are located on the NH 2 -BDC unit, especially on -NH 2 . [127] Nevertheless, the best photocatalytic activity was not obtained with a 100% degree of amination. The positive correlation between the photocatalytic activity and the degree of amination was maintained only up to 0-50%, and the linear relationship changed after the degree of amination exceeded 50% (Figure 9a). [128] Simulations based on advanced hybrid generalizations demonstrate that the electronic structures of MIL-125 and MIL-125-NH 2 are changed, the interaction energy of charge carriers of MIL-125-NH 2 is positive, and the compounding time of charge carriers is one order of magnitude larger than that of MIL-125, which reduces the possibility of photogenerated carrier compounding (Figure 9b). [129] This discovery implies that  [114] Copyright 2020, Elsevier B.V. d) Structure of MIL-125 and the Ti 8 (µ 2 -O) 8 (µ 2 -OH) 4 12+ SBU and the procedure of loading Cu I centers on the Ti-SBU and corresponding structure around the µ 2 -O-site in each step; e) C 2 H 4 productivity of MIL-125-NH 2 -Cu I with different Cu loadings. (d, e) Adapted with permission. [67] Copyright 2021, ACS. f) The calculated band structure of as-prepared samples and the Ni II dopants effect on the photocatalytic performance; projected density of states (PDOS) for Ni II doped Ti-MOF indicated as g) TiMOF and h) NT; i) photocatalytic performance of as-prepared samples. (f-i) Adapted with permission. [115] Copyright 2020, Elsevier B.V.
-NH 2 can stabilize holes and promote efficient separation of photogenerated electron-hole pairs, allowing MIL-125-NH 2 to hydrogen production under visible light conditions, while MIL-125 cannot. [130] In addition to -NH 2 , MIL-125 modified by other groups (such as -OH, -NO 2 , and -SH) also exhibited different photocatalytic properties. Olga's team analyzed the non-radiative electron-hole recombination in MIL-125(OH/ NH 2 /SH) affected by ligand functionalization using the firstnature principle. [131] Depending on the type of substituent, it will allow adjusting the type and strength of the carrier complex channel. For example, the introduction of -OH enhances the vibrational coupling between the ligand and metal node vibrations, thus increasing the electron-hole complexation rate while displacing -SH and -NH 2 to induce the splitting of influential spectral peaks and their shift to lower frequency modes, including soft phonon modes, which favor slower electron-hole recombination (Figure 9c). In another report, a pre-modification strategy of such ligands mainly affects the energy band structure of MOFs, especially the contribution of substituents becomes the main component of the VB. In contrast, the CB remains unchanged. [132] In UiO-66, Carcia et al. were the first to compare UiO-66 and UiO-66-NH 2 for hydrogen production by hydrolysis under UV irradiation. [137] And then, the first evidence of photoexcited electron transfer from the NH 2 -BDC to the Zr-oxo cluster using photoluminescence spectroscopy in Sun's study of UiO-66-NH 2 for the photoreduction of CO 2 (Figure 9d,e). [133] They also prepared UiO-66-NH 2 with NH 2 -BDC and 2,5-diaminoterephthalate (DTA) using a mixed-ligand strategy, which exhibited enhanced light absorption and adsorption conversion efficiency in the photocatalytic reaction. Except for -NH 2 , Shen et al. expanded the effect of H 2 BDC ligands modified with different electronegativity substituents on the photooxidation of As III and Cr VI by UiO-66-X (X = H, NH 2 , NO 2 , Br). [134] The order of the absorption fringe bands of the modified UiO-66-X on the UV-vis spectrum was: -H (320 nm) < -Br (360 nm) < -NO 2 (400 nm) < -NH 2 (450 nm) (Figure 9f). UiO-66-NH 2 showed a significant red shift compared to the other samples, which was attributed to the lone pair of electrons on the N atom of the amino group interacting with the π * -orbital of the benzene ring, increasing the electron density of the antibonding orbital. It resulted in the generation of absorption in the visible region due to the elevation of the HOMO energy level. In order to figure out the effect of the modified groups on the light absorption and catalytic properties of UiO-66-X, a combined theoretical and experimental study was conducted for the first time in 2015. [135] The electronic properties of UiO-66-X for monoand bifunctional junctions (X = OH, NH 2 , SH) were studied using static time-dependent density flooding theory (TD-DFT) of molecular dynamics simulations. Compared to monosubstituted ligands, double-substituted ligands show more promising variations, making them candidates for further photocatalyst studies (Figure 9g). It was further confirmed in the work of Qiu et al. The energy band structure and photocatalytic properties of Ce-UiO-66-X (X = H, NO 2 , NH 2 , Br) were optimized with the modification groups ( Figure 9h). [136] In addition to MIL-125 and UiO-66, ligand pre-modification strategies have been applied to other types of MOFs, such as PCN series. [138,139] The advantages of this strategy include ease of handling, high quantifiability, and ease of synthesis, which also provides the basis for post-ligand modification. However, the degree of modifiability, group type, and modification of ligands is limited, [22] and the pore blockage caused by transitional modifications will affect the performance of MOFs. Therefore, finding the balance between modifications and performance should be the focus of attention.
the LMCT process and expanded the MOF's photoresponse range. In the following year, Chen et al. prepared Zr-UiO-66-(SOCH 3 ) x (SCH 3 ) 2-x (x = 0, 0.4, 0.6, 2, indicating the ratio of the two ligands) with different band gap widths and structures by post-oxidation. [145] The energy band structure analysis reveals that the change from -(SCH 3 ) 2 to -(SOCH 3 ) 2 increases the number of energy bands near VB but has almost no effect on the orbital energy of VB, [121,146] which leads to a similar band gap for a series of Zr-MOFs, but exhibited enhanced photocatalytic efficiency (Figure 10i-l).
Besides coordination and covalent bonds, ionic bonds are also essential bonding modes for post-ligand modifications, such as ionic liquid-modified Cr-MIL-101-IMOH-Br − for the cycloaddition reaction of CO 2 and epoxide. [147] In conclusion, we can obtain the rule that in most cases, the post-ligand modification results from a re-modification based on a pre-modified moiety (e.g., NH 2 ). Although limited by the small variety of modification groups available in the pre-modification strategy, the PSM offers a promising approach to modifying MOFs thanks to the rich synthetic chemistry.

Ligands Substitution
Hybrid Linker Strategy: A hybrid linker strategy integrates two or more organic ligands into a MOF, which generally includes two types: iso-ligand mixing and unequal ligand mixing. A mixture of two or more different types of organic ligands, each with an unequal number of junctions, such as a dicarboxylic acid and a tri/tetracarboxylic acid, is also called multiple MOFs (MTV MOFs). MTV MOFs have been a critical approach to increasing complexity and improving functionality, first appearing in MOF-5. [148] It integrates multifunctional ligands with different geometries and connectivity into one frame, enabling the diversity of MOFs and functional expansion. [149,150] For example, in Zr-MOFs, based on the combination of terephthalic acid and its derivatives with tetra(4-carboxyphenyl)porphyrin (TCPP), it was possible to obtain 49 different functional MOFs, [124] even possible to integrate metal complexes with carboxylic acid connectors into the MOFs. [151] As the most commonly used photosensitive and large π-structured organic ligands, porphyrins have become important targets for ligand modification and hybridization strategies. [5,152,153] It is well known that the introduction of ligands with large π-structures or photosensitive units to enhance the photoresponse characteristics and charge transport efficiency of the parent MOF are two target directions of hybrid ligand strategies. [120,154,155] Zhou's [156] and Deng's group [157] constructed 3D frameworks with good porosity and stability by integrating H 3 BTB (4,4",4""-benzene-1,3,5-triyltris(benzoic acid)) and TCPP into Zr-PCN by one-pot method, respectively (Figure 11a,b). Such a combination strategy was used for the photoreduction of CO 2 and degradation of diclofenac (DF), respectively, and the enhanced catalytic efficiency was attributed to the efficient charge transfer between the photosensitive porphyrin fragment and the Zr-oxo cluster. In addition, two similar jobs were reported by Liu and Huang et al. [158,159] In another report by Lin and colleagues, a bidentate carboxylic acid complex of Ir, Re, and Ru with photocatalytic ability was introduced as the second ligand to construct MTV UiO-66 with high stability and porosity using UiO-66 as the parent one (Figure 11c,d). [160] Such a combination leading to enhanced photocatalytic performance is explained by a) the intermediate course involving Ir-, Re-, and Ru-complexes; [161] b) the oxidative activation of Ir-, Re-, and Ru-complexes by photo excitation, which participates in the catalytic reaction; [154] and c) the bimolecular pathway initiated by CO 2 molecules bridging [(CO) 3 (bpy)Re I ](CO 2 )[Re I (by)(CO) 3 ]. [162] In addition, a similar work was published earlier by Choi's group to stabilize ReTC (Re I (CO) 3 (BPYDC)Cl, BPYDC = 2,2"-bipyridine-5,5'-dicarboxylate) in Zr-MOP (metal polyhedra) for visible CO 2 →CO conversion. [163] Except for MTV MOFs, an isomeric mixed ligand strategy is one in which the number of ligand junctions is the same but with different functional groups, such as terephthalic acid and amino terephthalic acid. For example, Luis' and Kumar's team designed a hybrid ligand UiO-66 with binary and ternary ligands, including UiO-66-NH 2 -NO 2 , UiO-66-NH 2 -(OH) 2 , UiO-66-(OH) 2 -NO 2 , and UiO-66-NH 2 -(OH) 2 -NO 2 , respectively, for photocatalytic water remediation (Figure 11e). [164,166] In the former work, excellent adsorption performance depends on the electrostatic interaction between the functional groups on the ligand side chains and the chromate anion, particularly -(OH) 2 and -NH 2 . While the latter attributes the enhanced photoactivity to the different electronic properties of the substituents, this is because the photoactive performance of UiO-66-X maintains a good correlation with the Hammett constant. [164] In the work of Wang et al., a mixed [Ru(bpy) 3 ] 2+ -derived dicarboxylate ligand (H 2 L 1 ) and azidemodified dicarboxylate ligand (H 2 L 2 ) were constructed for functional UiO-MOFs. [165] Therein, H 2 L 1 was used as a photosensitizer to enhance the photo responsiveness of MOF (Figure 11f). At the same time, the Azide group on H 2 L 2 could be further anchored to the [Fe 2 S 2 ] catalytic site into MOF by post-covalent modification via Click Reaction, and a new hydrogen precipitation catalyst, UiO-MOF-Fe 2 S 2 , was obtained. Not only did this strategy obtain a stable framework structure and a new electron transfer pathway, but it also distinguished from other weak ligand strategies; Fe 2 S 2 was more easily controlled in terms of linkage ability and grafting number by covalent grafting onto H 2 L 2 . In addition, this report also provides a new way to combine hybrid ligands and PSM to construct new functional MOFs.
However, to obtain MOFs with excellent stability, the hybrid ligand strategy is limited by the geometry and connectivity of the second ligand to prevent the creation of a "mixed phase" during the synthesis process or to affect the integrity of the framework itself. Thus, the limited second ligand selectivity inhibits the broad applicability of hybrid ligand strategies, as the vast majority of studies focus on Zr-MOF with high stability, connectivity, and defect tolerance. [167] Ligands Exchange Strategy: The ligand exchange strategy refers to the partial/complete replacement of the original ligand by other similar ligands with decorated groups through an exchange process. [75,168] Like the metal exchange course in cluster engineering, ligand exchange can also effectively regulate the band gap structure and catalytic performance of MOFs. [169,170] This strategy can also be subdivided into two categories depending on whether the fundamental backbone of the organic ligand is the same: homologous and non-homologous ligand exchange.
For example, a x%-MIL-125-(SCH 3 ) 2 with H 2 BDC and H 2 BDC-(SCH 3 ) 2 used a solvent-assisted ligand exchange strategy (SALE) (x% is the ratio of BDC-(SCH 3 ) 2 exchanged into the MOFs) by Han and colleagues. [146] Among them, H 2 BDC and H 2 BDC-(SCH 3 ) 2 have the same basic skeleton, both are terephthalic acid, and the difference lies in the side end modification group, which is a homologous ligand. The S 3p -orbitals of the methylthio group can provide electrons for the aromatic ring and act as auxochromic and bathochromic groups to expand the photoresponse range of MOFs and increase the HOMO energy level (Figure 12f-i). For the non-homologous ligand exchange strategy, it can be understood with the help of Sun's work. Aminated linear dicarboxylic acid ligands of different lengths have different aromatic ring skeletons and are not homologous ligands. [171] Experimental and theoretical calculations showed that shorter dicarboxylic acid ligands and more amino side chains lead to smaller E abs (Absorption Energy), which is favorable for photocatalytic reactions (Figure 12a-e). In addition, linear dicarboxylic acid ligands anchored with molecular catalysts (Rh and Ru) were integrated into the Zr-MOFs by PSE, Figure 11. a) Structure of PCN-137 (left) and PCN-138 (right). Adapted with permission. [156] Copyright 2019, ACS. b) The topology (up) and the 3D framework of PCN-134 (down). Adapted with permission. [157] Copyright 2019, Elsevier B.V. c) Synthesis of doped UiO-67; d) structure model of MOF-1 showing doping of the L1 ligand into the UiO-67 framework (left) and SEM micrograph of intergrown nanocrystals of MOF-1 (right). (c, d) Adapted with permission. [160] Copyright 2011, ACS. e) Key functions needed in MOF for the Cr VI to Cr III capture and transformation (I) and multivariate encoding strategy of the UiO-66 framework developed (II). Adapted with permission. [166] Copyright 2022, Elsevier B.V. f) Modification of UiO-MOF via a click reaction was used to form new catalysts UiO-MOF-Fe 2 S 2 incorporating [Fe 2 S 2 ] catalytic sites. Adapted with permission. [165] Copyright 2019, Elsevier B.V. which distinguishes them from the one-pot synthesis in the hybrid ligand strategy. [172,173] The basic backbone of the metal molecule catalyst is bpydc (2,2″-bipyridine-5,5″-dicarboxylate), which is different from bpdc (biphenyl-4,4′-dicarboxylate), and the two are not homologous ligands. N atom-anchored metal molecule catalysts in bpydc act as photosensitizers to improve the photoresponsive properties of MOFs while constructing new electron transfer pathways (Figure 12j,k). The ligand PSE is an essential complement to ligand engineering but also has some disadvantages. Compared with the one-pot mixing strategy, it has the disadvantages of being time-consuming and labor-intensive, limited exchange degree, and harsh exchange conditions. However, it can avoid the appearance of "mixed phases".
Ligands engineering is one of the most popular means of MOFs modification due to its simplicity and versatility. Although ligand strategies were first applied to improve the performance of MOFs, they are still in the early stages of research. Many kinetic and thermodynamic factors are not yet clear, primarily that MOFs properties are controlled by their stability, and the majority of studies are still stuck in Zr-, Ti-, Zn-, Fe-, or Cr-MOFs, while few involve other metal-based MOFs.

Inclusions
MOFs are widely known for their high porosity, which allows for the occupation of a wide variety of inclusions, such as solvents, ions, and reactive substances. Inclusions are an awkward classification because, unlike normal composites, it only surface-loads, but "incorporates" the guest molecule as part of the catalyst. Such wrapping depends on the physicochemical properties of the MOFs' framework and pore environment. [174,175] On the other hand, the "engulfed" object does not belong to the MOF itself. From the former point of view, it should be considered a self-tuning strategy, but from the latter end of view, it is an assisted-tuning strategy. Actually, the encapsulation strategy differs from the loaded composites in that the encapsulated substance should be attributed to one of the components of the MOFs. Hence, in this paper, we attribute Inclusions to the self-tuning strategy in case the discussion is omitted or insufficient. The removal/addition of inclusions can bring exciting changes to MOFs structure/properties. [25,176] From weak interactions of light gases to strong coordination interactions with open metal sites, [177] the host-guest chemistry of MOFs reveals the importance of "Inc.@MOFs", which can affect the adsorption, [178] optics, [179,180] and thermal conductivity. [181] Moreover, the changes in host-guest interactions can also make a contribution to influencing the photoactivity of MOFs materials. [182,183] In photocatalysis by MOFs, the influence of the inclusions on the photoactivity includes three aspects: exposure of the active metal site, photoresponsive properties, and its catalytic activity. The ability of metal oxides or polymetallic oxides (POMs) to undergo rapid, reversible, and stepwise multiple electron transfer reactions without changing their structure  [146] Copyright 2018, Wiley-VCH. j) Synthesis of 3 by the PSE of 2 with bpdc in Zr-bpdc. Adapted with permission. [172] Copyright 2016, Wiley-VCH. k) Heterogenization of a rhodium complex into the framework of UiO-67 through post-synthetic linker exchange. Adapted with permission. [173] Copyright 2015, can open up new avenues of research in catalysis and other fields when embedded as inclusions in MOFs. [ [185] Due to the synergistic interaction between POM and OMPO, compound 2 exhibited an ultrahigh CO yield (10 852 µmol g −1 h −1 ) and selectivity (> 93%) (Figure 13a-d). Larsen's team encapsulated tetra(N-methyl-4pyridyl)-21H,23H porphine (TMPyP) in two Cd-MOFs, named CdTMPyP@MOM-11 and CdTMPyP@MOM-12. [186] The planar orientation of the porphyrin molecules is significantly affected by the MOM-11 and MOM-12 inner holes that fix the porphyrin S 1 -CT coupling (i.e., the singly excited state (S 1 ) and the compact charge-transfer state (CT) of TMPyP), leading to differences in steady-state emission (emission lifetime and absorption properties) (Figure 13e-g). It demonstrated how a specific cavity could modulate the excited state properties by regulating the conformation of the accessible inclusion. In a similar report by Warnan's group, the benchmark photocatalyst fac-ReBr(CO) 3 (4,4"-dcbpy) (dcbpy = dicarboxy-2,2"bipyridine) and the photosensitizer Ru(bpy) 2 (5,5"-dcbpy)Cl 2 (bpy = 2,2"-bipyridine) were synergistically wrapped in a cage of nontoxic and inexpensive MIL-101-NH 2 (Al) via noncovalent host-guest interactions. [187] The effect of regulating the reaction environment as a nanoreactor in MOFs was demonstrated and evaluated by adjusting the ratio of immobilized catalyst to photosensitizer. It illustrated the optimal efficiency of two photo sensitizers and one catalyst in each cage and further determined the relationship between the molecular complex size, the MOF pore window, and the number of molecules that can be carried in each cage (Figure 13h,i). In another study, host-guest and bimetallic synergistic effects were achieved by encapsulating noble metal or alloy NPs into the MOF. [174,188,189] The Pt NPs embedded in a flower cluster-like NH 2 -UiO-68 for enhanced photocatalytic CO 2 reduction by Sun's group. [190] The  [185] Copyright 2022, RSC. e) Diagram illustrating the formation of CdTMPyP@MOM-11 and CdTMPyP@MOM-12; f) Top: Diagrammatic representation of the energy levels for TMPyP relative to the rotational positions of the porphyrin pyridinium rings. Bottom: Effect of Cd 2+ insertion on the relative energy of the CT and S1 states of TMPyP with all four pyridinium rings nearly perpendicular to the porphyrin plane; g) Normalized UV/Vis absorption spectra of TMPyP, CdTMPyP, suspensions of CdTMPyP@ MOM-11, and CdTMPyP@MOM-12. (e-g) Adapted with permission. [186] Copyright 2019, Elsevier B.V. h) Schematic host-guest representation with photoinduced electron transfers between TEOA, the [Ru(bpy) 2 (5,5'-dcbpy)]Cl 2 photosensitizer (2), and the ReBr(CO) 3 (4,4'-dcbpy) catalyst (1) for CO 2 reduction, in a 25 Å cage; i) Accumulated TON versus time plot showing recyclability (cycle number in gray) with 1 and 2 as homogeneous complexes (black), Re-Ru@MIL(R MOF 7.9) (blue), Re@MIL (red), and blank tests of pristine MIL-101-NH 2 (Al) or Ru@MIL (green). (h, i) Adapted with permission. [187] Copyright 2021, RSC. j) Schematic diagram of the synthetic process of Pt(2)/NH 2 -UiO-68 (A), Pt(1)@NH 2 -UiO-68 (B), Pt(2)@NH 2 -UiO-68 (C), and Pt(4)-NH 2 -UiO-68 (D) hybrid photocatalysts; k) comparisons of photocatalytic CO production rates for different catalysts. (j, k) Adapted with permission. [190] Copyright 2019, RSC.
positive slope of the Mott-Schottky curve indicates that NH 2 -UiO-68 is an n-type semiconductor with LUMO and HOMO potentials of −0.60 and 2.27 V (vs NHE), respectively. The effective energy level of Pt NPs is 1.15 V versus NHE, according to the previous study. When Pt NPs are uniformly distributed into the NH 2 -UiO-68, the formation of tight-contact Pt-MOF Schottky junctions will guide rapid electron transfer and promote carrier transfer and separation. Thus, Pt NPs@NH 2 -UiO-68 exhibited better photoactivity than the simple mixture of Pt NPs/NH 2 -UiO-68 (Figure 13j,k). Supramolecular host-guest chemical systems generally contain multiple components. In MOFs systems, the interactions, size, and structure between the metal cluster, organic ligand, and inclusions play a crucial role in most design strategies. For the design of "Inc.@MOFs" type composites, the search for optimal intermolecular or interfacial interactions helps the guest to occupy the host. Thus, supramolecular host-guest chemistry is integral to understanding and designing MOFs modification strategies, which now presents some dilemmas. For the host, the thermodynamics and kinetics of the frame opening and closing processes are unclear, whether the in-pore molecule influences the stability. For the inclusion, the alternative material is limited by the frame pore structure and size, with a single/multiple guest molecules distributed in a single/ different cavity, which may become the future direction of MOFs' host-guest chemistry.

Defects Engineering
Defect engineering (DE) is mainly used to optimize conventional semiconductor materials and generally arises from interatomic or intermolecular dislocations, interlayer mismatches, crystal interfaces, and vacancies. [191] In addition, defective sites usually function as active sites and electron traps in photocatalysis. [42,192] Different from conventional materials where the number and location of defects are not controllable, the above problems can be solved using DE tools for MOFs optimization. [193] In this section, we classify the DE of MOFs into three types: cluster defects, ligand defects, and oxygen vacancies (OVs). The first two types are commonly found in most research or review works, which are the most common types of defects. [194][195][196][197][198] The OVs were once thought to be a type of ligand defect, [199] but we believe this is distinct, as detailed in the OVs section.
Although defects engineering is a proven strategy for optimal modulation, it is challenging to introduce a defined type and number of defects in a material. [200] According to the sequence of defect introduction, it can be divided into two methods: introduced after synthesis and introduced during the synthesis process. They differ in that the former is not limited by crystal growth kinetics and can readily introduce vacancies, doping, and disorder, yet is limited by the reaction depth and can only obtain defect sites in shallow layers of the material. The latter benefits from crystal growth dynamics and can readily introduce bulk defects in the interior, which is highly dependent on the growth dynamics of dislocations and boundaries. In addition, there are some other classification methods. Based on the different states of defects, they can be divided into point, line, surface, and body defects. According to the location of the defects, they can be divided into body and surface defects.

Ligand Defects
Defective sites are found everywhere in the routine synthesis procedure of MOFs. Zr-UiO-66 has been known for its high stability and tunability since it was developed and was often chosen as a model for DE MOFs studies. For example, according to structural information on Zr-UiO-66, the Zr ions form twelve-coordinated Zr 6 O 4 (OH) 4 clusters with no unsaturated coordination sites. After high-temperature activation, some coordination bonds will be broken because of the reversibility of coordination bonds, and the coordination numbers of Zr 6 -clusters are reduced to ten, or even lower. [222] The competitive ligands commonly used in general synthesis procedures include two significant groups: modulators and competitive ligands. Among them, modulators can be subdivided into solvents, fatty acids, and aromatic and inorganic acids. For example, the different equivalents of acetic acid as a modulator to obtain UiO-66-NH 2 -X (X = 0, 50, 100, 150, 200, representing the molar equivalents of acetic acid) with different degrees of ligand defects by competitive ligand coordination with NH 2 -BDC. [223] The optimal structural defect level of UiO-66-NH 2 -100 leads to the lower energy of the Zr-atom's unoccupied d-orbitals, favoring the photogenerated electron-hole separation. However, excessive defect levels may become the center of electron-hole complexes, which leads to reduced photolytic aquatic hydrogen activity. Different defect levels of UiO-66-NH 2 -X show a surprising trend of volcano-type catalytic activity (Figure 14a,b), which indicates that moderate-intensity defects help to turn on photocatalysis. Using benzoic acid (BA) as a modifier, UiO-66-NH 2 -XH with different defect levels (X = 6, 12, 18, 24, denoting different equivalents of BA) was prepared by Shi's group. [224] BA is introduced as a competing ligand to the defect in this case, while hydrochloric acid post-treatment removes BA to reduce the intrinsic energy gap (ΔE LMCT ). The exposed Zr-nodes act as adsorption and photocatalytic sites for Cr VI and show a trend of volcano-type catalytic activity with an increasing number of defects similar to Jiang's group. [223] In addition to monodentate carboxylic acid modulators, Figure 14. a) Pt@UiO-66-NH 2 -X exhibits an impressive volcano-type trend in photocatalytic H 2 production, b) maximizing at a moderate defect level. (a, b) Adapted with permission. [223] Copyright 2019, Wiley-VCH. c) Amount of the produced HCOOH over Zr-MOF-T and Zr-MOF-DF samples upon visible-light irradiation. Adapted with permission. [225] Copyright 2019, RSC. d) Crystal morphology revealed by SEM observation and simulated pore structures of UiO-66 and TCPP@UiO-66s (9-55%); e) schematic illustration of the photoreaction process in TCPP@UiO-66; f) the effect of TCPP ratio on photoreactivity in the UiO-66 system (initial DF concentration: 30 mg L −1 ; MOFs concentration: 0.1 g L −1 ). (d-f) Adapted with permission. [226] Copyright 2020, ACS. g) Schematic diagram of the defects obtained by photothermal treatment of MIL-125(Ti)-NH 2 ; h) a plausible mechanism of linker elimination by the photothermal treatment in a water medium containing TEOA. (g, h) Adapted with permission. [227] Copyright 2020, Elsevier Inc. i) Mechanism of hierarchically porous structure construction. Adapted with permission. [228] Copyright 2021, Wiley-VCH GmbH.
inorganic acids can also play a role in defect modulation. In the report of Zhu et al., a series of Zr-UiO-66-NH 2 -T (T = 353, 393, 433, and 473 K, synthesis temperature) with different defects used concentrated hydrochloric acid and temperature modulation. [225] As the synthesis temperature increased, HCl inhibited the deprotonation of NH 2 -BDC, leading to the formation of intrinsic defects due to misjoining or dislocation during crystallization. The formation of defects leads to a change in the coordination environment of the Zr-atom, exposing the OMS and promoting LMCT (Figure 14c). In addition, introducing structural defects through competitive ligands has received similar attention to the hybrid ligand strategy. Yu and colleagues prepared defect-controlled binary ligands TCPP@ UiO-66 by a one-pot method using the large mixed-linker (LML) approach. [226] The ligand sites on the Zr-oxo cluster can be reduced in connectivity in favor of binding other functional parts and allowing the presence of highly defective frameworks or subnetworks. The level of DE MOFs is controlled by TCPP equivalents, which further affects the material's phase purity, microcrystalline morphology, and properties. Based on the adsorption kinetics and photoreaction efficiency, the synergistic effect between the improved adsorption effect of defects and the high reactive substances (RS) generated by the ultralong-lived triplet state of TCPP resulted in TCPP@UiO-66 (25%) exhibiting the best DF removal rate (Figure 14d-f).
Ligand defects can reflect the intrinsic dynamic flexibility of the MOFs concerning changes in the coordination number of metal atoms; this internal deformation pattern points to de-liganding, translation, rotation, and re-liganding of the ligand. [229] For example, the preparation of MIL-125-NH 2 -T with different defect levels used water as an etchant at different temperatures (T = 25, 45, 55, 65, and 75 °C) by reported He et al. [230] Heat agitation of MIL-125-NH 2 in water systems leads to desegregation between ligands and metal clusters, exposing active Ti IV sites. Photocatalytic performance assays showed that 32.08% of the defective MIL-125-NH 2 -65 exhibited efficient NO purification (65.49%, vs 30.76% (the original MOF)) and hydrogen production (2.2-fold, vs the original MOF) under visible light. In Matsuoka's report, [227] Ti III was generated in Ti-oxo clusters using photothermal treatment, and since Ti III has a weaker coordination ability than Ti IV , decoordination or rotation can occur at a relatively low temperature (313 K). The exposed OMS on the Ti-oxo clusters exhibited plasmonic reduction sites comparable to Pt NPs co-catalysts (Figure 14g,h). In another report by Zhang, [228] Ti-MOF with graded porosity and defects were synthesized using a competitive coordination strategy with different electronegativity ligands (BDC 2− = 3.7 and NH 2 -BDC 2− = 3.2). The Ti IV was first coordinated with the more electronegativity NH 2 -BDC, and H 2 BDC was introduced for local coordination due to hydrogen bonding between -NH 2 and -COOH to obtain SBU with mixed linkers (Figure 14i). Along with the crystal growth, the ligands were misaligned or rearranged to form graded porosity and inherent defects, improving the photodegradation toluene adsorption performance and charge separation efficiency.
In conclusion, modulators, competitive ligands, etchants, and synthesis conditions can contribute to the customization of DE MOFs. The long-term goal of defect modulation is to obtain MOFs with optimal defect levels by easy and controllable means and extend them to other non-stable MOFs preparation procedures. Therefore, the route to fine-tailoring DE MOFs remains challenging.

Cluster Defects
Cluster defects have long been found to coexist with ligand defects in MOFs (Mixed DE MOFs). However, since the formation mechanism and controlled synthesis of cluster defects are not as simple as those of ligand defects, the vast majority of studies in this area have remained in theoretical calculations, [231] synthetic characterization, [232] and adsorption properties. [233] In particular, studies targeting adsorption properties have demonstrated that cluster defects bring about changes in material structure and properties, [234,235] which has attracted much attention from scholars. As an essential part of the defect engineering of MOFs, the study of cluster defects will help to understand the structure, coordination mechanism, and enrichment modification strategies. In this subsection, we will focus more on what changes cluster defects can bring to MOF photocatalysts, such as SSA, [236] multistage pore size, [237] and catalytic active site. [238] The mechanism of cluster defects is still unclear, making it either avoided in the vast majority of MOFs reports or generally attributed to DE along with ligand defects, with only a few reports explicitly describing it. For example, in Taddei's report, a series of bandgap-tunable DE MOFs were obtained by introducing aminated monodentate aromatic carboxylic acids at the cluster defect site of DE Zr-UiO-66 by a PSE method. [239] The reo defect topology associated with missing clusters is introduced in Zr-UiO-66 in TD-DFT calculations to model the electronic structure of Zr-UiO-66 containing cluster defects and the performance of photoreduced CO 2 . The results showed that the cluster defect causes the HOMO to become a set of two parsimonious local orbitals while the LUMO is unaffected (Figure 15a-e). However, many reports have illustrated the harmful effects of cluster defects relative to ligand defects. For example, Wang and colleagues synthesized a series of Zr-UiO-66-NH 2 with excellent connectivity, ligand, and cluster defects by controlling the modulating agent (formic or acetic acid), heat treatment, and synthesis time. [240] Theoretical calculations combined with experimental results showed that samples with ligand defects outperformed those without defects and cluster defects in photoinduced CO 2 (Figure 15f,g). Soon after, in a similar report, they further demonstrated that cluster defects are essential for improving MOFs photoactivity, but are not as pronounced as ligand defects. [241] In particular, Gao et al. provided more convincing evidence that repair of DE UiO-66 utilizing PSE demonstrates that cluster defects do not enhance photo-nitrogen fixation as significantly as ligand defects. [242] In conclusion, in MOFs photocatalysis, metal centers are more critical redox active sites than organic ligands, which is perhaps the direct reason why cluster defects have been little studied. However, we should not ignore the importance of cluster defects, even if the adverse effects are beneficial to help us better understand the defect chemistry of MOFs.

Oxygen Vacancies
The OVs were first discovered in 1960s while studying metal oxide and gas interaction mechanisms. Lattice oxygen atoms (oxygen ions) detach under specific conditions (e.g., high temperature and reduction treatment) to form OVs, which are a form of point defects in solid materials. As part of the materials, OVs have similar roles to other types of defects (modulating electronic structure, band gap modulation, opening active sites, and promoting carrier separation). [243] Therefore, in photocatalysts such as BiOCl, [244] TiO 2 , [245] Bi 2 O 3 , [246] In 2 O 3 , [247] and AgNbO 3 , [248] the moderate levels of OVs on photocatalytic performance is positive. Inorganic clusters in MOFs are multioxygen metal clusters (M-oxo clusters), and under specific conditions, M-oxo clusters detach from oxygen (atoms or ions) to form OVs that directly expose active metal sites. Unlike OVs, ligand-deficient sites are mostly occupied by solvent molecules, modulator molecules, or other groups in the coordination of unsaturated sites of metal clusters despite the lack of oxygencontaining carboxylic acid junctions. Such linkers' defective sites do not involve OVs. Therefore, we believe that OVs and linkers' defects are distinct.
It is well known that most MOFs are poorly stable on their own, so it is challenging to introduce moderate OVs levels. The introduction of OVs into MOFs that do not affect their basic framework has been demonstrated by a series of means, such as mixed ligand selective pyrolysis, [249] controlled heating time, [250] mixed metal doping, [251] and corrosion. [252] For example, in the report of Liu et al., they prepared OVs H-MIL-53 using ligandselective pyrolysis based on the different heat resistance of H 2 BDC and NH 2 -BDC. [253] After heat treatment, FeO bonds are broken to produce positively charged OVs, which increase light absorption while participating in the redox cycle of Fe II/III , thus improving the Photo-Fenton efficiency. Second, Zhang and Tang et al. obtained OVs TiCe-UiO-66 [254] and FeCe 2 -TA-MOF (TA = homo-trimethylbenzene) [255] with moderate OVs, respectively, with the help of a mild cation exchange strategy. In contrast, in previous studies, the introduction of OVs also required a high-temperature calcination treatment. [256] The OVs have been shown to reduce the localized state of the CB, which not only facilitates the internal LMCT of MOFs and enhances the photoresponsive properties, but also acts as a charge carrier trap and promotes the effective separation of photogenerated electrons and holes (Figure 16a,b). It is due to the large electronegativity of the O atom, which is equivalent to taking away two negative charges when the metal cluster loses an O atom. As a result, OVs become active sites with positive charges attractive to photogenerated electrons. In particular, the doped metal (Ti or Fe)  [240] Copyright 2021, Elsevier B.V. changed the order of the periodic coordination arrangement, leading to the creation of low crystalline structures that exposed more active sites (Figure 16c). According to XPS analysis, the mixed-valence Ce III/IV redox cycle will generate more OVs to maintain local charge balance and expose more reactive sites. In addition, Chang's group reported the preparation of H-MIL-125-NH 2 -Vo (H-MOF denotes graded porous MOFs) with OVs and graded pore size using etchant (PA = propionic acid) as a photocatalyst for the degradation of organic dyes. [257] The synergistic effect of OVs and graded porosity promotes the adsorption of dye molecules and complete contact with the active sites, generating more photogenerated radicals to participate in the degradation reaction. Furthermore, another novel strategy to modify acetylacetone (AA') on the amino side chain of MIL-125-NH 2 through post-synthetic ligand modification was reported by Zhang and colleagues, which can introduce a large amount of OVs in MOFs (Figure 16h). [258] As a result, the carbonyl oxygen atoms in the AA' structure can undergo further coordination interactions with Ti IV , leading to easier detachment of the coordination oxygen atoms of the Ti-oxo cluster and the generation of a large number of OVs. Interestingly, EIS shows that AA' introduction leads to lower interfacial charge transfer resistance but exhibits low transient photocurrent (Figure 16d). Furthermore, it indicated that the enhanced photocatalytic performance of MIL-125-AA' does not depend on the separation efficiency of photogenerated carriers (Figure 16e). Fluorescence and phosphorescence emission spectroscopy verified that the single-line state in MIL-125-AA' can be rapidly converted to the three-line state by inter-system crossover, promoting the production of 1 O 2 , resulting in high photoactivity (Figure 16f,g). Nevertheless, OVs have been little studied in MOF photocatalysts, and the detailed formation mechanism and mechanism of action are still unclear. However, it cannot be denied that OVs are a vital tool for modifying MOFs.
In summary, in this subsection, we review the feasibility of ligand defects, cluster defects, and OVs of MOF photocatalysis. In particular, we divorced OVs from ligand defects in the traditional sense and carefully analyzed their differences and connections. The defect engineering strategy proved to be a proven strategy for modifying MOF photocatalysts and should be widely studied. Even though the moderate level of defect types is all effective in improving the performance of MOFs photocatalysts within a suitable range, they are distinct. Of all defect types, ligand defects are the easiest to obtain and control, thanks to their tunability, stability, and pre-synthesis. As pointed out in the previous report, the tunable range of cluster defects is limited. The OVs, a newly discovered defect type in MOFs, play a vital role in the structure and performance of catalysts but are relatively poorly studied, which is attributed to the single method of oxygen vacancy introduction. As we all know, the "active site," "photosensitive antenna," and carrier separation occupy crucial parts in the photocatalytic reaction, which are the key factors affecting the photocatalytic reaction. Hence, defects are theoretically detrimental to photocatalysis. However, it was observed that moderate defects facilitate active site exposure and improve the pore structure/environment, promoting molecular exchange/binding and electron circulation within the framework, thus enhancing photocatalytic performance. Defect engineering will arise as an attractive tuning strategy for porous coordination chemistry.

Crystals Engineering
Crystal engineering has become an essential strategy for designing efficient photocatalysts. [259] It refers to using the energy and geometric properties of the interactions between different types of molecules to construct crystals with desired  structures and properties, which was first proposed in "supramolecular chemistry". The facets/internal structure of crystalline photocatalysts provide the active planes/sites in the catalytic reaction, and the atomic coordination profiles of different crystals parameters (including facets, sizes, and crystallinity) exhibit unique electronic and optical properties, [260,261] so their photocatalytic performance is heavily dependent on the structure of the crystals themselves. [175] For example, for TiO 2 , the photocatalytic activity of anatase is superior to that of rutile and slate, [262,263] which is attributed to the active facets exposed by the different crystal types. Among them, the highenergy facets {001} are usually more suitable for solar energy conversion [264,265] than the thermodynamically stable {101}, so increasing the exposure of {001} facets can significantly improve the photoactivity of anatase TiO 2 . [266,267] Besides, the size effect [268,269] and crystallinity [270,271] can also effectively tune the photocatalytic performance of TiO 2 . Such a structure-performance relationship applies to other crystalline photocatalysts, such as WO 3 , [272] Cu 2 O, [273] SrTiO 3 , [274] and BiTaO 4 . [275] Inspired by this, most MOFs are crystalline porous materials with periodic network structures, promising to open new doors in the field of photocatalysis through crystal engineering. In this section, we will systematically describe the recent progress of crystal engineering for MOF photocatalysts from facet engineering, [276][277][278][279][280] size engineering, [51,281,282] and amorphization engineering, [283,284] and present some humble opinions.

Facets Engineering
It was found that different crystalline states of the same MOF would lead to differences in crystallinity, pore effect, SSA, size effect, and facets such as polyhedral, spheres, rods, flakes, and other shapes, thus having important research value in adsorption, [285] catalysis, [286] separation, [287] and energy storage. [288] Furthermore, like TiO 2 , [289] the effect of the crystallographic engineering of MOFs on the catalytic efficiency depends on the exposure of high-energy facets, which means that more active sites can participate in the reaction. [276,277] However, nucleation and crystal growth of MOFs are extremely sensitive to conditions at multiple stages of the synthesis process, including a) incubation of the reaction solution; b) high-temperature heating to initiate nucleation, and c) cooling to low temperature to initiate growth. [290][291][292] According to the Gibbs-Wulff rule, the fast-growing high surface energy crystalline surfaces account for only a tiny fraction and even disappear slowly during crystal growth. Thus, the final morphology of the crystal depends on the slow-growing low-energy facets, which will eventually dominate. [260,293] As a final result, crystals obtained under natural or equilibrium conditions typically expose a few high surface energy crystal faces (i.e., active facets), which dramatically affects the photogenerated charge transport/separation process and is detrimental to solar energy conversion (Figure 17a). Therefore, crystallographic engineering was performed by screening metal precursors, [279,294] ligands, [295][296][297] modulators [293,298,299] with structure directing agents, [300,301] solvents, [302,303] as well as synthesis conditions (mainly temperature [304] and time [305] ) and methods (e.g., hydrothermal, solvent thermal and microwave-assisted [51] ) to obtain target crystals with high energy facets. The crystallographic engineering of MOFs is of great importance in photocatalysts, as evidenced by the successful practice of a wide range of non-MOFs-type crystalline photocatalysts. [306] In photocatalysis, studies on the facets engineering of MOFs remain scarce because the soft crystals of MOFs are conceptually distinct from the crystalline surface-dependent catalytic activity of hard materials such as metals and metal oxides. However, in MIL-125, crystalline materials with different exposed facets can be obtained by introducing competing secondary ligands. Secondary ligands will compete with Ti-oxo clusters for coordination, reducing the nucleation growth rate and controlling the exposure of high-energy facets (Figure 17b-d). [307] In particular, mixed ligand strategies with different electronegativities, such as NH 2 -BDC (NH 2 -BDC 2− = 3.2) and H 2 BDC (BDC 2− = 3.7), can be observed with changing ligand molar ratios in the evolution of crystal morphology between MIL-125 and MIL-125-NH 2 process. [228] It is since during the crystal nucleation stage, the lower electronegativity NH 2 -BDC first coordinates with Ti IV ions leading to an increase in the electron density of Ti-nuclei. Then, hydrogen bonding between -NH 2 and -COOH induces the continued coordination of H 2 BDC with unsaturated Ti IV ions to form the SBU of mixed joints. Thus, MIL-125-1 with two {001}, four {110} and eight {111} facets exhibit the best photocatalytic efficiency (Figure 17e-h). Modulator strategies developed from the controlled synthesis of other crystals also provide important lessons for MOFs crystal engineering. [308,309] In the report of Xu [314] Copyright 2020, Wiley-VCH GmbH. w) Schematic illustration of the fabrication of Ni-BDC NSs; x) CO 2 photoreduction performance under various reaction conditions; y) the proposed reaction mechanism for CO 2 photoreduction. (w-y) Adapted with permission. [315] Copyright 2021, Wiley-VCH GmbH.
their properties for photodegradation of methyl orange (MO) (Figure 17i,j). [310] In addition, Guo's group synthesized T001-1, T001-2, T110, T100, and T111 containing different facets by controlling the amount of capping agent CTAB addition. [311] Theoretical calculations show that the surface energy of {110} is higher than that of {001}, {100}, and {111} at 1.18 Jm −2 , which is probably due to the exposure of more metal clusters as active catalytic sites in this facet. In the photocatalytic test, the T110 sample exhibited the highest hydrogen evolution reaction (HER) activity (60.8 µmol g −1 h −1 ) and apparent quantum efficiency (AQY) (3.60%), which was about three times higher than the T111 sample (Figure 17k-n). In addition, by controlling the amount of acetic acid, a series of NM001, NMA, NMB, NMC, and NM111 with different facets ({001} and {111}) (A, B, and C denote different ratios) were synthesized by Sun's team (Figure 17o-q). [312] The gas adsorption tests showed that the SSA (MN111 = 1312.2 cm −3 g −1 ) and CO 2 adsorption (MN111 = 151.33 cm −3 g −1 ) of the samples increased with the increase of {111} exposure ratio and showed the best CO 2 photoreduction performance. Except for competitive ligands and modulators, precise control of MOFs' synthesis conditions can enable crystal engineering strategies. In the report by Song et al., they investigated two different crystalline forms of Ni-MOF obtained under different solvents (DMF/H 2 O = 1:1 or 9:1) and temperature (90 and 140 °C) conditions and tested their photoreduction of CO 2 (Figure 17r,s). [313] The catalytic experimental and characterization results explain that Ni-MOF(H 2 O) exhibits stronger CO 2 adsorption, lower photogenerated carrier transfer resistance, and more efficient electron-hole separation than the Ni-MOF system. In another study, two different facets of Ni-MOF were obtained by modulating the concentration of NaOH and the ratio of the starting material, respectively, Ni-MOL-100 rich in {100} and Ni-MOL-010 in {010}. [314] Theoretical studies have shown that the distance between two coordination-unsaturated Ni II on the Ni-MOL-100 surface is 3.50 Å (compared to 6.33 Å in Ni-MOL-010), which is energetically favorable for synergistic catalysis between ions in such proximity (Figure 17t-v). Interestingly, 2D Ni II -MOFs (Ni-BDC NSs) with high-energy facets {200} and ultrathin thickness (1.2 nm) were obtained with the help of a supercritical (SC) CO 2assisted technique. [315] The intervention of supercritical CO 2 (SC CO 2 ) competes with the liganded H 2 O molecules for exchange and directed chemisorption on the Ni-sites, leading to delamination by the breakage of interlayer hydrogen bonds. This strategy is based on the strong interaction between the active metal center and the reacting molecules to obtain 2D MOFs with high photoactivity (Figure 17w-y).
In conclusion, crystal facet engineering is an effective modification tool applicable to crystalline MOF photocatalysts. Although the strategy is still at an early stage of development, the highly efficient photoactivity exhibited by high-energy facets is desirable. Furthermore, even different facets exhibit different photoactivity reflecting the catalytic versatility of MOF photocatalysts, bringing an infinite future of reverie.

Sizes Engineering
According to literature research, in many reports on photocatalysis by crystalline MOFs, scholars prefer to explain the positive effect of crystal facets on catalytic performance enhancement. [309,316,317] In contrast, the SSA, hole effect, and small size effect due to crystal size are briefly elaborated. Furthermore, nanoscale crystals are distinguished from their bulk counterparts by the extreme dependence of their functional properties on size. [318] For example, the catalytic activity of metal NPs reflects the underlying electronic structure sensitive to subnanometer size variations. [319] Therefore, many researches have focused on exploring the synthesis of nanoscale MOFs (Nano-MOFs), which are expected to be useful in catalysis, [320] drug transport, [321] gas storage/separation, [322] analytical sensing, [323] and other performance enhancements. [324] Various strategies for preparing MOFs below 100 nm and their effects on the properties of nano-MOFs were reviewed by Usman et al. [325] By adjusting the synthesis parameters (e.g., temperature, time, heating rate, and grinding), [326,327] surface modification, [328] ligand modulation, [329] and solemnizations, [330] one is allowed to tailor the MOFs' size for specific applications.
Nano-MOFs offer significantly enhanced physical and chemical properties compared to its bulk counterparts. Due to their controlled diffusion kinetics and effective limitation of redox centers, Nano-MOFs are used for emerging applications requiring specific sizes and shapes. [331,332] For example, for the synthesis of ZIF-8 with sizes less than 100 nm and high SSA, the intervention of surfactants (Span 80 and Tween 80) in order to avoid the use of organic solvents and excessive bridging ligands [329,333] would yield ≈70 nm and ≈20 µm ZIF-8, the former exhibiting faster adsorption kinetics for I 3− and RhB. [334] In addition, the dimensional design of Nano-MOFs is critical for pore size and environment, active site exposure, and encapsulation of the other material (Figure 18a-c). [286] It is well known that the bulk performance of MOFs in natural systems will be profoundly affected by the differences between the stacking density and crystal density of MOFs. For example, the upper limit of ordered stacking efficiency for uniform spheres is 74%, while the upper limit of random stacking efficiency is 63% and the stacking efficiency varies for different crystal shapes. [335] Therefore, designing crystal shape/size characteristics and filling MOFs strategies with low porosity still need to be extensively explored. For example, by adjusting the metal/ligand molar ratio, reactant concentration, and additives, four different sizes (ranging from hundreds of microns to millimeters) of cubic MOF-5 were obtained, which can significantly improve the bulk hydrogen storage performance. [288] In the photocatalysis of MOFs, they regulated the concentration of the starting material to synthesize two different spindle sizes MIL-88A-1/2 by Fu's group. [336] The concentration of the starting material for MIL-88A-2 was twice that of MIL-88A-1, and the length and width of the crystals obtained were 500 and 300 nm, respectively (the sizes of MIL-88A-1 were 1000 and 500 nm, respectively). In addition, it was found that water molecules accelerate the deprotonation of fumaric acid and the hydrolysis of iron salts, promoting crystal nucleation, which leads to the immediate precipitation of brick-red crystals in aqueous solutions at room temperature. In contrast, ethanol solutions make no crystals visible to the naked eye. The result was that MIL-88A-2 performed better in the photodegradation experiment thanks to its smaller size and larger active sites (Figure 18c). In another report by Xue  with an average size of 50-100 nm by using polyacrylate (PA) as a template and growing MIL-88A in situ on its endospore surface. [337] In contrast, the separately synthesized MIL-88A exhibited a well-crystallized spindle morphology with an average size of 500-800 nm. Compared with the separately synthesized MIL-88A, the SSA of MIL-88A@PA with nanoscale surface roughness and internal void limitation increased linearly, gaining exposure to more catalytic active centers and thus increasing the photocatalytic degradation rate of RhB (≈ 9.4 times).
Since most MOFs stay between a few hundred nanometers and microns in size, it is extremely difficult to break the limit size up or down in the preparation process, which may be why crystal size engineering is not given much attention in the field of MOFs. In particular, the stability of MOFs is closely related to their size, and larger or smaller sizes may cause framework collapse and affect their performance. Therefore, although crystal size engineering can only modulate the performance of MOFs on a small scale, the value of size engineering may be better exploited with the help of template synthesis.

Amorphization
According to the International Union of Pure and Applied Chemistry definition of MOFs, most studies have focused on the long-range order scope, although there is no rigidity  [336] Copyright 2020, Elsevier Ltd. requirement for crystallinity. In contrast, amorphous MOFs (aMOFs) lack any long-range periodic order but retain their crystalline counterparts' basic building blocks and connectivity. As far as we know, defects of significant magnitude can also cause a collapse of the MOF structure, resulting in aMOFs. However, the aMOFs discussed here are not the same as defect-induced skeleton collapse or deformation. It should be noted that as referred to aMOFs here, refer to compression, expansion, distortion, or crystallographic transformation based on the original skeleton under external conditions, such as pressure, magnetic field, ultrasound, and irradiation, undergoing a transformation process from order to disorder. [338] The study of aMOFs is still in its infancy as there is no definitive knowledge of such transformation kinetics and specific transformation outcomes. [339] The preparation of aMOFs based on physical modulation (temperature, [340] pressure, [341] ball milling, [342] and electrical discharge [283] ) and chemical modulation (heterogeneous metals, [343] ligand competition, [344] solvents, [345] and modulators [346] ) is complicated. Moreover, in addition to the preparation difficulties, conventional crystallographic structural analysis methods are insufficient to study aMOFs. Amorphous MOFs are more difficult to characterize than their crystalline counterparts due to the lack of reliable structural data, making it difficult to establish a rational design of amorphous materials for specific applications. However, some advanced electron microscopes, such as the Synchrotron X-ray radiation at Diamond Light Source provide the high-resolution structural data needed to reveal the local and long-range structure of materials. [347] Furthermore, in XRD tests, amorphous MOFs do not have conventional Bragg diffraction peaks, which would prevent us from determining the exact location of individual atoms and hence the long-range ordered structure of the material. However, the disorder caused by this amorphization produces diffuse scattering, which contains critical information about the two-atom interactions that can be used to understand short-or medium-range structural pieces of information. [50] In addition, some specific good characterization tools enable the visualization of the local atomic environment (short-range order) by local probes, including XAS, pair distribution function (PDF), solid-state nuclear magnetic resonance spectroscopy (ssNMR), and high-resolution scanning transmission electron microscopy (SEM). [348][349][350] For example, the lack of long-range ordering in amorphous structures results in the absence of Bragg peaks in their X-ray or neutron diffraction clusters, which are dominated by broad "humps" caused by diffuse scattering (Figure 19). [50] Therefore, the difficulty in characterizing aMOFs explains why the field still lacks reliable structural data to support the development of the required structure-property relationships of aMOFs. [347] Retaining the common properties found in crystalline MOFs, aMOFs include excellent thermal and chemical stability, high SSA, high porosity, and tunable pore size, shape, and chemical environment, as well as unique properties as amorphous materials, such as isotropy, absence of grain boundaries, defects, and active sites, and flexibility. [50] Therefore, aMOFs have been considered the key to next-generation technologies such as plastics technology, communication, catalysis, and display technology, such as amorphous, liquids, and glasses MOFs. [350] In catalysis, amorphous MOFs in electrocatalytic oxygen evolution reaction (OER) [122,[343][344][345] and organic conversion [342,353] successfully practiced in MOFs photocatalysis seems to point a direction for the field of MOFs. A supercritical CO 2 drying technique to obtain amorphous metal-organic aerogels (MOA) as a substitute for Ti-MIL-125 and Ti-MIL-125-NH 2 with extraordinary performance in photoreduction CO 2 →CH 3 OH conversion was reported by Beobide et al. [351] They noted that MOA treated with supercritical drying technology could maintain the shape of the parent gel, while conventional solvent evaporation leads to the collapse of the gel structure. Both the low-angle broad diffraction peaks in the XRD results and the low-temperature TEM images indicated the successful preparation of amorphous MOA. BET tests showed that the abundant presence of tiny and large pores in the product promotes the rapid flow of solvent molecules, intermediates, and product molecules (Figure 19g-i). This feature indicates far superior performance in promoting photoreduction of CO 2 →CH 3 OH than MOF. In the report by Ma et al., the electrostatic interaction between the cationic surfactant CTAB and anionic polymer poly(acrylic acid) (PAA) was borrowed to co-assemble to form a composite structural guide to obtaining titanium phosphate-based MOF (HM-TiPPh) with multistage mesopores. [352] SEM and TEM provide the multilevel mesoporous behavior inside the nanospheres and spheres, respectively. The low-angle broad diffraction peaks presented in the small-angle XRD pattern demonstrate the amorphous state of HM-TiPPh (Figure 19j-l). The high surface areas, ordered mesostructure, and extensive secondary nanopores provide a smooth pathway for material transport during the reaction, resulting in excellent photolytic aquatic hydrogen performance. Similarly, Mondal et al. reported synthesizing a Cu/TiO 2 coreshell photocatalyst loaded on an amorphous TiO 2 substrate based on a Cu-MOF self-templating approach for facilitating photolytic HER. [354] Although the direct application of aMOFs in photocatalysis has not been developed yet, based on the above successful practice, we can boldly envision that aMOFs will make a big impact in photocatalysis shortly.
In conclusion, this section analyzes the current research status and the dilemma that crystal facets, dimensions, and amorphization face in the photocatalysis of MOFs. In particular, we believe that aMOFs will realize the opportunity to be fully demonstrated in photocatalysis, which will be an important part of the MOFs' transformation strategy. In addition, we can see from some reports that the crystal engineering strategy is not an independently existing modulation tool; it is perhaps the result of a combination of morphology, faceting, and size modulation. Therefore, the crystal engineering strategy is an important tool for improving the photocatalytic performance of MOFs.

Applications
Scientific research should be practical. In the context of "carbon peaking" and "carbon neutral," photocatalysis (or solar catalysis) has become one of the most promising alternative energies. For photocatalysts, MOFs-type self-tuning research aims to enhance performance and expand practical applications. This section will discuss the current research status of MOFs as a photocatalyst for water redox, CO 2 reduction, nitrogen reduction, organic transformation, and environmental treatment, starting from its modification ( Table 1). Figure 19. a) Unit cell and associated X-ray diffraction pattern of ZIF-1 [Zn(Im) 2 ] and b) comparably sized configuration of an amorphous ZIF and powder pattern. Zn, pink; C, green; N, blue; H atoms omitted; c) the ZnN 4 tetrahedra (blue) and bridging imidazolate linker common to both crystalline and amorphous species; d) the different amorphization routes possible using unsubstituted ZIFs (e.g., ZIF-4) and those incorporating substituted linkers (e.g., ZIF-8); e) nanoindentation studies on ZIFs, data taken from prior literature and for indentation depths over 150 nm; f) optical microscopy performed on a sample of ZIF-4 obtained by heating a single crystal to 300 °C. (a-f) Adapted with permission. [50] Copyright 2014, ACS. g) TEM images on aerogel fragments of B100A0 at (a) 25 kX and (b) 150 kX magnifications and B0A100 at (c) 25 kX and (d) 150 kX magnifications. TEM, transmission electron microscopy; h) PXRD patterns of referential MOFs (MIL-125 and MIL-125-NH 2 ) compared to B100A0 and B0A100 MOAs. MOA, metalorganic aerogel; MOF, metal-organic framework; PXRD, powder X-ray diffraction; i) N 2 physisorption isotherms at 77 K for B100A0 and B0A100 MOAs. (g-i) Adapted with permission. [351] Copyright 2022, Elsevier Ltd. j) SEM and TEM images of HM-TiPPh; k) Low-angle and (inset) wide-angle XRD pattern of HM-TiPPh; l) N 2 adsorption-desorption isotherm and (inset) the corresponding pore size distribution curve. (j-l) Adapted with permission. [352] Copyright 2018, Wiley-VCH.

Redox of Water
Using MOFs as a photocatalyst for photoconversion of hydrogen/oxygen evolution reactions (HER or OER) to yield H 2 or O 2 is considered one of the most promising strategies to solve energy and environmental issues. In the classical semiconductor photocatalytic process, MOFs absorb photons to produce photogenerated electrons (e − ), which are transferred to the CB while leaving holes (h + ) in the VB. As a result, the proton from H 2 O is reduced to H 2 by e − and oxidized to O 2 by h + (Figure 20). [34] Hence, the ideal MOFs photocatalyst should satisfy a CB bottom layer with a more negative redox potential than H + /H 2 (0.0 eV vs NHE) and a VB top layer with a more positive oxidation potential than that of O 2 /H 2 O (1.23 eV vs NHE). However, we know that achieving photocatalytic total water dissolution is difficult. Then, provided that the theoretical band gap of photolytic water (> 1.23 eV) is satisfied if the energy band structure of the semiconductor satisfies more negative (< 0.0 eV vs NHE) or positive (> 1.23 eV vs NHE), the halfreaction of photolytic water oxidation (O 2 ) or reduction (H 2 ) is theoretically possible. Therefore, most of the research has been carried out in this direction.

Reduction of H 2 O
According to the literature findings, the case of MOFs photolytic water HER involves components, defects, and crystal engineering.
Clusters Engineering: A solvent-assisted approach to clearly defining a mixed metal cluster AB 2 OX was proposed by Liu's. [358] The transition metal ion A is linked to the other two metal ions B via µ 3 -O, and X (-OH or H 2 O) is coordinated to A and B to form the terminal species. MTV-MIL-100 (Ti, B) was synthesized using a trinuclear AB 2 OX cluster (A = Ti, B = Co, Ni, Mn, or Fe) and 1,3,5-benzenetricarboxylic acid (BTC). The doping of metal ion B finely tuned the electronic structure of MOF and enhanced catalytic synergy. In the photolytic water test, MTV-MIL-100 (Ti, Co) exhibited the highest reaction kinetics with H 2 production at 6 min of irradiation and a turnover frequency (TOF) of 113.7 µmol g −1 min −1 (better than 8.6 µmol g −1 min −1 for commercial TiO 2 ) (Figure 21a-e). In another typical synthetic procedure of Kong and colleagues, they prepared multi-metallic (M/Zr)-UiO-66-NH 2 (M = Co, Ni, or Fe) in one-pot. [89] The HOMO-LUMO bandwidth was reduced due to the doping of transition metals. It was attributed to the MMCT process between the second metal ion and Zr ion, facilitating the photogenerated charge transfer. Thus, the yields of (Co/Zr)-UiO-66-NH 2 , (Ni/Zr)-UiO-66-NH 2 , and (Fe/Zr)-UiO-66-NH 2 in visible-light-catalyzed HER were 29.94, 11.76, and 1.34 µmol h −1 g −1 , for 35.22, 13.84, and 1.58 times of Zr-UiO-66-NH 2 , respectively (Figure 21f-i).
Ligands Engineering: In the report by Jiang's group, synthesized MIL-125-X (X = H, NH 2 , Br, NO 2 ) with different activities by introducing different functionalized ligands through a pre-modification strategy, exhibiting different dark-state photocatalytic properties. [359] After turning on the light, the different electron-donating/absorbing groups greatly affected the reduction of Ti IV →Ti III in MIL-125-X by the LMCT process to form an intermediate state with long-lived electrons. Moreover, the long-lived electron-releasing ability in the dark is maintained for a long time after the lamp is turned off, similar to natural photosynthesis. Especially for MIL-125-NH 2 , the dark photocatalytic HER can be as long as 1.5 h and reach a plateau period (Figure 22a,b). Among them, the hydrogen production rate was as high as 270.2 µmol g −1 h −1 in the first 30 min, and the average rate was 116.7 µmol g −1 h −1 in 90 min. In contrast, H, Br, and NO 2 , which have electron-absorbing properties, did not show dark photocatalytic activity.
In the report by Chen and colleagues, obtained Ti-MIL-125-NH 2 -Cu II (Cu-NM) with the help of a PSM strategy using NH 2 ligand-anchored Cu II to facilitate the transfer of e − from the ligand to the Cu II center instead of the Ti-oxo clusters. [360] The Cu II was partially reduced and formed a highly active Cu I/II hybrid center, which exhibited a hydrogen release capacity of 490 µmol g −1 h −1 (18 µmol g −1 h −1 for MIL-125-NH 2 , an enhancement of about 27-fold) under visible light (λ > 420 nm) irradiation with an apparent quantum yield of up to 6.6% (Figure 22c,d). A similar post-ligand modification strategy was also reported by Luo et al., [361] where the synthesized post-protonated MIL-125-NH 3 + Cl − x (x = 0, 2.4, 3.6, 4.8 mol L −1 , treatment concentration of HCl) exhibited enhanced photocatalytic properties (Figure 22e,f).
in HKUST-1 to obtain CuO@HKUST-1 for hydrogen production and degradation of MB under visible light conditions. [365] The highest hydrogen production rate of 667 µmol g −1 h −1 for CuO@HKUST-1 was higher than that of 70 and 510 µmol g −1 h −1 for both HKUST-1 and CuO, respectively. The enhanced photocatalytic rate can be attributed to the rapid transfer of photoexcited electrons between the CB of HKUST-1 and CuO (Figure 23e-i).
Defects Engineering: Wang and colleagues constructed the intrinsic linker defective structure of Cu-MOF using a defect engineering strategy. [375] First, defect-free MET-Cu or linker defective MET-Cu-D were prepared in different solvent systems (ethanol and ammonia mixed system or N, N-diethylamide (DEF) system). The defect vacancies were attributed to DEF competing for coordination with the triazole ligand. The presence of ≈ 8% ligand defects in MET-Cu-D was confirmed using inductively coupled plasma mass spectrometry (ICP-MS) and elemental analysis (EA). The variation of the weight loss curves of both materials in thermogravimetric analysis (TGA) also corroborates the ligand defects. In addition, the asymmetric Cu 2p 3/2 and 2p 1/2 peaks in XPS indicated the presence of mixed-valence Cu I/II at the defect sites (Figure 24a-c).
Crystals Engineering: In the work of Zhang's group, investigated the crystal facet for the first time and prepared three UiO-66-NH 2 samples (cubic (C), deca-tetrahedral (T), and octahedral (O)) with different exposed facets by controlling the addition of reactants and hydrofluoric acid (HF). [368] A series of photochemical characterizations and experiments showed that the T with co-exposed {100} and {111} exhibited the best photocatalytic HER performance due to the presence of more active sites and crystallographic junctions. The HER rate of T under full-spectrum irradiation is 64.06 µmol g −1 h −1 , which is 2 and 1.5 times higher than that of C and O. With DFT calculations and ultrafast spectroscopy, the crystal growth mechanism of UiO-66-NH 2 was revealed. At low concentrations (≤ 3 mm), the crystal growth is kinetically controlled, and the larger crystallographic spacing of {100} facilitates their in-plane expansion and eventual retention according to the Bravais-Friedel-Donnay-Harker (BFDH) law. However, when the concentration is too high (≥ 36 mm), the kinetically mediated growth rates of {100} and {111} are equal, while the {111} becomes more stable due to the thermodynamic advantage (Figure 25).

H 2 O Oxidation (WOR)
According to the literature findings, the case of water oxidation reaction (WOR) of MOFs involves clusters, ligands, defects, and crystal engineering.
Clusters Engineering: The comparative activity of UiO-66-M with five different multi-metal nodes (Zr, Zr/Ti, Zr/Ce, Zr/ Ce/Ti, Ce) for photocatalytic monolithic hydrolysis was investigated by Ferrer's group. [378] The mixed metal nodes were introduced by one-pot or post-synthesis cation exchange. As a result, the three-metal cluster UiO-66 (Zr/Ce/Ti) based on four-electron-four-proton process kinetics exhibited WOR (Figure 26a,b). Yamashita's team grafted hydrophobic octadecyl phosphate (OPA) onto Zr-clusters. [357] A mixed metal strategy prepared MOFs before alkylation with Zr 1-x Ti x -MOF (x = 5, 7.5, 10%, total Ti and Zr of 1.0 mmol, x = Ti/(Ti+Zr)). Ti is present as mixed valence (Ti III /Ti IV ) in the inorganic cluster, forming redox centers and promoting LMCT. A series of characterization experiments (e.g., XPS, FT-IR, water contact angle measurements) demonstrated the successful coordination of OPA on Zr-clusters. The hydrophobic OPA/Zr 92.5 Ti 7.5 -MOF showed an H 2 O 2 yield of 9.7 mmol L −1 h −1 in the two-phase  [375] Copyright 2021, Elsevier B.V. d) H 2 evolution after 24 h irradiation as a function of modulator addition; e) normalized H 2 evolution after 24 h irradiation as a function of modulator incorporation. Normalization was performed toward the 2 µm pristine MUV-10 sample as 100%; f) schematic representation of (I) MUV-10 and MUV-10 modified with (II) 5-hydroxy isophthalic acid (blue) and (III) 5-fluoro isophthalic acid (dark red). Green titanium, purple calcium, gray carbon, blue oxygen, and red fluorine. Hydrogen atoms have been omitted for clarity. (d-f) Adapted with permission. [376] Copyright 2022, ACS. system (BA/H 2 O) under visible light (λ > 420 nm), which is about 4.5 times higher than Zr-MOF (Figure 26c-h).
Ligands Engineering: Lionet et al. investigated the effect of nine ligands modified with different electron-donating/ absorbing groups on the photocatalytic WOR performance of MIL-88B. [379] The modified MIN-88B-X (X = the grafted group) exhibited enhanced photoresponsive activity in the order of -2OH > -NH 2 > -4CH 3 > -OH > -2CH 3 > -4F > -Br > -NO 2 = -4H. However, the photocatalytic activity of these MOFs did not conform to the photoresponsive order. Using AgNO 3 as an electron sacrificial agent, MIL-88B-4F showed the best WOR activity under visible light (λ > 420 nm), which was 5 times higher than MIL-88B. In contrast, MIL-88B-X modified with electron-donating groups such as -2OH, -OH, -NH 2 , -2CH 3 , and -4CH 3 exhibited weaker photocatalytic activity than MIL-88B. It is inferred that WOR is a four-electron oxidation process, and ·OH is involved in the side reaction as a key intermediate of the reaction, which will directly affect its photocatalytic activity. The electron-absorbing groups can reduce the electron cloud density of the benzene ring and prevent or mitigate the hydroxylation rate of the ligand, thus promoting the photocatalytic WOR (Figure 27a-d).
Isaka and colleagues introduced alkyl chain modified MIL-125-NH 2 -R (R 1 = acetic anhydride, R 4 = valeric anhydride, R 7 = capric anhydride) utilizing PSMs. [380] The hydrophobic alkyl side chain allows MIL-125-NH 2 -R 4 /R 7 to be dispersed only in the organic phase. The photocatalytic experiments were carried out in a two-phase (BA/H 2 O) system, where MOFs were selectively present in the BA phase, and the highly acidic (pH = 0.3) and ionic (NaCl) aqueous phase promoted H 2 O 2 production. Mechanistic studies showed that photogenerated h + was burst by benzyl alcohol reduction to produce benzaldehyde. Furthermore, Ti III reduced O 2 to ·O 2− , which migrated to the aqueous phase to produce H 2 O 2 by disproportionation reaction. This two-phase separation of MOFs photocatalytic H 2 O 2 production and chemical synthesis reaction not only solves the problem of MOFs intolerance to acids and bases in aqueous systems, but also achieves spontaneous product separation and avoids the trouble of subsequent separation (Figure 27e-m). In addition, Yamashita's team covalently modified PDI (an n-type organic c) hydrogen evolution rate of three samples and d) NH 2 -UiO-66 with specific morphology of cubes, e) tetra-decahedra, and f) octahedra, from left to right; g) UV-vis spectra; h) photocurrent-time curves, and i) EIS Nyquist plots with full-spectrum light irradiation. (a-i) Adapted with permission. [368] Copyright 2021, RSC. semiconductor) with a large π-system on MIL-125-NH 2 using a PSM strategy. [381] The presence of amide bonds allowed the formation of a more powerful electron leaving the domain system between PDI and the linker, which would facilitate the transfer of photogenerated charges and carrier separation. MIL-125 -PDI exhibited 4800 µmol g −1 h −1 H 2 O 2 yield under visible light (λ > 420 nm) irradiation without co-catalyst involvement, which was 4.3 and 3.2 times higher than that of MIL-125-NH 2 and PDA (perylene tetracarboxylic dianhydride, a PDI precursor), respectively (Figure 27n-q).
Defects Engineering: An HKUST-1 with planar ligand defects using vanillin (VAN) was developed by Guo et al. [366] The SEM observations revealed that the defects morphology of HKUST-1 was strongly influenced by the solvent compositions DMF: EtOH: H 2 O (1: 1: 1, v/v/v) and MeOH : H 2 O (1: 1, v/v), where the same solvent system was used to control the n(BTC): n(VAN). According to the electron paramagnetic resonance (EPR) test results, the concentration of Cu ion vacancies in the defective structure follows the order HKUST-1 < NR-HKUST-1 < NS-HKUST-1 (Figure 28a-f). The XPS and auger electron spectroscopy (AES) result also demonstrates the presence of Cu I/II in the defective material. Photocatalysis test showed that both vacancy-generated opening site and Cu I/II active centers were able to enhance WOR activity (where NR-HKUST-1 (3117.6 µmol g −1 h −1 ) > NS-HKUST-1 (2541.4 µmol g −1 h −1 ) > HKUST-1 (1785.8 µmol g −1 h −1 )), but Cu I/II centers are more favorable for the separation of photogenerated charges. Using an acetic acid modulator, Yamashita's team synthesized hydrophobic UiO-66-NH 2 -X with linker defects (X = 0.5, 1.0, 1.5 mol, equivalents of acetic acid) with the help of defect engineering. [382] The activity of UiO-66-NH 2 -X was superior to that of UiO-66-NH 2 in photocatalytic (λ > 350 nm) H 2 O 2 production experiments with either Triethanolamine (TEOA) or benzyl alcohol as the sacrificial agent. Interestingly, the relationship between benzaldehyde and H 2 O 2 yields was not theoretically proportional in the benzyl alcohol system. Thus, H 2 O 2 decomposition in the dark and water adsorption tests demonstrated that the Zr-clusters in the acetic acid-regulated DE UiO-66-NH 2 -X were occupied by hydrophobic acetate, and the pore surface hydrophobicity increased with acetic acid concentration, and H 2 O 2 separation was accelerated (Figure 28g-j). It solved the problem that the product H 2 O 2 was over-reduced by the disproportionation reaction upon contact with the Zr-oxo clusters and improved the yields. Crystals Engineering: In the study of Yamashita's group, [382] a sacrificial agent must be used to consume h + to produce H 2 O 2 , and this "single channel" route is not in line with the green chemistry concept. In contrast, MIL-xx x with single/ multiple facets (MIL-125-NH 2 as the parent) was obtained by Zheng and colleagues using different modulators. [369] Among them, BA modulation resulted in MIL-001 (pancake) with {001}, CTAB modulation gave MIL-111 (octahedron) with {111}. However, cross-sectional octahedra and MIL-111/001 with {001} and {111} co-existing facets and Z-scheme structures were synthesized without modulators. In other words, MIL-111/001 with a Z-scheme structure can produce H 2 O 2 in  [380] Copyright 2019, Wiley-VCH GmbH. n) The related H 2 O 2 production rate under visible-light irradiation (λ > 420 nm) using 5 mg of photocatalysts; ESR spectra recorded for o) DMPO-O 2 ·− and p) DMPO-· OH in the reaction system after visible-light irradiation with MIL-125-PDI and MIL-125-NH 2 as photocatalysts; q) proposed mechanism for photocatalytic H 2 O 2 production from H 2 O and O 2 under visible-light irradiation using MIL-125-PDI. (n-q) Adapted with permission. [381] Copyright 2021, RSC.
both CB and VB. The "dual channel" route is attributable to the generation of ·O 2− and ·OH on {111} and {001}, respectively, and the presence of faceted synergistic effects of the eutectic MIL-111/001 will promote the generation of ·O 2− and ·OH. Therefore, the order of the photogenerated H 2 O 2 activity under visible light (λ > 400 nm) is: MIL-111/001 > MIL-111 > MIL-001 (Figure 29).

CO 2 Reduction
One of the most promising approaches to achieving "carbon neutrality" and "carbon peaking" is photocatalysis. A series of energy crises have resulted from the steady depletion of fossil fuels, which has also intensified the global warming issue through carbon emissions. MOFs are a newly discovered photocatalysts whose physical and chemical characteristics suggest that they will be crucial in green chemical energy. According to the literature findings, the case of CO 2 photoreduction of MOFs involves components, defects, and crystal engineering.

Clusters Engineering
The Ti x Ce 1-x -UiO-66-NH 2 with Ce III/IV and Ti IV mixed metal centers using a one-pot hybrid metal strategy by prepared Roy's group, which exhibited high selectivity and photoreduction (a-f) Adapted with permission. [366] Copyright 2020, RSC. g) H 2 O 2 decomposition rate of UiO-66-NH 2 and UiO-66-NH 2 -X (X = 0.5, 1.0, and 1.5) in the dark; h) benzaldehyde and i) H 2 O 2 concentrations photocatalytically produced with benzyl alcohol utilizing UiO-66-NH 2 and UiO-66-NH 2 -X after 3 h of light irradiation (λ > 350 nm); j) proposed reaction mechanism for photocatalytic H 2 O 2 production using UiO-66-NH 2 -X. (g-j) Adapted with permission. [382] Copyright 2021, ACS. capacity for CO 2 . [383] The relative surface concentration of Ce III in Ti 0.5 Ce 0.5 -UiO-66-NH 2 was 52.32%, much higher than that of Ce-UiO-66 (27.53%), respectively, as shown by XPS analysis. It suggests that the doping of Ti IV promotes the Ce IV →Ce III transition, obtaining more active sites and enhancing the selective adsorption and photoreduction of CO 2 (Figure 30a-g). Yamashita's team used the same approach to get M x /Fe 3-x -MIL-88-NH 2 with mixed metals (M = Mn, Zn, Co, Ni) confined within its framework PdAu NPs. [384] Gas adsorption tests showed that the influence of the doping of hetero-metal atoms on the photocatalytic performance was mainly in the selective adsorption ability of CO 2 , especially the adsorption amount of MnFe 2 -MIL-88-NH 2 was 42.0 cm 3 g −1 , which was significantly higher than that of several other mixed-metal MOFs (Figure 30h-j). In addition, with the help of the PSM strategy, Fu et al. obtained Co/Ti-MIL-125-NH 2 with Co NPs modified Ti-oxo clusters for the photoreduction of CO 2 . [385] In particular, the use of benzyl alcohol as a sacrificial agent to consume the photogenerated holes enables the effective utilization of both halves of the reaction (Figure 30k-m).

Ligands Engineering
In fact, Ti-MIL-125-NH 2 is one of the first MOFs discovered for CO 2 reduction. [125] Ding et al. found that compared with Ti-MIL-125, Ti-MIL-125-NH 2 could exhibit efficient photocatalytic conversion of CO 2 to CH 4 in the absence of MeCN, sacrificial agent, and H 2 O as a reducing agent (Figure 31a-c). [386] From TiO 2 , Ti-MIL-125-NH 2 to the multi-ligand Ti-MIL-125-NH 2 , Li and colleagues have shown the advanced optimization of Ti-MOF as a photocatalyst. [387] MTV Ti-MOF (D-Ti-MOF) was derived by introducing Zn-TCPP as a second ligand into Ti-MIL-125-NH 2 . Wherein, D-Ti-MOF, Zn-TCPP, and NH 2 -BDC obtained by the one-pot method are uniformly coordinated to the Ti-oxo clusters, while Z&T, Zn-TCPP obtained by a simple PSE are attached to the outer superficial layer of Ti-MIL-125-NH 2 . Benefiting from Zn-TCPP doping, D-Ti-MOF achieves a broader optical response range, narrower optical band gap, and excellent CO 2 adsorption compared to Ti-MIL-125-NH 2 . The PL spectra and photocurrent curves showed that D-Ti-MOF exhibited enhanced photoelectron conversion, (a-e) Adapted with permission. [369] Copyright 2022, RSC. electron transfer, and charge separation. It is because both Ti-oxo clusters and Zn-TCPP can act as photoactive centers and enhance the electron transfer separation by LMCT. The results showed that the average CO 2 yields were 3.94, 5.36, 6.43, 6.72, and 59.55 µmol g −1 h −1 for TiO 2 , Ti-MIL-125-NH 2 , Zn-TCPP, Z&T, and D-Ti-MOF, respectively (Figure 31d-h). The same hybrid strategy with porphyrins as the second ligand appears in other photocatalytic reports, [388,389] and porphyrin ligands with different metal centers can also affect the photocatalytic performance of MOFs (Figure 31i).

Inclusions Engineering
The synergistic enhancement of MOFs photoreduction of CO 2 by encapsulating active substances between host and guest has received much attention. [390] In the work of Sun et al., encapsulated [Ru(2,2'-bipyridine) 3 ] 2+ (Rubpy) exchange into In-MOF by a post-synthesis for CO 2 reduction under visible light. [391] Compared with In-MOF, the conversion of Rubpy@In-MOF in a pure CO 2 atmosphere is improved by about 9 times. The homogeneously encapsulated Ruppy suppresses the thermodynamically favorable spontaneous agglomeration tendency, and it synergizes with the LMCT of In-MOF to promote the transfer and separation of photogenerated carriers. Even under a 10% N 2 /CO 2 mixed atmosphere, high catalytic activity is maintained (Figure 32a-f). In another report, the one-pot method was used to obtain POM@Cd-MOF encapsulated with POM (such as Keggin-type SiW 12 , PW 12 , and PMo 12 ). [392] In situ transient photovoltage tests showed that different POMs in the photoactive Cd-MOF body exhibited different electron transfer behaviors and charge separation pathways, thus showing different catalytic activities in the photoreduction of CO 2 . Wherein, SiW 12 @Cd-MOF and PW 12 @Cd-MOF exhibited CO yields of 4.35 and 3.60 mmol mol −1 h −1 , respectively, while PMo 12 @Cd-MOF showed mediocre performance. This report also reveals the precise electron transfer mechanism of molecules in photocatalytic CO 2 RR by DFT calculations, which provides essential guidance for the design of multiphase photocatalysts (Figure 32g-i).

Defects Engineering
DE can fine-tune the optical E abs and E LMCT of MOFs, which is crucial for photocatalytic studies. Sun's group modulates the molar ratio of FA and dicarboxylate ligands (ATA) to create a series of Pre UiO-66-NH 2 -ML/MC-X (X = 50, 100, 150, 200) with different defect types (cluster or ligand defects) and degrees of ligand vacancies. [367] Where Pre denotes unactivated by high temperature, X denotes the molar equivalent of FA relative to the ligand, and ML and MC denote the ligand and cluster defects, respectively. The original Zr-UiO-66-NH 2 cell consists of 12 fcu 4[Zr 6 (µ 3 -O) 4 (µ 3 -OH) 4 L 6 ]. In contrast, the ligand or cluster defects in the fcu network are denoted as ML-4[Zr 6 (µ 3 -O) 4 (µ 3 -OH) 4 L 6-x FA 2x ] or MC-3[Zr 6 O 4 (OH) 4 L 4 FA 4 ], respectively. Here, UiO-66-NH 2 is noted as (12,12,12,12) (12 denotes the number of joints in each node), and the number of removed ligands will be subtracted from the nodes after removing FA from the n-node (n = 3 or 4) cells after high-temperature activation. UiO-66-NH 2 -ML-50/100 and UiO-66-NH 2 -MC-150/200 were expressed as (10(1FA),10,10,11), (9,10,10,11), (8,8,12(4FA)) and (8,8,8), respectively, as determined by 1 H NMR. It means that the degree of defects in each node increases with the FA feeding ratio. The LUMO potentials of UiO-66-NH 2 -ML/ MC-X are higher than the CO 2 →CO redox potential (−0.53 V vs NHE) as seen from the Mott-Schottky plots, implying their ability of CO 2 photoreduction. UiO-66-NH 2 -ML-100 with ligand vacancies exhibited excellent photogenerated charge conversion and separation efficiency in photovoltaic tests and could be an ideal candidate for the photoreduction of CO 2 to CO. It also obtained the best CO and photoelectron conversions in photocatalytic experiments with 21.3 and 42.6 µmol g −1 h −1 , respectively, which is as much as 3 times higher than the parent UiO-66-NH 2 . The DFT calculations show that, on the one hand, the newly established coordination state of the defect node promotes the transfer of excited electrons, which preferentially accumulate at the defect site and greatly reduce the LUMO energy level of the Zr-nodes, contributing to the reduction of E LMCT . On the other hand, MC expands E abs and misses part of the active metal site. In conclusion, the total energy barrier (E LMCT and E abs ) of the ligand vacancy UiO-66-NH 2 -ML-100 is most favorable for CO 2 photoreduction (Figure 33).

Crystals Engineering
Combined ultrasonic-assisted techniques to synthesize 2D ultrathin Ti-MIL-125-NH 2 nanosheets (T110NS) with the largest exposed active surface {110} by Sun and colleagues. [370] In practice, the other regular 3D Ti-MIL-125-NH 2 are disc-shaped (T001, faceted {001}), rhombic dodecahedral (T110, faceted {110}) and octahedral (T111, faceted {111}), prepared by solvent method and surfactant, respectively. The order of the activities of the above four materials in the photoreduction of CO 2 to CH 4 and CO is T110NS > T110 > T001 > T111, which opens up a new idea of 2D Ti-MIL-125-NH 2 crystal facet engineering to enhance photocatalytic performance. According to Mott-Schottky curves and Tauc plots, the LUMO of different forms of Ti-MIL-125-NH 2 is all more harmful than the reduction potentials of CO 2 →CO (−0.53 V vs NHE) and CO 2 →CH 4 (−0.24 V vs NHE). For the ultrathin T110NS, the BET calculation is 2758.6 m 2 g −1 , which is about 3 times that of T111 (871.1 m 2 g −1 ). The extra-large surface area will expose more Ti III/IV sites, and the results of the DFT calculation also demonstrate that 2D T110NS has more exposed Ti III/IV -sites than 3D T110. The PL and time-resolved PL decay spectra indicate that the rapid burst and decay of T110NS can be attributed to the large surface and ultrathin thickness. The transient photocurrent response and EIS tests show that T110NS has the best charge separation and migration efficiency. In addition, in the three facets {001}, {110} and {111}, all test results indicate that Ti-MIL-125-NH 2 with {110} facets have the best photocatalytic performance (Figure 34a-h). However, in the other cases, Ti-MIL-125-NH 2 with {111} facets exhibited the best photoreduction performance of CO 2 . [312] The exact mechanism by which these two results are opposite is not precise. In addition, Sun's group synthesized three different morphologies of ZIF-67 by solvent induction, including rhombic dodecahedron (ZIF-67-1), dragon fruit morphology (ZIF-67-2), and 2D leaf-like (ZIF-67-3). [371] Solvent induction refers to controlling different ratios of multiple solvents to limit the growth of crystals in different directions guided by kinetic and thermodynamic factors during syntheses, such as different volume ratios of methanol and water. The powder X-ray diffraction (XRD) demonstrated that the three different morphologies of ZIF-67 have different space groups and crystal structures. Compared to ZIF-67-1 (1698.877 m 2 g −1 ) and ZIF-67-2 (835.704 m 2 g −1 ), (a-f) Adapted with permission. [391] Copyright 2022, RSC. g) Crystal structure and energy band structure of POM@CdMOF; h) comparison of CO, HCOOH, and CH 4 production rate of POM@CdMOF and Cd-DTAB complex; i) gaseous (left) and liquidous (right) production selectivity of POM@CdMOF and Cd-DTAB complex. (g-i) Adapted with permission. [392] (Figure 34i-n).

Nitrogen Reduction Reaction (NRR)
Biological nitrogen fixation enzymes have evolved a subtle and specific nitrogen fixation capacity over time, which depends on ordered metal clusters and active metal centers for targeted electron transfer and molecular activation under atmospheric pressure and temperature conditions. [393] It alleviates the disadvantages of the conventional industry that relies mainly on the Haber-Bosch process for nitrogen fixation at high temperatures and pressures (≈573-773 K and ≈100-200 atm). [394] The nitrogen-fixing enzyme consists of organic ferritin (including the F-cluster [4Fe-4S]), bimetallic MoFe protein (containing the P-cluster [8Fe-7S] and M-cluster [7Fe-9S-Mo-C-homocitrate]). [395] The F-cluster is responsible for generating electrons that are transferred to the M cluster via the P cluster to react with the N 2 molecule (Figure 35a). This "synergistic" electron transfer pathway between different protein molecules effectively reduces the electron loss in the transfer process. [396] The bimetallic M cluster acts as the reactive center of the N 2 molecule, and the transition metal Fe donates its available occupied d orbital electrons to the π-N-N antibonding system. [355] A strong NN (941 kJ mol −1 ) bond is significantly weakened by this mechanism, allowing biological nitrogen fixation to proceed under ambient conditions. As modular compositions and processability with metal nodes and organic linkers, MOFs are ideal platforms for constructing structural compositions and bimetallic centers similar to those of biological nitrogen fixation enzymes. According to the literature findings, the case of N 2 photoreduction of MOFs involves components and defects engineering.

Clusters Engineering
In the report by Zhao et al., designed and synthesized MIL-53 (Fe II/III ) with the hybrid Fe II/III clusters. [355] In this case, Fe II is similar to the active site of the low-valent metal Fe II in the M-clusters of nitrogen-fixing enzymes and is responsible for binding and activating N 2 . On the other hand, Fe III was engineered for a non-catalytic function to form and maintain framework, similar to the high-valent Metallo-Mo III site in the M cluster of nitrogen-fixing enzymes. However, unlike MIL-53 (Fe III ), the Fe III in MIL-53 (Fe II/III ) is reduced to Fe II by ethylene glycol (EG) in situ during the process of solvent heat. Therefore, the ratio of Fe II/III in MIL-53(Fe II/III )-X (X indicates the volume content of EG in the mixture) can be easily adjusted (from 0.18/1 to 1.21/1) by adjusting the amount of EG added. EG plays the following roles during crystal growth: a) as a reducing agent; b) as a structure inducer adsorbed in certain lattice planes to control the crystal growth direction; and c) as a competitive ligand causing linker defective sites in the structure, exposing active sites that can bind N 2 (Figure 35b-e). The band gaps of MIL-53(Fe II /Fe III )-0/0.05/0.1/0.2 are 2.77, 2.47, 1.91, and 2.00 eV, respectively, calculated from the Kubelka-Munk equation, which indicates that the mixed Fe II/III metal clusters enhance the photoresponse properties of MIL-53. [101] Under visible light irradiation, MIL-53(Fe II/III ) exhibited an NH 3 release rate of 306 µmol h −1 g −1 , and the AQY was calculated to be 0.12% at 420 nm, while no product was detected for MIL-53(Fe).
The construction of bimetallic centers for efficient nitrogen fixation under environmental conditions using UiO-66 as a platform was carried out by An's group. [397] U(Hf x Zr 1-x )-yX was obtained by a one-pot hydrothermal method from bimetallic Zr/Hf-nodes and functionalized organic ligands (x is the molar ratio of the two metal precursors, X = OH, Cl, SH, and y is the number of functional groups). The Zr acts as the active site for nitrogen fixation in the Zr-Hf bimetallic cluster, while Hf acts as an electron buffer tank to optimize electron transfer and utilization. It was demonstrated by the grand canonical Monte Carlo (GCMC) method of simulation and nitrogen adsorption experiments. The band gap calculations and Nyquist plots show that U(Hf 0.5 Zr 0.5 )-2SH exhibits the smallest band gap (2.62 eV) and the lowest charge transfer internal resistance. It is well known that the electron transfer path from ligand to metal in UiO-66 is an energy "climbing" process. [83] In addition, the CB of the Zr-oxo and Hf-oxo clusters were 2.44 and 2.70 eV, respectively, based on the analysis of Mott-Schottky measurement curves. The electron transfer path in U(Hf 0.5 Zr 0.5 )-2SH is linker-Hf-Zr (LMMCT). In the light fixation N 2 experiment, U(Hf 0.5 Zr 0.5 ) did show the best NH 3 release rate under total spectrum irradiation (351.8 µmol h −1 g −1 ), which was much higher than that of U(Zr) (23.2 µmol h −1 g −1 ) and U(Hf) (11 µmol h −1 g −1 ) (Figure 35f-i).

Ligands Engineering
Inspired by the nitrogen fixation reaction using OVs on TiO 2 , [398] Huang et al. used Ti-MIL-125 to investigate the effect of different functional group-modified X-BDCs (X = H, NH 2 , OH, CH 3 ) on the Ti-MIL-125-X catalyzed conversion of N 2 to NH 3 . [399] Not only is there a high density of immobilized Ti IIIsites in Ti-MIL-125-X, but its band gap and catalytic properties can also be modulated with the help of ligand derivatives. [125] According to Mott-Schottky curves, Ti-MIL-125-H/CH 3 /OH/ NH 2 are 3.85, 3.72, 3.37, and 2.88 eV, respectively, where the CB and VB of NH 2 -MIL-125 (Ti) are −0.47 and 2.39 eV (vs NHE), respectively. Thermodynamically, the CB of Ti-MIL-125-NH 2 cannot directly reduce N 2 (N 2 /N 2 · = −4.2 eV vs Ag/AgCl), while the VB is sufficient to oxidize H 2 O to ·OH. For N 2 adsorbed on Ti III , photogenerated e − can be injected into the π antibonding orbitals to achieve a weakening of the NN bond (N 2 /N 2 H = −3.2 eV). Thus, Ti-MIL-125-X (X = H, NH 2 , OH, CH 3 ) exhibited NH 3 production rates of 0.07, 12.25, 5.03, and 1.39 µmol g −1 h −1 under visible light, respectively (Figure 36a-f). It is also the first reported case of visible light-assisted catalytic nitrogen fixation by MOFs at room temperature, atmospheric pressure, and without sacrificial agents.
a Fe-centered porphyrin MOF (PMOF) as an artificial photocatalyst for nitrogen reduction reaction (NRR) under ambient conditions, called Al-PMOF (Fe). [400] The anchoring of Fe III aims to construct a high density of N 2 adsorption active sites in the framework by making d-orbital electrons available to the π*-orbitals of NN (i.e., π-backdonation). [401] Fe III was postsynthetically inserted into Al-PMOF, and successful loading was demonstrated using the synchrotron PXRD and inductively coupled plasma emission spectroscopy (ICP-OES) with 100% loading and 7 wt.% Fe-atom concentration, respectively. In addition, the X-ray absorption near-edge structure (XANES) of the Fe K-edge was analyzed using XAS and XPS. The results verified that Fe III exists in the framework as a high density of single atoms without Fe-Fe agglomeration. The N 2 adsorption of Al-PMOF(Fe) (0.191 mmol g −1 ) at 1 bar was approximately twofold higher than that of Al-PMOF (0.109 mmol g −1 ), indicating the positive effect of the Fe II -sites on N 2 selective adsorption. Various photovoltaic test results (like photocurrent density, EIS, and PL spectra) also support that Al-PMOF (Fe) is the superior NRR photocatalyst. The photocatalytic NRR results also verified that the NH 3 yield of Al-PMOF (Fe) (127 µg h −1 g −1 ) was higher than that of Al-PMOF (84.5 µg h −1 g −1 ) (Figure 36g-l). In particular, Ti-MIL-125-NH 2 readily undergoes skeletal degradation in aqueous solution, while Al-PMOF (Fe) with Al-oxo clusters exhibits good stability for improved reusability. [402]
reported by Luo and colleagues. [403] Based on the PL and EIS spectroscopic analysis, SiW 12 @MIL-101(Cr)-S has lower interfacial resistance and higher carrier separation efficiency compared to MIL-101(Cr) and SiW 12 @MIL-101(Cr)-I. It may be attributed to forming a more extensive hydrogen bonding network during the solvothermal synthesis. In the photo-nitrogen fixation (PNF) reaction, SiW 12 @MIL-101(Cr)-S and SiW 12 @ MIL-101(Cr)-I achieved yields of 75.56 and 33.93 µmol g −1 h −1 with TOF values of 1.95 and 0.597 h −1 , respectively, which were better than MIL-101(Cr) and SiW 12 . This work is the first to achieve the PNF reaction of POM@MOF without adding any sacrificial agents (Figure 37).

Defects Engineering
An ultrathin 2D bimetallic CuCo-MOFs nanosheets OVR-CuCo-MOFs NS with a large number of OVs were prepared by Wang's group, which opened up the coupled redox reaction of air (N 2 and O 2 ) and water to NH 3 under visible light, also known as direct air redox reaction (ARR). [404] The abundance of OVs is derived from the coordination disorder of bimetals and the synergistic effect of the surfactant PVP. For OVR-CuCo-MOFs NS, the synergistic effect between OVs and the bimetals (Cu, Co) leads to strong absorption properties in the visible and near-infrared regions. A series of optoelectronic characterizations (PL, TRPD, photocurrent test, and EIS) demonstrated excellent photoelectron transfer and separation performance on OVR-CuCo-MOFs NS. The NH 3 production performance was evaluated under visible light with the following production rates in different gaseous environments: Ar (2.16 ± 0.61 µmol g −1 h −1 ) < N 2 (51.94 ± 3.61 µmol g −1 h −1 ) < N 2 : O 2 = 4: 1 . It can be seen that the high concentration of O 2 can accelerate the oxidation reaction activation and decomposition of NN, producing more NO for further reduction to NH 3 . Possible reaction pathways of OVR-CuCo-MOFs NS were explored under air (N 2 : O 2 = 1: 1) conditions using Vienna Ab Initio Package (VASP) simulation calculations. Compared to the conventional direct hydrogenation NRR process * N 2 → * N-NH (with an energy barrier spanning from −0.04 to 1.48 eV) (black routes in the figure), the initiation step * N 2 → * N-NO of the novel ARR (red route in the figure) releases 0.04 eV of free energy (from −0.04 to −0.08 eV), suggesting that ARR has the advantage of spontaneous cleavage and activation of NN. Although * NO→ * NHO (from 1.02 to 1.23 eV) exhibits a crossing of the maximum energy barrier (0.21 eV), the next hydrogenation step generally tends to give NH 3 by spontaneous conversion. In other words, * NO→ * NHO is the rate-determining step in the ARR route. The rate-determining step of the conventional NRR route, H 2 N-NH 2 → * NH 2 , requires crossing an energy barrier of up to 3.81 eV (from −2.52 to 1.29 eV). More importantly, the new ARR route (−6.84 eV) also has priority over the conventional NRR route (−4.83 eV) in the final step * NH 2 → * NH 3 . Therefore, the novel ARR route is more advantageous in photocatalytic N 2 →NH 3 reaction than direct hydrogenation (Figure 38a-f).
The π-backdonation is a central process for breaking through the kinetic complexities and energy barriers for efficient catalytic NH 3 synthesis but occurs only for certain transition metals (Fe, [405] Ti, [398] and Mo [406] ) with empty and filled d-orbitals. By simulating π-backdonation and using the nucleophilic attack of polar water molecules to dissociate the coordination bonds between metal species and ligands, [407] Chen's group prepared (a-f) Adapted with permission. [403] Copyright 2022, Elsevier Inc.
rod-shaped MOF-76(Ce) with a large number of coordinationunsaturated Ce species (Ce-CUSs). Ce has a unique electronic structure (unoccupied and occupied d-orbital) that has the proper energy and symmetry to accept and feedback electrons to N 2 . Thus, the π-backdonation of the Ce-MOF simulation has the following advantages: a) the flexible coordination environment of the Ce-nodes and the stable spatial structure of the Ce-MOF; [408] b) the LMCT process is only applicable to Ce-based materials from low vacancy 4f-orbital and negative electron mobility energy (E LMCT ). [83] The nucleophilic attack of polar water molecules under illumination causes a dramatic change in the coordination environment of Ce-nodes. Except for an O atom of the H 2 O molecule, the 4f-orbital of Ce III is occupied by three semi-alignments and four ordinary O atoms of BTC 3− , forming a CeO 8 polyhedron. When the crystal size is reduced to the nanoscale, a large amount of Ce-CUSs is exposed to become N 2 active adsorption sites due to defects. The DFT calculations showed that VB and CB consist mainly of O 2p , C 2p , and Ce 4f orbitals, while the density of states calculations for the crystal model with N 2 adsorption suggested a sufficient overlap of Ce 4f -orbital and NN. These calculations directly demonstrated that the null 4f-orbital of Ce-CUSs can accept electrons from the σ-orbital in NN. Then the highly reducible Ce III could also return electrons to the π*-orbital of NN. The π-backdonation process promotes the activation and conversion of N 2 . MOF-67(Ce) nanorods exhibited an NH 3 release rate of 36.4 µmol g −1 h −1 , which was more than 8 times higher than the derivative CeO 2 of the same precursor (Figure 38g-j).

Organic Transformation
It is well known that MOFs have been widely used as a photocatalyst for the redox reaction of organic compounds (e.g., olefins, alcohols, cycloalkanes, and heterocyclic compounds) by offering the advantages of both homogeneous catalysts (the high-density active sites) and multiphase catalysts (a-f) Adapted with permission. [404] Copyright 2021, Elsevier B.V. Theoretical calculation to demonstrate π electron backdonation: total and projected density of states for MOF-76(Ce) g) without or h) with N 2 adsorption; i) optimal structure for adsorption geometry of N 2 on Ce-CUS; j) photocatalytic nitrogen fixation over MOF-76(Ce) compared to CeO 2 . (g-j) Adapted with permission. [407] Copyright 2019, ACS. © 2023 The Authors. Advanced Energy Materials published by Wiley-VCH GmbH (recyclability). [27,409] According to the literature findings, the case of photocatalytic organic conversion of MOFs involves components and crystal engineering.

Ligands Engineering
A series of functionalized UiO-66(Ce)-X (X = H, CH 3 , Br, NO 2 ) were prepared by one-pot for the oxidative coupling of benzylamine to N-benzylamine under visible light, in the report of Chen and colleagues. [410] The LUMO of UiO-66(Ce)-X (X = H, CH 3 , Br, NO 2 ) were −0.32, −0.35, −0.34, and −0.50 V versus NHE, respectively, as shown in the Tauc plot. Concerning the potential of oxygen (E (O2/·O2−) = −0.28 eV vs NHE), UiO-66(Ce)-X has the thermodynamic potential to reduce O 2 to ·O 2− and further participate in the selective oxidation reaction of benzylamine. Furthermore, photoelectric tests characterized that -CH 3 with electron-donating properties showed positive performance in photogenerated charge transfer and separation processes, showing the best photocatalytic performance. Mechanistic studies have shown that benzylamine is oxidized to benzylamine cation radicals via -Ce⋯N-coordination to unsaturated reactive Ce-sites by photogenerated vacancies. At the same time, the O 2 anchored on OVs captures photogenerated electrons to be reduced to ·O 2− , and the benzylamine cation radical further reacts to produce benzaldehyde. Benzaldehyde undergoes an amine-formaldehyde condensation reaction with free benzylamine to make the coupling product N-benzenemethanamine (Figure 39a-f).
ions, denoted as UiO-66(SM) 2 (M = Pd, Pt, Au), for the selective oxidation of BA to benzaldehyde (BAD) under visible light. [411] In addition, linker functionalization, such as -NH 2 , -CH 3 , -NO 2 , and -OH, [135] is an effective way to expand the light absorption range. Analogous to the -NH 2 effect, the electron-donating -SH not only expands the light absorption edge of UiO-66-(SH) 2 into the visible region but also exhibits excellent coordination anchoring ability for metal ions. The FT-IR spectra and XPS demonstrated the successful introduction of -SH. Based on DFT calculations and Tauc plots, the HOMO of UiO-66 (3.17 V) is more favorable than the oxidation potential of BAD (2.50 V). However, the HOMO of UiO-66-(SH) 2 (2.43 V) is more positive than the oxidation potential of BA (1.98 V) and more damaging than that of BAD (2.5 V). It indicates that UiO-66(SH) 2 is thermodynamically allowed to oxidize BA to BAD selectively. Under an O 2 atmosphere at 298 K, UiO-66 showed weak photocatalytic activity (conversion of 4.1% and selectivity of 26.8%) under UV light. In contrast, UiO-66-(SH) 2 exhibited 8.9% conversion and 99% selectivity when irradiated for 6 h under visible light, indicating that a suitable energy band structure is beneficial for improving photocatalytic selectivity. Compared with UiO-66-(SH) 2 , the conversions of UiO-66-(SPd/Pt/Au) 2 with anchored metals were 33.4%, 30.6% and 29.2%, respectively, with selectivity greater than 99%. In contrast, the simple physical mixture of Pd/UiO-66-(SH) 2 exhibited only 15.6% BA conversion, much lower than that of UiO-66(SPd) 2 . Both photocurrent response and EIS also demonstrated that UiO-66(SPd) 2 has better electron transfer and separation efficiency as a superior photocatalyst (Figure 39g-l).
A multiphoton-responsive photochemical process for CN and CC oxidative coupling reactions was reported by Duan et al. [412] With the mixed ligand strategy, binary ligands ZJU-56-x (x = 0, 0.2, 04, 0.6, denoting the molar proportion of H 4 L 1 -OH) with different molar ratios of H 4 L 2 (bis(3,5-dicarboxyphenyl) pyridine) and H 4 L 1 -OH (bis(3,5-dicarboxyphenyl) methylpyridinium) were prepared in one-pot for photocatalytic oxidation of benzylidene amine CN coupling to form benzylidene-1-phenylmethylamine (reaction I) and CC coupling of N-phenyltetrahydroisoquinoline with nitromethane (reaction II). Notably, the photocatalytic efficiency not only increased with the increase of the H 4 L 1 -OH ligand ratio, but also ZJU-56-x (x = 0, 0.2, 0.4, 0.6) showed excellent photocatalytic efficiency in the near-infrared region (660 nm). It indicates that the twophoton excitation process can also achieve the excited states reached by the single-photon excitation process, hopeful of solving the excessive activation of substrate molecules by highenergy photons (e.g., UV light) and improving product selectivity. In addition, intensity-dependent experiments were used to investigate the specific photocatalytic mechanism of ZJU-56-x to demonstrate a nonlinear correlation between benzylamine conversion and photon power density (Figure 39m-o), verifying that the two-photon response nature and the two-photon process can accumulate sufficient energy for visible-light-driven redox reactions. Similar mixed-ligand strategies were also reported for Iglesias' group ( Figure 39p) [413] and Li's group (Figure 39q). [414]

Inclusions Engineering
A simple hydrothermal method was used to encapsulate Keggin-type H 3 PMo 12 O 40 into stable MIL-100 (Fe) by Liang's team for the photocatalytic selective oxidation of BA and reduction of Cr VI . [415] HPMo@MIL-100(Fe) with 30% POM content exhibited the most extended photogenerated charge carrier lifetime and the minor interfacial charge transfer resistance compared to MIL-100(Fe) according to PL and EIS spectro scopy results. The energy band structure analysis shows that the matching energy band between MIL-100(Fe) and H 3 PMo 12 O 40 under visible light allows the transfer of photogenerated electrons from CB to H 3 PMo 12 O 40 (Figure 40).

Crystals Engineering
In the work by Cheng and coworkers, 2-methylimidazole (2MI) was used to tune the NH 2 -MIL-101(Fe) crystal size, morphology, and structure. [416] The 2MI accelerates the deprotonation of carboxylic acids and promotes the rapid coordination of linkers to metal clusters to form different nuclei. As a result, the original Fe-MOF gradually evolved from an irregular polygon to NH 2 -MIL-101(Fe)-2MI with a needle-like structure. However, the parent Fe-MOF is unsuitable for efficient photocatalytic onepot hydrogenation and N-alkylation reactions of nitrobenzene/ benzonitrile with alcohols. For that, a two-solvent impregnation method was used to introduce Pd NPs and calcine Pd/NH 2 -MIL-101(Fe)-2MI under an N 2 atmosphere to obtain a "quasi-MOFs" material (Pd/NH 2 -MIL-101(Fe)-2MI (300 °C)). The photocurrent response curves and EIS demonstrated the charge transfer and separation advantages. In the photocatalytic reaction, Pd/NH 2 -MIL-101(Fe), Pd/NH 2 -MIL-101(Fe)-2MI, and Pd/ NH 2 -MIL-101(Fe)-2MI (300 °C) showed 82%, 100% and 100% conversion of nitrobenzene and 64%, 84% and 96% selectivity for N-alkylated products, respectively. It shows that the crystal engineering strategy is applicable to enhance the photocatalytic efficiency of MOFs itself and can be extended to the field of MOFs derivatives and composites (Figure 41).

Environmental Restoration
Photocatalytic degradation of organic pollutants and heavy metal ions is a powerful tool for solving environmental problems. Many different semiconductor photocatalytic materials have been discovered in the last few decades with promising applications for environmental remediation problems. However, quantum efficiency and solar energy utilization are the culprits limiting the industrial application of semiconductor photocatalysts. MOFs, a class of photocatalytic candidates with great potential, have been extensively studied for improving environmental problems. According to the literature findings, the environmental management cases of MOFs involve components, defects, and crystal engineering.

Clusters Engineering
A doped Cu II by one-pot on NH 2 -MIL-125 (Ti) for visible light degradation of MO and phenol, called CuMILx by Liu's group. [417] The competitive coordination differences between Cu II (0.72 Å) and Ti IV (0.68 Å) with different ionic radii in the lattice are manifested by the low-angle characteristic peak of NH 2 -MIL-125 (Ti) in the XRD pattern shifting to a lower angle with increasing Cu II doping. With the increase of Cu II content, the photo absorption of CuMILx was also enhanced gradually.
It is known that the reduction potential of Cu I /Cu II (0.16 V vs NHE) is lower than the HOMO of NH 2 -MIL-125 (Ti) (−0.72 V vs NHE), so the Cu II -site can be used as a trap to inhibit the c) The UV-vis diffuse reflectance spectra; d) the plots of (αhν) 2 versus photon energy; e) photocurrent responses and f) EIS Nyquist plots for MIL-101(Fe)-X. (a-f) Adapted with permission. [416] Copyright 2020, Elsevier Inc. complexation of photogenerated electron-hole pairs. Under visible light, NH 2 -MIL-125(Ti) degradation rate for MO was 37.0%, which was much lower than those of CuMIL1.5 at 98.2%. According to the pseudo primary kinetic model, the degradation rate constants of CuMIL1.5 for MO were 10.4 times higher than those of NH 2 -MIL-125(Ti) (Figure 42a-d). The active substances during photocatalytic reactions are essential information for understanding the reaction mechanism. It revealed that ·OH and h + are the main active substances in the photocatalytic degradation process, while · O 2− is irrelevant. Therefore, Cu II preferentially captures LUMO electrons from NH 2 -MIL-125 (Ti) and is reduced to Cu I . The unstable Cu I releases electrons to oxidize to Cu II for the catalytic cycle. Again, using a one-pot strategy, Yang's team combined mixed metal (Ni/Co) and ligand (H 3 btc/ H 2 bdc) strategies to synthesize a series of bimetallic (NC-series) and hybrid metal/ligand (ML-series) MOFs for photodegradation of MB (Figure 42e,f). [418] It is well known that the conventional solvothermal form of PSE is time-consuming and energy-intensive. [419] A new skeleton Ti/Zr-UiO-66-M by promoting the rapid insertion of Ti IV into Zr 6 O 4 -clusters with microwave assistance was obtained by Tu et al. [86] As a result, it can obtain Ti/Zr-UiO-66-M with an exchange rate of more than 50% in 4 h compared to the solvothermal method, which is much less than the former. Furthermore, the simulated bandwidths of Ti/Zr-UiO-66-M and UiO-66 are 3.75 and 4.00 eV, respectively, which indicates that the introduction of Ti IV effectively interferes with the energy band structure of UiO-66. In addition, the probability of electron transfer from BDC to Ti IV is higher than that of Zr IV , which explains why the photocatalytic activity of Ti-MIL-125 is higher than that of Zr-UiO-66. The reducing radicals generated by formic acid under UV irradiation can eliminate the strongly oxidizing holes in the VB of Ti/Zr-UiO-66-M, reduce Se VI to H 2 Se, and produce CO 2 . Nevertheless, Zr-UiO-66 did not produce a significant effect (Figure 42g-i).
In the study of Gao and colleagues, they have obtained exciting results targeting the modification of metal clusters and introduced bimetallic central nodes and Ti-N clusters with N-pyrrole coordination to obtain N/Zn-doped MIL-125 (N-Ti 9 Zn 1 ). [356] The internal defects and OVs caused by Zn II and N-doping reduce the BE of Ti-oxo clusters, [420] increase the electron density of Ti IV , and promote the generation of Ti III and reactive radicals (·OH and ·O 2− ). Compared with MIL-125 (Ti) and MIL-125 (Ti 9 Zn 1 ), the total degradation of gaseous CH 3 [86] Copyright 2017, RSC. j) Schematic pathway of photogenerated electron-hole pairs, modification of band gap and conduction/VB alignment using Zn/N doping of a MOF; k) degradation activity and l) apparent reaction rate constants (k, min −1 ) of vaporous CH 3 CHO over MIL-125(Ti), NH 2 -MIL-125(Ti), MIL-125(Ti 9 Zn 1 ), and MIL-125(N-Ti 9 Zn 1 ) under vis-light irradiation. (j-l) Adapted with permission. [356] Copyright 2020, Elsevier B.V.

Ligands Engineering
A fascinating phenomenon was found by Chen 2 is close to that of RhB * and is not favorable for electron transfer (Figure 43a-f). The enhancement of the electron separation efficiency dominated by the CB potential reminds us that the electronic effect and bandwidth of the modified groups are not the determining factors of photocatalytic efficiency. Adapted with permission. [424] Copyright 2021, Elsevier Ltd.
The H 2 TCPP has been a popular photosensitive organic ligand with a 18π-electron conjugated macrocycle that responds to visible light from 400 to 800 nm. He's team easily synthesized Ti-TiTCPP MOF with high conjugation and visible light response for visible light degradation of RhB using H 2 TCPP. [424] Analysis of photogenerated carrier dynamics during photoreactions based on time-resolved fluorescence decay spectroscopy. A triple exponential function fitted the decay curve of Ti-TiTCPP MOF with an average lifetime of 0.875 ns, which is much longer than that of NH 2 -MIL-125(Ti) (i.e., 0.170 ns), [312] which has a similar crystal morphology. The increase in ring conjugation promotes the localization of photogenerated charges and further improves the transfer and separation efficiency of holeelectron pairs. Although NH 2 -MIL-125(Ti) showed a more vital ability to adsorb RhB selectively, the photodegradation performance of Ti-TiTCPP MOF for RhB was much higher than that of NH 2 -MIL-125(Ti) when both materials reached adsorption equilibrium under dark conditions. Therefore, this enhanced catalytic performance is independent of the adsorption performance (Figure 43g-k).

Inclusions Engineering
In the work by Li et al., encapsulated photosensitive RhB in Zr-MOF to find a new way to break through the poor performance of Zr-MOF as a photochemical probe and catalyst in terms of detection sensitivity, selectivity, and photosensitivity. [425] Due to the successful phototransfer of electrons from the RhB to the main body backbone, it is very favorable to facilitate the electron transport from the LUMO of RhB to the CB of Zr-MOF, which in turn accelerates the electron transfer and separation. A more negative CB potential than the standard redox potential of Cr VI/III and a more positive VB than MO also make RhB@Zr-MOF perform well in the photocatalytic removal of Cr VI and MO. In addition, the coordination reaction between the Cr VI center and the carboxyl O atom on RhB@ Zr-MOF plays a crucial role in effective fluorescence sensing and photochemical reduction ability (Figure 44). This work provided a new way to tune the sensitivity of fluorescence detection and photochemical removal of heavy metal ions and dyes.

Defects Engineering
In the structural design of MOFs, the ligand design is critical to the frame structure, dimensionality, hole environment, and defects. The hybrid ligand strategy based on thermodynamic guidance is essential for transforming 2D MOFs into rigid and stable 3D MOFs. Zhou and colleagues introduced "column" linkers (DCDPS = 4,4′-dicarboxydiphenyl sulfone (PCN- 133) or TCPP (PCN-134)) between the Zr 6 -clusters and the 2D kdg layer formed by BTB, thus expanding the framework from 2D to 3D. [155] Due to the structural differences between DCDPS and TCPP, PCN-133 was not studied with much significance, a side aspect that demonstrates the importance of second ligand selection. For PCN-134, the BTB/TCPP feeding ratio in the synthesis procedure is crucial for synthesizing controlled crystals since TCPP and Zr 6 -clusters can also generate ordered framework crystals (PCN-224) alone. Therefore, to obtain pure phase crystals without impurities, the optimal reaction parameters for the target crystals can be obtained by high throughput methods (Figure 45a-c). With the variation of TCPP content, N 2 adsorption experiments demonstrated the presence of different 2− in water, PCN-134-22% TCPP showed the best performance, representing a delicate balance between stability, porosity, and defects. Such easy synthesis of Zr-MOF with controllable density defects showed us the precise control of the multi-ligand strategy in terms of MOFs structure and properties. A similar ligand design strategy appears in the report by Fun and colleagues (Figure 45d,e). [426] It was the first case of introducing MOF defect engineering into the photodegradation of organic dyes, and DE ZnIr-MOF was the most effective MOF-based photocatalyst for degrading RhB reported at that time.
For the Photo-Fenton reaction, the number of Lewis acid sites is critical. For example, in the study by Wang et al., selected different monodentate modulators (i.e., BA, pyrrole (Py), and pyrrole-2-carboxylic acid (Pca)) to construct linker defective sites in MIL-88 (Fe) that expose Fe-CUSs. [428] The Bac coordinates with Fe-oxo clusters via -COOH, but the O atom is weaker than pyrrole-N. While the Pca, due to the presence of the pyrrole ring, can provide more electrons to -COOH and enhance coordination. Thus, among the DE Bac/Py/Pca-MIL-88(Fe),  [155] Copyright 2016, ACS. d) The 2D structure of ZnIr-MOF (top), and the molecular structures of Ir-AH 3 and Ir-BH 3 (bottom); e) time-dependent concentration changes of solutions of RhB with and without ZnIr-MOF or ZnIr-MOF-d0.3 catalyst under the same conditions. (d, e) Adapted with permission. [426] Copyright 2017, Wiley-VCH GmbH. f) Single crystal X-ray structure of TMU-4 and TMU-6 (Left) and summary of the SALE reactions performed on TMU-4 and TMU-6. Adapted with permission. [427] Copyright 2017, RSC. g) Schematic illustrations of ligand vacancies in defected Bac/Py/Pca-MIL88(Fe); h) kinetic plots of ln(C o /C ta ) versus adsorption time for adsorption of ACTM under dark; i) kinetic plots of ln(C o /C tr ) versus irradiation time for the degradation of ACTM under visible light. (g-i) Adapted with permission. [428] Copyright 2020, Elsevier B.V.
Pca-MIL-88(Fe) produced more ligand vacancies and exposed more Fe II/III -sites. Based on the adsorption kinetic curves and photodegradation tests, the adsorption rate and Photo-Fenton efficiency of Pac-MIL-88(Fe) on ACTM were 7.3 and 5.5 times higher than MIL88(Fe), respectively. In addition, Pac-MIL88(Fe) achieved a TOC removal rate of 97.0% in only 90 min under visible light, which is 1.7 times higher than MIL-88(Fe) (Figure 45g-i).

Crystals Engineering
β-cyclodextrin (β-CD) was introduced as a morphological regulator to regulate the synthesis of a series of polyhedral Fe-MOF, in Liu's report. [429] It gradually extends from the original hexagonal bipyramidal crystal to a bipyramidal hexagonal prism with the increase of β-CD (Figure 46a-e). The β-CD with poly-OH groups is prone to form a stable system with metal ions through ion-dipole moment interactions, thus limiting the opening direction of Fe-oxo clusters. [430] In addition, the interaction mechanism is controlled by temperature, and there is a significant change in the conductivity of the DMF solution mixed with β-CD and FeCl 3 after heating and cooling with increasing β-CD concentration, and there is a significant critical concentration. It indicates that the concentration of β-CD and the stimulation of temperature affect the migration rate of FeCl 3 in solution and thus control the growth of crystals. In order to verify the effect of crystal shape change on the catalytic performance, two different types of dye molecules (positive Crystal Violet (CV) and negative MO) were tested separately for photodegradation. The results show that Fe-MOF-1 (TPA/β-CD = 1:4) has stable catalytic properties for both dyes but is more effective for anionic MO molecules. Similar strategies for modulators to control crystal growth are reported in Jiang [431] and Yang (Figure 46f ), [432] respectively, where HCl or BA not only inhibit the protonation process of carboxylic acid ligands but also act as capping agents to limit crystal growth.
Bedia et al. prepared Zr(A/B)-UiO-66-NH 2 (A = ZrOCl 2 ·8H 2 O, B = Zr(OC 4 H 9 ) 4 ) in a rare combination of microwave-assisted synthesis techniques from metal precursors and temperature. [433] The crystal sizes of the MOFs prepared with ZrOCl 2 as precursors were more prominent than Zr(OC 4 H 9 ) 4 , 3.7 ± 0.4 and 10 ± 2 nm, respectively. In addition, UiO-66-NH 2 NPs synthesized by the conventional solvothermal method generally showed an octahedral shape, while the intervention of the microwave-assisted technique resulted in smoother crystal edges, which may be attributed to the acceleration of the nucleation phase by microwaves. In the photoactivity test, UiO-66-NH 2 synthesized with ZrOCl 2 as a precursor exhibited the best activity in degrading antibiotic SMX at 140 °C (Figure 46g,h). Moreover, the charge separation efficiency and photoactivity of A-UiO-66-NH 2 were generally higher than those of B-UiO-66-NH 2 at all temperature conditions, indicating that

Summary and Outlook
MOFs are ideal candidates for various separation, gas storage, catalysis, and energy applications, with more than 20 000 cases have been reported and studied in the last decades. However, in addition to the general dilemma of poor stability and limited production scale, technological breakthroughs in applying MOFs from a photocatalytic perspective depend highly on the light response properties and redox activity. In other words, MOFs need to solve more of their problems if they want to make a big splash in photocatalysis. In this review, we explore the root of the problem from the perspective of the composition, defects, and morphology of MOFs. This is because exploring MOFs per se will be more conducive to further expansion into composites and derivatives, as MOFs are the root. Also, this is the first review that summarizes the self-tuning strategy of MOFs from a photocatalytic perspective and extends it to photocatalytic performance.
Although the previous cases have all shown that the modification of MOFs in terms of composition, defects, and crystals is feasible, these explorations are also full of unknowns. At present, MOFs photocatalysis is still a new field with many opportunities and challenges, and more fundamental studies are needed to complement the current status. Here are some of our insights.

Clusters Engineering
As one of the most promising materials for multiphase catalysis, tuning the ingredient complexity of MOFs is a crucial way to promote catalytic reactions. Multi-metallic MOFs are more complex systems than monometallic MOFs, so they have more advantages in terms of performance and applications. The insertion of second metal ions can modulate the stability, pore structure, defects, and electronic structure of bimetallic MOFs, resulting in MOF modification. In particular, the introduction of the SBU approach facilitates the extension of precision chemistry from molecular chemistry to 2D or 3D frame structures, which gratifyingly allows moving from the construction of functional units to the rational design of target frame structures. However, such multi-metallic MOFs are currently facing two main problems: First, how to select a suitable second metal ion and how to obtain a stable framework structure; and then, how to characterize the arrangement position of the metal ions, that is, the distribution of metal centers in homo/hetero SBU. Bimetallic MOFs are still in their infancy, and much research is needed to continue exploring this complex mechanism. In contrast, bimetallic MOFs with well-defined crystal structures are suitable for atomic-level structural characterization and modeling calculations, which will give us a fundamental understanding of structure-property relationships. In catalytic reactions, clusters often play a role in determining the active site, which makes cluster engineering particularly compelling. The ideal bimetallic MOFs must have a transition state with two metal sites adjacent to each other to form an intermediate. With the development of single-atom catalysts (SACs) and highentropy alloy NP catalysis, such mixed-metal materials have attracted much attention in practical research and applications.

Ligands Engineering
The linker strategy is the earliest studied and most attractive strategy for the modification of MOFs. The tunability of ligands differs from that of metal clusters, and as one of the pre-prepared components, the means of ligand modification seems to be more mature and stable. The ligand modification strategy around photoresponse expansion, carrier separation, and catalytic performance extension has been a favorite and facile tool for scholars. Especially for MOFs design, linkers of different shapes, sizes, electronic structures, and coordination features allow us to obtain versatile MOFs with tunable topologies, pores, functions, and surface environments from a single material. For functionalized ligand strategy, the introduction of functional groups not only leads to the tuning of optical regions and electronic structures, but these reactive groups help further extend to MOFs composites, such as photosensitized organic dyes and grafting of chiral macromolecules. However, it appears that the influential functionalized groups are limited to some electron-giving groups (such as -NH2, -SH, and alkyl chains), and there is a paucity of studies on the degree of substitution of groups on ligands on the photocatalytic performance of MOFs. Diversified ligand strategies are also one of the main approaches to improving the complexity of MOFs. In some cases, this diversification can produce complementary or contradictory properties. In particular, the competition of the second ligand allows the creation of partial structural defects in the framework while maintaining framework stability. In addition, primarily seen in layered MOFs, the introduction of multiple ligands can expand the topology from 2D to 3D or even further form larger cages on the original MOFs. However, this diversification is achieved at the expense of MOFs' structural stability, and finding a suitable second ligand is challenging. So structural designers must be imaginative.

Inclusions Engineering
It has been proved that the active inclusions will primarily affect the photocatalytic performance of MOFs. However, as tightly connected long-range ordered structures, it is imperative how the inclusions can obtain good dispersion and encapsulation rates in MOFs. One-pot and PSE methods have their advantages and disadvantages, respectively. Even the MOFs are used as templates to encapsulate the active inclusions to obtain composites instead of focusing only on MOFs as the primary active substance, which is the direction of the development of Inc.@MOFs.

Defects Engineering
According to Pearson's soft/hard acid/base (HASB) theory, a stable MOF is composed of a hard Lewis base (carboxylate ligand) and a hard Lewis acid (high-valent metal salt). Influenced by thermodynamic and kinetic factors, defects are ubiquitous, and their stability is even more governed by external stimuli such as chemical, optical, thermal, electric fields, and mechanical. Despite the SSA and pore opening that defects engineering brings to MOFs, cluster defects often bring a host of adverse effects as Lewis and Brønsted acid catalytic sites. In contrast, ligand defects are often accompanied by the opening of active sites and the introduction of functional ligands, which is positive for the performance enhancement and expansion of MOFs. The OVs can be regarded as an extension of the ligand defects, but unlike the ligand defect sites where various groups or molecules occupy the metal center, which is exposed to the OVs. Defects engineering obtains massive changes in hole environment/size and electronic structure at the cost of destroying internal structure and stability. In addition, characterizing and quantifying defect sites and their electronic and spatial properties at the molecular level is difficult, which makes the relationship between defect levels and properties elusive. Although several characterization methods for defects have been described, such as TGA, potentiometric acid-base titration, XRD, NMR, nitrogen adsorption, probe FTIR and HRTEM, reliable techniques for defects identification, quantification, and distribution are still lacking. Therefore, with further knowledge of the defect structure of MOF, defects engineering will become an essential tool for designing and tuning the structure and properties of porous materials.

Crystals Engineering
As a class of long-range ordered structures, MOFs involve the reaction of metal precursors and organic ligands in solution in conventional synthesis, and both the nucleation and growth phases are difficult to achieve precise control by pre-design. In particular, the physicochemical properties of MOFs are closely related to their crystalline facets, size, and crystallinity, which need to find a harmonious balance between these aspects. In this review, we detail the active role of crystal engineering of MOFs in photocatalysis from the above three aspects. The explanations for the enhancement of the crystal strategy in terms of catalytic performance include facet effects, dimensional effects, and crystallographic junctions. Although we are actively seeking to understand the intrinsic mechanisms of crystal engineering, there are no widely accepted conclusions at this time, limited by the level of knowledge and technical means. Therefore, it is not surprising that the same facets of the same MOFs obtained in different cases have different reaction properties. In addition, aMOFs, although not yet prominent in the field of photocatalysis, maintain the basic building blocks and connectivity that their crystalline counterparts should have, that is, they retain the primary reaction sites and porosity. Such a non-periodic arrangement of atoms can be obtained by off-site stimuli, like temperature, pressure, mechanical forces, and irradiation. The partial collapse of the framework during amorphization can bring about an increase in ion transport capacity, release of active material from the pores, or a more flexible framework structure. These changes are expected for photocatalytic reactions, especially for the release and separation of products in organic synthesis. We believe that crystal engineering is an essential tool to enhance the catalytic performance of MOFs.
However, there are limited ways to adjust MOFs' composition, defects, and crystal. For example, by tuning the properties of MOFs by modulators, the final result may be a balance point where composition, defects, and crystals are jointly tuned. Thus, microwave, ultrasound, electric and magnetic fields, and even template synthesis methods provide essential support for the targeted design and modification of MOFs. The development of theoretical modeling and characterization techniques will also bring new insights to our understanding of the structure and mechanism of MOFs. In conclusion, the future of MOFs is bright, and the quest for exploration is never-ending.