Principles of Design and Synthesis of Metal Derivatives from MOFs

Materials derived from metal–organic frameworks (MOFs) have demonstrated exceptional structural variety and complexity and can be synthesized using low‐cost scalable methods. Although the inherent instability and low electrical conductivity of MOFs are largely responsible for their low uptake for catalysis and energy storage, a superior alternative is MOF‐derived metal‐based derivatives (MDs) as these can retain the complex nanostructures of MOFs while exhibiting stability and electrical conductivities of several orders of magnitude higher. The present work comprehensively reviews MDs in terms of synthesis and their nanostructural design, including oxides, sulfides, phosphides, nitrides, carbides, transition metals, and other minor species. The focal point of the approach is the identification and rationalization of the design parameters that lead to the generation of optimal compositions, structures, nanostructures, and resultant performance parameters. The aim of this approach is to provide an inclusive platform for the strategies to design and process these materials for specific applications. This work is complemented by detailed figures that both summarize the design and processing approaches that have been reported and indicate potential trajectories for development. The work is also supported by comprehensive and up‐to‐date tabular coverage of the reported studies.


Introduction
Nanostructured materials have long been used as catalysts due to their ultrahigh surface areas and tunable properties. The active site concentrations available to nanomaterials often exceed those of their bulk counterparts by several orders of magnitude and these can be manipulated significantly with only minor modifications to the composition, size, and/ or morphology. [1] This variability leads to a plethora of properties characteristic of nanomaterials, although strict control of these features is required to produce consistent properties required for reliable performance. The potential of nanomaterials, therefore, is limited by the synthesis techniques available. There are many synthesis techniques for metal derivatives (MDs), such as oxides and sulfides, but most are complex, expensive, and can be applied only to specific materials. [2] Thus, further development of synthesis routes is required to expand the potential of MD nanoparticles. To this end, metal-organic frameworks (MOFs), which effectively are coordination polymers (CPs), are under development because they can be used as precursors to a wide range of complex morphologies of unique surface characteristics. A second advantage is that they can be generated reproducibly using relatively simple techniques. Although MOFs as precursors for MD synthesis of oxides and sulfides have been reported widely, there are other MDs that have received less attention; these are summarized in Figure 1. While there are a number of reviews that examine the applications of MOFderived materials including energy storage, [3,4] energy conversion, [5][6][7][8] catalysis, [9] and electrochemical sensors; [10] the following review focuses on the synthesis and design of MOFderived materials, with focus specifically on MDs. 1

.1. MOF Design and Synthesis
MOFs are made up of ordered metal-containing clusters, which are known as secondary building units (SBUs), that are held together by linkers composed of organic ligands. [11] The natures of the linkers depend on the number of available ligand bonds, where generally the linkers may be ditopic, tritopic, or tetratopic. The coordination bonding between the SBUs and linkers enables the formation of complex crystalline 1D, 2D, or 3D structures with BET surface areas reported Materials derived from metal-organic frameworks (MOFs) have demonstrated exceptional structural variety and complexity and can be synthesized using low-cost scalable methods. Although the inherent instability and low electrical conductivity of MOFs are largely responsible for their low uptake for catalysis and energy storage, a superior alternative is MOF-derived metalbased derivatives (MDs) as these can retain the complex nanostructures of MOFs while exhibiting stability and electrical conductivities of several orders of magnitude higher. The present work comprehensively reviews MDs in terms of synthesis and their nanostructural design, including oxides, sulfides, phosphides, nitrides, carbides, transition metals, and other minor species. The focal point of the approach is the identification and rationalization of the design parameters that lead to the generation of optimal compositions, structures, nanostructures, and resultant performance parameters. The aim of this approach is to provide an inclusive platform for the strategies to design and process these materials for specific applications. This work is complemented by detailed figures that both summarize the design and processing approaches that have been reported and indicate potential trajectories for development. The work is also supported by comprehensive and up-to-date tabular coverage of the reported studies.
resents different MOF crystallographic structures that influence their 1D, [15] 2D, [16] and 3D [17] morphologies. Although it is possible to design the processing to synthesize monodisperse particles, it is more common to form morphologies, particularly rods and sheets, that agglomerate. In such cases, surfactants and/or specialized processing conditions are required in order to prevent or reverse agglomeration. Further, with suitable primary processing, the weak coordination bonding of the 3D morphologies can be leveraged to modify the morphology to 1D [18] or 2D. [19] Also, secondary processing, such as partial decomposition, can result in morphological modification, leading to the alteration of 1D rods [20] or, more commonly, conversion of hollow 3D particles [21] into 2D sheets.
The variety of synthesis techniques for MOFs and CPs enables a wide range of MOFs to be produced. [22] Solvothermal synthesis using an autoclave is the simplest and most commonly used technique to produce MOFs. However, the process has the potential to produce mixed-phase assemblages and they are relatively slow, typically requiring >10 h. While electrodeposition is a relatively new technique that can yield both phase-pure samples and is relatively rapid, typically requiring only several minutes, [23] it also has the significant advantage of enabling the synthesis of phases not possible using solvothermal techniques. [20] However, electrodeposition requires careful control of the pH in order to enable formation of the desired phase [24] and the required infrastructure can be expensive. Additionally, the design of MOFs includes a number of other challenges. First, at present, there exists no protocol or system to allow prediction of the composition and structure of MOFs. Second, once MOFs have been synthesized, there exists no protocol or system to easily modify them in a systematic manner. Third, the determination of this organometallic phase's composition and structure are challenging and time-consuming.
Early research focussed on divalent metal MOFs and simple ligands with labile bonds, which facilitate the achievement of highly crystalline structures. [25] However, weak bonding in chemical storage, separation, and catalysis conditions leads to structural degradation from exposure to reagents such as water, acids, and bases. [26] In order to improve the stability, weakweak or hard-hard acid-base pairs should be used to increase the metal-ligand bond strength. [27] Such strategies to improve the stability of MOFs to thermal and chemical degradation must be balanced against the tuneability and crystallinity enabled by the labile bonds. More recently, research has aimed at structural or macro defect engineering in order to increase the activity by, for example, removing a ligand to expose a metal ion or increasing the surface area, respectively, while maintaining structural stability. [28] More broadly, recent research has focussed on the achievement of more complex structures aiming at a wider range of performance parameters. To this end, using different metals and/or ligands can yield different structures through alteration of the connectivity between SBU ligands. MOFs are defined by the strict ABAB coordination bonding pattern between SBUs (A) and ligands (B), where SBUs and ligands are of varied sizes and shapes, altering the number and orientation of the bonds. [29] However, examination of the literature reveals that even greater complexity can be achieved by varying: 1) the number of metal atoms in the SBU, which can range from one to eight, 2) varying the number of ligands, which can range from 3 to 24, and 3) varying the number of SBUs that can bond to a single ligand, which can range from one to six. Four examples of different SBU-ligand pairs and their resulting MOF topologies are shown in Figure 3. Further complexity can be engineered from the blending of different SBUs and/or ligands to produce multivariate MOFs. The SBU/ligand bond connectivity can be leveraged through these three variabilities to control the structure and morphology of MOFs. However, while the design of such complex MOFs offers considerable potential to achieve specific properties and applications, it tends to preclude their use as simple precursors for tunable alteration of their properties and applications.

MOF Conversion
There are three different methods to convert precursor MOFs or CPs into MDs, although some may be material-specific; these are summarized in Table 1:

Pyrolysis
The simplest route is temperature-induced destruction of the structure under a non-reactive gas, which leads to the conversion of less stable MOFs into more stable composites comprised of MDs and carbon-based derivatives. At temperatures above ≈600 °C, all MOFs will decompose. [42] The coordination bonding between SBUs and ligands breaks down and is replaced by the covalent/ionic or metallic bonding of the MDs while the ligands carbonize, releasing any gases deriving from the hydrocarbons (e.g., N 2 , NO x , H 2 O, CO/ CO 2 ). Although mass loss occurs, only the elements in the precursor are present in the product. That is, if there is sufficient nitrogen, oxygen, and/or sulfur in the MOF, the corresponding metal oxides, [43] nitrides, [33] or sulfides [32] may form. The abundance of carbon in MOFs results in the formation of composites of MDs combined with carbonaceous secondary phases: Amorphous, [31] graphene/graphite-based, [44] and/or carbides. [45] The resultant MD-carbon composites generally exhibit bulk properties superior to those of the individual phases. Critically, unlike most decompositional phase transformations, the MOF-to-MD conversion entails preservation of the original MOF morphology even though the original crystal structure is lost. Further, the use of pyrolysis allows the potential retention of the bulk chemical composition, albeit as a composite. Although the pores associated with the crystal structures of MOFs nearly always result in ultrahigh surface areas, [46] the maintenance of the original morphologies upon loss of the original crystal structure can result in the retention of these surface areas. The principal disadvantage of pyrolysis is that it involves a closed system and so the bulk chemical composition of the products can be tuned only by modifying the composition of the precursor MOF. Consequently, the variety of MDs that can be synthesized by pyrolysis is limited.

Solvent-Based Synthesis
MOF precursors can be converted into MDs through ion exchange with reactive ions in solution. [56] The process generally requires temperatures in the range 90-150 °C to produce the reactive ions from decomposition of salts, such as thioacetamide (TAA), [57] although it also can be done at room temperature using reactive salts. [58] However, in the absence of heat to facilitate the process, the alternative is the use of high reagent concentrations and/or suitable pH conditions [40] (MOFs typically are more stable in acidic environments [59] ). This low-temperature route has been used to successfully synthesis oxides and sulfides, and has the significant advantage of greater retention of the MOF precursor morphology. While the more stable 3D MOFs are able to retain their morphologies after conversion at temperatures as high as ≈900 °C, the morphologies of the less stable 0-2D MOFs are more likely to be retained at lower temperatures.

MOF Stability
The stability of MOFs arises from the stabilities of their individual components, which are the SBU and the ligand, and the strength of the coordination bonds between them, which follows Pearson's hard-soft acid-base (HSAB) theory. [26,27] This theory states that ligands with relatively high pK a values (e.g., azoles) will form strong bonds with lower valence metals (viz., 1 or 2) while ligands with relatively low pK a values (e.g., carboxylates) will form strong bonds with higher valence metals (viz., 3 or 4). [59] Well known MOFs (e.g., UiO-66, MIL-101, ZIF-8) follow this theory, forming robust 3D networks that establish chemical and mechanical stability even in the presence of large pores. However, strong MO bonding in high-stability MOFs often leads to the formation of amorphous materials. [60] In such cases, the kinetics of metal and ligand bonding is too rapid to allow sufficient diffusion and associated rearrangement for the achievement of structural equilibrium. The means of overcoming this shortcoming is the addition of modulators that compete with the ligand for bonding with the metal ion, thereby decelerating the synthesis rate and increasing the crystallinity. While most research aims to improve the stability of MOF structures, a converse strategy is the synthesis of unstable MOFs of lower stability in order to facilitate conversion into MDs. This allows the conversion of unstable 3D structures into unstable 2D precursors, facilitated by the low-strength coordination bonds between SBU and ligand. Additionally, the range of post-synthesis modifications of the structure by metal and ligand exchange is expanded owing to the weaker coordination bonds and the resultant capacity to modify the structure further before thermal and solvothermal conversion. It has been shown that modulators can be used to promote phase selection and defect generation in MOFs and so facilitate their conversion into MDs. [61,62]

Metals
A key factor that determines the potential for the synthesis of MOF-derived MDs is the nature of the metal. For example, copper, zinc, and zirconium have been studied extensively owing to the varieties of possible MOF compositions, structures, morphologies, and resultant applications. Less commonly investigated metals, such as titanium, manganese, and indium, have been examined using the same ligands as applied to the more commonly investigated metals. The associated limited number of permutations of metals and ligands consequently constrains the range of MDs that currently can be synthesized. An example of such constraint is Ti-based MOFs, the MDs of which are photocatalytically active and environmentally stable, [63] owing to the incompatibility between the pH stability conditions for the metal versus the ligand. [64] A summary of the different metals used in the synthesis of MOFs that can be converted to diverse types of MDs is shown in Figure 4, wherein examples from refs. [20,32,33,41,49,51,52,54, are compiled (see also Supporting Information for more details). MOFs that have been synthesized with an element have been categorized as "MOFs exist", where the metal may be present in either the SBU or ligand. Those that have been shown to form complexes but have not yet been made into MOFs are labelled as "MOFs likely". Finally, non-metals and highly radioactive metals that have not formed complexes are designated as "MOFs unlikely".

MOF-Derived Metal-Based Derivatives (MDs): General Design, Synthesis, Key Features, Characteristics, and Properties
The majority of research in precursor MOFs involves the straightforward synthesis of carbon nanomaterials, including graphene, graphene oxide (GO), reduced graphene oxide (rGO), and carbon nanotubes (CNTs), and so extensive reviews of these materials already have been published. [166][167][168][169][170][171] Since these reviews are limited to carbon nanomaterials and do not focus on the potential for forming MDs in situ with or without carbon, the present work examines the synthesis of MDs, both with and without carbon as a secondary phase, from MOF precursors. The present work also reviews mixtures of metal derivatives, which consist of chemical and mechanical mixtures of two or more different MDs. [172] In effect, while MDs occur as single-phase systems of single types of MDs and as well as solid solutions, they also occur as binary, ternary, and higherorder systems, where the latter is composed of combinations of lower-order systems. Table 2 provides a shorthand nomenclature for these building blocks.

Carbon in MD Composites
Since all MOFs have organic ligands, therefore they contain carbon, which under high-temperature conversion in a nonoxygen atmosphere yields an MD-carbon composite. The ratio   [20,32,33,41,49,51,52,54, See Supporting Information for further information.
of carbon to MD determines whether the carbon acts as the matrix phase or as a mixture component. Since the dispersed or mixed phase plus carbon in the MD both derive from the MOF precursor itself, their nanostructural distributions tend to be homogeneous. With dispersions, the carbon matrix is continuous, and the MD is present as isolated nanoparticles, i.e., a 0-3, or 1-3 composite. [173] With mixtures, there may be a single dispersed phase but it is also possible for both phases to consist of interpenetrating networks, i.e., a 3-3 composite. In general, the presence of carbon improves three key parameters: 1) surface area, 2) charge-carrier conductivity, and 3) extent of retention of the MOF morphology. This general behaviour is substantiated in the data of the tabulated summaries that follow. Surface Area : High surface areas are advantageous in catalytic applications as they increase the distribution density of adsorption sites. [174] Chen et al. [150] heated  in N 2 at various temperatures to produce C(ZrO x ) with surface areas up to 469 m 2 g −1 . The catalytic performance tracked closely with the surface area, which was assumed to derive largely from the carbon. However, the authors speculated that ZrO x was the active species and carbon both minimized agglomeration and increased electrical conductivity. Wang et al. [175] synthesized Co 3 O 4 with and without carbon by thermal conversion of ZIF-67(Co) in two-and one-step processes, respectively. In the onestep process, direct heating at 350 °C in air produced Co 3 O 4 but also removed carbon, thereby leading to agglomeration of the MD. In the two-step process, carbonization at 550 °C in Ar produced C(Co) free of agglomeration, which was followed by heating in air at the lower temperature of 150 °C, and C(Co 3 O 4 ) was produced. This composite outperformed pure Co 3 O 4 in all energy-storage tests. For both processes, while the pores originally present in the MOF collapsed, relatively high surface areas were retained owing to hindrance of agglomeration by the carbon.
Charge -Carrier Conductivity: The charge-carrier conductivity is improved by the incorporation of carbon, which can provide a conductive network for electrochemical applications. [176] In addition, the performance of the MD likely depends to some extent on the nature of the carbon-MD bonding and its effect on maintaining charge carrier conductivity. That is, chemisorbed nanoparticles exhibit direct electrical conductivity, while physisorbed nanoparticles must exhibit electron tunneling. [177] Therefore, metals in the MD that readily form carbides are more likely to have coherent or semicoherent interfaces with carbon, thereby promoting charge transfer. However, incoherent interfaces also can exhibit significant charge transfer owing to their direct apposition resulting from the in situ synthesis of MDs and carbon. TiO 2 was synthesized with and without carbon from MIL-125 by conversion at 600 °C in Ar and air, respectively. It was demonstrated that the presence of carbon improved the reversible capacity by ≈280%. [176] Retention of MOF Morphology : The continuity of the carbon matrix architecture allows the monodisperse MDs to retain their original morphologies more readily by reducing the tendency to agglomerate. [128] As shown in Figure 5, Shao et al. [47] synthesized Co 3 O 4 in two different morphologies through direct and indirect oxidation of ZIF-67(Co) precursor. Direct conversion at 350 °C in air produced a discrete dodecahedral MD shell around a rounded MD agglomerate. This dense morphology was avoided by indirect oxidation, where the first step was pyrolysis, in which the carbonization stabilized a concaved dodecahedral morphology. In the second step, the MD was heated in air at 350 °C, thereby completely oxidizing the Co and the carbon while retaining a porous concave dodecahedral morphology.

Pre-Conversion MOF Design
The structural chemistry of the ligand in the precursor MOF can alter the final properties of the MD. Even in otherwise identical MOFs, a simple change in ligand functionality can alter both the stability and chemical composition of the MOF. Silva   [54] demonstrated this for three electrodeposited Cu-based MOFs using 1,3-H 2 BDC, 1,4-H 2 BDC, and 5-NH 2 -1,4-H 2 BDC ligands (MOF-I, MOF-II, and MOF-III, respectively). Although all the ligands were similar, conversion in air at 650 °C for 3 h yielded differential results, where MOF-I produced more agglomerated CuO particles relative to MOF-II and MOF-III, leading to lower catalytic activity owing to the lower active area in MOF-I. It is probable that the bent shape of the ligand in the MOF reduced the distance between SBUs and hence increased the ease of agglomeration. In contrast, the catalytic activity and agglomeration of the MDs from MOF-II and MOF-III, which have straight shapes, were indistinguishable. This was owing to the amine group present in MOF-III being a functional group rather than forming a part of the ligand backbone. These results suggest that a higher order of ligand symmetry packs the metal ions more closely and hence increases the tendency to agglomerate during direct conversion.
Since the conversion of MOFs into MDs generally terminates the potential for further manipulation of the morphology, such changes must be implemented prior to the conversion of MOF into the MD. Yu et al. [178] followed this strategy and modified the morphology of mixed Ni/Co-MOF-74 by alteration of the ratio of water/ethanol used as the solvent. With increasing ratio from 0.5:0.5 to 1:0, the particle morphology shifted from sharp solid particles to hollow nanowires. These were calcined at 350 °C to produce mixed NiO/Co 3 O 4 with morphology identical to that of the MOF precursor. Further, the ratio of different metals in the precursor MOF is known to influence the particle size of the MOF and the resultant MDs. This was demonstrated for Co-doped ZIF-8(Cu). [179] It also is possible to obtain MDs of unique morphologies by direct conversion of MOFs and CPs that have been electrodeposited, although this more tedious procedure has not been investigated frequently. Despite the greater demands on process control, there are two primary advantages of this approach. [180][181][182] First, for certain MOFs and CPs, electrodeposition is the only possible method of synthesis owing to the capacity to overcome energy barriers in specific pH conditions. Second, owing to the controlled growth from an electrode, the nucleation sites of the MOF are controlled and thus 1D morphologies can be grown from an electrode to produce nanorods. Mofarah et al. [20] synthesized a highly unstable MOF consisting of nanorods of Ce-TCA using electrodeposition. The 2D MOF contained relatively strong coordination bonds within sheets and weak van der Waals bonds between sheets. Owing to the weak bonds between sheets, the unstable CP could be exfoliated easily at the liquid/air interface to form nanosheets, which were collected using a stamping process. The thickness of the nanosheets could be controlled by varying the rate of liquid evaporation, which drives the nanosheets toward the interface. Additionally, by increasing Ce-TCA concentration, other 3D structures can be formed, including the hollow spheres and hollow polyhedra shown in Figure 6.

Sacrificial Templates
The self-assembly of MOFs into complex morphologies during solvothermal synthesis is one of the principal reasons to utilize them as precursors for MD synthesis. However, MOFs can also be grown on substrates and non-reactive sacrificial templates to produce applicable MD morphologies, as shown in Figure 7. [183] During MOF synthesis, when PVP is mixed with a Mn/Co source and a BTC ligand is added, the MOF structure precipitates on the PVP template and subsequent annealing causes the MO to adopt the form of the original coated PVP fibers. [184] Additionally, PVP and other organics that are nonreactive with MOFs can function as additional sources of carbon. When the thermal conversion is done in air, oxidation of the carbon increases the porosity; when it is done in reducing atmospheres, the pyrolysis contributes additional carbon to the matrix. Both approaches can potentially improve the catalytic performance of the resultant MDs.
Although uncommon, MOFs can be used as templates for non-MOF precursors for the construction of MDs. In such cases, the metal of the MOF precursor is arbitrary and serves  [20] Copyright 2020, Royal Society of Chemistry.
only to generate a specific MOF structure. Gao et al. [185] modified the synthesis of C(Co,N) [186] by adding Si-functionalized PVP to precursor ZIF-67(Co). The MOF functioned as a template for the carbon-based structure surrounding the Si associated with the PVP. The Si was not distributed homogeneously within the MD owing to the PVP's clumping and the MOF's encasing, as shown in Figure 8. Upon conversion in Ar at 800 °C, the carbon surrounding the Si clusters retained intimate contact across the matrix-agglomerate interface, facilitating effective charge transfer. Other primary doping techniques could be used to distribute such secondary phases (Si) more evenly throughout the MOF, with or without non-reactive sacrificial templates.
Finally, MOFs can be grown directly from architecturally designed sacrificial templates that act as a source of metal that reacts with ligands in the immersive solution, thereby producing a MOF. However, these templates are limited to lowchemical-stability metallic salts, including some oxides, [187] carbides, [188] carbonates, [189] and hydroxides. [190] Since the ligands used to produce MOFs generally are of low reactivity, then there is only limited reaction-front distance, leading to thin MOF surface layers on bulk materials. Cai et al. [191] grew CoO nanoneedles on NF before adding the ligand (MIM), which reacted with the surfaces of the CoO nanoneedles to produce ZIF-67(Co)/ CoO/NF. Finally, metal meshes also can be used as metal sources for MOF growth but they have the shortcomings of . Adapted with permission. [183] Copyright 2019, American Chemical Society.  [185] Copyright 2020, Elsevier.
providing only minor dopant concentrations [192] or they require etching agents to transform the surface layer into hydroxide. [193] Although rare, vapor-phase synthesis of precursor MOFs to duplicate the porous substrate morphology has been done. As shown in Figure 9, Hu et al. [49] used a low-temperature (350 °C) multistep process as an alternative to high-temperature nitriding or plasma nitriding of Ni using N 2 . They grew a reactive Ni(OH,F) coating on nickel foam (NF) before MIM vapor was introduced to produce Ni-MIM MOF in the original morphology of Ni(OH,F). It is probable that the vapor phase synthesis and the use of the higher temperatures (>250 °C) increase the ligand-substrate reactivity and kinetics, thereby providing the driving force for the complete conversion of the reactive layer into MOF. While liquid-phase synthesis of MOFs from metallic salts is limited to the surface layer, vapor phase synthesis has the capacity to achieve complete conversion. [194] The MOF was then nitrided in an NH 3 atmosphere to produce NiN of identical morphology. Although the low temperatures of such processes can avoid degradation of the substrate's mechanical properties, their complexity has limited the extent of applicability. Consequently, there is a much larger body of research on precursor MOFs that undergo direct conversion on the non-reactive substrate, viz., without the reactive coating and its conversion. In principle, the multistep process can be applied to form MOF structures from which the MDs are fabricated in vapor phase(s). [195,196]

Substrates
MOFs also can be grown onto non-sacrificial substrates. Catalytic and electroactive materials are attached to substrates with binder agents but underperform compared to active materials grown directly from substrates. [197] MOFs can be grown nonepitaxially on substrates, such as NF and carbon cloth (CC), in 1D, 2D, and 3D morphologies, as shown in Figure 10. While the interfaces may be incoherent, precipitation of a MOF on a substrate, usually with complete coverage, occurs when nucleation is effected owing to the requirement that the interfacial energy between a MOF and solid surface is lower than that between a MOF and immersive solution. [198] When these MOFs grown on surfaces are thermally converted into MDs, thermal stresses can cause delamination. To avoid this, temperatures 400-600 °C are generally used but solvothermal conversion (<200 °C) can also be applied to eliminate thermal stresses at the cost of potentially etching at the interface between poorly coated substrates. The latter method has been used to synthesize Co-MOFs on 3D graphene foam in order to produce Co 9 S 8 nano-arrays. [199] While thermal conversion methods do not allow partial conversion of MOFs, solvothermal conversion allows greater control of the extent of the MOF-to-MD conversion. Since solvothermal conversion involves diffusion of ions from the MOF surface to the core, halting the conversion before full conversion allows fabrication of MD/MOF composites. Zhao et al. [200] synthesized (Fe,Ni)-MOF in the presence of TAA to form (Fe,Ni)-MS/MOF. As mentioned previously, solvothermal conversion avoids carbonization; hence, MD/ MOF composites exhibit relatively low electrical conductivities. In the absence of a conductive network for electrochemical applications, these composites must be grown on external conducting substrates. The presence of even small amounts of MS in these MS/MOF composites led to two-and four-fold improvements in capacitance compared to that of MS without MOF and MOF without MSs, respectively. For many electronic, photonic, and sensing applications, high interfacial electrical conductivities require the epitaxial growth of MOFs. [201] To achieve this, metal sources for reaction with ligands to form MOFs can be used as substrates. As previously discussed, this process tends to limit MOFs to thinfilm morphologies. As many metals are too stable to react with the ligand or they form a stable passivating layer that inhibits reaction with the ligand in solution, metal hydroxides typically are used as surfaces for MOF growth. Okada et al. [202] demonstrated multiple methods of converting copper 2D patterns and 3D architectures into Cu(OH) 2 , which was then used to achieve controlled growth of HKUST-1(Cu).
Unfortunately, owing to the coordination bonding and the very large lattice parameters of MOFs, incoherent interfaces between MOFs and non-MOF substrates are common. However, MOFs readily form coherent or semicoherent interfaces with other MOFs. Ikigaki et al. [203] grew a Cu-MOF on a Cu(OH) 2 substrate before growing either a functionalized version of the same Cu-MOF (forming a coherent interface) or a different Cu-MOF with smaller ligands (forming a semicoherent interface). Although both cases involved epitaxial MOF-MOF growth from the same metal (and SBU coordination), homogenous and heterogenous epitaxial growth from MOFs with dissimilar metals is possible and is discussed in further Figure 9. Multistep synthesis route to grow MOF on NF with subsequent nitriding. Adapted with permission. [49] Copyright 2020, American Chemical Society.
detail in Section 2.6.7. These MOFs can then be converted into heterogenous MD with excellent interfacial properties.

Complementary Conversion
The direct conversion of MOFs also can be done in conjunction with other processes, with the aim of achieving good interfacial connectivity in composites. The reduction of GO is one such technique as rGO is a very common composite component for many applications of MDs. [204][205][206][207] In these studies, the MOF was either grown on GO [204] or GO was mixed with the MOF before MD synthesis; [205] both types of composites underwent and subsequent thermal conversion to reduce GO to rGO, resulting in well-dispersed MD on the rGO nano-sheets.

Carbon-Removal
There are three variables that determine the retention of carbon in an MD. In the case of oxidation temperature, the temperature required to induce thermal conversion of the MOF to MD (>300 °C) invariably facilitates complete oxidation of carbon. However, ex situ oxygen is required for complete oxidation; hence, carbon removal is dependent on the oxidation atmosphere. In this case, while many MOFs contain sufficient oxygen for self-oxidation of the metal, there rarely is enough for complete oxidation of the carbon, and thus ex situ oxygen is required. Fang et al. [208] synthesized (Zn,Co)-MIM nanosheets at room temperature in deioinized (DI) water in order to minimise the growth of thicker stacked sheets. This precursor was then oxidized in air at 400 °C for 30 min to produce ZnO • Co 3 O 4 nanosheets. Since carbon typically is not catalytically active, its absence improved the electrochemical performance of these oxide MDs. In the case of such a carbonization temperature, the thermal conversion of MOFs always results in carbonization. This can be avoided through solvothermal conversion, which also has the advantage of using temperatures that avoid oxidation. He et al. [199] solvothermally converted ZIF-67(Co) using TAA to produce Co 9 S 8 without carbon or oxygen. More broadly, for single-step conversions, carbon-free MDs are limited by the preceding three variable constraints.

Carbon-Free MDs
Some carbon-free MDs cannot be converted from MOFs in a single step. In these cases, an MO or solvothermal synthesized MD is produced as an intermediary before a second conversion step. Hsu et al. [57] and Feng et al. [209] converted precursor MOFs into MO or LDH intermediates, respectively. These then underwent sulfuration or phosphidising into CoS 2 /FTO and FeNiP, respectively. This multistep process can be used to leverage specific characteristics of different conversion techniques. For example, solvothermal conversions strongly degrade hollow 3D structures into nanosheets by cleaving and opening the structure, as shown in Figure 11. Xiao et al. [210] synthesized Me-doped CoP nanosheets (Me = Mn, Ni, Cu, Zn) using an identical phosphidising process. [209] The precursor MOF was prepared by rapid precipitation (15 min) in methanol of CoMe-MIM, which then was deconstructed solvothermally in a solution containing excess Co 2+ and Me 2+ to form CoMeLDH nanosheets. This ultrarapid CoMe-MIM synthesis produces thin MOF-polyhedra that can convert high ionic strength into nanosheets in media. The resultant LDH was then phosphorated into CoMeP nanosheets.
In other cases, using MOs as intermediates to generate MDs offers different morphological outcomes. When MOFs are converted into MOs by heating in air, the carbon oxidizes, resulting in significant mass loss and associated porous morphology. Hence, MOFs that generally produce low-porosity MDs can be induced to produce high-porosity MOs, which then can be converted into other MDs. For example, relatively low porosity MOFs, such as PBAs, can still generate porous MOs owing to the release of CO/CO 2 and NO 2 /NO 3 during thermal conversion in air. [36] Nanorods of MOF-74 were produced by heating in air at 350 °C for 2 h to produce (Ni,Co) O. [211] The MMO was then converted into porous nanorods of (Ni,Co)P by thermal conversion under the same conditions but under an atmosphere of decomposed Na 2 HPO 4 in carrier N 2 .
Although it was claimed that the MOF-74 precursor underwent carbonization in air at 350 °C, elemental mapping showed only trace amounts of carbon. Other studies of this MOF heat treated under the same conditions indicated complete oxidation had occurred. [178] When substrates are used as platforms, the MDs will utilize the same substrate on which the MOF was grown while retaining the morphology and composition of the MOF. While single-step conversion of a MOF to MD on NF or CF has been demonstrated, [41] multistep conversion is more common. sulfide/metal shell-core nanoparticles have been synthesized on CFs. [212] In this work, a Co-MOF on CF was converted into (Ni,Co)-LDH by solvothermal conversion in a solution containing Ni 2+ . The NiCo-LDH then was reduced in an H 2 /Ar (10/90 v/v) stream at 250 °C to produce (Ni,Co)metal. This then was surface-sulfurized by solvothermal conversion with TAA to synthesize (Ni,Co)S/(Ni,Co)-metal shell-core nanospheres on CF. The degree of sulfuration was controlled by varying the concentration of TAA and the solvothermal temperature and time. The retention of the metal as core was confirmed by XPS analysis, which confirmed Ni and Co in their metallic forms. This work suggests that partial sulfuration can be applied generally for conversion of MOFs to MSs. and CoMeP (f). Adapted with permission. [210] Copyright 2017, Royal Society of Chemistry.

Carbon-Containing MDs
Although Yang et al. [212] synthesized a carbon free (Ni,Co)S/ (Ni,Co) that was not highly conductive, it remained applicable for electrochemical applications owing to the carbon substrate on which it was grown. Consequently, non-substrate and MOFderived carbon is capable of improving catalytic performance. Yang et al. [213] synthesized C(N,Ni,Ni 3 S 4 ,CNT) composites produced in situ using CNTs to produce 3D conductive networks. Although similar to a conductive substrate, where the CNTs effectively provide conductive matrix, additional carbon generated during pyrolysis provided the three benefits previously discussed: Increased surface area, charge carrier conductivity, and retention of MOF morphology, all of which are beneficial for electrocatalysis [214] and heterocatalysis. [215] Following MOF conversion, carbon generally can be removed only by oxidation at >300 °C as it is resistant to other conversion methods, including solvothermal conversion [213] and acid etching. [216] MOF conversions performed in air that inevitably remove carbon can utilize ex situ carbon sources during subsequent conversion for its reintroduction. Wang et al. [217] used Co and FcDC to produce hollow MOF spheres and, upon conversion in air at 340 °C, the MOF was completely converted to CoFe 2 O 4 . The electrochemical properties of this material were improved by adding dopamine hydrochloride or GO to a suspension containing this MD, followed by conversion in N 2 at 450 °C to produce a nitrogen-doped carbon or r-GO coated MD, respectively. In both cases, the carbon may have blocked the active surface of the MMO, although the thickness of the porous surface layer was limited to ≈10 nm, and it did not have a detrimental effect on the electrocatalytic performance of the MD.
When the MOF is the source of in situ carbon structures, they can be impregnated using ex situ sources of metal. Shi et al. [216] produced SnS 2 on porous carbon by decomposing ZIF-67(Co), thereby forming C(Co), followed by removal of Co by acid etching. The SnS 2 MD was grown directly on the porous carbon by solvothermal synthesis with SnCl 4 and TAA. However, the introduction of ex situ carbon [217] or metal [216] results in MDs only at the surface, covering up the porous carbon, reducing the available surface area. For example, the BET surface area of the SnS 2 MD synthesized by Shi et al. [216] decreased from 306 m 2 g −1 for the MOF-derived porous carbon to 28 m 2 g −1 for the SnS 2 /C.

Doping
The catalytic and electrochemical performance of MDs are often improved with dopants. [218] These dopants can be categorized into three distinct types: Primary dopants are metal dopants either in solid solution or as distinct phases forming heterostructures and mixtures. While non-metals can dope the MD, these cases are generally not considered as doping since they completely change the local chemistry (i.e., oxide to sulfide after introduction of S). Secondary dopants are nonmetal dopants in the carbon that can improve MD-carbon composite performance. Finally, surface dopants are heterogenous dopants that are isolated to the surfaces of MDs and can modulate properties such as adsorption and absorption. However, as many MDs are nanostructures with small dimensions, bulk properties are often identical to surfaces, and thus surface doping is not largely considered. Additionally, many surface doping techniques are post-synthesis methods and are therefore not specific to MOF-derived materials and already have been covered extensively. [219][220][221] The limited occurrences of surface and heterogenous dopants are discussed in greater detail in Sections 2.6.6 and 2.6.7.

Primary Dopants
As shown in Figure 12, primary dopants in solid solutions can either appear in the interstitial sites, where dopants sit in the spaces between ions in a crystal structure, or at substitutional sites, where the dopants will replace existing ions within the crystal structure. The dopant concentrations are limited by the solid solubility limit, which can change depending on the host-dopant compatibility, where solid solution dopants are typically at low concentrations (<10 mol%). When the solid Figure 12. Schematics of substitutional (green and blue spheres) and interstitial (red spheres) solid solubility, where primary doping involves metal dopants in MD and secondary doping involves non-metal dopants in carbon. solubility limit is exceeded (or the dopants are not soluble in the host), dopants exist as distinct minor phases that can either modify the properties of the major phase (host) through interfacial interactions or provide independent performance. These dopants can either be included within the MOF and are retained post-conversion in the MD, or are introduced during conversion, which is discussed in more detail in Section 2.6.

Secondary Dopants
Since the carbon in MD composites is important to a range of properties, as described previously, performance of MD composites can be optimized by doping the carbon. Carbon chemistry allows only non-metal ions to be integrated into their structure, [222] so these ions represent secondary dopants. An exception is the oxygen ion, which causes oxidation during thermal conversion in air. While the effects of secondary dopants are not fully understood, dopants introduce perturbations to the electronic distribution of carbon sheets, leading to changes in conductivity, catalytic site density, degree of graphitization, etc. The most common dopant, nitrogen, has been shown to improve charge conductivity of MD-carbon composites. Nitrogen added is through an active nitrogen-containing source, including reactive nitrogenous gas, [158] ligand in precursor MOF, [223] and added powders than thermally decompose in the reactive gases [224] but not through N 2 , which is often used as an inert gas in pyrolysis. The only example found in literature that used N 2 gas in thermal conversion for producing nitrogen-doped carbon and MS composite showed only trace amounts of nitrogen-doping with no quantitative data. [225] In the work, FeS x and carbon composites were synthesized by mixing MIL-88A and sulfur in 2:1 mass ratio with a solvent method and then thermally treated in N 2 . Given the relatively low nitrogen content from EDS and weak XPS signal, comparative studies using Ar as a pyrolysis gas should be used as a control to justify the claim that nitrogen doping was present and/or significant. Based on all current evidence, nitrogen is not effectively doped when pyrolysis is conducted in an N 2 atmosphere. Other potential non-metal dopants, such as S or P, can also be included in the final material using similar techniques.

MOF Thermal Stability
During thermal conversion processes, the temperature is as important as the atmosphere but it is readily controlled for greater effect. In general, increasing temperature above MOF stability ranges (>300 °C) leads to increasing diffusion, recrystallisation, morphological development, crystallinity, grain growth, and agglomeration. These trends often are significant for MD-carbon composites. Zhang et al. [226] observed that, with increasing conversion temperature, the morphology of the precursor was progressively degraded, where the polyhedral shape of the precursor gradually converted to irregularly shaped agglomerates.
The minimal temperature for thermal conversion for a particular atmosphere is determined by the intrinsic structural stability, defect chemistry, and surface energy (area) of the precursor MOF. Hu et al. [227] and Zhang et al. [226] calcined Cu-BTC to produce CuO • Cu 2 O hollow polyhedra by thermal conversion in air. Considering the raw materials, processes, and XRD data reported by both sets of authors, it is likely that the synthesised MOF was HKUST-1(Cu). However, thermogravimetric analysis (TGA) data for both MOFs indicated differential decomposition temperatures, where the higher temperature was consistent with the greater degree of crystallinity of the MOF synthesized by Hu et al. [227]

Valence
Although the temperature is critical to thermal conversion, it also can alter the redox state. These effects are suggested unambiguously by thermodynamic stability diagrams, which plot stability regions for different valences as a function of temperature and oxygen partial pressure. [228] Hu et al. [227] reported that increasing the heating temperature of HKUST-1(Cu) in air resulted in an increase in the CuO/Cu 2 O ratio up to 400 °C; no Cu 2 O was detected at 500 °C. The thermodynamic stability diagram for this system [229] shows that CuO is the thermodynamically stable phase under these conditions. However, the valence ratio is not limited to this single variable as there also was an atmospheric effect from the reduction of carbon during heating, which would have lowered the oxygen partial pressure and thus facilitated the transition into the Cu 2 O stability region. A further consideration is that examination of the Cu-O phase diagram [230] reveals that both of these oxides are compatible up to 1091 °C, so the Cu valence ratio also is dependent on the overall composition (i.e., valence ratio), which changed during heating. Consequently, the elimination of Cu 2 O >400 °C can be attributed to complete oxidation of carbon, and the valence effect observed was a function of carbon reduction ≤400 °C rather than simply of the atmosphere-temperature conditions suitable for the stabilities of particular valences. These observations make clear that such processes are consistent with metastable equilibrium rather than the conditions of the thermodynamic equilibrium of the respective stability diagrams and phase diagrams.

Composition
Although ultrahigh-temperature conversions (>700 °C) are limited by morphological stability and cost, they can facilitate mechanisms not possible at lower temperatures. Since the melting and boiling points of metals of MDs are significantly higher than the temperatures of MOF conversion (≈350-700 °C), chemical treatments generally are the only methods for metal removal. However, metals with high vapor pressures, viz., Zn, [231] can be removed by volatilization at high temperatures during thermal conversion. Further, this potential also applies to heat treatments of MDs post-conversion. Chai et al. [232] removed Zn from C(ZnO,Mo 2 C,N) by reductive pyrolysis (ZnO → Zn + ½O 2 ) at 800 °C for 0.5 h. In this case, the resulting C(Mo 2 C,N) structure did not agglomerate at high temperatures owing to the low MD content, structurally stabilizing carbon, and possibly the nanostructural disturbance caused by gas exudation. Conversely, agglomeration would be more likely when the MD content is higher and carbon is not present.

Polymorphs
The temperature of thermal conversion affects both crystallinity and the polymorph that recrystallizes. Although the TiO 2 polymorphic transformations have been examined extensively, the kinetics are variable and give rise to uncertainties, particularly the observed conversion temperature. [233] Zhang et al. [153] synthesized TiO 2 short cylindrical particles of MIL-125(Ti) by conversion in air at 380 °C for 5 h and room-temperature XRD analysis revealed the anatase polymorph of low-crystallinity. When the conversion temperature was increased to 500 °C, the anatase crystallinity increased but the rutile polymorph also was observed. The proportion of rutile steadily increased until 800 °C, at which point anatase was no longer present. [234] When contrasted with an alternative TiH 4 O 4 precursor, rutile was found to appear only at 700 °C. This can be explained by the presence of carbon in the MOF and its reducing effects at these temperatures, which would be likely to result in oxygen vacancy formation and the resultant lowering of the phase transformation temperature owing to the structural distortion deriving from these defects (lattice contraction) and the charge compensating Ti 4+ → Ti 3+ reduction (lattice expansion), which increases the Ti radius in sixfold coordination by ≈9% (0.0745 nm to 0.081 nm). [235] Oxygen vacancies concentrations can be increased further through thermal conversion under reducing atmospheres. In contrast, Zou et al. [236] heated a Ti-MOF under Ar up to 900 °C for 2 h and observed a mixture of anatase and rutile at 800 °C but only rutile at 900 °C. The preceding studies suggest that oxygen vacancies stabilize the low-temperature anatase polymorph, which may explain the wide range of temperatures reported for this transformation (390-1190 °C). [233] The lower range of transition temperatures is likely to reflect surface area and purity. Consequently, in order to synthesize a specific polymorph, the complementary contributions of both carbon and atmospheres as reducing agents should be considered.

Room-Temperature Conversion
When the retention of the MOF morphology is intended, thermal conversion generally is unsuccessful owing to MD structural distortion and agglomeration. However, MOF precursors can be converted without structural alteration in concentrated acidic or basic solutions at room temperature. This approach has been used successfully for MOFs of both unstable (Ce-TCA) [20] and stable (Ce-BTC) [40] 2D morphologies of even a monolayer thickness. [20] The disadvantage of this approach is that the resultant MD typically is of low crystallinity and is limited to carbon-free MDs. Abney et al. [58] also used highly stable UiO-66 and MIL-125 to synthesize amorphous ZrO x and TiO x , respectively, at room temperature. In contrast, this approach can be applied using solvothermal synthesis (<200 °C) and less-concentrated acids and bases, which suggests the potential to obtain more crystalline MDs.

Carbon Modification
Although the impact of carbon on MD structures and properties has been explored extensively, there is a converse effect in the form of MD-facilitated catalysis of morphological alteration of the carbon structures. The derived carbon can be present in five forms: Amorphous, interfacial semicrystalline carbon, crystalline carbide, carbon nanotubes, or graphene. The long-range order associated with graphite formation has not been reported, probably owing to the structural inconsistency imposed by the presence of the MDs. Energy requirements to modify amorphous carbon to crystalline carbon or carbides generally are excessive compared to the temperatures used for MOF conversion. [237] However, the presence of the metal can facilitate the formation of core-shell structures through simple surface reaction, which then makes available the MD that is essential to catalyze further carbon modification to form carbon nanotubes or graphene. The morphologies of these core-shell structures can take the forms of MD particles with an interfacial layer of semicrystalline carbon [31] or CNTs wrapped around single-ion or particulate metals. [238] Metals such as Co, which forms a closefitting carbon shell around the MD, appear to have a stronger effect on formation of the carbon structure than other metals such as Ni, [31,152] which have been found to form a looser shell. However, only limited research has been done in this area, [239] and further research, including computational modeling is required to better understand this mechanism and to compare the effect across different metals.
Even the presence of alloys versus pure metals may influence CNT formation from MOFs. Work by Pan et al. [240] suggested that the catalyzing effect of Co alone is not enough to form CNTs, hypothesizing that volatilized Zn is required. However, it is likely that this aspect of their work involved the presence of residual oxygen, which caused surface oxidation of the Co metal in the ZIF-67(Co) and its consequent inactivation, as suggested by a recent review published by Wang et al. [241] However, when ZIF-67(Co)/ZIF-8(Zn) was converted in N 2 , the vaporized Zn acted as an oxygen scavenger and potential CoO reducing agent, thus ensuring the outcome of metallic Co, which subsequently catalyzed CNT formation. It also is possible that the Zn alloyed with the Co, thereby forming a superior catalyst. This possibility is suggested by the work of Shi et al., [242] shown in Figure 13a-c, where thermally converted MOFs with Co and Fe, Co and Ni, and Co alone resulted in progressively reduced extents of CNT formation.
It also is possible to modify the composition of the MD by further processing post-carbonization while retaining the carbon structures, as shown in Figure 13d. Zhang et al. [238] produced CoS 2 and carbon nanotube composites from, in the first step, ZIF-67(Co) under reducing hydrogen during pyrolysis in order to expose the core Co, which then catalyzed the formation of carbon nanotubes. In the second step, sulfuration resulted in CoS 2 formation as well as sulfur doping of the carbon nanotubes. Finally, it is likely that the presence of sufficient oxygen in the ligand to form an oxide, rather than a carbide, can prevent this catalytic action owing to the mismatched chemistries and structures.
In addition to the effect of MDs in modification of the morphology of the carbon nanostructures derived from MOFs, the surface area is also affected. Shang et al. [243] coated ZIF-67(Zn,Co) with porous SiO 2 , thermally converted it into SiO 2 / C(Co,N) and the etched away the SiO 2 with HF while retaining the C(Co,N). It was reported that coating the MOF with SiO 2 prevented agglomeration of the MD-carbon composite during thermal conversion at high temperature (1000 °C). Additionally, the final MD was found to not include any Zn, likely caused by its volatility. Although not considered in the publication, metal vapor likely facilitated the formation of porous nano channels in the carbon matrix, which was calculated to have a BET surface area of 1203 m 2 g −1 . As shown in Figure 14, the surface area of a MOF-derived MD-carbon composite linearly increases with Zn content; at the potential detriment of reduced MD content in the MD-carbon composite.

Intrinsic Mixed-Metal-Organic Frameworks
Within a closed system, decomposition of MOFs will result in compositionally consistent MDs (as a practice, hydrogen is not considered) and so conversion of mixed-metal-organic frameworks (MMOFs) will produce mixed-metal derivatives (MMDs). Prussian blue analogues (PBAs), with a chemical formula of A x [B(CN) 6 ] y (where A and B = metals, x = 2-4, y = 1-3), are among the most common MMOFs. PBAs are nonporous MOFs that contain small molecules rather than long ligands, which provides them with three distinct features. [245] First, PBAs are oxygen-free. Second, the small coordinating molecules require metals of smaller radii to allow ligand structural balance and continuity. Third, PBAs contain two different metal sites (A and B), which can be filled by identical or metals of similar size (these are limited largely to the row 4 metals K, [246] Mn, [247] Fe, [248] Co, [249] Ni, [250] and Zn [251] ). Although intrinsic MMOFs are obvious candidates for conversion into MMDs, their restricted compositions compared to those of doped MMOFs limit their applicability in synthesis of MMDs.

Extrinsic (Bulk-Doped) Mixed-Metal-Organic Frameworks
SBU Doping: MMOFs can be synthesized by doping the bulk using the SBU or the ligand of the MOF, as shown in Figure 15. SBUs of common single-metal MOFs can be doped by utilizing mixed-metal salt solutions of two or more metals during solvothermal MOF synthesis, [252][253][254] where the metal ratios are controlled by varying the molar ratio in the raw materials. [255] The process is uncomplicated but structural constraints at high dopant concentrations generally require MOFs to be grown from similar metals in which the valences are identical and the ionic radii are similar. The most common example of such doping is 50 mol% Ni 2+ (sixfold ionic radius of 0.83 nm) [235] and 50 mol% Co 2+ (sixfold ionic radii of 0.885 nm (high-spin) and 0.79 nm (low-spin)). [235] Given weak-field ligand oxygen would favor high-spin [256] and the strong-field ligand nitrogen would favor low-spin, [257] it can be assumed that SBU doping can be affected by the ligand (O or N bonding), however computational modeling on the effect of doping within the SBU is needed to fully elucidate the mechanisms at work. However, the feasibility of doping at high concentrations, typically >10 mol%, is contingent not only on these valence and size issues of the metals in the raw materials. Another requirement is chemical compatibility of the metals prior to MOF growth, at which point the potential for metal-metal contact would be precluded. Further, with the changing pH conditions of solvothermal synthesis and electrodeposition, the valence and hence the ionic size also could change. Finally, at these high dopant concentrations, the morphology of the MMOFs can be altered owing to increased structural disorder. This was demonstrated by Hou et al., [258] on a Co x Fe 1−x -MOF, where x = 1 was composed of agglomerated large rods, x = 0 of small nanoparticles of indistinct shape, and the optimized x = 0.8 produced high surface area nanorod flowers.
Although the SBU is doped most commonly during synthesis, post-synthesis doping can be achieved in MOFs by transmetallation. [259] The latter process is performed by soaking a prototype MOF in solution containing the desired dopant metal, and depending on the soaking time, complete or partial exchange of the metal within the SBU can occur, producing MMOFs that may not be synthesized directly. Goswami et al. [260] soaked a 2D Co-MOF in a Cu-rich solution for 48 h to achieve complete exchange. It is likely that shorter soak times would have resulted in partial transmetallation. Transmetallation is possible for dissimilar metals although this would be expected to be limited to lower dopant concentrations of <10 mol%. Su et al. [223] introduced Ru into Co-MOF nanospheres by partial transmetallation using an Ru-PBA (from RuCl 3 ). Interestingly, this MOF was Co 3 (Co(CN) 6 ) 2 and so contained 3Co 2+ , 2Co 3+ , and 12 (CN) − , where the sixfold ionic radii would be 0.885 nm (Co 2+ , high-spin), 0.75 nm (Co 3+ , high-spin), and 0.82 nm (Ru 3+ ). In this case, it appears that the intermediate size of Ru 3+ may have been the basis for its selection as a dopant. Even so, the optimal concentration for the HER was found to be ≈3.4 mol% (metal basis), although the maximal dopant level was ≈5.1 mol% and no exsolution was reported at this level.
Ligand Doping: As shown in Figure 15, rather than internally doping the SBU, metalation of the ligand, where the metal is incorporated into the ligand structure, and transmetallation of the existing metal in the ligand are a second method of bulk doping. These approaches offer greater flexibility in the type and amount of metal dopant(s) because the structural constraints imposed by SBUs are replaced by those of the ligand, which accommodate distortion more readily and, in fact, can be designed specifically to this end. Kassie et al. [261] attached diphosphine pincer complexes to the backbone of a ligand to incorporate large dopant metal ions in a Zr-MOF structure. The original Co ions in the ligand were then removed and replaced entirely with Rh or almost entirely Pt. The exchange required the application of heat (85 °C) in order to effect replacement, which probably resulted from expansion of the Co 2+ metal site and coordinating ligand to accommodate the larger Rh 2+ and Pt 2+ ions. Although such processes greatly expand the flexibility of doping of MMOFs, a requirement is for the MOFs to be constructed from ligands sufficiently long to allow bonding to the ligand backbone by the large functional groups required to contain the dopant(s). This is exemplified by the researchers' use of the large Zr-MOF.

Extrinsic (Surface-Doped) Mixed-Metal-Organic Frameworks
Pore Decoration: In cases when doping with a dissimilar metal is restricted by the small size of the MOF, the alternative strategy of occupation of the pores by the metal ion is possible, as shown in Figure 15. Despite the potential that is allowed by the ultrahigh surface areas and associated porosities of MOFs, this approach has not been reported frequently. Similar to transmetallation, this method involves doping a MOF by soaking it in a metal salt solution. However, important considerations include the effects of the medium on the MOF in terms of solubility, chemical stability, wetting, and swelling. This method also offers the advantage that the pore-filling process is such that the concentration of the dopant in the MD can be moderated through control of the soak time and/or the metal salt concentration. Wang et al. [262] synthesized ZIF-67(Co) by soaking the MOF in solution of RuCl 3 for partial replacement and hence formation of ZIF-67(Co,Ru). Following thermal conversion in air at 350 °C for 2 h, the product consisted of Co 3 O 4 (RuO 2 ). Although the authors described this as doped Co 3 O 4 , which requires ionic Ru as a dopant (compounds cannot be solutes in substitutional or interstitial solid solutions), they followed a common practice of referring to the material as Ru-doped Co 3 O 4 , which is misleading as their TEM images clearly show the presence of a mixture of RuO 2 and Co 3 O 4 , which would be Co 3 O 4 (RuO 2 ), which is consistent with their description of this material as a heterostructure. Although the authors did not discuss the role of porosity, this was clarified through similar work by Koo et al., [263] who soaked ZIF-67(Co) in a solution of K 2 PdCl 4 , which resulted in its encapsulation within the MOF pore network; the salt then was reduced to Pd metal by soaking in a solution of NaBH 4 . Following thermal conversion in air at 400 °C for 1 h, the product consisted of what was described correctly as a Co 3 O 4 (PdO) heterostructure. In a similar study, Liu et al. [264] synthesized a different MOF, Co-BTC(Ru), by soaking the MOF in a solution containing RuCl 3 and, after thermal conversion in air at 600 °C, produced heterojunction Co 3 O 4 (RuO 2 ). Surface Decoration: As shown in Figure 15, doping also can be done by chemisorption and possibly physisorption on internal and external MOF surfaces. In the former case, this would be through conjugation with a (non-MOF) ligand attached to a dopant. In the latter case, this would be through surface adsorption or soaking, which feasibly could involve all of the previously described doping mechanisms. Chen et al. [265] anchored different metal acetylacetonate complexes on Mn-BTC and heated these in air at 500 °C to produce MMOs. This process was handicapped by poor compositional control, as evidenced by outcomes of different MOF dopant concentrations despite the use of identical precursor dopant concentrations. In contrast, Tan et al. [266] claimed to have achieved ≈12 mol% Mn (metal basis) doping through physisorption of MnCl 2 on MIL-68(In), followed by sulphuration by solvothermal synthesis using TAA in ethanol at 180 °C for 3 h, to obtain the heterojunction MnS • In 2 S 3 . However, as immersion of physisorbed species often results in de-adsorption, then this result may derive from SBU doping and/or pore decoration that occurred during solvothermal synthesis; it would not derive from ligand doping as this ligand cannot be metalated. Alternative mechanistic interpretations are that the physisorbed Mn 2+ ions were dislodged upon immersion, the Mn 2+ ions dissolved in the SBUs and/or pores, and: a) the S 2− ions in the TAA also dissolved or b) the Mn 2+ ions reacted with the S 2− ions in the TAA upon MOF conversion; in both cases, the reaction between the two ions resulted in widespread formation of MnS heterojunction particles in an In 2 S 3 matrix.
As suggested, multiple simultaneous doping processes may occur, thus raising the possibility of increasing dopant concentrations, which is difficult for dissimilar metals. However, there appears to be little information about doping mechanisms. For example, in addition to the doping methods previously described, additional sources of dissolution may be present in the form of vacancies generated by absent SBUs and/ or linkers, incomplete SBUs, and surface disorder (established by the vacancies in the grain boundaries and outer surfaces). Although these are fundamental defects that have been characterized, there do not appear to be any works that associate these defects with solubility mechanisms. [267]

Conversion of MMOFs into MMDs
The conversion of MMOFs into MMDs results in the formation of homogenous single-phase (solid solution) and multiphase MMDs (mixture). In both cases, dopants are homogeneously dispersed throughout the MMDs. An alternative means of generating MMOFs in addition to the preceding formalism is a composite MOF, in which a homogeneous mixture of two different MOFs is formed. As shown in Figure 16, Xu et al. [268] synthesized ternary (Cu,Zn,In) sulfides by converting a composite MOF containing three metals (Cu, Zn, and In) and two ligands (MIM and PTA). It was proposed that the Cu and Zn preferentially bonded with MIM (likely ZIF-8(Cu,Zn)) while In bonded with PTA (to form MIL-68(In)) and thus the optimal the Cu:Zn:In proportions could be achieved by varying the molar ratio of two different ligands at fixed metal concentrations. However, the optimal proportion of ligands only utilized 1.2 mol% of PTA (compared to 98.8 mol% MIM) while still retaining significant proportions of In in the MD. Therefore, although Cu and Zn may not be reacting with the PTA, In is present in both MOFs (as a dopant or otherwise). Since carbon was lost during solvothermal conversion, the composition was compensated by addition of GO and g-C 3 N 4 nanosheets to the solvothermal suspension containing the MOFs. The product consisted of a graphene-type heterostructure decorating a (Cu,Zn,In) sulfide MMD; where the photocatalytic HER performance was improved 61, 42, and 37 fold compared to non-MOF derived, pure PTA, and pure MIM derived MMDs, respectively.

Conversion of MOFs into MMDs with Ex Situ Metals
Although converting MMOFs into MMDs is typically an in situ process, external metal sources can also be utilized with solvothermal conversion. In common with the incorporation of non-metal in the MD during solvothermal conversion processes, metal ions in the solution also can be assimilated to produce MMDs. Nagaraju et al. [192] used an NF substrate on which to grow Co-MIM nanotubes, which then were converted solvothermally in a Ni 2+ solution to produce (Co,Ni)LDHs of the same morphology These then were sulfurated solvothermally in the presence of a Cu foam, which was exfoliated, thus freeing Cu 2+ ions. The outcome was nanotubes of a Cu-doped Co-Ni MD of approximate composition (based on EDS spectra) CoNi 3 S 4 (Cu 0.3 ) (the stoichiometry indicated was Cu(Co,Ni) 2 S 4 ). The higher Ni 2+ fraction probably derives from its presence in excess in solution, whereas the lower Cu 2+ probably derives from the requirement of having to dissolve following exfoliation. More broadly, in principle complex doping of MMDs could be achieved through the use of volatilization of phases with high vapor pressures, such as alkalis, alkaline earths, Group 12 transition metals (Zn, Cd, Hg), pnictides, chalcogenides, and halogens.

Stepwise Conversion of Single-Metal MOFs into Hierarchical MMDs
Hierarchical MMD heterostructures are synthesized in multistep processes by stepwise layered (with ex situ substrate) or concentric (in situ particulate) growth. Although this can be done by the alternating deposition of a MOF-derived MD and a non-MOF-derived MD, [269] the deposition of non-MOF-derived MDs is often non-porous and leads to significant reductions in surface area. However, superior nanostructures can be generated through the conversion of two MOF-derived MDs. Wang et al. [270] thermal converted ZIF-67(Co)/CC to produce nanowire arrays on CC. This system then was used as a substrate for the alternating stepwise cathodic electrodeposition of Ni metal and CoNi 2 S 4 . The Ni metal facilitated charge transfer between active surfaces of CoNi 2 S 4 and the CC electrode in this MD-carbon composite. Hierarchical MMDs also can be achieved by nonuniform stepwise conversion processes. Le et al. [269] immersed ZIF-67(Co)/NF in an Fe 2+ solution for 5 min in order to convert the MOF surface only to an (Fe,Co)LDH. The entire MD/ MOF composite then was solvothermally sulphurated in TAA in order to produce a hierarchical (Fe,Co)S/CoS MD.

Conversion of Hierarchical MMOFs into Hierarchical MMDs
Homogeneous Growth: Rather than growing an MD on the surface of a MOF or another MD, hierarchical MMDs can be synthesized by the direct conversion of hierarchical MMOFs. This approach generally involves the epitaxial growth of one MOF on another. [271] Structurally identical MOFs with alternative metal compositions (viz., ZIF-8(Zn) and ZIF-67(Co)), which utilize identical ligands and are crystallographically similar), can be grown homogenously and epitaxially on one each another to form hierarchical MMOFs. These MMOFs then can be converted to form analogous MMDs. Chen et al. [272] repeatedly and sequentially grew ZIF-8(Zn) and ZIF-67(Co) to achieve four bilayered polyhedral MMOFs, as shown in Figure 17. The authors concluded that, instead of forming a hierarchical C(N,Co)/C(N,Zn) structure, thermal conversion resulted in the complete removal of ZIF-8(Zn) from the MMD, forming C(N,Co)/C(N) hollow shells. The effective removal of Zn occurred owing to vaporization of this low-vapor-pressure metal while retaining the carbon. These observations suggest that the thermal conversion coupled with the Zn vaporization cause deposition of the carbon from ZIF-8(Zn) on the shell of the decomposing ZIF-67(Co), resulting in a layered hollow MMD with hierarchical porosity. Since residual Zn potentially could have beneficial or deleterious effects on, for example, catalysis, the absence of EDS data for Zn to confirm or deny its presence precludes assessment. Examination of the TEM images in Figure 17 suggests that the interlayer regions are not completely Figure 16. MMOF-derived Cu 0.5 Zn 0.5 In 2 S 4 • rGO • gC 3 N 4 synthesis. Adapted with permission. [268] Copyright 2020, Royal Society of Chemistry.
free of residual carbon and/or Zn and that this residuum is responsible for providing the volume to support retention of the concentric shell composite structure.
Heterogeneous Growth: The complexity of hierarchical structures can be increased through the synthesis of heterogeneous hierarchical structures. While heterostructured MOFs have Reproduced with permission. [272] Copyright 2019, American Chemical Society. yet to be used as precursors for heterostructured MMDs, they show significant potential for development. Recent work by Liu et al. [271] demonstrated several different configurations of heterogeneous hierarchical MOF structures. To achieve this, rather than using MOFs of identical ligands but similar metals, [272] dissimilar ligands were used to synthesize layered MOFs. MIL-125(Ti) containing carboxylate-based ligands was used as a substrate for selective growth of amide-based ZIF-8(Cu) on the corners of the MIL-125 tabular structure, as shown in Figure 18.
Owing to their dissimilar crystal structures, growth of the ZIF-8 was limited to the crystallographic planes exposed at these corner sites. ZIF-67(Co) (similar to ZIF-8 but still dissimilar to MIL-125) then was grown on this MMOF to create a third layer. The similar structures of ZIF-67 and ZIF-8 as well as the unavailability of the favorable crystallographic planes of MIL-125 resulted in growth of ZIF-67 only on ZIF-8. The application of conventional solvothermal synthesis in a sequential process using common MOFs provides the capacity to produce heterogenous hierarchical structures and further higher-order complex structures with numerous MOF and MMD layers greater than those achieved for ZIF-67(Co)/ZIF-8(Cu)/MIL-125(Ti).

MOF-Derived Metal-Based Derivatives: Specific Aspects of Individual Types
Many of the procedures to make one type of MOF-derived MD can be applied to produce different types of MD. However, since the principal determinant of the structure of an MD is the conversion process, the differences across the range of MDs are considered in detail, with focus on the unique conversion agents used.  [271] Copyright 2020, The Authors, published by Springer Nature.

Metal Oxides
Metal oxide (MO) nanoparticles are promising materials for many applications, including, inter alia, catalysis, optoelectronics, magnetism, and sensing. [273] The high reactivity of oxygen is responsible for the formation of numerous stable MOs, ranging from simple cubic oxides to complex pyrochlores and perovskites. Additionally, many modifications are possible for MOs as well as composites containing carbon (MO + C) through the application of strategies applied to MOF precursors, including such variables as selection of metal, oxide variants, doping, temperature, atmosphere, defect chemistry, etc. Figure 19 illustrates the synthesis route of MOs from MOF precursors. In the assembly of MOFderived MOs, oxygen can be introduced by three distinct modes: 1) in chemical groups attached to the ligands within the MOF (e.g., carboxylates), 2) in the gas phase during thermal conversion (i.e., air), and 3) in solution as ions during solvothermal conversion. Since most MOFs are constructed from oxygen-containing ligands, oxygen is so readily available in air, and any solution sufficiently rich in hydroxide ions can be used for solvothermally conversion, MOF-derived MOs have the simplest routes relative to those of other MDs. This flexibility has facilitated large-scale research in this approach since the expansion of research in MOF-derived MDs starting in ≈2013, [227] which is summarized extensively in Table 3.

In Situ Oxygen from MOFs
As most MOFs utilize carboxylate groups to link SBUs and ligands, pyrolysis typically produces MO-rich MD-carbon composites. Although the oxygen content in principle may be sufficient for complete internal oxidation of the metal, carbon also reacts with oxygen, which creates an environment conducive to partial reduction of the metal. Although increasing temperatures of thermal conversion in inert atmospheres can increase the extent of internal oxidation of metal, combined metals have been observed to decrease oxidation. Sun et al. [31] used thermal conversion of MOF-74(Ni) under N 2 to synthesize Ni/NiO nanoparticles encapsulated in carbon. It was found that full oxidation of the Ni occurred at ≥550 °C. When MOF-74(Ni) was then physically mixed with MOF-74(Co) and pyrolyzed together, no change to the oxidation state of Ni was found. However, when an MMOF synthesised from an equimolar metal precursor solution, viz., MOF-74(Ni,Co), was pyrolyzed, it was observed that the MMD was not completely oxidized, as seen from the retained metallic Ni. This outcome suggests that 1) the activation energy for oxidation of the Ni-Co solid solution [299] was increased, thereby improving the oxidation resistance or, 2) the Co preferentially oxidized, forming a passivating layer similar to the effect of Co doping in stainless steel. [300] These works also indicate the importance of homogeneity, where the atomic-scale mixing, associated minimal diffusion distance, and equilibration deriving from the use of precursor solutions are advantageous compared to those from mechanical mixing.
Even up to conversion temperatures of 900 °C, Razaee et al. [295] demonstrated the presence of metallic Ni and Co in an MD-carbon composite. However, in contrast to the work by Sun et al., [31] the MMOF synthesized by the former authors lacked the two hydroxyl groups that were present on the same ligand used by the latter authors. These results indicate the importance of the oxygen content of the MOF as this is a direct cause of internal oxidation. Consequently, these data emphasize the importance of choice of ligand in MOF synthesis and the potential to blend ligands to tune the extent of oxidation.   The preceding studies suggest three generalizations that can be used to increase the oxidation resistance MDs: 1) A higher onset temperature of oxidation can result from the alloy system of MMDs owing to higher activation energy, although this will depend on the metal combination selected. 2) The extent of oxidation of MDs can be reduced through the use of ligands of low or nil oxygen content, thereby reducing internal oxidation.
3) The extent of oxidation can be reduced through the application of lower pyrolysis temperatures.

Ex Situ Oxygen from Air
Owing to its simplicity, thermal conversion in air is the most common synthesis method for MOs from MOFs and MDs.
The high reactivity of oxygen can enhance interparticle chemical bonding and hence hard agglomerate formation, as opposed to physically bonded agglomerates. In order to avoid such agglomeration of MOs, low conversion temperatures generally are used. Wu et al. [156] synthesized Co-MIM nanospheres and thermally converted them in air at 400 °C for 30 min to produce Co 3 O 4 hollow nanospheres with high BET surface area of 124 m 2 g −1 . However, this low oxidation temperature resulted in the MO's relatively low crystallinity. Although the use of high temperatures can overcome the latter problem, there is an alternative strategy. At the initial conversion of a MOF into an MD, the morphology is most susceptible to degradation, so pyrolysis is used to preclude oxidation of the carbon and associated damage. In the absence of sufficient internal oxidation, then an external The following acronyms were used to summarize the above information: Carbon dioxide adsorption(CO 2 Ads), supercapacitor (SC), lithium-ion battery (LIB), Oxone-assisted salicylic acid degradation (Oxo-SAD), methylene blue degradation (MBD), sodium-ion battery (SIB), photocatalytic hydrogen evolution reaction (P-HER), oxygen evolution reaction (OER), carbon monoxide oxidation (CO Ox), zinc-air battery (ZAB), lithium-sulfur battery (LSB), photocatalytic oxygen evolution reaction (P-OER), methanol oxidation (Meth Ox).
Adv. Mater. 2023, 35, 2210166 source, viz., air, must be used. Consequently, the extent of change in structural chemistry upon direct oxidation of the MD to form the MO is minimal, thereby retaining the original morphology while achieving high crystallinity using this multistep conversion process. Liu et al. [52] produced thermally converted MIL53(Al) in N 2 at 550 °C and DUT-5(Al) at 700 °C, followed by oxidation in air at 650 °C, thus synthesizing Al 2 O 3 nanosheets and nanorolls, respectively. The initial pyrolysis allowed carbon to stabilize the morphology of the MOF during conversion; further conversion in air removed the carbon, fully oxidized the metal, and increased the crystallinity. Despite the relatively high conversion temperatures, which are ≈150-300°C above typical, they did not result in agglomeration, which can be attributed to the high thermal stability of alumina. [301] In cases when an ex situ oxygen source is required, but carbon needs to be retained, a two-step process of selective oxidation must be used to produce MOs. As before, the precursor MOF is pyrolyzed initially in order to retain both the carbon and the MOF morphology. In the second step, the MD-carbon composite is heated in air at <350 °C, where the MD is oxidized but not the carbon. While this has been demonstrated for Co 3 O 4 -carbon composites, [283] as shown in Figure 20, it also is applicable to other MO-carbon composites. [152] Depending on the temperature used (≈150-250°C), the time required for complete oxidation of the metal ranges from 1 to 12 h, [175] although the carbon is retained in the structure. Partial oxidation can be achieved by prematurely terminating this oxidation step. [152]

Oxygen from Solution
While thermal conversion into MOs by internal and external oxidation is common, solvothermal conversion offers some unique advantages over these. The process uses lower temperatures, ranging from room temperature to that of the limitation of autoclaves (≈220 °C). These minimize both the temperature used for conversion of the MOF to MD and potential agglomeration during conversion. Both of these facilitate greater retention of precursor nanostructures, even including nanosheets of single unit cell thickness. [20] While the solvothermal conversion of other MDs is limited by the range of potential reactant bases, MOs in principle can be produced from any oxygen-containing salt, (e.g., carboxyl, acetate, etc.) of sufficient concentration. [20] However, at present, the only successful salt for this process is hydroxide, and no clear advantage from the use of different hydroxides has been observed (NaOH, [58] KOH, [40] and TEAOH [275] ). Bases of higher concentration, typically ≥2 m, or of higher strength generally allow the use of lower conversion temperatures ranging from 25 to 180 °C.

Metal Sulfides (MSs)
The research breadth of MOF-derived metal sulfides (MSs) is similar to that of MOs, with a focus on optimizing surface area and resultant activity. These MSs are typically paired with conductive substrates or synthesized as composites (MS + C) Figure 20. a) Schematic of multistep conversion process from MOF to C(MO) dispersion, with b,c) corresponding SEM images (b) and TEM images (c). Adapted with permission. [283] Copyright 2017, Wiley-VCH.
because they exhibit excellent electrochemical [302] and catalytic [303] performances. The general synthesis routes used for MOF-derived MSs and their composites are shown in Figure 21. In the assembly of MOF-derived MSs, sulfur can be introduced by three distinct modes: 1) In chemical groups attached to the ligands within the MOF (e.g., sulfonates), 2) in the vapor state during thermal conversion, and 3) in solution as ions during solvothermal conversion. The scope of synthesis routes is as broad as that for MOs owing to the high reactivity of sulfur in its different forms. An extensive summary of the MSs produced from precursor MOFs is given in Table 4.

Sulfur from MOFs
Although research in sulfonate-based ligands is limited, recent work has demonstrated that ligands can provide a sufficient source of sulfur during pyrolysis of MOFs into MScarbon composites. Chen et al. [32] used six different metals to produce MOFs, which were converted to MSs by pyrolysis; the MOFs contained both BPI (non-sulfur-containing ligand) and ARM (sulfur-containing ligand). Although the molar ratio of metal to sulfur in the MOF synthesis was identical in all six cases, this ratio in the MSs varied significantly. This outcome suggests an origin in: 1) variable reactivity of the SBU with the ligand(s) in the MOF, 2) variable reactivity of the metal with the sulfur in the MS, 3) dissolution of unreacted sulfur in carbon, and 4) loss of sulfur through vaporization. Of these, the first is poorly understood, the second has been identified and is well understood, the third is difficult to determine, and the fourth is determined easily with characterization of the exhaust gases. These uncertainties have hindered the development of MSs synthesized entirely from MOFs despite their potential.

Vaporized Sulfur
The thermal conversion of MOFs in the presence of vaporized sulfur powder is the most common MS synthesis method. The low cost of sulfur has promoted the use of excess sulfur to ensure complete conversion. However, this also may lead to carbon doping, potentially improving the catalytic performance. The disadvantage of excess sulfur is the common outcome of the formation of multiphase MS mixtures, e.g., MS + MS 2 . In order to avoid this potential shortcoming, Ahn et al. [53] carefully synthesized phase-pure CoS 2 porous nanoparticles through sulfuration of a PBA. In order to ensure the absence of Co, Co 9 S 8 , and S impurities, careful control of the sulfuration process was required. This was done through the use of relevant binary and ternary phase diagrams to guide the control of the proportions of vaporized sulfur and precursor MOF.

Sulfur from Solution
Sulfur deposition in the cold zones of furnaces and the low reactivity of some metals and MDs with sulfur are potential problems associated with the use of vaporized sulfur. These issues can be avoided with the use of solvothermal conversion. This generally is performed in solutions containing sulfur ions that react with the SBUs, converting them into MSs and degrading the organics. As solvothermal temperatures are insufficient for carbonization, MOF-derived MSs converted this way are always inherently carbon free. However, carbon structures are stable in the sulfurizing agents; therefore, the existing morphology is maintained during conversion while also being doped with sulfur. Thus, MOFs thermally converted to produce MD-carbon composites then can be converted solvothermally to MS-carbon composites. One of the first solvothermal conversions of MOF-derived MDs into MSs utilized aqueous sodium sulfide  solution; however, almost all research since then has utilized TAA. TAA is an organo-sulfur compound that is highly soluble in ethanol and releases the active sulfur ions only at elevated temperatures (>90 °C). Most solvothermal conversions performed with TAA are completed over long periods of time (>10 h) but work by Nagaraju et al. [192] showed that the morphologies of some structures, such as hollow nanorods, are sensitive to conversion times. These authors found that sulfuration for 1 h provided insufficient MOF etching and MS regrowth time, while 5 h led to structural collapse caused by overgrowth. This time frame correlates with work by Guan et al., [41] which indicated that solvothermal treatment for 3 h was optimal. Despite the great stability of MDs such as oxides, such approaches potentially allow the synthesis of a wide range of MSs from stable MDs.

Metal Phosphides (MPs)
Many metal phosphides (MPs) have metallic properties that make them particularly useful in HDN, HDS, and HDO applications; they also have demonstrated potential in many other applications. [318][319][320] The general synthesis routes available to MOF-derived MPs and their composites (MP + C) are shown The following acronyms were used to summarize the above information: Dye-sensitized solar cell (DSSC), supercapacitor (SC), hydrogen evolution reaction (HER), carbon dioxide reduction (CO 2 Red), oxygen evolution reaction (OER), lithium-ion battery (LIB), lithium-sulfur battery (LSB), sodium-ion battery (SIB), hybrid magnesium-and lithium-ion battery (Mg/LIB), apparent quantum efficiency (AQE), zinc-air battery (ZAB). Mo-dopamine* does not consist of coordination bonds and may not qualify as a MOF but displays many of the same properties as a precursor and thus is relevant.  Figure 22. While it is possible for phosphorus to be introduced using chemical groups attached to the ligands within the MOF, [165] this has not been demonstrated for MOF-derived MPs. Additionally, the solvothermal conversion of MOFs into MPs appears to have been demonstrated only once, which involved the use of phytic acid. [209] Thus, phosphidation is achieved almost entirely through thermal conversion in an atmosphere containing reactive P-containing species or vaporized P. An extensive list of MPs produced from MOFs is given in Table 5.

Vaporized Phosphorus
The least complex method of producing MPs involves the thermal conversion of MOF in atmospheres of vaporized phosphorus. The approach was used by Xia et al. [48] and involved thermal conversion of a mixture of ZIF-67(Co) and red phosphorus (RP) in a 2:1 weight ratio in N 2 at 800 °C for 2 h. At this temperature, RP is vaporized completely and so reacts with the Co to form the MP. This process could be expanded by control of the principal variables, viz., composition and temperature. The authors explored the latter and observed that CoP, CoP + Co 2 P, and Co 2 P were formed at 700, 800, and 900 °C, respectively. In such an approach, the relevant considerations include batch ratios, particle sizes, mixing versus separate sources of raw materials, temperature ranges of stability of the phosphides, kinetics of phosphidation of the different phosphides, and maintenance of phosphorus gas throughout the process. Although most studies involved isolation of the sources of metal and phosphorus, another study of mixed such sources is that of Li et al. [342] As shown in Figure 23, this study involved ball milling of the RP, dispersion in a solution containing Fe 3+ , mixing with a suitable ligand to form an Fe-MOF gel with finely dispersed RP, and pyrolysis below the sublimation temperature of the RP (≈450 °C) [346] to produce C(FeP x • P). The use of the relatively low temperature resulted in only partial vaporization of the RP, which yielded partial phosphidation of the Fe, generation of a phosphorus-deficient Fe phosphide (FeP x ), voids, and residual phosphorus. These microstructural features were proposed to improve penetration of electrolytes and shorten the diffusion path for Li + insertion. This work appears to be the only study in which macroscopic voids were generated from removing raw materials.

Decomposed Phosphorus Sources
Given the prevalence of sulfur powder as a sulfur source for MS synthesis, RP could be inferred as the most common phosphidation agent for MOF-derived MPs. However, there are two key advantages of sodium hypophosphite (NaH 2 PO 2 ) that position it as the primary source material for phosphorus compared to RP. First, NaH 2 PO 2 initiates decomposition at a much lower temperature of ≈200 °C [347] according to Equation (1), producing vaporized phosphine (PH 3 ), which reacts readily with MOFs to form MPs.
2NaH PO (s) Na HPO (g) PH (g) Second, the reactive PH 3 species is a strong phosphidation agent at relatively low temperatures (<500 °C), allowing costeffective conversion for MOF-derived MP synthesis, as shown in Figure 24. Although the role of Na 2 HPO 4 is unknown, its apparently greater stability suggests that it is not active. Further, the hydrogen of PH 3 can act as a reducing agent, thus increasing the reactivity through removal of passivating oxygen. Phosphoric acid (H 3 PO 4 ) also has been used in thermal phosphidation to generate PH 3 , albeit with no obvious differences between NaH 2 PO 2 . [18] Although most studies involved isolation of the sources of metal and phosphorus, the alternative of mixing the raw materials appears to have the capacity to affect the morphology. In what appears to be a unique study, Kavian et al. [332] mixed NiCo-BDC and NaH 2 PO 2 raw materials, pyrolyzed them in Ar at 350 °C for 3 h, and observed that, in addition to the majority equiaxed MOFs, nanowires formed. Although the composition and structure of the nanowires were not disclosed, the authors suggested that this morphology may have originated from the liquefaction of NaH 2 PO 2 ; even so, it is clear    The following acronyms were used to summarize the above information: Hydrogen evolution reaction (HER), oxygen evolution reaction (OER), lithium-sulfur battery (LSB), supercapacitor (SC), lithium-ion battery (LIB), photocatalytic hydrogen evolution reaction (P-HER), sodium-ion battery (SIB), photocatalytic oxygen evolution reaction (P-OER). Adapted with permission. [342] Copyright 2018, Elsevier.
that the growth process would have involved the solution-precipitation of the nanowires. This conclusion suggests the potential to generate specific euhedral morphologies through the use of such reactive liquids but also through the addition of non-reactive liquids, which would serve purely as diffusion media. However, in the case of such liquids, species other than phosphorus may be important. That is, they may lead to chemical contamination from Na [246] and microstructural alteration from residue to lower the surface area, [341] which are potential disadvantages.

Metal Nitrides (MNs)
Metal nitrides (MNs) can provide catalytic and electrochemical properties superior to their oxide counterparts owing to their higher electrical conductivities, although they may be less stable and more difficult to form from precursor MOFs. [348] The general synthesis routes for MOF-derived MNs and their composites (MN + C) are shown in Figure 25 and the known studies of the synthesis of MOF-derived MNs are summarized in Table 6. There are three general methods to supply nitrogen to produce MNs from precursor MOFs: 1) from within the ligand itself (e.g., imidazoles), 2) as a reactive nitrogen-based gas (e.g., ammonia or NH 3 ), and 3) as a solid that undergoes decomposition into reactive-nitrogen-containing species during thermal conversion (e.g., DCD). Thermal conversion under pure N 2 does not result in the formation of MNs owing to its low reactivity with the MOF or MD components at practicable processing temperatures. Consequently, N 2 is typically used to produce oxygen-free environments for thermal conversions of Figure 24. a) Schematic of synthesis process of CoP • C(N)/CoO/NF by phosphidation of ZIF67 using sodium hypophosphite (NaH 2 PO 2 ), with b-d) corresponding SEM images. Adapted with permission. [55] Copyright 2020, Pergamon Press. MDs at reduced costs relative to the use of the alternative inert gas Ar.

Nitrogenous Gases
Owing to the difficulty in nitriding of MOFs with M-O bonds, oxygen-free MOFs are used commonly; these include imidazole-based MOFs (e.g., ZIF-67 and ZIF-8) and PBAs. However, when using other MDs (i.e., MOs or LDHs) for thermal conversion to MNs, strong nitriding gases must be used to overcome the M-O bonds. These gases invariably contain H 2 , which can reduce and volatilize the passivating oxygen layer as well as cause metal ion reduction. NH 3 is used most commonly owing to its accessibility and effectiveness. Feng et al. [39] calcined MOF-74(Co,Ni) at 400 °C for 2 h to generate an MO before nitriding in NH 3 at 500 °C to synthesize Co x Ni y N. An alternative to NH 3 gas is DCD, which has been used a number of times for N doping. [353] The thermal decomposition of DCD is reported to commence at ≈250 °C, producing principally NH 3 and hydrogen cyanide (HCN) gases, both of which act as nitriding agents. [353] However, DCD has yet to be used to produce a MOF-derived MN.

Parallel Nitrogen Sources
Unstable MN species can benefit from the use of multiple different nitrogen sources during thermal conversion. The most common approach is the use of both an imidazole-based MOF and NH 3 for MN synthesis. Liu et al. [51] synthesized N-containing V-MOF nanospheres that were thermally treated in N 2 / NH 3 (40/60 v/v) at 650 °C to produce VN nanospheres. The authors identified an atomic-thickness passivating surface layer of VO 2 , which protected the bulk of the nanospheres from further oxidation, thereby enabling assessment of the properties of the VN nanoparticles. Excess nitrogen is dissolved within the carbon lattice as secondary dopants due to the use of parallel nitrogen sources to ensure complete nitridation. As discussed in Section 2.3.2, nitrogen dissolution has been shown to improve electrical properties of MN-carbon composites. [350]

Stable Nitrides
Metals that form more stable nitrides, such as Mn, typically do not require parallel nitrogen sources during conversion and their extents of surface oxidation after thermal conversion are reduced. [354] . Hu et al. [33] synthesized Mn 2 N x from thermal conversion of MET-2 in N 2 . Although this gas may have assisted in the nitriding process, repeated thermal conversion in Ar produced similar results. Thus, the nitrogen present in the precursor MOF was sufficient for nitridation. As specific nitrogen contents can be sourced entirely from the MOF itself, individual nitride phases can be targeted by controlling the conversion temperature. The authors determined that heating at 525 °C for 4 h produced crystalline Mn 2 N 0.86 . The following acronyms were used to summarize the above information: Supercapacitor (SC), oxygen evolution reaction (OER), hydrogen evolution reaction (HER), photocatalytic hydrogen evolution reaction (P-HER), oxygen reduction reaction (ORR).

Metal Carbides (MCs)
Transition MCs have demonstrated catalytic properties similar to those of Pt-group metals, even when present as relatively large particles (1-3 µm). [355] The basic synthesis route for MOFderived MC-carbon composites (MC + C) is shown in Figure 26. Since solvothermal conversion occurs at temperatures insufficient for carbonization, the thermal conversion of MOFs into MCs is the only possible method at present. While abundant carbon for MC synthesis always is present through the ligands of the MOF, the excess carbon over that required for MC synthesis is extremely difficult to remove while retaining the MC. Consequently, the resultant MDs are MC-carbon composites.
While the presence of carbon in all MOFs suggests that MCs might be present in all MDs, this is not the case because many metals do not readily form stable carbides and others do not form carbides at all. [356] Even when an MC is formed, it can be difficult to distinguish its presence from metal-carbon composites. In such cases, XPS and XRD are the principal analytical methods for confirmation of the natures of the respective bonding and structure. A comprehensive list of MCs produced from MOFs is given in Table 7.

Promotion of Carbide Formation
The formation of bonds between metal and carbon typically requires high thermal conversion temperatures (>600 °C). Further, as the presence of oxygen is a significant hindrance to MC growth, a reducing agent is typically used. This usually is external H 2, but solids that decompose to produce reducing agents also are used. Jia et al. [34] demonstrated the synthesis of an MC using a synthesis route similar to that used by others [156] to synthesize the analogue MO by the strategy of including H 2 (5 vol%) in the pyrolysis atmosphere. Kong et al. [50] used solid MAE (C 3 H 6 N 6 ) to produce an N-doped Co 3 InC 0.75 and carbon composite. Since the gaseous decomposition products of MAE Figure 26. General method to produce MC-carbon composites from precursor MOFs. The following acronyms were used to summarize the above information: Oxygen evolution reaction (OER), oxygen reduction reaction (ORR), hydrogen evolution reaction (HER), photocatalytic oxygen evolution reaction (P-OER), lithium-ion battery (LIB).  [361] then the former acted as dopant and the latter two acted as reducing agents.
Of the metals that readily form MOF-derived carbides, Mo appears to be the only metal examined to date. Such MCs can be synthesized without reducing agents [232] and at relatively low temperatures. [45] The difference in conditions required for MC synthesis between Mo and other metals can be leveraged to produce MC(Mo)-MD composite in a single step. This advantage is the simple formation of heterojunctions formed in situ, maximising interfacial surface area; potentially promoting catalytic performance. However, as addition of non-metals may impede MC(Mo) formation, the paired MD is generally limited to TMs but may be include MOs, MSs, and MNs, where the non-metal is sourced only from the MOF. [357] Unreacted metal can be identified by XPS, XRD, and HRTEM. [34,119] Although MCs always can be synthesized solely from the carbon constituting the ligands, MD-carbon composites using in situ carbon may be subject to insufficiency. Consequently, the addition of ex situ carbon from conventional carbonizing agents, such as coking coal, would be required. However, such agents typically require temperatures (>1000 °C) above those of typical thermal conversion (≈400-900 °C). This limitation raises the possibility that reactant gases, such as CH 4 , may be used not only as sources of carbon but also as simultaneous reducing agents. Further, a significant advantage is the relatively low cracking temperature of 545 °C [362] for CH 4 , which would reduce the tendency to agglomerate. More importantly, these attributes offer the potential advantage of MC synthesis at temperature less than those used to date (600-1600 °C). Even lower carbiding temperatures may be achieved through solid and liquid hydrocarbon feedstocks, similar to work shown for CVD. [363]

MXenes
Closely related to 2D MCs are MXenes, which are 2D layered carbides, nitrides, or carbonitrides. [364] Since most MCs are produced by high-temperature pyrolysis, the particle size and morphology are difficult to control. Consequently, the lower temperatures used to synthesize some MXenes, e.g., those derived from acid etching at room temperature, [365] offer an alternative to pyrolysis. Since MOFs have been synthesized from MXenes, [366] then the reverse process of synthesizing MXenes from 2D MOFs may be possible.

Metal Oxycarbides (MOCs)
A subcategory of MCs is metal oxycarbides (MOCs), which involve oxygen ion dissolution in the carbide structure. Oxycarbides derived from MDs typically are high-density and amorphous, the former of which hinders synthesis from low-density MOFs at low temperatures and the latter of which hinders characterization. Since to date, no MOCs have been produced with BET surface areas >10 m 2 g −1 , then this disadvantage limits their use in catalytic and energy conversion and storage applications despite their suitability. Even so, Xiu et al. [151] obtained a promising lithium-ion battery (LIB) performance of 793 mA h g −1 at 0.1 A g −1 with a surface area of only 3 m 2 g −1 . The ultrahigh synthesis temperatures are typically required to represent another technological challenge. David et al. [360] found that crystalline carbide and oxycarbide phases develop only at very high temperatures above which MOs are not stable. Figure 27 shows that recrystallization of MOC commenced at 1400 °C and required 1500 °C for full development. Other work [290] has proposed oxycarbide synthesis at the low temperature of 350 °C, but this material appeared to be an MO-carbon composite.

Transition Metals (TMs)
It is well established that the particle size and morphology can be tuned to alter the activities of TMs. [367] MOF-derived synthesis enables TMs and TM-carbon composites with highly controlled compositions, nanostructures, sizes, and morphologies. Excluding the few examples of pure TMs, the conversion processes are essentially identical to those used for MC synthesis. The general synthesis routes for TMs and their composites (TM + C) are shown in Figure 28. The sole or primary source of metals for the TMs is the SBUs of the MOFs; this can be supplemented by metals from dopants in the MOF and in solution during solvothermal synthesis, as described in Section 2.6. Since TMs often are found in the presence of other MDs, these specific materials are difficult to categorize, then Table 8 provides an extensive list of the MDs that are most relevant to TMs.

Pure TMs
As TMs exist as pure metals, most are highly reactive and relatively unstable. Consequently, TMs generally are encapsulated by other MDs (typically MOs) or carbon shells. As mentioned in Section 2.5, metals that form carbides can form close-fitting carbon shells that stabilize the TMs. Hu et al. [44] produced nanospheres of Fe-Co solid solution encapsulated in N-doped graphene layers after pyrolysis of PBA in N 2 at 900 °C for 4 h. Highly stable noble metals, such as Pd [36] and Ag, [100] are stable in air and so do not require encapsulation. In such cases, carbon-free TMs can be produced by electrolytic reduction of a MOF [100] or MOF-derived MO, [36] which avoids carbonization altogether or eliminates carbon through oxidation, respectively. As shown in Figure 29, after oxidation to form a mixture of PdO and NiO, the latter was removed by dissolution in H 2 SO 4 . The residual PdO was reduced electrochemically to form metallic Pd of the same but more porous morphology as the precursor MOF and MMO. This appears to be the only report of MOderived TM synthesis, which probably is the case owing to the importance of the avoidance of both in situ and ex situ oxygen.

TMs for Anchoring Doping
Although the high reactivity of TMs can complicate synthesis, this characteristic also can generate some unique features. When TMs are synthesized directly from MOFs, they often exist as nanoparticles (≈5 nm in diameter) [186] in TM-carbon composites. These unreacted metallic nanoparticles are finely dispersed throughout the carbon matrix and can be used as anchor sites for doping. Other species that interact selectively with existing metal sites can be mixed with TM-carbon composites to form, for example, highly distributed intermetallics that otherwise cannot be synthesized directly. Wang et al. [378] pyrolyzed Co-doped ZIF-8(Zn) in N 2 at 900 °C for 1 h. Following removal of the Zn by thermal evaporation, the resultant carbon matrix contained finely dispersed Co nanoparticles that, when loaded with Pt using an ethylene glycol reduction method, produced finely dispersed the CoPt 3 intermetallic in a carbon matrix. It is likely that the highly dispersed Co dopant in the SBU was responsible for the ultrafine size (5 nm) and unagglomerated state of the Co nanoparticles, which are unlikely to have been feasible through the synthesis of a pure Co-MOF. The advantages of superior dispersion and its effect on performance have been observed by others. [378]

Other MDs
Although there are many other types of MDs that may be synthesized from MOF precursors, they have not received as much attention as have other MDs, including MOs, MSs, and MPs. These lesser-known MDs include those synthesized from heavier anions and mixtures of non-metals. A summary of these MDs is given in Table 9.

Metal Borides (MBs)
As with most uncommon MDs, MOF-derived metal borides (MBs) only recently have been synthesized. They show considerable promise in energy conversion and storage applications. [409][410][411][412] At present, there appear to be only three MOF-derived MBs that have been synthesized. [384][385][386] The sources of boron are solvent-based H 3 BO 3 and NaBH 4 , the latter of which has been used for other MD conversions as it is a strong reducing agent capable of dissolving non-metals in solution; it also has been used for boron doping.

Metal Fluorides
Much less common than other MDs, metal fluorides (MFs) also are a relatively recent subject of research. Its potential in energy generation derives largely from it having the highest electronegativity of all elements, so its dissociation is very easy in alkaline electrolytes, thus benefitting the adsorption of important species during OER. [387] Its potential in energy storage is owing to its very high theoretical capacity (≈600 mA h g −1 ). [389,390] Currently, there appear to be only four different MOF-derived MFs that have been synthesized, although these have involved three different fluorination techniques. Further, complete fluorination occurs during thermal conversion at temperatures as low as 250 °C owing to the reactivity of F. [388] Solvothermal conversion using hexafluorosilicic acid (H 2 SiF 6 ) has been demonstrated and found to be significantly safer than with the use of hydrofluoric acid (HF), although it has the disadvantage of producing impurities such as MeSiF 6 . [389] Ammonium fluoride (NH 4 F) is the only known fluoride source for thermal conversion but its fluorination mechanism is similar to that of solvothermal conversion using HF as it thermally decomposes according to Equation (2). Tan et al. [390] compared MF synthesis by thermal conversion (NH 4 F) and vapor solvothermal  synthesis (HF) and found that, even though similar reactive species were fluorinating the MD, the required higher temperature for the former led to less retention of the morphology and thus poorer energy-storage performance.

Metal Selenides
Similar to MSs, metal selenides (MSEs) are stable owing to the high reactivity of Se and can be synthesized by mixing precursor MOFs with volatile Se powder and heating. The simplicity of using volatile Se powder has led to the selenisation by the hot gas method for almost all known MSEs synthesized from MOFs. Solvo-selenisation is possible but involves the ionization of Se in strong bases, such as NaBH 4 [394] or NaOH, [391] which produce active Se 2− ions that convert the MDs into MSEs.

Metal Tellurides
Of the chalcogenides, metal tellurides (MTs) have small band gaps (E g ≤ 1.2 eV) owing to the large ionic radius and broadening of the d-bands in Te. [413] MTs also typically exhibit more structural variations and associated variant electronic properties. [414] Crystals of WTe 2 have demonstrated large non-saturating magnetoresistance [415] and MoTe 2 thin films have exhibited ohmic homojunctions. [161,162,416] Similar to other chalcogenides synthesized from MOFs, MTs are synthesized easily using thermal conversion with vaporized Te. In common with Se, the boiling point of Te is well above normal thermal conversions, so reduced vapor pressures enable volatilization. [417] Solvothermal conversion of MOFs into MTs appears to have been demonstrated only using K 2 TeO 3 . However, since this was achieved at room temperature, this suggests it has the potential to tellurize most, if not all, stable MDs.

Layered Double Hydroxides
The solvothermal conversion of MOFs in metal-rich solutions generally produces LDHs as intermediates for the synthesis of carbon-free MDs. Dissimilar metals in the MOF and solution can be used to achieve MMDs. However, LDHs themselves have shown considerable promise for a diverse range of applications owing to their highly active 2D structure. [418] Although the synthesis of LDHs from MOFs is not well understood because the mechanisms that occur during conversion are complex, recent The following acronyms were used to summarize the above information: Carbon dioxide reduction (CO 2 Red), reduction of 4-nitrophenol (Red-4), oxygen reduction reaction (ORR), zinc-air battery (ZAB), oxygen evolution reaction (OER), hydrogen evolution reaction (HER), lithium-ion battery (LIB), lithium-sulfur battery (LSB). Figure 29. Synthesis route for porous Pd nanosheets from MOF precursor. Adapted with permission. [36] Copyright 2019, Elsevier.  in situ TEM analysis has revealed that there are effectively two competing processes that occur. [417] The first is the decomposition of the MOF, which is degraded by metal-ion intercalation from the outer surface-inward. The second is the reconstruction of the LDH from the degraded MOF. As the latter occurs almost instantaneously, the LDH is established as a stacked 2D artifact of the MOF morphology. The rate of each competing mechanism can be adjusted by controlling the salt concentration, which allows manipulation of the MOF morphology. Further, LDHs can be produced from hydrolyzation reactions of MOFs in water-containing solutions. Zhang et al. [56] solvothermally converted MIL-88A(Fe) in a Ni-rich solution of ethanol and water. By carefully increasing the volumetric ratio of ethanol to water from 2:3 to 3:2, the rate of solvothermal conversion could be slowed and multilayered morphologies could be generated from the precursor 3D MOF; as shown in Figure 30.

Metal Phosphates
Phosphate and oxyphosphate nanomaterials are used only rarely for catalytic applications but they can be useful in radioactive environments. The phosphoryl present in both types of MPOs and the hydroxyl groups present in the oxyphosphates are effective in the adsorption of radionuclides. Abney et al. [58] fabricated phosphate and oxyphosphate MPOs using solvothermal synthesis by converting MIL-125(Ti) and UiO-66(Zr) with respective phosphoric acid concentrations of 1 m and 0.21 mm, respectively. It was proposed that the phosphate and oxyphosphate groups exchanged with the ligands of the MOFs to create a new covalent 3D network, albeit completely amorphous. MPOs can be analysed by FTIR to confirm the presence of Me−OH, P−O, and P−O−H stretching peaks. The following acronyms were used to summarize the above information: Oxygen evolution reaction (OER), supercapacitor (SC), asymmetric supercapacitor (ASC), sodiumion battery (SIB), lithium-ion battery (LIB), aluminum-ion battery (AIB), hydrogen evolution reaction (HER), ethylene glycol oxidation reaction (EGOR). Mo-dopamine* does not consist of coordination bonds and may not qualify as a MOF but displays many of the same properties as a precursor and thus is relevant.

Summary and Outlook
With the development of MOFs over the last two decades, the increase in research in MDs has been exponential, resulting in numerous articles in the last half decade. The unique conversion method from MOF to MD enables the large-scale bottomup production of nanostructurally complex MDs that are difficult, if not impossible, to replicate by other means. Additionally, MOF-derived MDs exhibit three principal advantages in terms of the materials and their design: ○ Metal Variety: Current research generally focusses on the conversion of MOFs to produce Mn-, Fe-, Co-, Ni-, Cu-, Zn-, and Mo-based MDs. However, the scope of the applied synthesis methods is such that it has the potential be used to generate numerous other MDs based on other metals, as projected in Figure 4. Additionally, these MOFs and MDs can be converted using a wide range of different conversion methods, as conceptualized in Figure 31, to produce new MDs, including pure phases, solid solutions, and heterojunctions with applications in catalysis and other functionalities. ○ Composite Structures: While the principal foci of current research are carbon-based MOF-derived materials, an emerging area is the synthesis of composites using MDs to catalyze the formation of carbon nanostructures, as in carbon nanotubes and graphene shells using in situ conversions of MDs. A second emerging area is the synthesis of composites composed of MD/MD solid solutions or heterojunctions with or without carbon matrices. ○ Universal Substrate Compatibility : Since MOFs preferentially nucleate non-epitaxially on surfaces and so precipitate readily from solution, then there are many avenues for the homogeneous formation of MOFs, regardless of substrate chemistry, reactivity, electrical conductivity, topography, and shape.
Even non-adhesive surfaces such as Teflon are coated readily with MOFs during solvothermal synthesis. This approach offers an alternative to conventional methods of deposition of coatings in that the conversion of MOFs into MDs on the substrate results in retention of the contact between MD and substrate. This is particularly advantageous in electrochemical applications that require conductive substrates, such as HER, OER, and electrodes in batteries, which thus eliminates the necessity of binding agents.
The extensive range of synthesis methods for MDs facilitates equivalently large ranges of these materials and their associated applications. Since the majority of research focuses on the optimization of thermal (pyrolysis and hot-gas synthesis) and solvothermal conversion methods, the biggest hindrance to achieving intrinsic performance of nanoparticulate MDs is agglomeration during conversion, the principal affecting factors (exclusive of interfacial stresses with substrates) of which are as follows: ○ Composition: A fundamental and specific factor affecting MD agglomeration is the inherent thermal stability of the MD or composite. Although monodisperse MDs of high structural stability and surface area are feasible, as in the case of Al 2 O 3 , most materials require intervention in order to prevent agglomeration. The most common such method by far is coprecipitation of carbon along with the MD. This carbonization is effected by thermal conversion under non-oxidizing conditions at temperature >300 °C. Even with the implied potential for agglomeration during heating, this process actually can increase the surface area, even up to ≈1000 °C. Although this strategy has the benefit of suppressing agglomeration, it has the disadvantage that the application of elevated temperatures enhances the grain growth of the MDs. ○ Conversion Temperature: An applied but universal factor affecting MD agglomeration is control of the conversion Figure 30. Synthesis route for controlled etching of MIL-88A(Fe) and subsequent growth of single-and multilayer NiFeLDH. Adapted with permission. [56] Copyright 2020, Wiley-VCH.
parameters to limit or prevent agglomeration. As previously discussed, increasing conversion temperatures increases diffusion and hence increases agglomeration. This can be effected by solvothermal conversion, which limits conversion to low temperatures (<200 °C). However, following dissolution of the MOF, precipitation of the MD may result in unanticipated and potentially unfavorable structural modifications. However, the use of such low temperature limits the range of feasible MD compositions but it also precludes tempera-tures sufficient to effect carbonization. A compromise strategy is low-temperature thermal conversion (<400 °C), which typically produces MDs with low levels of agglomeration but with most carbon present in amorphous form leading to lowsurface-areas. These limitations can be mitigated by some MDs that catalyze graphitization at temperatures much lower (≈300-400 °C) than normal (>900 °C), although the graphene is limited to proximity of the MD surface. As the properties of the MD and carbon are optimized at different temperatures, one possibility to address this issue is the application of a single compromise temperature. A second more novel approach is the use of a selective heating method, such as microwave irradiation, which couples with graphene but not the MD. The temperature of the graphite would be lower than that in the MD as the latter's heating would depend on conductivity. Even greater control of the process could be achieved by using the microwave heating for progressive recrystallization of the graphene, where amorphous graphene would not couple as effectively as crystalline. ○ Conversion Time: In the preceding cases, the assumption of the priority of composition and temperature represents a simplification in that the kinetics of the process also depend on the time. That is, the effects of composition and temperature can be refined through moderation of the time applied for heating. This results from the well known dependence of diffusion kinetics on time and temperature for specific materials. One such approach is sequential heating that targets the specific time-temperature conditions conducive to independent conversion of the MD and carbonization. A more novel second approach is the use of rapid heating to cause conversion while suppressing grain growth; this strategy also could be applied subsequently to carbonization.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.