Keywords

2.1 Introduction

Metal–organic frameworks (MOFs) are a family of crystalline porous materials, where their framework structures are constructed by the coordination interactions between metal ion/cluster nodes and organic linkers. Thanks to their unique structural and chemical properties, MOFs have been under intensive studies for their applications in modern energy and environmental technologies, such as fuel cells [1], metal-air batteries [2], water electrolysis [3], supercapacitors [4], hydrogen storage [5]. However, the utilisation of MOFs in the above-mentioned energy and environmental technologies is not limited to their pristine forms. After thermal or chemical treatments in either inert or reactive environments, pristine MOFs can be converted into many different derivative materials, including MOF-derived porous carbon (such as metal-doped or metal-free carbon) and MOF-derived metal compounds (such as metal oxides, sulphides and phosphides). These MOF derivatives do not just partially inherit the structural and chemical characteristics of their MOF precursors, but also exhibit some more interesting porous networks and chemical components, leading to even improved performance of the materials.

In order to describe the characteristic advantages of MOF-derived materials, it is inevitable to recapture the unique structural and chemical characteristics of their MOF precursors. The distinct merits of pristine MOFs are mainly reflected in the following aspects: (1) Crystalline porous structure with variable structures from two-dimensional (2D) to three-dimensional (3D) constructions. (2) Diverse combinations of metal cations with organic ligands, leading to the formation of a variety of porous framework structures and the possession of different metallic and non-metallic active elements (such as N, P and S). (3) Metal cation/cluster nodes are separated by the surrounding organic ligands, forming atomically dispersed metallic sites. Thanks to the above-mentioned merits of pristine MOFs, those advantageous characteristics may be not just partially inherited by MOF derivatives but can also evolve into some more interesting structural and chemical features, which may be beneficial in terms of their utilisation in energy and environmental applications [6]. The key attractions of MOF derivatives are summarised as follows:

  1. (1)

    Hierarchically porous structures

    Due to the removal of carbon or non-carbon elements, thermal treatments of pristine MOFs usually result in the collapse of their crystalline porous structures. This process can change the arrangement of the remaining atoms (such as carbon, metal or metal compounds) and thus the original pore sizes, leading to the formation of pores of different sizes, covering the pore size range of macropores, mesopores and micropores. According to the classification by International Union of Pure and Applied Chemistry (IUPAC), the sizes of macropores are above 50 nm, those of mesopores are between 50 and 2 nm, while those of micropores are below 2 nm [7]. These pores of different sizes are usually interconnected with each other, leading to the formation of hierarchically porous networks. When compared with many MOFs, which mainly compose of micropores, MOF derivatives with hierarchically porous structures may be more favourable in the circumstance when mass transfer within the pore network is an important consideration for material design. For example, in the case of MOF-derived porous carbon for gas adsorption, macropores more likely play the role of “lobby”, which provide a spacious entrance for adsorbent molecules, while mesopores offer smooth “passageways” for the diffusion of adsorbent molecules within the pore networks. The adsorption of those molecules usually take place in the connected micropores due to their comparatively large adsorption potentials [8]. The significance of hierarchically porous structure is also widely recognised in many electrochemical studies in terms of mass transport of reactants and reaction products within pore networks (such as the design of electrocatalysts and electrode materials) [9].

    In addition, the variable porous framework structures of pristine MOFs also lead to the formation of variable porous structures of their derivatives. For example, 2D MOFs are usually used for the manufacture of 2D porous membranes, while 3D MOFs are sometimes used for the construction of porous carbon materials with 3D porous networks as well. Furthermore, it is important to note that well-developed porous structure is not a unique merit for pristine MOFs or their derivatives, since it can be found in many other porous functional materials, such as activated carbon produced by activation methods. However, in the latter case, it is difficult to keep the resulting porous structures the same in different batches of products, because it is hard to control the distribution and dispersion of activating agents within carbon precursor materials. In the case of MOFs, the formation of their crystalline porous structures is a direct consequence of metal–organic coordination. This makes the ordered structures of MOFs designable and predictable prior to synthesis. In this case, the porous structure of MOF derivatives can also be designed and predicted prior to thermal treatments, which is an important factor be considered for the future stable mass production.

  2. (2)

    Nanoscale metallic active sites

    The diversity of MOFs is attributed to the diverse combinations of metal modes and organic linkers. This also makes it possible to synthesise a wide range of MOF derivatives with different metallic active sites to fulfil various functions of the materials. More importantly, it is previously mentioned that metal cation/cluster nodes are separated by the surrounding organic ligands, forming atomically dispersed metal sites. After thermal treatments, metallic elements are usually still preserved in the resulting MOF derivatives. In this case, the orderly arranged and isolated metallic sites in pristine MOFs can promote the formation of highly dispersed and uniformly distributed metal nanoparticles in their corresponding MOF derivatives. This is particularly true in terms of MOF-derived carbon materials. The carbonisation of organic ligands lead to the formation of carbon frameworks, where metal nanoparticles are embedded, which prevents them from agglomerating into larger particles. The formation of large agglomerates of metal particles results in the reduced number of exposed active sites because only those on the surface of metal particles can take the effects of enhanced interactions between active sites and reactants/adsorbents. In contrast, the formation of highly dispersed and uniformly distributed metal nanoparticles as active sites is beneficial in terms of enlarged active interface for adsorption and reaction to take place. Particularly in the case of transition metal-based MOFs and their derived materials, transition metal nanoparticles can catalyse the formation of graphitic carbon shells around them, which does not just prevent them from agglomeration but also protect them from external corrosion and poisoning. In recent years, there is a fast-growing interest in the design and synthesis of advanced materials with atomically dispersed metallic active sites. In this case, metallic active sites are dispersed and distributed in the porous substrates in the form of single atom sites, which maximises their exposure to the surrounding interaction environment. These atomically dispersed metallic sites can also interact with neighbouring atoms or functional groups, such as carbon and nitrogen. This can cause the redistribution of electrons at the metallic sites and thus tailors their reactivity [10]. It is aforementioned that orderly arranged and isolated metal nodes are a unique characteristic of MOFs. Therefore, it offers the possibility to utilise this characteristic to produce atomically dispersed metal sites in the resulting MOF derivatives [11].

    Furthermore, it is also possible to design and synthesise MOFs with orderly arranged bimetallic sites, which provides an opportunity to produce MOF-derived carbon with highly dispersed and uniformly distributed bimetallic active sites. The possession of bimetallic active sites in MOF derivatives are favourable in terms of achieving the following two functions:

    On the one hand, the two metallic elements may exhibit different affinities/activities towards different adsorbents/reactants, which grants MOF derivatives bifunctionalities in some specific applications, for instance, oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) for metal-air batteries, hydrogen evolution reaction (HER) and OER for water electrolysis.

    On the other hand, synergistic effects may occur between the two metallic elements, which can improve the performance of MOF derivatives through facilitating their spatial distribution, electron distribution and electron transfer pathway.

  3. (3)

    Diverse choices of organic ligands

    Sometimes, organic ligands are also called organic linkers because they play the role of “bridges” to link metal nodes and form the “skeletons” to build framework structures of MOFs and separate metal nodes from each other. In addition, organic ligands are the source of carbon to produce MOF-derived carbonaceous materials by means of high temperature thermal treatments in inert gas atmospheres. These processes are usually referred to as the carbonisation of MOFs, which leads to the formation of carbon frameworks with embedded metal nanoparticles. Apart from the role of carbon source, organic ligands can also possess other non-metallic elements, such as oxygen, nitrogen, sulphur and phosphorus, which may incorporate into the structures of the resulting MOF derivatives as active sites in the following two forms:

    On the one hand, non-metallic elements can incorporate into the resulting carbon structures of MOF derivatives as defect dopants. A frequently applied non-metallic dopant is nitrogen element. It can incorporate into the carbon structure in the form of pyridinic, pyrrolic, pyridonic and graphitic nitrogen. The activities of the above-mentioned nitrogen functional groups are different from each other due to varied electron distributions at these active sites. For example, pyridinic nitrogen possess lone pair electrons, which offers this type of nitrogen comparatively higher affinity and activity [12].

    On the other hand, non-metallic elements can also react with the metallic elements during the thermal treatments of MOFs, converting or partially converting metal nanoparticles into metallic compounds, such as metal phosphides and metal sulphides. These chemical transformations can help to enhance the activities of metallic sites through the modification of electron distribution and the facilitation of electron transfer at active sites. Moreover, the association with non-metallic elements may tailor the affinities of active sites towards specific adsorbents or reactants, which is also an effective way to introduce bifunctionalities to MOF derivatives [13]. In the past, conventional modification methods usually involve multi-step thermal treatments in inert and reactive gas environments. This is a common approach for many studies on the modification of metallic active sites in MOF derivatives. However, with the purpose choice among diverse organic ligands, the modification of metallic active sites can be carried out in a single step-thermal treatment, which is favoured in terms of reduced material and energy consumptions.

    In summary, the attractions of MOF derivatives originate from the unique structural and chemical characteristics of their pristine MOF precursors. These advantageous characteristics of pristine MOFs can evolve into more favourable porous structures and chemical constituents, including hierarchically porous structures, nanoscale metallic active sites and non-metallic functional groups. In this chapter, selected examples from earlier published research works are used to demonstrate the advance of design and synthesis strategies for a variety of MOF derivatives, most of which can be classified into MOF-derived carbon materials (metal-doped carbon, metal-free carbon and MOF-derived carbon with atomically dispersed metallic active sites) and MOF-derived metal compounds (metal oxides, hydroxides, phosphides, sulphides, selenides and nitrides). Particular attentions are paid to the design of metal–organic coordination structures and chemical components of MOFs and its influence on the resulting porous structures and chemical compositions of MOF derivatives, in order to demonstrate the “uniqueness” of using MOFs as precursor materials.

2.2 MOF-Derived Porous Carbon

Thanks to the crystalline porous structures and carbon-rich organic ligands of MOFs, porous carbon is a major form of MOF derivatives, which can be produced by means of thermal treatments at high temperatures and in inert gas atmospheres (such as Ar and N2). This process is well-known as the carbonisation of MOFs. On the one hand, metal and most of carbon elements can be preserved in the resulting MOF derivatives after carbonisation, leading to the formation of porous carbon frameworks with embedded metal nanoparticles, that is, metal-doped carbon. In the meantime, the other non-carbon elements, such as O, N and P, are removed or partially removed from the resulting carbon structures during the carbonisation process. The remaining non-metallic elements either incorporate into carbon structures as active sites or react with metal nanoparticles to modify metallic active sites. The doping of both metallic and non-metallic elements in MOF-derived porous carbon contributes to the formation of heteroatom-doped carbon. On the other hand, metallic elements can be removed by either acid washing after thermal treatments or vaporisation at high temperatures during thermal treatments, leading to the formation metal-free carbon with non-metallic dopants or missing carbon vacancies as active sites. Herein, this section focuses on the summary of recent advance in the design and synthesis of MOF-derived heteroatom-doped carbon and metal-free defective carbon.

2.2.1 MOF-Derived Heteroatom-Doped Carbon

The diverse combinations of metals and organic linkers make it possible to produce MOF-derived porous carbon with different combinations of heteroatom dopants. Zeolitic imidazolate frameworks (ZIFs) are a distinctive group of MOFs, which are constructed by the coordination between metal cations and imidazolate ligands [14]. As the term suggests, ZIFs possess structures similar to those of aluminosilicate zeolites, where imidazolate ligands play the role of “bridges” to connect two metal cations (M-IM-M, IM = imidazolate), similar to the Si–O–Si bonds in zeolites [15]. Apart from the common characteristic of crystalline porous structure, the attractions of ZIFs are also reflected in their isolated metallic cation sites and nitrogen-containing imidazolate ligands. Because of the above-mentioned two characteristic features, it is possible to make use of ZIFs as precursor materials to produce ZIF-derived, metal- and nitrogen-doped porous carbon derivatives.

ZIF-67 is one of the most intensively studied ZIFs, which composes of cobalt cation as the metallic node and 2-methylimidazole as the organic ligands [16]. It usually possesses purple colour and polyhedral shape. Thanks to the isolated metallic sites and N-containing ligands, ZIF-67 is also widely used as the precursor materials to produce Co- and N-doped porous carbon materials for a wide range of energy and environmental applications, including gas adsorption and separation, electrochemical energy conversion and storage [16]. ZIF-67 can be synthesised by means of either solvothermal methods [17] or simple mix and stir methods [18]. Different synthetic methods and conditions can influence the sizes and morphologies of ZIF precursors and their carbon derivatives. Xia et al. synthesised a series of ZIF-67 crystals by means of both the solvothermal method at 120 °C and the mix and stir method at different temperatures (60 and 25 °C, respectively) and in different solvents (water and methanol, respectively). Figure 2.1a shows that ZIF-67 prepared with the mix & stir method at room temperature in methanol exhibits the smallest particle size of 300 nm, which can be attributed to the comparatively slower crystal growth at a lower temperature and a less polar solvent. This ZIF-67 nanocrystal sample is chosen to be further carbonised in the argon flow in the temperatures ranging from 600 to 900 °C. Figure 2.1c–e show the scanning electron microscope (SEM) and transmission electron microscope (TEM) images of the 750 °C carbonised ZIF-67, which suggests that the polyhedral shape of pristine ZIF-67 is relatively well-preserved even after carbonisation. Figure 2.1d also reveals a relatively uniformly dispersed black dots, that is, Co nanoparticles, embedded in the carbon polyhedron. A higher magnification TEM image (as shown in Fig. 2.1e) further shows that the size of Co nanoparticles is about 10 nm. In addition, due to the catalytic effect of Co, a few layers of carbon shells are also formed around these nanoparticles, which helps to improve the stability of Co nanoparticles. The X-ray photoelectron spectroscopy (XPS) surface chemistry analysis shows that nitrogen from the ligands incorporates into the resulting carbon structure mainly in its pyridinic forms, while Co tends to form Co–N active sites due to the abundant Co-N4 moieties in the ZIF-67 precursors (as shown Fig. 2.1b). The specific surface area of both pristine ZIF-67 and their carbon derivatives are investigated by the nitrogen sorption methods. The characterisation results show that the specific surface area of pristine ZIF-67 decreases with the increasing crystal size. The 300 nm ZIF-67 shows a high Brunauer–Emmett–Teller (BET) surface area of 1512 m2 g−1. This value drops to 386 m2 g−1 after thermal treatments, however, the original crystalline porous framework evolves into a hierarchical structure as well, which is beneficial in terms of mass transfer within the pore network.

Fig. 2.1
figure 1

a Synthesis of ZIF-67 with different methods and conditions. b Schematic illustration of ZIF-67 polyhedron, its microscopic porous framework structure and the coordination structure with Co cation and 2-methylimidazolate ligands. c SEM image of ZIF-67-derived carbon polyhedrons after thermal treatments at 750 °C. d and e TEM images of ZIF-67-derived polyhedron with Co nanoparticles (black dots). Reproduced with permission [18]. © 2014 Royal Society of Chemistry

Full size image

With the advance in the design and synthesis of innovative MOFs, more Co-based MOFs can become the candidate precursor materials for the production of MOF-derived Co- and N-doped porous carbon. Jagadeesh et al. reported a cobalt nanoparticle-based porous carbon material, which is derived from a cobalt-diamine-dicarboxylic acid MOF precursor [19]. Figure 2.2a illustrates the synthetic procedure of this Co-MOF-derived porous carbon. In this case, cobalt nitrate is used as the source of metal cation and,4-diazabicyclo[2.2.2]octane (DABCO) and terephthalic acid (TPA) are applied as the organic ligands. The Co-MOF precursor is synthesised by the simple mix & stir method in dimethylformamide (DMF) and immobilized on Vulcan XC 72R carbon substrate. The corresponding Co-doped carbon is then produced by the carbonisation at 800 °C in Ar. The aberration-corrected scanning transmission electron microscope (STEM) image (as shown in Fig. 2.2b) does not just reveal Co nanoparticles with sizes from 5 to 30 nm, but also shows that graphitic layers and short-range ordered graphitic shells are formed around the nanoparticles due to the catalytic effect of Co. More interestingly, the high-angle annular dark field (HAADF) image demonstrates the existence of single Co atom sites in short-range ordered carbon regions. By combining TEM imaging and XPS analysis, it is discovered that some Co nanoparticles exhibit a Co–Co3O4 core–shell nanostructure and nitrogen from DABCO ligands incorporates into the resulting Co-doped carbon in the form of pyridinic nitrogen and bonding with Co cation (CoNx).

Fig. 2.2
figure 2

a An illustration of the synthesis of Co nanoparticle-encapsulated Co-DABCO-TPA-derived porous carbon. b Aberration-corrected STEM image of Co nanoparticles encapsulated in the graphitic carbon shells. c HAADF image of single Co atom sites (white dots) in MOF-derived carbon. Reproduced with permission [19]. © 2017 Science

Full size image

Apart from heteroatom-doped carbon derived from Co-based MOFs, there are numerous published research works on MOF-derived porous carbons embedded with many other metallic nanoparticles. Zou et al. reported a hierarchical hollow Ni/NiO nanoparticle-graphene composite derived by means of carbonisation and oxidation of a Ni-based MOF (ligand: trimesic acid) [20]. Yun et al. introduced a Fe/Fe2O3-doped porous carbon, which is derived from another famous group of MOFs, that is, MIL-101 (Fe) (MIL = Materials Institute Lavoisier) [21]. Many other metal-doped carbon derivatives can also be produced from MILs, such as MIL-101(Cr) [22], MIL-125(Ti) [23] and MIL-53(Al) [24]. Cai et al. synthesised Cu/Cu2O-supported graphitic carbon composite from the widely studied MOF-74 (Cu) [25]. Thanks to the large number of possible combinations between metal and organic ligands, a wide range of heteroatom-doped porous carbons can be produced from MOF precursors. Moreover, the existence of bimetallic MOFs makes it possible to synthesise bimetal-doped porous carbon materials. By controlling the molar ratio of the bimetallic elements in MOF precursors, it is possible to control the molar ratio of the bimetallic elements in the resulting carbon derivatives.

An interesting example was a ZnCo co-doped carbon nanotube networks derived from a ZnCo bimetallic MOF, which was published by Li et al. [26]. It is previously mentioned that ZIF-67 is an intensively studied Co-based MOF, which is widely used as the precursor to produce Co- and N-doped porous carbon. ZIF-8 is another well-known ZIF-type MOF, which possesses the isostructure of ZIF-67 but substitutes Co with Zn as the metallic cation centre. Therefore, it brings the opportunity to use both Co and Zn salts (Co(NO3)2 and Zn(NO3)2) to react 2-methylimidazole ligands, leading to the formation of ZnCo bimetallic MOFs (as shown in Fig. 2.3a). In this work, ZnCo bimetallic MOF is synthesised with the 1:1 molar ratio of Co(NO3)2/Zn(NO3)2. The as-prepared bimetallic MOF is then pyrolysed at 700 °C in Ar flow. Since the boiling point of Zn is 908 °C [27], Zn is preserved in the resulting porous carbon with Co as the co-dopant, leading to the formation of ZnCo bimetallic nanoparticles, which catalyses the graphitization of carbon and the growth of carbon nanotubes on the surface of carbonised bimetallic MOF (as shown in Fig. 2.3b). Figure 2.3c and d show the ZnCo nanoparticle at the tip of the carbon nanotube. It is worth mentioning that ZnCo bimetallic MOFs based on ZIF-8 and ZIF-67 frequently appear in many other studies on MOF-derived heteroatom-doped carbon as well. However, in many cases, Zn plays the role of “structure modifier”, where the corresponding carbonisation temperature is set to be above its boiling point, resulting in the evaporation of Zn and the formation monometal-doped carbon. On the one hand, the volatilisation of Zn can may induce the formation of pores and thus improve the porous structure of MOF-derived carbon [28,29,30]. On the other hand, the evaporation of Zn can also help facilitate the distribution and dispersion of the other metal nanoparticles, preventing the formation of large agglomerates [31, 32].

Fig. 2.3
figure 3

a The framework structure of ZnCo bimetallic MOF. b The growth of carbon nanotubes on the surface of the carbonized bimetallic MOF. c and d ZnCo bimetallic nanoparticle at the tip of the carbon nanotube. Reproduced with permission [26]. © 2018 Royal Society of Chemistry

Full size image

In summary, thanks to the unique structural and chemical characteristics of MOFs, they can be utilised as precursors to produce a wide range of MOF-derived porous carbon materials, which are doped with a variety of metallic and non-metallic heteroatoms in the form of nanoparticles, functional groups and moieties. These highly dispersed and uniformly distributed active sites can help to improve the performance of MOF-derived carbon in energy and environmental applications. With the further pursuit of optimised performance, MOF-derived porous carbon materials with atomically dispersed metal sites have attracted great attention from chemists and materials scientists.

2.2.2 MOF-Derived Carbon with Atomically Dispersed Metal Sites

The significance of metal nanoparticles in porous carbon is to promote the exposure of highly dispersed and uniformly distributed active sites for optimised material performance. In this case, the formation of atomically dispersed metal sites (ADMSs) can maximise the exposure and interfacial contact of active sites with adsorbents/reactants, and thus release the “full potential” of an active site. In addition, the single-atom metallic site may interact with the neighbouring atoms (such as N, S and P), which can modify the electron distribution and transfer at the active site and thus tailor the material performance as well. However, the production of porous carbon with ADMSs usually involves the careful control over the concentration of metal source and the complicated synthetic method to induce the formation of M–N–C coordination structure. Owing to isolated metal sites and active non-metallic elements (particularly nitrogen) in organic ligands, the carbonisation of MOFs becomes a more convenient and controllable route to produce porous carbon with ADMSs. It is previously mentioned that, besides Co nanoparticles, Jagadeesh and co-workers also found atomically dispersed Co sites in a cobalt-diamine-dicarboxylic acid MOF-derived porous carbon [19].

Yang et al. reported the synthesis of a carbon framework with atomically dispersed Mn sites by means of the carbonisation of a Mn-BTC MOF (BTC = 1,3,5-benzenetricarboxylic acid), followed by the HCl etching treatment and NH3 annealing (as shown in Fig. 2.4a) [33]. Aberration corrected high‐angle annular dark‐field scanning transmission electron microscope imaging (HAADF‐STEM, Fig. 2.4b) and Extended X-ray absorption fine structure spectra (EXAFS, Fig. 2.4c) demonstrate the existence of single Mn atom sites and the formation of Mn–O and Mn-N coordination in the MOF-derived carbon framework. In Fig. 2.4c, Mn foil is used as the reference material to show the peak for Mn–Mn bond (black curve), which is absent in the sample with ADMSs (orange curve). This apparent difference further proves that Mn metallic sites exist in the form of single atoms.

Fig. 2.4
figure 4

a A flow chart of the synthesis procedure for the carbon framework with atomically dispersed Mn sites from Mn-BTC, which are coordinated with O and N atoms at the edges of pores and frameworks. b Aberration-corrected HAADF‐STEM shows single Mn atom sites (red circles). c Fourier transform of Mn K-edge x-ray EXAFS spectra indicate the existence of Mn-N and Mn–O coordination and the absence of Mn–Mn bond in Mn-BTC-derived carbon (Mn/C-NO, orange curve). Reproduced with permission [33]. © 2018 WILEY–VCH Verlag GmbH & Co. KGaA

Full size image

Similar to the case of bimetallic MOF-derived bimetal nanoparticle-doped carbon, it is possible to produce porous carbon atomically dispersed bimetallic sites with bimetallic MOFs as well. Han et al. presented a hollow carbon nanocubes with atomically dispersed CoNi bimetallic sites derived from a CoNi bimetallic MOF (as shown in Fig. 2.5a) [34]. This bimetallic MOF is fabricated by the chemical precipitation of Ni2+ cations and Co(CN)63− and coated with polydopamine for the formation of the core–shell hollow carbon nanotube after 500 °C annealing in NH3 flow. The last step of acid etching is to remove the excessive Co and Ni nanoparticles, leaving only the diatomic Co–Ni sites, which form coordination structures with the neighbouring N atoms. In this way, the Co–Ni diatomic site still fulfil the purpose of atomic dispersion, which acts like a “single-atom” site. These atomically dispersed bimetal sites are imaged by the aberration-corrected HAADF-STEM in Fig. 2.5b. Figure 2.5c shows the Fourier transformed EXAFS spectra of MOF-derived porous carbon nanocubes with bimetallic ADMSs (red curve) and nanoparticles (black curve). It evidences the existence of only Co–N and Co–Ni interactions in the sample with ADMSs, while the sample with bimetallic nanoparticles shows a clear peak for Co–Co bonds.

Fig. 2.5
figure 5

a An illustration of the synthesis of CoNi bimetallic MOF-derived carbon nanotube with atomically dispersed bimetallic Co–Ni sites. b HAADF-STEM image of atomically dispersed Co–Ni sites (red circles). c Fourier transformed Co K-edge EXAFS spectra of the sample with atomic Co–Ni sites (red curve), indicating Co–Ni and Co–N bonds. The sample with CoNi nanoparticles (black curve) is used as the reference material for Co–Co bond. Reproduced with permission [34]. © 2019 WILEY–VCH Verlag GmbH & Co. KGaA

Full size image

Apart from intrinsic isolated metal cation centres, ADMSs can be produced by the introduction of external metal sources in accompany with the carbonisation of MOFs. For example, Zhu et al. demonstrated a hierarchical carbon architecture with atomically dispersed Fe sites derived from FeCl3- and DCD-encapsulated MIL-101-NH2 (DCD = dicyandiamide) [35]. In this case, DCD does not just play the role of additional source of carbon and nitrogen, but also prevent the collapse of porous structure and the agglomeration of metallic components. This is further in favour of the formation of ADMSs in addition to the merits of the MOF precursor. Again, the HAADF-STEM imaging and EXAFS analysis demonstrate the existence of single Fe atom sites with the formation of FeNx coordination structure.

Besides the above-discussed examples, there are a large number of reported works on MOF-derived porous carbon with either mono- or bimetallic ADMSs [11, 36]. In most cases, ADMSs in MOF-derived carbon can be achieved by either the one-step carbonisation of MOF precursors with intrinsic metal cations or the encapsulation and dispersion of external metal sources in crystalline porous frameworks of MOF precursors before carbonisation. The formation of ADMSs can be attributed to isolated metal sites, well-developed porous structures and coordination interactions with neighbouring non-metallic sites from organic ligands. Since ADMSs represent the non-stopping pursue of higher performance of functional materials, it can be expected the on-going research on ADMSs will continue to bring excitement and pleasant surprise with the exploitation of MOF-derived porous carbon.

2.3 MOF-Derived Metal-Free Defective Carbon

Apart from the rich progress of MOF-derived metal-doped carbon, there is also a growing interest in the development of MOF-derived metal-free carbon. The functional performance of metal-doped carbon heavily relies their metallic active sites. However, this also makes them susceptible to acidic and alkaline environments due to metal leaching and etching [37]. In the absence of metallic elements, the activity of metal-free carbon depends on the defective sites in the carbon structures, which frequently exist in the form of chemical defects (non-metallic dopants, such as N, P and S) and structural defects (vacancies, edges and holes) [38]. The formation of these defective sites can modify the local charge distribution and thus play the role of active sites in metal-free carbon. Furthermore, these active sites may possess higher stabilities when compared with those of metallic sites in acidic and alkaline environments. When MOFs are applied to produce metal-free carbon, the diverse choices of organic ligands make it possible to introduce multiple non-metallic dopants as chemical defects, while the rearrangement of carbon atoms during thermal treatments may create all kinds of structural defective sites in the resulting carbon structures. Metallic elements from the MOF precursors are removed either by means of acid washing or volatilisation at high temperatures.

In the case of the acid washing method, Wang et al. produced a nanoporous carbon from ZIF-67 [39]. It is previously introduced that ZIF-67 is a Co-based MOF with 2-methylimidazolate linkers. Therefore, the pyrolysis of ZIF-67 in the inert gas atmosphere (Ar in this work) leads to the formation of MOF-derived carbon with embedded Co nanoparticles. As illustrated in Fig. 2.6a, in order to produce metal-free carbon, this ZIF-67-derived carbon is further washed with aqua regia. This acid washing process does not just remove Co nanoparticles but also promotes the formation of new nanopores. Figure 2.6b and c are the high resolution TEM images of ZIF-67-derived porous carbon before and after acid washing, respectively. It clearly shows the formation of new nanopores due to the removal of Co nanoparticles by acid washing. Porosity analysis results show that both specific surface areas and pore volumes of ZIF-67-derived carbon increase after acid treatments.

Fig. 2.6
figure 6

a An illustration of Co-doped nanoporous carbon derived from ZIF-67, followed by the acid treatment to produce metal-free carbon. TEM images of ZIF-67-derived carbon (b) before and (c) after the acid treatment by aqua regia, showing the removal of Co nanoparticles and the formation of nanopores. Reproduced with permission [39]. © 2016 Royal Society of Chemistry

Full size image

It is mentioned in the previous example that Zn can be used as a “structural modifier” in bimetallic MOF-derived heteroatom-doped carbon due to its volatilisation at high temperatures above its boiling temperature (908 °C). Based on the same consideration, Zn-based MOFs are widely used as precursor materials to produce metal-free porous carbon [40,41,42]. In contrast, if the temperatures of thermal treatments are below the boiling point of Zn, Zn element is still preserved in MOF-derived carbon in the form of metallic nanoparticles. In this case, acid washing is required to produce metal-free carbon [43]. When compared with the acid washing process, the volatilisation of Zn may be recognised as a more convenient one-step method and a more environmentally friendly way to produce MOF-derived metal-free carbon. As early as 2014, Zhang and co-workers reported the synthesis of metal-free and N-doped porous carbon from ZIF-7, which is a Zn-based ZIF-type MOF with the benzimidazole ligands [44]. In the authors’ work, the ZIF-7 precursor is mixed with glucose and then carbonised at 950 °C, where glucose is used as an additional source of carbon. The analytical results of XPS, X-ray diffraction (XRD) and Energy Dispersive X-ray (EDX) confirm the removal of Zn after the carbonisation of ZIF-7 due to the evaporation of Zn. In this case, nitrogen becomes the only dopant and the main active sties in the resulting carbon structure to fulfil its functional performance (i.e. electrocatalysis in this work).

Apart from ZIF-7 from the above-discussed case, ZIF-8 is another Zn-based MOF, which is widely studied for the production of MOF-derived metal-free carbon due to its merit of easy synthesis at ambient conditions and the volatilisation of Zn as well [45]. It is previously mentioned that ZIF-8 possesses the same framework structure of ZIF-67 and use the same ligands of 2-methylimidazole. Recently, Pan et al. reported such a metal-free and heteroatom-doped carbon derived from phytic‐acid‐functionalized ZIF‐8 by means of 1050 °C annealing in NH3 (as shown in Fig. 2.7a) [46]. Again, Zn is removed from ZIF-8-derived carbon due to its evaporation, where no apparent Zn nanoparticle is shown in the high magnification TEM image in Fig. 2.7b. N and P elements are introduced into the resulting carbon structure as dopant defects from the 2-methylimidazolate ligands and phytic acid functional groups, respectively (as shown in Fig. 2.7c). In another work by Zhao et al., a metal-free N-doped graphitic carbon photocatalyst is produced from ZIF-8 as well [47]. In this case, besides the role of N as dopant defect active site, N dopants also promote the formation of vacancy-type defects in this MOF-derived carbon (as shown in Fig. 2.8).

Fig. 2.7
figure 7

a A schematic flow chart of metal-free and N, P-co-doped carbon derived from phytic‐acid‐functionalized ZIF‐8. b A high magnification TEM of ZIF-8-derived carbon shows no apparent Zn nanoparticles. c Simulated N and P doping in carbon structures. Reproduced with permission [46]. © 2020 WILEY–VCH Verlag GmbH & Co. KGaA

Full size image
Fig. 2.8
figure 8

An illustration of the synthesis of ZIF-8-derived carbon with vacancy-type defects. Reproduced with permission [47]. © 2016 WILEY–VCH Verlag GmbH & Co. KGaA

Full size image

It is also worth mentioning that, although the volatilisation of Zn is a more effective and environmentally friendly way to produce MOF-derived metal-free carbon. Many laboratory works still carry out acid washing after thermal treatments, in order to to remove any possible residual Zn in MOF-derived carbon for the purpose of research preciseness [48, 49]. Furthermore, Zn-based MOFs can also be used as hosts and templates to load external metal elements as active dopants, where Zn is vaporised during carbonisation and only monometal-doped carbon is formed [50].

In summary, thanks to the porosity and diversity of MOFs, they can be utilised to produce a variety of porous carbon materials with or without metallic dopants. On the one hand, in the case of MOF-derived metal-free carbon, metallic elements can be removed by means of either acid washing or high temperature volatilisation. In this case, the performance of MOF-derived carbon heavily relies on its porous structure, non-metallic dopant and structural defect sites in the resulting carbon structure. On the other hand, in the case of MOF-derived metal-doped carbon, metallic elements can exist in the form of nanoparticles or ADMSs, which become the major active sites to fulfil the functions of the materials. Non-metallic elements from organic linkers can co-dope with metallic elements in MOF-derived carbon as active sites. They can either modify the electron distribution and transfer at metallic active sites or form coordination interactions with metal cations to promote the formation of ADMSs. Furthermore, it is also possible for non-metallic elements to react with metallic elements during thermal treatments, which can lead to the formation of metal compound active sites and thus tailor the activities of metallic sites.

2.4 MOF-Derived Metal Compounds

In the case of MOF-derived metal-doped carbon, the majority of metal dopants frequently exist in their elemental forms due to their reduction reactions with carbon. These elemental metals can further react with non-metallic elements, such as O, N, P, S and Se, forming metal compounds, which can improve or tailor the activities of metallic sites. These non-metallic elements can come from external chemical sources (such as NH3, CS2, S powder and NaH2PO4) or organic linkers of MOF precursors. In the former case, MOF precursors are subject to thermal treatments in either inert or oxidative gas atmospheres to produce MOF-derived metal-doped carbon or metal oxides. Then, metallic elements in MOF derivatives can further react with the above-mentioned chemical sources of non-metallic elements in the form of either continuous gas flows, decomposed products in inert carrier gas flows or solid mixtures. In the latter case, non-metallic elements are released upon thermal decompositions of organic linkers. Some of them may be lost in the form of decomposed gas molecules, while the other may dope into the resulting carbon structures or react with metallic elements to form metal compounds. Apart from the methods of thermal treatments, MOF-derived metal compounds can also be produced by means of chemical etching methods at ambient conditions. To date, there are reports of metal oxides [51], hydroxides [52], phosphides [53], sulphides [54], selenides [55] and nitrides [56] derived from MOF precursors, which are discussed in the following subsections, respectively.

2.4.1 MOF-Derived Metal Oxides

Metal oxide is a common form of active metal compounds. Thanks to their porous frameworks and isolated metal nodes, MOFs can be used as precursor materials to produce both metal oxides and metal oxide-doped carbon with inherited porosities and metallic active sites. In the case of MOF-derived metal oxides, they can be produced by means of high temperature calcination in air, which completely removes the carbon content from the original MOFs. In the case of MOF-derived metal oxide-doped carbon, they can be produced by either post-oxidation treatments in air at comparatively lower temperatures after carbonisation or reactions with intrinsic oxygen elements from decomposed organic linkers during carbonisation.

A typical example for MOF-derived metal oxides was presented by Yu and co-workers. In their work, a bimetallic oxide hollow nanowire (NiO/NiCo2O4) is synthesised from MOF-74 with Ni/Co bimetallic nodes and 2,5-dihydroxyterephthalic acid (DHTA) linkers [57]. Similar to ZIF-67, MOF-74 is another group of frequently studied and utilised MOF due to its high specific surface area and a variety of tailorable metal cation nodes, including the possibility of bimetallic cation nodes. In this case, Ni and Co acetates are simultaneously added to the reaction solution for the synthesis bimetallic MOF-74. The MOF precursor is then subject to oxidation in air at 350 °C in air. The mechanism for the formation of hollow nanowires is proposed to the result of Kirkendall effect [58], which can be explained by the different diffusivities of inner and outer components during the high temperature oxidation treatment [59]. In the case of MOF-derived metal oxide-doped carbon, Zhou et al. synthesised a range of transition metal oxide-doped carbon materials from a variety of transition metal-based MOFs, including Co-BTC, Co-BDC (BDC = 1,4-dicarboxybenzene acid), Co-DHTP (DHTP = 2,5-dihydroxyterephthalic acid) and the same MOFs but with Ni nodes [60]. Since all the MOF precursors are carbonised in high-purity N2 gas flow, organic linkers become the only source of oxygen for the formation of metal oxides. In addition, as shown in Fig. 2.9, the use of different organic ligands can also lead to the formation of 1D, 2D and 3D MOFs and corresponding MOF derivatives as well.

Fig. 2.9
figure 9

The direct production of various Co3O4-doped carbon materials from MOFs with different O-containing organic ligands, leading to the formation of 1D–3D structures. Reproduced with permission [60]. © 2018 American Chemical Society

Full size image

Zhang et al. reported the synthesis of Fe2O3 with microboxes and hierarchical shell structures from a Prussian blue (PB)-type MOF, which is an iron(III) hexacyanoferrate(II) compound (Fe4[Fe(CN)6]3) that possesses a cubic shape [61]. Prussian blue is an accidently synthesized and discovered cubic crystal materials, where substitutional and interstitial modifications lead to the formation of a variety of structures analogous to that of PB [62,63,64]. Therefore, this group of MOFs are well-known as Prussian blue analogues (PBAs). In this work, Fe2O3 microboxes are produced by the calcination of Fe4[Fe(CN)6]3 in air. The authors conclude the 3-stage formation of hierarchical Fe2O3 microboxes, which is illustrated in Fig. 2.10a. Figure 2.10b–j show the SEM and TEM images of PB-derived Fe2O3 microboxes, which are calcined at 350, 550 and 650 °C, demonstrating different structures of PB-derived Fe2O3 at different stages illustrated in Fig. 2.10a.

Fig. 2.10
figure 10

a An illustration of hierarchical Fe2O3 microboxes from Fe4[Fe(CN)6]3 PBA precursors. SEM and TEM images of PB-derived Fe2O3 microboxes after calcination in air at (b–d) 350 °C, (e–g)550 °C and (h–j) 650 °C. Reproduced with permission [61]. © 2012 American Chemical Society

Full size image

2.4.2 MOF-Derived Metal Hydroxides

Metal hydroxide is another common form of active metal compound, which is frequently synthesised and utilised in the form of layered double hydroxides (LDHs). LDHs is a large group of layered compound materials, which consist of positively charged layers (e.g., metal cations) and interlayer regions filled with charge balancing anions and water molecules [65]. Similar to both structural and chemical diversities of MOFs, LDHs also possess the merit of flexible combinations of metal cations and charge compensating anions. This common advantage interlinks these two large families of functional materials, and make it possible to use MOFs as precursors to produce LDHs with a variety of nanostructures and chemical properties [66, 67], and vice versa [68]. Differing from the high temperature oxidative treatments for MOF-derived metal oxides, MOF-derived metal hydroxides can be produced through alkaline hydrolysis reactions between MOFs and hydroxide ions in the alkaline solutions at ambient conditions. The framework structures of MOFs collapse as a result of chemical etching. Organic linkers are removed during the etching process, while metal cations react with hydroxide ions or water molecules to form metal hydroxides.

Zhang et al. reported an ultrathin Co(OH)2 nanoarray derived from a Co-based MOF with 2-methylimidazolate linkers, which is grown on a piece of Ni foam [69]. The growth of MOFs is carried out by simply immersing the Ni foam in the mixture aqueous solution of cobalt chloride and 2-methylimidazole. Then, the synthesis of Co(OH)2 nanoarrays is carried out by the further immersion of Co-MOF/Ni foam in the cobalt chloride ethanol solution with a pH value between 5 and 6, allowing the alkaline hydrolysis reaction to take place. The porous nature of MOF precursors ensures the easy access of cobalt chloride solution, leading to the formation of ultrathin layers of Co(OH)2 sheets. Figure 2.11 also indicates that Co(OH)2 nanosheets also inherit the nanoarray structure of MOFs, leading to the formation of Co(OH)2 nanonetworks on Ni foam. Furthermore, as emphasised in Fig. 2.11, all the above-mentioned reactions are carried out at room temperature, which is beneficial in terms of easy synthesis and energy saving.

Fig. 2.11
figure 11

An schematic synthesis procedure of Co(OH)2 nanonetworks from the growth of a Co-based MOF nanoarrays on Ni foam. Reproduced with permission [69]. © 2018 American Chemical Society

Full size image

It is demonstrated in the above section of “MOF-derived metal oxides” that PB-type MOF Fe4[Fe(CN)6]3 is used for the production of Fe2O3 microboxes. Later on, the same research group reported their continued work on the synthesis of Fe(OH)3 from the same MOF precursor [70]. In this case, other than calcination in air, Fe(OH)3 with different structures are produced by either simple “mix & shake” reactions at ambient conditions or hydrothermal reactions at 80 °C, both with NaOH. The former method leads to the yolk-shelled Fe(OH)3 microboxes, while the latter induces the formation of multishelled Fe(OH)3 microboxes. The different architectures of Fe(OH)3 microboxes can be attributed to varied diffusion of OH− and precipitation of Fe(OH)3 at different temperatures and concentrations of NaOH solutions.

2.4.3 MOF-Derived Metal Phosphides

Owing to the diverse bonding types in metal phosphides (MxPy), the combination of metal and phosphorus can bring about many interesting physical and chemical properties [71]. Thanks to the high electronegativity of phosphorus, they can draw electrons from metallic elements, which may tune MxPy into “electron sinks”, making them more attractive to positively charged protons [58, 72]. This characteristic is particularly useful in catalysing proton-involved reactions, such as HER for water electrolysis. This arouses interests from researchers to develop and study a wide range of MxPy either in the form of pure compounds or active dopants in carbon materials. Similar to the case of metal oxides and hydroxides, MOFs have been studied as precursor materials because of the unique advantages of their structural and chemical characteristics. Both MxPy compounds and MxPy-doped carbon can be produced by the two-step thermal treatment processes, which involve high temperature calcination in air or carbonisation in inert gas atmospheres as the first step, followed by the additional step of phosphorisation at elevated temperatures as well. The phosphorisation reaction is usually carried out by the decomposition of phosphorous compounds (sodium hypophosphite in most cases) and subsequent reactions between metallic elements and P-containing decomposed gaseous products in carrier gases (e.g., N2 or Ar). Phosphorous compounds are placed at the upstream of the carrier gas, while MOF derivatives are placed at the downstream. Therefore, decomposed products follow the carrier gas flow and reach MOF derivatives, where phosphorisation reactions take place. In some other cases, metal phosphides are produced just in a single-step phosphorisation of MOFs with phosphorous compounds.

In the case of MOF-derived MxPy compounds, Chu et al. reported the synthesis of Zn-Co phosphide porous nanosheets (Zn0.33Co0.67P), which was derived from a ZnCo bimetallic MOF with the 2-methylimidazolate linkers [73]. In this work, the authors calcine this bimetallic MOF in air at 400 °C to produce ZnCo2O4, and thus proceed with the phosphorisation treatment at 300 °C. Sodium hypophosphite (NaH2PO2) is used as the source of phosphorus, which decomposes and releases phosphine (PH3) upon heating. N2 is used as the carrier gas for PH3 to reach the calcined MOF and convert Zn-Co oxides into Zn-Co phosphides. In some other cases, for example, in Xiao et al.’s work [74], they adopted a one-step phosphorisation process to convert a wide range of CoM bimetallic ZIFs (M = Ni2+, Mn2+, Cu2+ and Zn2+) into a variety of CoM bimetallic phosphide nanosheets. In the case of MOF-derived MxPy-doped carbon, Liu et al. reported a Ni2P-doped porous carbon derived from a Ni-MOF with the BTC linkers [75]. This Ni-MOF is firstly carbonised in Ar at 600 °C. Again, sodium hypophosphite (NaH2PO2) is used as the source of phosphine for the phosphorisation of carbonised Ni-MOF at 350 °C. Chemical analysis on the carbonised and phosphorised Ni-MOF indicates that Ni2P and Ni co-exist in the resulting carbon framework and form hybrid nanoparticles. Synergistic effects can occur between Ni2P and Ni, which can help to optimise the electronic structure and charge transfer within this MOF-derived Ni2P/Ni-doped carbon.

The above-mentioned research works adopt the straightforward carbonisation and phosphorisation method to produce MxPy and MxPy-doped carbon. With some clever material design strategies, it is possible to introduce both interesting structural and compositional features to MxPy with the utilisation of MOFs. He and co-workers demonstrate their effort to design and synthesise such a carbon-incorporated Ni–Co mixed metal phosphide nanoboxes through a two-step modification of ZIF-67 [76]. In their work, ZIF-67 is not subject to direct carbonisation and phosphorisation treatments to produce NixPy-doped carbon. Instead, ZIF-67 nanocubes are firstly immersed in a Ni(NO3)2 ethanol solution, which is a previously mentioned method to produce MOF-derived metal hydroxides. As suggested by Fig. 2.12a, the reaction between ZIF-67 and Ni(NO3)2 takes place at the surface of the nanocubes in the first place, forming Ni-Co bimetallic LDH on the surface. In the meantime, the diffusion of ZIF-67 from inside out leads to the formation of ZIF-67@LDH nanoboxes. Then, ZIF-67@LDH nanoboxes are phosphorised with NaH2PO2 in N2 flow at 350 °C and organic linkers of ZIF-67 are also carbonised at the same time, leading to the formation of carbon-incorporated Ni-Co mixed metal phosphide nanoboxes. Again, the existence of bimetallic elements in the phosphide compound brings the potential of bimetallic synergy in terms of improved performance (enhanced electrocatalytic activity in this work), while the incorporation of carbon enhances the structural stability, electron transfer efficiency and electrical conductivity of this MOF-derived bimetallic phosphide material. Apart from He et al.’s work, a similar approach was adopted by Guan and co-workers to produce hollow Mo-doped CoP nanoarrays from a 2D Co-based MOF [77]. In this case, the 2D Co-based MOF firstly reacts Na2MoO4 to produce Mo-Co LDH arrays, and is then phosphorised with NaH2PO2 to generate hollow CoP nanoarrays.

Fig. 2.12
figure 12

a An illustration of the two-step formation process from ZIF-67 nanocubes to carbon-incorporated NiCoP nanoboxes: I. LDH formation; II phosphorisation. b HAADF-STEM image of a carbon-incorporated NiCoP nanoboxe and its corresponding elemental mapping images of Ni, Co and P. Reproduced with permission [76]. © 2017 WILEY–VCH Verlag GmbH & Co. KGaA

Full size image

It is previously mentioned that non-metallic elements can be released from the organic linkers during the thermal treatments of MOFs and take part in the reaction with metallic elements to form metal compound active sites. This is a more favourable route based on the consideration of material and energy saving. In the case of MOF-derived metal phosphides, P-containing organic ligands can be applied as an intrinsic source of phosphorus. In addition to the merits of material and energy saving, considering the toxicity of PH3, the use of P-containing ligands also become a less toxic and more environmentally friendly way to produce MOF-derived metal phosphides. A representative example was demonstrated by Wang and co-workers to produce FeNi bimetallic phosphide (FeNiP) nanoparticles anchored on hollow carbon structures, which are derived from a Ni-based MOF with H3TPO (H3TPO = tris-(4-carboxylphenyl)phosphine oxide)) and DABCO (DABCO = 1,4-diazabicyclo[2.2.2]-octane) as the linkers, denoted as BMM-10 [78]. As illustrated in Fig. 2.13a, after the synthesis of this Ni-based MOF BMM-10, it is acid-etched by the hydrolysis of Fe(NO3)3 in ethanol solution to form Fe–Ni hollow nanoparticles (Fe–Ni-HNP), where the crystalline structure of BMM-10 is destroyed. As depicted in Fig. 2.13b, chemical analysis suggests that the uncoordinated P = O bonds in TPO-2 linkers tend to be transformed into P-O during the Fe(NO3)3 etching process, while the coordinated P = O bonds in TPO-1 linkers tend to decompose during carbonisation at low temperatures. Then, the decomposition of a large amount of P-O and C-P bonds promotes the conversion from FeNi to FeNiP. It is worth mentioning that the direct carbonisation of MOFs with P-containing organic linkers is not always applicable for the formation of functional MOF-derived metal phosphides. In one of the authors’ earlier work [79], they developed a cage-like Zn-based MOF with the same H3TPO and DABCO as organic linkers. However, its direct carbonisation results in the formation of corrosive phosphates, which causes the collapse of the cage-like structure and the formation of large agglomerates of metallic phosphide particles.

Fig. 2.13
figure 13

a An illustration of the synthesis of a FeNi bimetallic phosphide hollow carbon composite from a Ni-based MOF with P-containing organic linker H3TPO as the intrinsic source of phosphorus. b A schematic of structural change from P = O to P–O bonds during Fe(NO3)3 etching and low temperature treatments, and the breaking of C–P bonds at high temperature carbonisation. Reproduced with permission [78]. © 2017 Elsevier

Full size image

2.4.4 MOF-Derived Metal Sulphides

Metal sulphides (MxSy) are another group of frequently encountered in MOF derivative-related studies. S atom in MxSy plays a similar active role in comparison with that of P atom in MxPy due to its high electronegativity as well. However, S is even more electronegative than that of P, which is not necessarily an advantage for S as an active site [58]. The stronger interaction can enhance the adsorption of reactants (e.g. protons in HER) on S active sites. However, it can also make it more difficult for the reaction products being desorbed from S active sites (e.g., H2 in HER) [80]. The overall activity of MxSy can be tailored by the choice of different metals, metal/sulphur ratios and material structures. In this case, thanks to the diversity of MOFs, they can be used to produce MOF-derived MxSy with varied structural and compositional characteristics. The synthesis methods for MOF-derived MxSy are very similar to those for MOF-derived MxPy. They can exist in the form of either MxSy or MxSy-doped carbon, which can be obtained by means of either thermal treatments with external sources of S (such as CS2 and S powders) or direct calcination/carbonisation of MOFs with S-containing organic linkers. There are also reports of converting MOFs to MxSy through chemical reaction routes at ambient conditions without the need for thermal treatments.

In the case of MOF-derived MxSy, Zhou and colleagues adopted the two-step LDH formation and sulfuration method to produce MOF-derived NiCo bimetallic sulphides [81]. In their work, ZIF-67 is firstly grown on Ti3C2Tx nanosheets (Tx stands for surface terminal groups such as -Fx and -OHx), followed by the immersion of ZIF-67@ Ti3C2Tx nanosheets into a Ni(NO3)2 ethanol solution, where ZIF-67 is etched and converted into NiCo bimetallic LDH. Then, this bimetallic LDH is sulfurised by means of thermal treatments at 400 °C in the CS2 gas flow. In the case of MOF-derived MxSy-doped carbon, Shao et al. synthesised FeS2-doped porous carbon by means of the two-step carbonisation and sulfuration of MIL-88 (Fe), which is a Fe-based MOF with the 2-aminoterephthalic acid linkers. The authors firstly carbonised MIL-88 (Fe) at 600 °C in the N2 gas flow to produce MOF-derived porous carbon with Fe nanoparticles [82]. Then, this Fe-doped porous carbon is further mixed with sulphur powder, which are placed at the downstream and upstream of the carrier gas (N2 in this work), respectively. The sulfuration reaction is carried out at 500 °C.

Apart from the use of external sources of S, MxSy can also be produced by the direct carbonisation/calcination of MOF precursors with S-containing linkers. A representative work was published by Zhao et al., where they designed and synthesised a 2D Co-based MOF with two organic ligands H2BDC and SPDP (SPDP = 4,4′-(sulfonylbis(4,1-phenylene))dipyridine) [83]. As illustrated in Fig. 2.14, since SPDP is a S-containing ligand, a simple one-step carbonisation is able to promote the formation of Co9S8 nanoparticles encapsulated in MOF-derived carbon matrix.

Fig. 2.14
figure 14

An illustration of the synthesis of a Co-based MOF with both H2BDC and S-containing SPDP as the bridging ligands, making it possible to produce Co9S8 in the resulting porous carbon in a single-step carbonisation. Reproduced with permission [83]. © 2019 Royal Society of Chemistry

Full size image

In addition to the thermal treatment methods, the synthesis of metal sulphides can be carried out by means of chemical reactions as well. Similar to the cases of MOF-derived metal oxides and hydroxides, PBA-type MOFs can be utilised to produce metal sulphides with interesting structures and morphologies. Again, the same research group, which presented their works on PB-derived Fe2O3 and Fe(OH)3, reported another continued research on the synthesis of NiS nanoframes from Ni-Co PBA nanocube templates (Ni3[Co(CN)6]2) [84]. In this work, NiS is synthesised by the direct reaction between the Ni-Co PBA and the Na2S solution, which is carried out by means of the solvothermal method in autoclaves at 100 °C. As illustrated in Fig. 2.15a, an anisotropic chemical etching and anion exchange mechanism is proposed, where the reaction tends to start at the edge of the PBA cube due to higher curvature and roughness with more defects. SEM images in Fig. 2.15b–e demonstrate gradual structural change from the PBA cubic form to NiS nanoframes at different passing hours. Dong et al. employed a similar approach with ZIF-8 but using a different chemical source of S, that is, thioacetamide (TAA). In this work, the as-prepared ZIF-8 is mixed with TAA in the ethanol solution and sealed in an autoclave for the solvothermal reaction at 85 °C, which converts ZIF-8 into ZnS [85]. This solvothermal sulfuration method was also adopted by Yilmaz et al. to produce Co9S8 by the sealed reaction between ZIF-67 and TAA at 100 °C [66].

Fig. 2.15
figure 15

a An illustration of the synthetic procedure of NiS nanoframes by means of the reaction between Ni3[Co(CN)6]2 and Na2S. SEM images show the gradual structural change from Ni3[Co(CN)6]2 nanocubes to NiS nanoframes in (b) 0 h. c 0.5 h. d 2 h. e 6 h (scale bar: 100 nm). Reproduced with permission [84]. © 2015 WILEY–VCH Verlag GmbH & Co. KGaA

Full size image

2.4.5 MOF-Derived Metal Selenides

Owing to their unique electronic structure and electrical properties, in recent years, metal selenides (MxSey) arise like a “rising star” in the field of transition metal chalcogenides for energy conversion and storage applications, once which mainly focused on MxSy- and MxPy-based materials. When compared with many other transition metal chalcogenides, MxSey demonstrates more appealing performance on the aspects of charge transfer and electronic conductivity [86, 87]. In addition, while MxSey can also attract protons, Se sites in MxSey exhibit weaker bonding strengths with reactants, which is favourable for the desorption of reaction products in some catalytic reactions. For example, Se–H show a lower bonding strength (276 kJ mol−1) than those of P–H (322 kJ mol−1) and S–H (363 kJ mol−1), which is in favour of hydrogen desorption from the active sites [80]. The excellence of MxSey motivates the enthusiasm of researchers to look into the design and synthesis of MxSey with different structural and compositional features, in order to the tailor the functionality of MxSey for a wide range of applications. The above-mentioned merits of MxSey encourages material researchers to take the advantages of MOFs as effective precursors to fabricate MxSey-based materials, in order to fully explore their potentials.

The synthesis methods of MOF-derived MxSey almost resemble those for MOF-derived MxSy and MxPy, which mainly involve single- or multi-step carbonisation/calcination and selenisation processes. For example, in the case of MOF-derived MxSey compounds, the two-step calcination and selenisation process was utilised by Zhao et al. to produce CoSe2 with a hierarchical nanosheet structure from a Co-based MOF with methylamine and formic acid as organic linkers. At first, this Co-based MOF is calcined at 400 °C in air to produce Co3O4 microcubes. Then, CoSe2 microcubes are produced through another two-step hydrothermal reaction, where Se reacts with NaOH to generate Se2− ions at 200 °C, followed by the subsequent anion exchange reaction between Se2− and Co3O4 to generate CoSe2 at different temperatures from 140 to 200 °C. 160 °C is found to be the optimised hydrothermal temperature because of the formation of the hierarchical nanosheet structure, which maximises the surface to volume ratio of the material and increase the interfacial contact area. A lower hydrothermal temperature can result in an incomplete selenisation process, while a higher hydrothermal temperature promotes the formation of nanorods rather than nanosheets. Besides hydrothermal reactions, selenisation reactions can also be carried out by means of carbonising the mixture of MOFs and Se powders. For example, in Yuan et al.’s work [88], the mixture of Fe-Zn bimetallic MOF-5 (ligands: H2BDC) and Se powder is carbonised at 350 °C in Ar atmosphere to fabricate ZnSe-FSe2.

Despite the wide adoption of the above-mentioned thermal selenisation treatment, it faces a major issue with the loss of Se due to its evaporation if the heating temperature is above its boiling point (~685 °C) [89]. A large amount of vaporised Se is carried away by the carrier gas before they can react with MOFs or MOF derivatives. This issue is not just applicable to the synthesis of metal selenides but also in the case of phosphorisation and sulphurisation. In this case, as illustrated in Fig. 2.16, Yang and colleagues proposed a confined thermal treatment method to fabricate Se/CoSe2/C composite from ZIF-67 [90]. In their work, the single-step carbonisation and selenisation process is carried out in a vacuum-sealed glass vessel. This confined reaction space helps to preserve vaporised Se, which can maximise the Se loading in the MOF-derived carbon composite. This does not just mean a thorough selenisation reaction between Co and Se to generate CoSe2 but also facilitates the vapor deposition of elemental Se, leading to the formation of Se/CoSe2/C composite.

Fig. 2.16
figure 16

An illustration of the single-step carbonisation and selenisation treatment of ZIF-67 with the Se powder in a vacuum-sealed glass vessel, leading to the formation of MOF-derived Se/CoSe2/C composite. Reproduced with permission [90]. © 2019 Elsevier

Full size image

2.4.6 MOF-Derived Metal Nitrides

Metal nitrides (MxNy) are a group of interstitial compounds, which means nitrogen atoms in MxNy usually occupy the interstitial sites of the parent metals. Thanks to their characteristic chemical structures, metal nitrides demonstrate promising electrical conductivity, chemical stability and electrocatalytic activity, particularly in terms of their application in electrochemical energy conversion and storage [91]. Therefore, it is inevitable for material researchers to take the structural and compositional advantages of MOFs to engineer the nanostructures and chemical compositions of MxNy.

NH3 was the most frequently used the source of nitrogen to react with metals and form MOF-derived MxNy [92, 93]. For example, in Liu et al.’s work [94], The authors firstly convert a Co-MOF (ligands: 2-methylimidazole) into NiCo2O4 in two steps: the hydrolysis of Co-MOF by Ni(NO3)2 ethanol solution to generate Ni-Co LDH and the oxidation of Ni-Co LDH in air at 350 °C to generate NiCo2O4. Then, NiCo2O4 is nitridised by means of thermal treatments with NH3 to form a heterostructure of Ni-doped Co–Co2N at 350 °C. Lower NH3 treatment temperatures of 250 and 300 °C lead to the formation of Ni-doped CoN and Co2N, respectively. This is due to the simultaneous reduction and nitridisation of NiCo2O4 by NH3 at these comparatively lower temperatures. However, when the treatment temperature increases to 350 °C, NH3 decomposes into N2 and H2, leading to an enhanced reduction effect and thus elemental Co appears to form a heterostructure with Co2N. When the temperature is further elevated to 400 °C, the nitridisation effect is superior than that of reduction, and all Co components are nitridised into the Co3N phase. Lai et al. synthesised a Ni–Co nitrides/carbon hybrid nanocages from ZIF-67 by using NH3 as the source of nitrogen and following the same two-step of LDH formation and nitridisation process [95]. Instead of multiple stages of transformation, there are also reports of fabricating MOF-derived MxNy through just a in situ single carbonisation and nitridisation step in NH3. For example, a porous carbon doped with Co5.47 N nanoparticles was directly formed by the in situ carbonisation and nitridisation of ZIF-67 at 700 °C in Chen et al.’s work [96]. Another example was demonstrated by Kang et al. for the in situ formation of Co3N-doped carbon nanocubes derived from a PBA-type MOF Co3[Co(CN)6]2 in a single-step carbonisation and nitridisation in NH3 at 450 °C [97] (Fig. 2.17).

Fig. 2.17
figure 17

a The two-step oxide formation and nitridisation process in Liu et al.’s work [94]. Reproduced with permission. © 2018 American Chemical Society. b The single-step carbonisation and nitridisation process in Kang et al. work. Reproduced with permission [97]. © 2019 Springer

Full size image

In terms of the formation of metal nitrides with intrinsic nitrogen from organic linkers, there is a report of the derivation of Co/CoN/Co2P ternary carbon composite from a Co-based MOF ([Co3(pimda)2(H2O)5]) with H3pimda (H3pimda = 2-propyl-1H-imidazole-4,5-dicarboxylic acid) as the bridging ligands [98, 99]. Although the formation of Co2P is carried out by the frequently applied thermal treatment method with NaH2PO2, since H3pimda is a N-containing ligand, the formation of CoN purely relies on this intrinsic source of nitrogen from the H3pimda ligand rather than any external chemical source. Apart from the above-discussed example, there is another report of the synthesis of CoN3-doped carbon derived from Zn-ZIF-67, that is, CoZn bimetallic MOFs with a range of Co/Zn ratios [100]. The authors propose that the formation of CoN3 species is attributed to the release of NH3 from decomposed 2-mehylimidazole ligands. The formation of the microporous channels by Zn evaporation is also important for the diffusion of NH3 to the Co sites, where nitridisation reactions can take place. In order to demonstrate the determining role of ZIF-67 in the formation of CoN3, the authors pyrolysed a mixture of Co and Zn salts with the 2-methylimidazole ligands, where no CoN3 is found in the pyrolysis products.

In summary, the structural and chemical diversities of MOFs make them ideal precursor materials to fabricate a variety of MOF-derived metal compounds with designed nanostructures and chemical compositions, in order to fulfil their functions in a wide range of applications. Metal compounds can be derived MOF precursors by means of either high temperature thermal treatments or chemical reactions at comparatively lower temperatures (such as room temperature). In the case of high temperature thermal treatments, on the one hand, non-metallic elements in MOF-derived metal compounds can come from extrinsic chemical sources in the form of reactive gas atmospheres, decomposed products from reactive solid chemicals and elementary substances in powdery form. On the other hand, they can come from decomposed organic linkers during thermal treatments. In the case of chemical reactions at comparatively lower temperatures, non-metallic elements can convert MOF precursors into metal compounds by means of either hydrothermal or solvothermal reactions based on the hydrolysis and anion exchange mechanisms. In comparison, the latter material synthesis route is more favourable due to the consideration of material and energy savings.