Structure-directing effects of ionic liquids in the ionothermal synthesis of metal–organic frameworks

The ‘ionothermal’ synthesis of metal–organic frameworks (MOFs) in ionic liquids (ILs) has led to a large number of new MOFs, often by incorporation of the IL cation as a templating counter-ion for an anionic MOF framework.


Introduction
Metal-organic frameworks (MOFs) are crystalline framework materials that consist of metal atoms or cluster 'nodes' joined by multifunctional organic linking 'struts'. The prototypical example, Yaghi's MOF-5 (Li et al., 1999) (see Fig. 1), serves to illustrate the important features of MOFs in general. MOF-5 has the formula Zn 4 O(BDC) 3 (BDC 2À = benzene-1,4-dicarboxylate, the dianion of terephthalic acid) and contains two 'secondary building units' (SBUs), i.e. an oxygen-centered Zn 4 O cluster and the terephthalate dianion linker. The connection geometry of the SBUs that comprise a MOF often lead to a predictable structure for the MOF itself (Kim et al., 2001), and that is the case for MOF-5. The Zn 4 O cluster connects to six carboxylate groups in an octahedral arrangement, while the BDC 2À dianion connects to two Zn 4 O clusters in a linear arrangement. The combination of the two SBUs leads to the cubic open-framework structure of  There are large open pores in MOF-5: the yellow and orange spheres of Fig. 1 represent these open spaces, with diameters of 15.1 and 11.0 Å , respectively, and the aperture between pores is large enough to allow passage of a sphere of diameter 8.0 Å (Li et al., 1999). This accessible pore structure endows MOFs with their most interesting and potentially useful properties, and in fact MOFs are often defined to include only porous materials. MOFs have been examined as materials for hydrogen storage (Suh et al., 2012), carbon dioxide capture (Sumida et al., 2012;Liu et al., 2012), gas separations (Banerjee et al., 2015), chemical sensing (Kreno et al., 2012), biomedical applications (Horcajada et al., 2012), and many other applications (Zhou et al., 2012).
There is some disagreement about what materials should be defined as MOFs. Omar Yaghi, who actually coined the term 'metal-organic framework', has suggested that it should refer to networks in which the bonds between metals and the organic linkers are formed with a functionality on the organic linker that would have a negative charge when not bound to the metal, as in a carboxylate, rather than be neutral, as in the N atoms of the pyridyl groups in 4,4 0 -bipyridine (Tranchemontagne et al., 2009). Other functionalities that can form such bonds include imidazolate (and other azolates) and phosphonates. Such bonds are usually significantly stronger than the coordinate covalent or dative bonds formed by a pyridyl or similar moiety, which means that metal-carboxylate or similar bonds generally lead to more stable frameworks, but at the same time makes obtaining crystalline materials more challenging (Jiang et al., 2016). Yaghi has suggested that the term coordination polymer be used to describe network materials formed by the weaker dative metal-ligand bonds, as in, for example, the diamondoid framework formed by the reaction of Cu + and tetra(4-cyanophenyl)methane (Hoskins & Robson, 1989). On the other hand, an International Union of Pure and Applied Chemistry (IUPAC) recommendation on nomenclature suggested that a coordination polymer refer to any coordination compound with extended bonding in one, two, or three dimensions (with bonds of either type referred to above), a coordination network refer to a coordination polymer with extended bonding in two or three dimensions, and a metal-organic framework refer to a porous coordination network (Batten et al., 2013). In any case, the vast majority of materials in this review are formed via metal-carboxylate linkages, and thus have the type of bonding appropriate for MOFs under Yaghi's definition. Many are porous (or potentially porous) and are therefore MOFs under the IUPAC definition; however, many are not porous. Nevertheless, the nonporous inorganic-organic framework materials (usually based on carboxylate linkages) are called MOFs by their creators, and are included in this review.
The topic of this review is the use of ionic liquids (ILs) in the synthesis of MOFs. Ionic liquids are usually defined as materials that are composed entirely of ions and that have a melting point below 373 K (Rogers & Seddon, 2003). Some of the most common ILs are based on 1,3-dialkylimidazolium cations and weakly coordinating inorganic anions, such as tetrafluoroborate (BF 4 À ), hexafluorophosphate (PF 6 À ), trifluoromethanesulfonate (triflate, OTf À ) and bis(trifluoromethanesulfonyl)imide (NTf 2 À ) (see Fig. 2). Many of the 1,3dialkylimidazolium cations are derived from 1-methylimidazole and thus have a methyl group as one of the alkyl substituents; a common abbreviation for such cations uses a C x designation for the other alkyl substituent (x = the alkyl-chain length) and 'mim' for the methylimidazole portion, such that 1-ethyl-3-methylimidazolium is [C 2 mim] + and 1-butyl-3methylimidazolium is [C 4 mim] + (see Fig. 2). While weakly coordinating and relatively chemically inert anions such as those listed above are most commonly used in ILs, several of the ILs in this review contain halide anions, such as chloride or bromide, which can result in the anions themselves being incorporated into the MOF framework.
A number of the properties of ILs are quite different from those of traditional solvents, and they can offer practical advantages and also access to new materials in the synthesis of MOFs. For example, ILs have extremely small or zero vapor pressure, which means that moderate-to-high-temperature synthetic reactions can be carried out in ILs with no need for either reflux condensers or a sealed system that can contain the autogenous pressure that would be generated with a traditional solvent. In contrast, a MOF synthesis in dimethylformamide (DMF), for example, would be limited in temperature range to the boiling point of DMF (425 K) with a reflux condenser, or would require a sealed vessel that can withstand the pressure generated by DMF at temperatures above its boiling point. The traditional synthesis of MOFs occurs in organic solvents at elevated temperatures, termed 'solvothermal' synthesis; by analogy, synthesis in ionic liquids is termed 'ionothermal' synthesis.
The most important property of ILs in the synthesis of MOFs is the presence of high concentrations of ions, and the opportunity to tune the properties of those ions to obtain new materials. Because charged ions are present, it is possible to create MOF frameworks with a net charge, with that charge topical reviews IUCrJ (2017 Hermes et al. (2006). Published by The Royal Society of Chemistry. Used with permission.

Figure 2
Structures of two common 1,3-dialkylimidazolium cations, i.e. [C 2 mim] + and [C 4 mim] + , and two common anions, i.e. OTf À and NTf 2 À , in ionic liquids. Table 1 Metal-organic frameworks and the ionic liquids in which they were synthesized. counterbalanced by ions of the IL incorporated into the structure. For example, the 1,3-dialkylimidazolium cation of an IL can be incorporated into the pores of an anionic MOF framework. The anions of an IL can also become part of a MOF, usually coordinated to the metal center(s) of the MOF. In both cases, the ions have an important structure-directing effect. It is also possible for an IL to act simply as a solvent for MOF synthesis, with no part of the IL incorporated into the final structure. All of these situations will be discussed. There were reviews published in 2007 and 2009 concerning the ionothermal synthesis of both MOFs and zeolites (Parnham & Morris, 2007;Morris, 2009), a review covering the use of ionic liquids in the synthesis of a variety of inorganic materials (including MOFs) (Freudenmann et al., 2011), a review on the synthesis of MOFs in ILs, supercritical CO 2 , and IL/supercritical CO 2 mixtures (Zhang et al., 2016a), and a review of the crystal engineering of MOFs, including a section on synthesis in ILs (Seoane et al., 2016). However, none of these reviews were comprehensive on the topic of the ionothermal synthesis of MOFs, and this review will serve to update and expand upon the previous reviews.
This review is roughly organized by the type of structuredirecting effect the IL has upon the MOFs, generally based on the part of the IL (cation, anion, neither, or both) that is incorporated into the MOF structure. However, some MOFs are discussed in a section to which they do not technically belong according to this categorization, such that they can be discussed in conjunction with related MOFs. Table 1 lists all the MOFs included in this review, along with the ILs in which they were synthesized, categorized by the section of the review in which they are discussed.
2. Ionothermal synthesis in which neither the cation nor anion of the IL is incorporated into the MOF As mentioned in the Introduction (x1), ILs have properties that can be advantageous over traditional solvents, such as their extremely low vapor pressures. They can thus be a useful replacement for organic solvents such as DMF, even when there is not a clear structure-directing role for the IL. Those syntheses will be explored in this section.
In our first example, the MOF Zn 4 (BTC) 2 ( 4 -O)(H 2 O) 2 was synthesized by the reaction of Zn(NO 3 ) 2 Á6H 2 O with benzene-1,3,5-tricarboxylic acid (H 3 BTC) in [C 2 mim][Br] (Xu et al., 2008c). It has a three-dimensional structure in which the [Zn 4 ( 4 -O)(H 2 O) 2 ] 6+ SBU connects to six carboxylate groups, but not in the highly symmetric octahedral geometry observed in MOF-5 and other MOFs. Instead, two of the Zn 2+ ions have coordinated water ligands and the asymmetry of the SBU leads to Zn 4 (BTC) 2 ( 4 -O)(H 2 O) 2 crystallizing in a noncentrosymmetric space group. It forms overall a (6,3) network in three dimensions. Neither the [C 2 mim] + cation nor the Br À anion is incorporated into the structure, but the choice of cation is important, as will be seen in later sections where other Zn-BTC MOFs do incorporate other imidazolium cations and adopt very different structures.
A set of MOFs containing polyoxometalate anions and other metal cations were synthesized in [C 2 mim][Br] (Fu et al., 2012(Fu et al., , 2011 Two MOFs have been reported that incorporate the neutral molecule 1-methylimidazole (1-mim) into their structure, in both cases coordinated to a metal center and therefore clearly with a structure-directing effect. In one case, 1-mim was purposefully added as a reactant in the synthesis, where the reaction of Ni(NO 3 ) 2 Á6H 2 O, H 2 BDC, 1-mim, and NEt 3 in [C 3 mim][Br] at 443 K yielded yellow-green (1-mim)Ni(BDC) (Hogben et al., 2006). The nickel-containing nodes consist of two Ni(1-mim) units bridged by four carboxylate groups, forming a 4-connected square SBU, which combine with the linear BDC linkers to form a square net. The IL was critical to the formation of the MOF in significant yield and good purity -molecular solvents were unsuccessful. In a different synthesis, the three-dimensional (3D) MOF Zn 3 (BDC) 3 (1mim) 2 was synthesized in [C 4 mim][BF 4 ], where the source of 1-mim is apparently decomposition of the [C 4 mim] + cation .
Reports occur fairly often in which the use of ILs in synthesis are emphasized, but the actual role of the IL is the straightforward replacement of another reagent in an otherwise standard solvent-based synthesis (Kelley & Rogers, 2015). In MOF synthesis, this can occur when an IL is dissolved in a conventional auxiliary solvent, making its role similar to any other linker ligand. Nevertheless, there are cases when the IL form of a reagent may confer advantages.   (Mondal et al., 2014). The reaction is essentially an in situ ligand synthesis, and neither IL ion is incorporated in the final structure. However, the use of neutral 4,5-dicyano-2-methylimidazole in this reaction failed to generate the ZIF, indicating that the role of the IL was to supply and stabilize the azole in its more reactive deprotonated form. Changing the counter-ion from [NEt 4 ] + to [C 8 mim] + resulted in the crystallization of a more porous variant of the same structure with better gas-sorption properties, indicating that the cation is able to provide tunability despite not being involved in the reaction or directly incorporated in the structure (Mondal et al., 2016). A set of four zinc-imidazolate MOFs/ZIFs were synthesized ionothermally in [C 2 mim][NTf 2 ] (Martins et al., 2010). In no case was either the cation or the anion of the IL included in the MOF or ZIF, but varying the reaction conditions and zinccontaining precursors resulted in different products. Two of the compounds, i.e. Zn(C 2 O 4 )(C 3 N 2 H 4 ) (C 2 O 4 = oxalate and C 3 N 2 H 4 = imidazole) and Zn(C 3 N 2 H 3 ) 2 , were known previously, but synthesized ionothermally for the first time. The MOF Zn(OAc)(C 3 N 2 H 3 ) resulted from the reaction of Zn(OAc) 2 Á2H 2 O and imidazole in [C 2 mim][NTf 2 ] at 423 K for 48 h. Zn(OAc)(C 3 N 2 H 3 ) has a 3D framework resulting from the bridging of both imidazolate and acetate linkers between tetrahedral Zn ions. The final compound is Zn 4 (C 3 N 2 H 3 ) 8 -(C 3 N 2 H 4 ), formed from the reaction of Zn(NO 3 ) 2 Á6H 2 O and imidazole in [C 2 mim][NTf 2 ] at 403 K for 48 h. The structure of Zn 4 (C 3 N 2 H 3 ) 8 (C 3 N 2 H 4 ) is a fairly complex 3D framework, but is notable mainly for the presence of terminal imidazole ligands. Here it is the reaction temperature that likely controls the structure formation, as reaction at 423 K leads to the fully condensed structure Zn(C 3 N 2 H 3 ) 2 .
In another series of reactions, nanoparticles of Prussian blue, Fe 4 [Fe(CN) 6 ] 3 , and a number of Prussian blue analogs, M 3 [Fe(CN) 6 ] 2 (M = Ni, Cu, Co), were synthesized in imidazolium ILs [C n mim] [BF 4 ] and [C n mim][Cl] (n = 2, 4, 10) (Clavel et al., 2006;Larionova et al., 2008) 3À to form a 3D cubic network, as dictated by the octahedral coordination geometry of each metal center and the linear linkage of the cyanide ligands. Similar Prussian blue analogs have long been known (Entley & Girolami, 1994, 1995, but this synthesis in ILs leads to stabilized nanoparticles of the materials, likely through some type of electrostatic interaction of the IL ion(s) with the exterior of the particles.

Introduction
Long before the first MOFs were synthesized, zeolites were well known and industrially significant framework materials with porosity on the molecular scale (Sels & Kustov, 2016). Some zeolites are naturally occurring aluminosilicate minerals and many others have been synthesized in the laboratory and on an industrial scale. Within the aluminosilicate framework  of a zeolite, both Si and Al atoms are bonded to four O atoms in a tetrahedral geometry, which places a net negative charge on the framework for each Al atom present. The negative charge of the framework is countered by weakly bound cations, often alkali cations such as Na + or K + , that reside within the galleries, such that a typical zeolite formula would be Na x [(AlO 2 ) x (SiO 2 ) y ]ÁzH 2 O. The alkali metal cations are often exchangeable with other cations, which leads to one of the important applications of zeolites, namely ion exchange. Zeolites are usually synthesized hydrothermally from reactants such as sodium silicate and sodium aluminate in varying proportions. If a zeolite is synthesized with a tetraalkylammonium cation in place of the alkali-metal cation, the organic cation can act as a structural template, as it is incorporated into the zeolite galleries (Lok et al., 1983), and it is possible to computationally design appropriate tetraalkylammonium cations for the synthesis of specific zeolites (Davis et al., 2016). More recently, ILs have been used as the solvent in the ionothermal synthesis of zeolites, with the IL cations acting as structure-directing agents (Cooper et al., 2004).
In an analogous manner, the ionothermal synthesis of MOFs in ILs allows the creation of anionic MOF frameworks with the cation of the IL incorporated into the galleries of the MOF to balance the charge and act as a template around which the framework forms. This is the most common way in which an IL acts as a structure-directing agent in MOF synthesis, and several examples will be described in this section.
3.2. Benzene-1,3,5-tricarboxylate (BTC 3À ) MOFs  . The anionic [Cd(BTC)] À framework contains Cd 2 units with six carboxylate groups coordinated to them, four of which coordinate only one Cd cation and two of which bridge Cd cations. The overall connectivity can be viewed as an MX 2type, with M representing a Cd 2 unit and X as the BTC 3À ligands. The authors of this study attempted ion exchange of the [C 2 mim] + cations with Cu 2+ and other metal cations, but those attempts were unsuccessful.
A set of Mn-BTC MOFs demonstrate the interesting effects of the IL cation on the structure obtained, and also an influence of the anion even though the anion is not incorporated into the MOF structure (Xu et al., 2013). [C 3 mim][Mn(BTC)], with no halide present, and somewhat different framework structures, and either 2 or 3 must be the thermodynamically more stable structure. However, with no halide present in the product, it must be a kinetic effect that leads to the formation of 2 for X = Cl À or Br À and 3 for X = I À . The syntheses were performed under identical conditions (except for the identity of the halide in the IL), so it is the coordinating ability of Cl À and Br À versus I À that likely leads to the mineralization of 2 versus 3.
If reactions similar to those above are performed with nickel or cobalt precursors (reaction of Ni (OAc)   . This compound contains Co 2 units with two carboxylate groups bridging the Co atoms and two monodentate carboxylate groups and one water on each Co atom such that the Co 2 unit forms a 6-connected node. These link to form an anionic [Co 2 (BTC) 2 (H 2 O) 2 ] 2À framework, with [C 3 mim] + cations within the galleries. This product can be contrasted with the [C 2 mim] 2 [Co 3 (BT-C) 2 (OAc) 2 ] obtained from a reaction in [C 2 mim][Br], as discussed above (Lin et al., 2006 Fig. 4). The six BDC 2À linkers thus project outward roughly in one plane, leading to 2D nets of [Co 3 (BDC) 3 Br 2 ] 2À , as shown in Fig. 4. Because the [Co 3 (BDC) 3 Br 2 ] 2À framework is 2D, it can accommodate a range of cation sizes in the cavities, from [C 2 mim] + to [C 4 mim] + , with some changes in the in-plane structural parameters and expansion in the third dimension. However, at the largest cation size examined, [C 5 mim] + , a different MOF is obtained, [C 5 mim] 2 [Co 3 (BDC) 4 ], which has a three-dimensional framework. It contains Co 3 units very similar to the other compounds, with six BDC 2À linkers similarly directed outward to form a 2D framework. In place of the terminal bromide ligands are now two BDC 2À carboxylate groups, and those serve as pillars to connect the 2D layers into a 3D framework, as shown in Fig. 4. It is not obvious what causes the change in favored structure from 2D [C 4 mim] 2 [Co 3 (BDC) 3 Br 2 ] to 3D [C 5 mim] 2 [Co 3 (BDC) 4 ], but the change does show the strong structure-directing effect of the IL, since the addition of one methylene group to the cation results in a significant structural change. When similar reactions were performed in ILs with chloride or iodide anions, all the products were 2D MOFs [C n mim] 2 [Co 3 (BDC) 3 X 2 ] (n = 2, 3, 4, 5; X = Cl, I) isostructural with compounds 1-3 in Fig. 4 (Zhang et al., 2016b).
Another example of the reaction of Co(NO 3 ) 2 Á6H 2 O with H 2 BDC in [C 2 mim] [Br], in this case with the addition of imidazole, shows the influence that additional additives can have on the MOF formed . The product is Co 3 (BDC) 3 (imidazole) 2 , which contains Co 3 units with a total of six bridging carboxylate groups, as in the MOFs of Fig. 4, but with the ends of the Co 3 units capped by imidazole rather than bromide ligands, resulting in a neutrally charged framework. In addition, these 6-connected nodes are joined by the BDC 2À linkers to form a 3D network, in contrast to the 2D anionic framework of [C 2 mim] 2 [Co 3 (BDC) 3 Br 2 ] in Fig. 4 ], which is isostructural with the analogous cobalt compound described above   (Cao et al., 2014). In both cases, there are chloride anions coordinated to the lanthanide (Ln) atom (originating from the LnCl 3 Á6H 2 O reactant), rather than the bromide ions that are much more abundant from the IL, presumably due to the fact that the hard Lewis-acid Ln 3+ ion favors the harder Cl À base over the softer Br À base. In the [Sm 2 (BDC) 3 (H 2 BDC)Cl 2 ] 2À framework, there are Sm 2 (RCOO) 6 (RCOOH)Cl 2 nodes, with all of the carboxylate groups bidentate and several bridging, such that each Sm center is nine-coordinated. The eight connections from the Sm 2 SBU do not form the commonly observed bcu net topology, but instead form a new type of 8-connected net (Cao et al., 2014). For the Eu and Tb MOFs, there are infinite chains of Ln 2 ( 2 -Cl)(RCOO) 6 SBUs, with carboxylate groups bridging to form the chains and with six BDC 2À linkers (per Ln 2 unit) extending in two dimensions, forming an overall hex network.
Blue crystals of the MOF [C 2 C 2 im][NaCu(BDC) 2 ] (C 2 C 2 im = 1,3-diethylimidazolium) were isolated by the reaction of Cu(NO 3 ) 2 Á3H 2 O and H 2 BDC at 453 K in the mixed IL solvent system of [C 2 mim] [BF 4 ] and [C 2 mim][l-lactate] . This raises the obvious question of the origin of the C 2 C 2 im + and Na + cations. According to the authors of the study, Na + is present as an impurity in the [C 2 mim][BF 4 ] IL. The C 2 C 2 im is formed in situ during the ionothermal synthesis by dealkylation and re-alkylation of [C 2 mim] + cations. The structure of [C 2 C 2 im][NaCu(BDC) 2 ] consists of infinite chains of alternating Na + and Cu 2+ ions bridged by carboxylate groups, with the BDC 2À linkers bridging to other Na + -Cu 2+ chains to form a 3D framework of pcu topology. The formation of [C 2 C 2 im][NaCu(BDC) 2 ] rather than [C 2 mim][NaCu(BDC) 2 ] or an alternative [C 2 mim] +containing MOF shows again the strong structure-directing effect of the IL cation, as the thermodynamic drive to form [C 2 C 2 im][NaCu(BDC) 2 ] apparently overwhelms the fact that [C 2 mim] + is present in much higher concentration, especially at the beginning of the reaction.  . The nodes comprise Mg 3 (RCOO) 8 (1-mim) 2 (H 2 O) 2 SBUs, with three Mg ions in a linear arrangement bridged by carboxylate groups and water molecules and a 1-mim ligand on each of the terminal Mg ions. The nodes are linked through eight 1,4-NDC units, forming a bcu network topology. [C 3

Other linking groups
0 mim][Cl] is known to thermally decompose to allyl chloride and 1-mim (Hao et al., 2010), and that is the presumed source of 1-mim in the MOF. The

Anion structure-directing effects
In an ionothermal synthesis, it is more common for the cation of an IL, rather than the anion, to be incorporated into the MOF structure and act as a template and structure-directing topical reviews IUCrJ (2017). 4, 380-392 agent. That could be viewed as a consequence of the fact that the metals (or metal clusters) that comprise the nodes of the MOF will usually accommodate a coordination number that is higher than their cationic charge (for example, Zn 2+ with a coordination number of 4 in a tetrahedral complex). It is thus possible for the metal atom to bond to more negatively charged linkers than its positive charge, leading to a node (and a framework) with a net negative charge. That negative charge can be balanced by the cation of the IL if the MOF forms as an open framework that accommodates the cation (and the cation thus acts as a template). However, there are many cases in which the IL anion becomes part of the MOF, usually when the anion is somewhat coordinating (such as a halide) and binds to the metal center in the MOF. Examples of anion incorporation, and other forms of anion structure-directing effects, are discussed in this section.
In one interesting study, the IL anion is not incorporated into any of the MOFs synthesized, yet the identity of the IL anion controls the MOF that is formed (Lin et al., 2007b) (Xu et al., 2013), both also discussed above. [Co 5 (OH) 2 (OAc) 8 ]Á2(H 2 O) has an extended framework structure with most of the bridges between Co atoms formed by acetate ligands and could be considered a MOF for that reason, even though it incorporates none of the intended BTC 3À linking ligand. In fact, [Co 5 (OH) 2 (OAc) 8 ]Á2H 2 O had been synthesized previously by the hydrolytic decomposition of Co(acac) 3 (acac = acetylacetonate) in THF/H 2 O at 528 K (Kuhlman et al., 1999). In [C 2 mim][Co 2 (H 2 BTC) 3 (HBTC)(2,2 0bipy) 2 ], the 2,2 0 -bipy ligands leave four coordination sites available on each Co atom, which are filled by monodentate carboxylate groups from the BTC ligand, forming 2D (4,4)nets, joined by hydrogen bonds.
Two  (Jin et al., 2002). The framework has a +1 charge per formula unit; note that the Cu 2+ ions have been reduced to Cu + during the ionothermal reaction, most likely by the bpp ligand. In [Cu(bpp)][BF 4 ], the Cu + ions are effectively two-coordinated, with a linear coordination of the two pyridyl groups, forming one-dimensional polymers of Cu(bpp) + . There are short Cu + -Cu + contacts at 3.002(2) Å and, if those are considered, there is a 2D framework with BF 4 À anions interstitial between the layers. A similar synthesis involves the reaction of Cu(NO 3 ) 2 Á3H 2 O and 2,4,6-tris(4pyridyl)-1,3,5-triazine (tpt) in [C 4 mim] [BF 4 ] to yield [Cu 3 -(tpt) 4 ][BF 4 ] 3 Á 2 3 tptÁ5H 2 O (Dybtsev et al., 2004). Once again the copper has been reduced from Cu 2+ to Cu + . All of the Cu I ions are tetrahedrally coordinated by the tpt ligands, which function as trigonal planar linkages. The overall net is of ctn-a topology.

Cation and anion incorporation/combined control
There are several examples of ionothermal syntheses of MOFs in which both the cation and anion of the IL are incorporated into the MOF, or otherwise have some structure-directing effect. The most common way that both IL ions become part of the MOF is when the anion coordinates to the metal center of the resulting anionic MOF framework, while the IL cation acts as a template for the open galleries for the MOF, where it acts as a counter-ion to balance the charge of the framework. Examples of this type and others are discussed below.
A set of six Zn-BTC MOFs synthesized in [C n mim][Br] ILs demonstrate the structure-directing effects of the IL cation and also the anion (in comparison with another synthesis in [C 4 mim][I]), and also the importance of the reaction stoichiometry (Xu et al., 2007) (Xu et al., 2008b). The presence of iodide leads to the creation of a novel SBU, namely the tetrazinc node Zn 4 (OH) 2 -I 2 (RCO 2 ) 6 .
Moving one position downward in the periodic table from Zn to Cd, three Cd-BTC MOFs were prepared ionothermally in [C n mim][X] (n = 2, 3; X = Cl, Br, I) (Xu et al., 2008a) Fig. 5). In both structure types, there are linear Ni 3 units bridged by BTC 3À carboxylate groups, which are then linked to form 2D layers. Those 2D layers are in turn linked to form 3D frameworks, as shown schematically in Fig. 5. The main point of interest in this set of MOFs is the formation of two different MOF structure types, in some cases with the same cation. The cavities within the anionic framework of the B structure are larger than those in the A structure, so it is unsurprising that the B structure tends to form for the larger cations. However, the formation of two different structures containing the same cation (A2 and B1; A3 and B2) indicates that there may be a kinetic effect where the IL halide anion induces the crystallization of one structure type in favor of the other. Another possibility is that the relative basicity of the halide anions of the IL exerts thermodynamic control on the structure type obtained: structure B, in which the BTC is singly protonated, is favored when less basic anions are used.

Other structure-directing effects
In one set of Co-HBTC MOFs, it is the hydrogen bonding of various added amines that directs the structure that is formed (Lin et al., 2008) (Ji et al., 2008). The porous anionic framework of [Cd 3 -(BTetC) 2 ] 2À contains Cd 3 units bound by eight carboxylate groups of four BTetC 4À linkers. These Cd 3 units are linked in three dimensions to leave open one-dimensional (1D) channels, which contain [K 2 (H 2 O) 8 ] 2+ chains. Each K + is eightcoordinate, with two monodentate waters of coordination, four bridging waters to form the chains, and two carboxylate groups that are also part of the [Cd 3 (BTetC) 2 ] 2À network.
There are two reported ionothermal syntheses of a single enantiomer of chiral MOFs using ILs containing chiral and enantiomerically pure anions. In neither case is the IL anion incorporated as part of the MOF, but the chirality of the anion directs the MOF structure toward only one enantiomer. In the first example, Ni(OAc) 2 Á4H 2 O was reacted with H 3 BTC in [C 4 mim][l-aspartate] to form the homochiral MOF [C 4 mim] 2 [Ni(HBTC) 2 (H 2 O) 2 ], which crystallizes in the space group P4 1 2 1 2 (Lin et al., 2007a). The nodes of the structure consist of Ni coordinated by four monodentate carboxylate groups and two water molecules, and these 4-connected nodes are linked to form a diamondoid (dia) network. The chirality of the MOF results from the 4 1 screw axis, which leads to a helical arrangement of the Ni SBUs. If the reaction is performed in [C 4 mim][d-aspartate], the same compound is obtained with the opposite chirality. In the second example of this type of chiral induction, the reaction of Cu(NO 3 ) 2 Á3H 2 O with H 2 -1,4-NDC in a mixture of [C 2 mim][l-lactate] and [C 2 mim][BF 4 ] yielded [C 2 mim][NaCu(1,4-NDC) 2 ] . The Na + originates from an impurity in the [C 2 mim][BF 4 ] IL. The compound crystallizes in the space group P4 1 , and it is once again the 4 1 screw axis that is the origin of the chirality.

Conclusions
A large number of MOF and MOF-like framework materials have been synthesized in IL solvents. The ILs very often exert structure-directing effects, commonly through the incorporation of the IL cation (as a charge-balancing ion within the galleries of the MOF), anion (usually coordinated to a metal center within an overall negatively charged MOF framework), or both. Even when no part of the IL is incorporated into the MOF, the IL can exert a structure-directing effect, as occurs when different IL anions direct the crystallization of different MOF polymorphs, or when a chiral IL anion directs the formation of one enantiomer of a chiral MOF.
In the ionothermal syntheses described herein, the ILs very often contain halide anions (most commonly chloride and bromide, but also iodide), certainly much more than is common in other applications of ILs, where weakly coordinating and chemically inert anions, such as PF 6 À and NTf 2 À , have found more widespread use. The moderate coordinating ability of the halides makes them useful in the solubilization of topical reviews the metal-containing precursors for MOFs and crystallization of the MOFs themselves. The coordinating ability of the halides can also lead to their incorporation into the MOF framework. The cations of the ILs in ionothermal MOF syntheses are most commonly of the [C n mim] + type, which is likely due to their easy availability and the fact that their asymmetric structure leads to melting points below 373 K for most [C n mim] + halides. This points to new opportunities for non-imidazolium ILs in MOF synthesis, where different cations would impose new steric constraints on the MOFs and likely lead to new materials. The room-temperature ionic liquid trihexyltetradecylphosphonium chloride, for example, contains a cation that is significantly larger and of a different shape from [C n mim] + cations, and it and related ILs are good candidates for future syntheses of new MOFs by ionothermal methods.