Electrically Conductive Metal–Organic Frameworks

Metal–organic frameworks (MOFs) are intrinsically porous extended solids formed by coordination bonding between organic ligands and metal ions or clusters. High electrical conductivity is rare in MOFs, yet it allows for diverse applications in electrocatalysis, charge storage, and chemiresistive sensing, among others. In this Review, we discuss the efforts undertaken so far to achieve efficient charge transport in MOFs. We focus on four common strategies that have been harnessed toward high conductivities. In the “through-bond” approach, continuous chains of coordination bonds between the metal centers and ligands’ functional groups create charge transport pathways. In the “extended conjugation” approach, the metals and entire ligands form large delocalized systems. The “through-space” approach harnesses the π–π stacking interactions between organic moieties. The “guest-promoted” approach utilizes the inherent porosity of MOFs and host–guest interactions. Studies utilizing less defined transport pathways are also evaluated. For each approach, we give a systematic overview of the structures and transport properties of relevant materials. We consider the benefits and limitations of strategies developed thus far and provide an overview of outstanding challenges in conductive MOFs.


INTRODUCTION
Metal−organic frameworks (MOFs) are a class of porous materials with high surface areas consisting of inorganic nodes connected by organic linkers. 1−3 Since the inception of the field in the late 1990s and early 2000s, these materials have been investigated extensively for applications, such as gas storage, 4−7 separations, 8−11 and catalysis, 12−15 because of their high porosity and chemical tunability. In contrast, the electronic properties of MOFs have received comparatively little attention until recently, and the potential of this class of materials as electrically conductive porous materials has only begun to be realized. Despite the relative youth of this research area, many advances in designing, synthesizing, and characterizing electrically conductive MOFs have been made, and the field has experienced considerable expansion in just the last five years, the focus period for this Review (Figure 1) Electrically conductive MOFs are of considerable interest from many perspectives. They are relevant as active materials for many applications, including electrocatalysis, 16−20 chemiresistive sensing, 21−26 and energy storage technologies. 27−30 Their fundamental transport properties merit further study, as certain conductive MOFs have been predicted to host topologically nontrivial electronic structures. 31−36 Additionally, they may contain arrangements of inorganic or organic moieties that are unprecedented in other materials, potentially giving rise to new physical phenomena. 37,38 They also present exciting new avenues for creativity in ligand design, selfassembly, and crystal engineering.
This Review covers recent advances in the chemistry of this class of materials, with emphasis on design principles, experimental materials discovery, and characterization of transport properties. We focus the majority of our discussion on crystalline coordination networks with potential voids 39 with two-and three-dimensional connectivities for which electrical conductivity values are reported. Where appropriate, we also review chemically related materials (as demonstrated by compositional and structural characterization) that do not exhibit permanent porosity. While it is important to understand the mechanisms underlying the transport properties of the most conductive MOFs, we believe it can be equally instructive to examine structure−property relationships among materials that may not exhibit record conductivity values. In an effort to capture the granularity necessary to establish broader trends, the conductivities of the materials included in this Review span more than 17 orders of magnitude.
Many other reviews pertaining to electrically conductive MOFs have appeared in recent years. Sun et al. first provided an overview of porous conductive MOFs in 2016, 40 and Medina et al. provided an update to this work. 41 Several other perspectives on conductive MOFs, their applications, and outlooks have been written. 42−46 Ko et al., 47 Zheng et al., 48 and Solomos et al. 49 have focused on two-dimensional (2D) conductive MOFs, and Murase et al. analyzed mixed valency in conductive MOFs. 50 Application-focused reviews have covered electrocatalysis 51 and chemiresistive sensing. 52−54 Stassen et al. reviewed MOF integration into electronic devices. 55 Sun et al. described techniques for measuring the conductivity of MOFs. 56 For a critical review of mostly nonporous conductive coordination polymers, we direct readers to the work of Givaja et al. 57

Design Strategies and Mechanisms
Electrical conductivity is a characteristic of a material that quantifies the efficacy of the transport of electrical charge. It is a convolution of the charge carrier concentration, n, and the carrier mobility, μ. 58 Control over either of these quantities is nontrivial and requires deep understanding of the electronic structure of the material, as well as the mechanisms underlying charge transport.
From a fundamental perspective, charge transport in solids can be described by one of two general mechanisms: hopping transport and ballistic (or band-like) transport ( Figure 2). 40 In the former, the charges transfer between discrete, nonbonded sites, where the charge carriers are localized. This mechanism is common to disordered materials, glasses, and organic semiconductors. In the latter, strong interactions between the sites allow for formation of continuous energy bands with delocalized charge carriers. This mechanism is often found in crystalline inorganic materials.
In chemical terms, an analogy can be made between these two mechanisms and the Robin−Day classification of mixedvalence compounds. 59 In a material with strong covalent bonding, where we would expect band-like transport, the electrons are delocalized between the different statessimilar to that of a class III mixed-valence compound. In contrast, charge transport in more ionic materials is better described by the hopping mechanism. In these cases, distinct sites with differing valences exist, and additional energy has to be applied to transfer the electron between them−similar to a class II mixed-valence compound. (Insulating materials with very low conductivities are analogous to class I mixed-valence compounds, in which the valences are trapped and even hopping is inaccessible.) The key difference in observable parameters for the two mechanisms lies in the temperature dependence of conductivity. Hopping-type conductivity is always thermally activated: higher temperatures lead to higher conductivities. This dependence follows an exponential law and is described by σ = σ 0 exp[−(T 0 /T) 1/d ], where σ is the conductivity, T is the temperature, T 0 and σ 0 are constants specific to the material, and d is the dimensionality of the sample. 60,61 In contrast to hopping, band-like transport can be both thermally activated and deactivated. In metallic materials with sufficiently high carrier concentrations, the conductivity is limited by various scattering processes, which are enhanced at higher temperatures, leading to less efficient transport. In semiconductors, the conductivity is generally limited by the charge carrier concentration. At higher temperatures, more free holes or electrons are generated, giving higher conductivities. Because very few MOFs are metallic, the specific temperature dependences for metals will not be discussed. The conductivity of semiconductors often follows an Arrhenius dependence, σ = σ 0 exp[−(E a /kT)], where E a is the activation energy. The value of E a is tied to the width of the band gap, as well as the energy levels of dopants. 58 The generalized electronic band structures of an insulator, semiconductor, and metal are illustrated in Figure 3. Generally, an insulator is defined by a large band gap (E g ), usually in excess of ∼4 eV, which results in very low population of the conduction band. A semiconductor has an intermediate E g that allows promotion of free charge carriers to the conduction band (e.g., by thermal or optical excitation). A metal contains at least one band crossing the Fermi level (E F ), resulting in a continuum of allowed electronic states that leads to high conductivities.
In practice, many factors complicate the determination of charge transport mechanisms. Specifically for conductive MOFs, two key issues are involved. First, most measurements in the literature have been carried out on polycrystalline pellets, and not on single crystals. This adds an inherent thermally activated contribution to the resistance from hopping between different grains of the material. 62 Second, significant disorder or defects can lead to creation of trap states, where charge carriers will be localized and would need to hop between. 60,61 It is also easy to imagine, for instance, how a missing linker defect could break up a charge transport pathway and create a barrier to charge transport; other types of defects would have similar effects. 63 Ultimately, a decisive determination of the charge transport mechanism in a material would involve a combination of variable temperature single-crystal conductivity studies, electronic structure calculations, and optical spectroscopy, at the minimum. Few studies, however, provide such in-depth investigation of charge transport in MOFs, making the actual underlying mechanisms difficult to discuss in the context of this Review. Therefore, we have opted to discuss the approaches that researchers have applied to designing conductive MOFs and how successful these strategies have been.
Much of the early work on electrical transport in MOFs took inspiration from similar work on coordination polymers, 64 covalent organic frameworks (COFs), 37 and molecular conductors. 65 In fact, conductive MOFs can be thought of as porous molecular conductors, given that they are indeed composed of molecular building blocks. The general approaches taken toward achieving high conductivities in MOFs also originate from these more established fields. Two key strategies in particular are the so-called through-bond and through-space approaches. The former focuses on improving bonding between the metal and the ligands to achieve stronger charge delocalization; the latter targets noncovalent interactions between organic fragments, particularly π−π stacking, to introduce continuous charge transport pathways. We anticipate that the nature of the transport pathways may affect the transport properties in MOFs, coordination polymers, COFs, and other conductors in similar ways despite other differences in the nature of these materials.
A number of unique characteristics of MOFs, however, tend to limit direct applications of these strategies. In particular, the inherent porosity of MOFs precludes close intermolecular contacts in many structural types, particularly in isotropic structures (i.e., high-symmetry structures where charge transport pathways are identical in all three crystallographic directions). In addition, the majority of MOFs are formed with carboxylate linkers, which usually form relatively ionic bonds with the metals, leading to large energy gaps and trapped valences/confined electronic states. That said, the chemical and structural tunability of MOFs, as well as their unique host−guest chemistry, have allowed for these approaches to evolve and produce materials with a variety of motifs favorable for conductivity (encompassing inorganic and Figure 2. Schematic representations of (a) ballistic band-like charge transport and (b) hopping charge transport. In both modes of transport, electrons move from high to low electric potential energy (denoted as E). In band-like transport, movement of electrons proceeds along a smooth energy landscape, while in hopping transport, movement of electrons is gated by activation barriers (denoted as E a ). organic components). Below, we give a broad overview of the current form of these approaches toward achieving electrical conductivity in MOFs ( Figure 4).
1.1.1. Through-Bond Pathways. Networks of coordination bonds in MOFs can form continuous routes for charge delocalization. In this work, we refer to bonding motifs involving metals and the ligand functional group (but not the organic core) as through-bond pathways. MOFs with continuous SBUs comprising metals and ligand moieties with well-matched energy levels and good orbital overlap can result in small band gaps and high charge mobilities, both of which are favorable for conductivity. Examples of 1D (most common) 66−70 and 3D pathways (much rarer) 68,71,72 have both been reported.
Many reports in this category focus on softer, more electropositive linkers, with sulfur or nitrogen coordinating to the metals, in which the energy matching and metal−ligand orbital overlap are improved. 66−68 Fe-based MOFs, particularly those with azolate ligands containing mixed-valent Fe II/III centers, exhibit the highest conductivities in this class of materials. [68][69][70]72 These directions follow principles previously demonstrated for inorganic chalcogenide chains, 73 as well as conducting metallopolymers. 74 1.1.2. Extended Conjugation. Pairing transition metals with ligands containing chelating functional groups (e.g., ortho-diols, diamines, and dithiols) that are conjugated with the organic core results in frameworks with extended conjugation encompassing both the organic and inorganic components. Some of these MOFs with 2D structures have been proposed to be metal−organic analogs of graphene, referring to the similarity between their extended π−d conjugation within the ab plane, and the sp 2 hybridization of graphene. This π−d conjugation allows for efficient delocalization of charge carriers within the plane, with such frameworks exhibiting the highest conductivities for MOFs. 21,30,75−79 The organic linkers found in MOFs with extended conjugation include dihydroxybenzoquinone, chloranilic acid, and hexa-substituted triphenylenes and benzenes. The redoxactive nature of these aromatic cores is also important for enhancing conductivity: many of these ligands are partially oxidized during solvothermal growth, introducing charge carriers into the as-synthesized frameworks.
For the majority of the 2D-connected frameworks, charge transport within the π−d conjugated planes has been invoked to be the dominant mechanism behind conductivity. We note, however, that π−π interlayer interactions in the bulk structures may also be important for contributing to charge delocalization.
1.1.3. Through-Space Pathways. Organic components participating in π−π interactions with one another can form through-space charge transport pathways. For organic semiconductors, it is well-established that the spatial separation is inversely correlated with the magnitude of the transfer integral and hence the charge mobility. 80 In an analogous way, linkers that have a high propensity for π−π interactions can yield MOFs with high conductivities originating from the organic components. Work in this direction can be seen as extension of prior research on dense coordination polymers based on tetracyanoquinodimethane (TCNQ), which show electrical conductivities above 0.1 S/cm. 81−83 Several well-known building blocks in organic electronics have been incorporated into conductive MOFs with throughspace charge pathways. For example, MOFs with ligands based on tetrathiafulvalene (TTF), an electron-rich organosulfur molecule, exhibit conductivities varying systematically with the stacking distance between the ligands. 84−88 Conductive naphthalene-, anthracene-, and naphthalenediimide-based MOFs also exhibit π−π stacking. 38,89−91 For conductive 2D MOFs that are fully conjugated within the ab plane, the role of through-space conductivity (i.e., transport along the c direction) remains to be fully elucidated. However, recent studies on triphenylene-based MOFs suggest charge transport through π−π stacking indeed contributes to the high bulk conductivities observed. 79  The extended conjugation pathway also involves π−d conjugation including the ligand core (both are shown here for M = a transition metal). (c) The through-space pathway involves π−π stacking of organic moieties (E = S for tetrathiafulvalene, a common component in MOFs).

Redox
Hopping. In practice, it is difficult to unambiguously distinguish between the two general mechanisms through which charge transport in MOFs can proceed (i.e., band-like transport and charge hopping transport). Even in structures with ostensible high-mobility transport pathways, real defect concentrations are unknown, and grain boundary resistances in polycrystalline samples may obfuscate the intrinsic behavior. Hence, hopping conduction cannot be ruled out, especially at lower temperatures. However, for frameworks in which no crystallographic pathways that could facilitate band-like transport can be identified, it is reasonable to assume that conductivity proceeds via a hopping mechanism ( Figure 5). In such materials, the presence of redox-active metals or linkers, as well as small spatial separation between these components, can promote charge hopping.
1.1.5. Guest-Promoted Transport. The porosity of MOFs can present both a challenge and an opportunity with respect to conductivity. Many researchers have exploited the porosity of MOFs to induce conductivity by postsynthetically loading electroactive guest molecules into the framework. These guests then form charge transport pathways throughout the material through guest−guest or guest−framework interactions. A diverse set of approaches to introducing conductivity pathways into such materials exist, encompassing both the aforementioned strategies translated to host−guest interactions, as well as entirely new directions, such as incorporation of extended solid state or polymeric materials within the pores. In all cases involving this strategy, it should be noted that porosity is either significantly reduced or even completely eliminated, as may be expected by occupying the pores with bulky guests.

Electronic Structures of MOFs
Electronic energy levels in MOFs, like other extended solids, are conveniently represented as band structures, where the energies of orbitals are plotted versus k-space (the crystal momentum). 93 Although experimental band structures and densities of states (defined as the number of energy levels versus energy) can be probed using photoemission spectroscopies, in practice, these techniques have only rarely been applied to MOFs. 94,77 Therefore, the vast majority of band structures for MOFs are computed using density functional theory (DFT) calculations. They are commonly referenced to the electrostatic potential inside a pore. 95 The calculated band structures for representative MOFs with insulating, semiconducting, and metallic electronic structures are shown in Figure 6. Zn 4 O(BDC) 3 (BDC = benzene-1,4-dicarboxylate), or MOF-5, has a band gap of 4.6 eV, indicating that it is an insulator. 45 The majority of reported MOFs are expected to be insulators as well due to large energy gaps between their frontier orbitals. The computed band gap of Zn 2 (TTFTB) (TTFTB = tetrathiafulvalene tetrabenzoate) is approximately 1.75 eV, 84 a typical value for a semiconductor. We note that, in the case of Zn 2 (TTFTB), the computed electronic band gap and the optical band gap from a Tauc plot (2.0 eV) are in reasonable agreement. 96 However, in the absence of transport measurements, optical band gaps alone are not sufficient to determine that a compound is a semiconductor. Finally, the calculated band structure of Ni 3 (HIB) 2 (HIB = hexaiminobenzene) contains bands crossing the Fermi level, suggesting metallic character. 77 Ultraviolet photoemission spectroscopy showed a nonzero density of states at the Fermi level, consistent with metallicity. Although temperature-dependent conductivity for polycrystalline samples of this compound indicated thermally activated transport, in seeming contradiction with the metallic nature predicted by computations, the authors proposed that scattering at grain boundaries dominated transport, which is otherwise metallic within each particle. The accuracy of a computed electronic band structure is necessarily determined by the quality of the input crystal structure and the functional used. Significant structural disorder and large unit cells in many MOFs can introduce considerable challenges for band structure calculations. The relevance of computed band structures to the real transport properties of a MOF should, therefore, always be considered in relation to the structural integrity of the model and practical constraints on computational resources.

Practical Considerations
To achieve fundamental understanding of charge transport in MOFs, it is necessary to deconvolute carrier concentration and mobility. Mobility can be obtained directly with noncontact measurements such as time-resolved terahertz spectroscopy (TRTS) or flash-photolysis time-resolved microwave conductivity (FP-TRMC). Hall effect measurements and fieldeffect transistors (FET) also allow determination of carrier mobility and identity (i.e., whether electrons or holes are the majority carriers). However, all of these measurements are challenging and have steep requirements for sample quality and preparation and, thus, have been rarely undertaken in MOFs. Furthermore, results obtained from one technique are not easily compared with those from other techniques. For instance, whereas FP-TRMC reports intrinsic mobility values, these are alternating current mobility values that cannot and should not be compared on a one-to-one basis with mobility values obtained from FET devices, which normally operate under direct current. As a result of all these factors, the majority of publications in this area simply report conductivity values without attempting to probe mobility or carrier concentration. Ideally, mobility values from contact (Hall or FET) measurements are preferable to supplement conductivity values because they provide additional insight into transport properties.
A number of previous reports described in detail the various methods of measuring electrical conductivity on MOFs and other coordination polymers. 56,57 Briefly, conductivity can be measured on either single crystals or on polycrystalline samples in the form of films or powders. Although clearly more technically challenging, single-crystal measurements are preferred because they eliminate contributions from grain boundary resistance, sample anisotropy, and other factors apart from the intrinsic conductivity of the material. In practice, however, because of particle size considerations and the considerable skill and effort required to produce single-crystal devices for many samples, powder or film measurements are used instead. Powders are generally pressed into compact pellets, while films are usually grown on or transferred onto insulating substrates to ensure that only the MOF material contributes to the measured conductivity.
Conductivity measurementsregardless of whether they are carried out on powders, films, or single crystalsare typically performed in either a two-probe or four-probe configuration. The former (Figure 7a) is simpler and involves simply contacting the material with two wires. In the conventional implementation, a known voltage is supplied to the sample, and the current that passes through it is measured to obtain the resistance through Ohm's law. Significant issues may arise when the contact between the wires and the sample is poor, as this configuration does not compensate for contact resistance. The use of the four-probe configuration, in which four wires are attached to the sample, eliminates contributions from contact resistance from the values measured. In the most common linear four-probe configuration, the outer two wires are used to supply a known current, and the inner two wires are used to measure the potential drop caused by the applied current. This method allows for precise determination of the intrinsic resistance of the sample and elimination of contact effects, so long as all contacts are identical and the geometry of the sample is rigorously determined. Other specific configurations for four-probe measurements exist, depending on the morphology of the studied material: for example, the van der Pauw method uses four wires placed around the outer perimeter of a sample.
When four-probe measurements are executed on single crystals, typically the four-contact configuration is used ( Figure  7b). In this method, two sets of wires are attached to the material along the direction where the conductivity is being   56 I and V denote current and voltage, respectively, and L, w, and t denote the length, width, and thickness of the samples, respectively. s denotes the spacing between contacts. R AB and R AC are the resistances measured between contacts A and B and A and C, respectively. F 1 and F 2 are correction factors depending on the geometry of the sample and the values of R AB and R AC , respectively. evaluated. It is important that the contacts cover the full width of the crystal in the orthogonal direction to allow good estimates of the cross-section area through which current is passed. Here, the resistance of the device, R, is the numerical derivative of the measured voltage drop with respect to the applied current, in accordance with Ohm's law. The resistivity can be estimated using the formula L wtR /( ) σ = , where L is the distance between the two inner contacts, w is the width, and t is the thickness of the sample.
On extended films and pellets, contact resistance can be analogously minimized with the four-point measurement (Figure 7c). In this case, four small contacts are attached to the surface, equally spaced along a line. The conductivity can be calculated from sFR , where s is the spacing between the contacts and F 1 is a correction factor dependent on the geometry of the sample.
The van der Pauw method (Figure 7d) is generally less sensitive to the shape of the sample and should be preferred to the linear four-probe configuration when working with nonrectangular samples. For van der Pauw measurements, four small contacts are attached to four corners of the sample. Then, two resistances are estimated, with current applied along one side, and voltage measured along the opposite. Then, the condu ctivity o f t h e sam ple is e st ima ted with , where t is the thickness of the sample, R AB and R AC are the resistances along two orthogonal sides, and F 2 is a correction factor dependent on the two resistances. The contacts are usually made of precious metals, such as Au, Ag, or Cu, and are attached with conductive adhesive, including carbon, gold, or silver paint, or by depositing a thin layer of metal on the surface of the sample to attach the wires to it. Generally, contact resistances can be minimized by matching the work functions of the sample of interest and the contact materials.
Accurate and precise determination of the conductivities of MOFs is the first step toward evaluating their utility in a range of applications. 51−53 For example, electrode materials in energy storage applications, such as supercapacitors or batteries, typically require relatively high conductivities on the order of 0.1 S/cm or higher. 27−30 High electrical conductivity alone is not sufficient, however, as ionic conductivity is crucial to enable operation at application-relevant currents. 27,29 Engineering composites that incorporate less conductive MOFs may also be a promising approach for energy storage applications. 97−99 For chemiresistive sensing, the requirements for conductivity are more flexible, as the electronics and device design can be tailored to accommodate different conductivities. MOFs showing promise as chemiresistive sensors have conductivities ranging from 10 −4 to 8 S/cm. 25,100 Generally, the change in conductivity upon interaction with an analyte is more important than the absolute conductivity itself. 52−54 For electrocatalysis, conductivity requirements are likely to vary significantly depending on the rate of the reaction being studied. The higher the rate of the electrochemical reaction, the more stringent the requirements on the MOF conductivity, because electron transport within the electrode (here, the MOF) should not be the rate limiting step in an electrochemical system. MOFs with conductivities between 10 −5 to 60 S/cm have been explored for different reactions, including oxygen reduction, 17,20 hydrogen evolution, 16 and carbon dioxide reduction. 101

THROUGH-BOND PATHWAYS
The through-bond approach focuses on investigating frameworks with continuous coordination/covalent pathways that can promote efficient charge transport. A number of studies report control over both the carrier concentration and mobility and, thus, the electrical conductivity, owing to the intrinsic tunability of MOFs (Table 1).
In most cases, this approach is realized in frameworks with 1D inorganic SBUs, also known as rod-based MOFs. 102 The majority of such materials employ carboxylate linkers and (−M−O−) ∞ chains. The relatively ionic bonding between transition metals and carboxylates generally leads to poor orbital overlap and low band dispersion−metal and ligand energy levels are relatively localized, leading to low conductivity.
The more successful studies have attempted to solve this issue by increasing the metal−ligand covalency. The general approach so far has been to work with softer ligands, in which the metal-binding atoms are sulfur and nitrogen rather than oxygen. One clear illustration of this concept is substitution of (−M−O−) ∞ chains for (−M−S−) ∞ SBUs in MOF-74/CPO-27 analogs, which led to increased conductivities and mobilities and decreased band gaps. 66,67 As in most other approaches, the carrier concentration is generally controlled through introduction of mixed-valent species, although it is rarely precisely quantified. Specifically, in pyrazolate, triazolate, and tetrazolate MOFs, Fe II/III mixed valency has resulted in high conductivity values that were tunable and correlated with the concentration of charge carriers. 69−72 Interestingly, the abundance of detailed and systematic structure−function relationship studies for throughbond materials allows us to speculate that the limiting factor for electrical conductivity in many of these MOFs is indeed the carrier concentration: the mixed-valence Fe II/III materials show vastly higher conductivities than those with other d-metals, and the conductivities correlate strongly with the extent of mixed valency.
Although the majority of studies investigate rod MOFs with 1D conductivity pathways, select examples with isotropic 3D pathways exist, most notably the framework Fe(1,2,3triazolate) 2 . 68,71,72 That said, the vast majority of MOFs containing 1D rod SBUs or other continuous inorganic coordination motifs have not yet been studied for electronic transport, even though stable mixed valency has been demonstrated in some of these other 1D SBUs. 103−105 Applying principles to increase conductivity in through-bond pathways to these unexplored materials is likely to yield new conductive MOFs with even more structural diversity and tunability.

MOF-74-Type Frameworks
The family of M 2 (DOBDC) (M = divalent metal; DOBDC = 2,5-dioxidobenzene-1,4-dicarboxylate) frameworks, 102,106−112 colloquially known as MOF-74 or CPO-27, exhibit 1D hexagonal channels and infinite SBUs consisting of (−M− O−) ∞ chains ( Figure 8). In the as-synthesized frameworks, which have the formula M 2 (DOBDC)(DMF) 2 , the metal ions occupy a pseudo-octahedral environment, with one DMF molecule coordinated to each metal center. Removal of this DMF molecule through activation allows for direct modification of the metals' coordination geometry, as well as access to coordinatively unsaturated metal sites.  68 The conductivities of the Mg, Mn, Co, Ni, Cu, and Zn materials were very low, between 1.4 × 10 −14 S/cm for Cu 2 (DOBDC)(DMF) 2 and 3.0 × 10 −13 S/cm for Mn 2 (DOBDC)(DMF) 2 . (Similar conductivity values for these materials were also reported by Scheurle et al. 113 ) Sun et al. hypothesized that replacing the hydroxy group of the linker with a thiol group would increase the conductivity of the resulting frameworks due to better energy matching between the frontier orbitals of the thiolate and the metal. Indeed, the framework Mn 2 (DSBDC)(DMF) 2 (DSBDC = 2,5disulfidobenzene-1,4-dicarboxylate) exhibited a higher conductivity by about 1 order of magnitude (2.5 × 10 −12 S/cm) than Mn 2 (DOBDC) (3.9 × 10 −13 S/cm). 66,67 The conductivities of the Fe analogs were approximately 6 orders of magnitude higher than those of the Mn materials, with the DSBDC MOF again being more conductive than the DOBDC MOF (3.9 × 10 −6 S/cm for Fe 2 (DSBDC), and 3.2 × 10 −7 S/ cm for Fe 2 (DOBDC)). 67 The charge mobilities of methanolexchanged and activated Mn 2 (DSBDC), as measured by FP-TRMC, were 0.02 and 0.01 cm 2 V −1 s −1 . 67 These results suggest that replacing (−M−O−) ∞ chains with (−M−S−) ∞ chains creates a more efficient pathway for intrachain charge transport. The authors attributed the dramatically higher conductivities of the Fe MOFs to the loosely bound electrons in the β-spin d band of Fe II (the analogous state is empty for high-spin Mn II ), effectively lowering the band gap for the Fe materials compared to the Mn analogs. Although these original studies did not detect Fe III by Mossbauer spectroscopy, oxidation of a small fraction of framework Fe II centers to Fe III may also explain these differences in conductivity. 68 A similar effect was observed in the M 2 (DOBDC)(DMF) 2 materials, where the conductivity of Fe 2 (DOBDC)(DMF) 2 was significantly higher than for the other metals, at 4.8 × 10 −8 S/cm. 68 Oxidation of some Fe II centers to Fe III because of the low reduction potential of the Fe II/III couple was deemed likely from electron paramagnetic resonance (EPR) spectroscopy studies, despite the handling of Fe 2 (DOBDC)(DMF) 2 under inert conditions. For these materials, the authors attributed the higher conductivity of the Fe II MOF to the relatively higher energy valence orbitals of the high-spin Fe II compared to other metals, leading to a smaller band gap.
Sun et al. also investigated the effects of pore guest DMF molecules on the electrical conductivities of M 2 (DOBDC)-(DMF) 2 and M 2 (DSBDC)(DMF) 2 (M = Fe II , Mn II ). 67 Guestfree Mn and Fe DOBDC and DSBDC frameworks each exhibited lower conductivities by about 1 order of magnitude compared to the as-synthesized materials. This difference was hypothesized to be the result of defects introduced by the solvent exchange and evacuation process. Removal of coordinated DMF in Fe 2 (DSBDC) was found to further reduce 114 the conductivity by an additional 2 orders of magnitude to 1.5 × 10 −9 S/cm. DFT calculations indicated that coordinated DMF accepts electron density from Fe centers, effectively hole-doping the material and increasing charge density. Studying different solvation conditions for other MOFs with coordinated solvents may reveal broader trends related to the effect of solvents on their conductivities. For example, a rhodium iodide coordination polymer showed reversible conductivity changes of up to 5 orders of magnitude on removal of water solvent. 115 Scheurle et al. incorporated anthracene into the MOF-74 structure by reacting the 4,4′-(anthracene-9,10-diyl)bis(2hydroxybenzoate) linker with divalent Mg, Mn, Co, Ni, and Zn. 113 The anthracene moieties are separated by the relatively long distance of 5.7 Å, thus making the charge transport through π−π stacking unlikely. Despite this, all of these MOFs were several orders of magnitude more conductive (between 5 × 10 −9 S/cm for Mg and 4 × 10 −7 S/cm for Ni) than their DOBDC analogs. This increase in conductivity was attributed to the presence of the electron-rich anthracene core, which likely affects the charge density of the material. Also of note is the high porosity of these materials, with Brunauer−Emmett−-Teller (BET) surface areas above 1100 m 2 /g.

Azolate Frameworks
MOFs with pyrazolate, triazolate, and tetrazolate linkers tend to exhibit pronounced electronic coupling between the metals and the linkers by virtue of the strong σ-donating and πaccepting character of azolates. 116 In such structures, 1D chain SBUs with bridging azolates are a common motif. 117 The short bridging lengths of azolates, along with their covalent bonding to the metal, are responsible for notable cooperative electronic properties. For example, many 1D coordination polymers containing (−Fe−N−N−) ∞ motifs based on triazolate and tetrazolate ligands are spin-crossover materials. 118−120 Cooperative binding events in metal−triazolate MOFs also trigger spin-state transitions 121 and reversible charge transfer events 122 at the metal centers. These electronic considerations motivated the initial studies of the transport properties of such materials.
High conductivities have been reported in Fe−azolate frameworks with Fe II/III mixed valency, which effectively introduces mobile charge carriers into the (−Fe−N−N−) ∞ chains ( Figure 9). Intervalence charge transfer (IVCT) Chemical Reviews pubs.acs.org/CR Review processes in these MOFs are expected to resemble those in molecular mixed-valence complexes of d 5 −d 6 metal centers with π-acid bridging linkers. 123 The accessibility of metalcentered mixed valency likely results from the low reduction potential of Fe III coupled with small reorganization energies of the coordination spheres, which in turn lead to faster electron transfer rates compared to other transition metals. 59 In other words, both thermodynamic and kinetic factors can be expected to play a role in enabling facile charge transport.
In their report of the M II (1,2,3-triazolate) 2 series of isostructural MOFs (M = Mg, Fe, Co, Cu, Zn, and Cd), 124 Gańdara et al. indicated that the Fe(1,2,3-triazolate) 2 material Chemical Reviews pubs.acs.org/CR Review was conductive (7.7 × 10 −5 S/cm) in its as-synthesized form. 71 In this structure, each triazolate linker bridges three octahedrally coordinated Fe II centers, forming a 3D network with diamond topology containing (−Fe−N−N−) ∞ chains (Figure 9a,d). The material is permanently porous, with a BET surface area of 450 m 2 /g. Sun et al. later reported that the Mg, Mn, Co, Cu, Zn, and Cd MOFs with 1,2,3-triazolate are insulating (conductivities <2 × 10 −13 S/cm) and measured a conductivity value of 3.0 × 10 −6 S/cm for Fe(1,2,3-triazolate) 2 , similar to the original study. 68 In these reports, Fe(1,2,3-triazolate) 2 was handled in ambient atmosphere because no changes in powder X-ray diffraction (PXRD) patterns were observed after several weeks of air exposure (suggesting stability to these conditions). However, several pieces of evidence suggested that oxidation of the Fe II centers is accessible and contributes to the conductivity. Gańdara et al. found that the conductivity increased to 1.0 × 10 −3 S/cm after exposure to I 2 vapor with retention of the structure. They surmised that some Fe II sites were oxidized to Fe III , enabling IVCT. Sun et al. showed that the room temperature EPR spectrum contained signals at g ≈ 2.0 and 4.2, consistent with the presence of high-spin Fe III centers. DFT electronic band structure calculations of Fe(1,2,3-triazolate) 2 with Fe III defects revealed unoccupied Fe III midgap states that could facilitate the formation of mobile valence band holes. Moreover, the calculated spin density distribution also indicated partial delocalization of unpaired electrons among Fe sites.
Definitive evidence of the importance of mixed valency on the conductivity of Fe(1,2,3-triazolate) 2 was shown by Park et al. via the systematic postsynthetic introduction of Fe III sites. 72 By handling all reagents under rigorously air-free conditions, they obtained the Fe(1,2,3-triazolate) 2 material with a pink color, in contrast to the orange or brown powders reported previously. The material contained only low-spin Fe II by Mossbauer spectroscopy, and its electrical conductivity was only <7 × 10 −9 S/cm, about 3 orders of magnitude lower than previous reports.
Upon oxidation with stoichiometric amounts of thianthrene tetrafluoroborate, materials with the formula Fe(1,2,3triazolate) 2 (BF 4 ) x were obtained, in which BF 4 − counteranions charge-balance the Fe III centers introduced into the framework. Critically, increasing the BF 4 − content (and hence the Fe III concentration) increased the conductivities, which in the study was explained by a polaronic hopping model. The maximum achieved BF 4 − loading of x = 0.33 corresponds to one anion per pore; accordingly, Fe(1,2,3-triazolate) 2 (BF 4 ) 0.33 showed a very low BET surface area of 50 m 2 /g. The conductivity of Fe(1,2,3-triazolate) 2 (BF 4 ) 0.33 , 0.3(1) S/cm, is among the highest reported for 3D frameworks, and represents an increase of nearly 8 orders of magnitude over the starting material. In addition, detailed Mossbauer spectroscopy measurements revealed that charges were delocalized between Fe II and Fe III on the measurement time scale (∼10 −8 s) at 290 K.
Correlation between introduction of Fe II/III mixed valency into (−Fe−N−N−) ∞ chains and increased conductivity was also demonstrated for an Fe II framework with benzene-1,4ditetrazolate (BDT). 70 This material, Fe 2 (BDT) 3 ( Figure  9b,e), was initially reported as an all-Fe II compound with mixed spin states and a black color. 125 Xie et al. found that crystals synthesized under anaerobic conditions were orangered in color, gradually darkening to black after several days in ambient atmosphere with no changes in the PXRD patterns. The air-exposed material remained porous, with a BET surface area of 614 m 2 /g. The single-crystal conductivities measured along the direction of the (−Fe−N−N−) ∞ direction increased from 6(2) × 10 −5 S/cm for the as-synthesized material to 1.8 S/cm after 30 days in air. UV−vis−NIR spectroscopy and DFT band structure calculations further confirmed IVCT between Fe II and Fe III centers as the likely mode of charge transport in this material. The method of oxidation and results Chemical Reviews pubs.acs.org/CR Review observed in this work are similar to a study of a porous vanadyl analog of Prussian blue. 126 The Fe II framework with bis[1,2,3-triazolo [4,5- [1,4]dioxin (BTDD), Fe 2 Cl 2 (BTDD), also contains (−Fe−N−N−) ∞ chains as part of the rod SBUs in a hexagonal structure similar to MOF-74. Sun et al. reported the conductivity of this material to be 1.1 × 10 −7 S/cm, 68 almost 6 orders of magnitude higher than the Mn II , Co II , and Ni II isostructural materials. 127 Unintentional doping with Fe III was again invoked to explain the higher conductivity.
Achieving Fe II/III mixed valency with an all-Fe III starting material was demonstrated by Aubrey et al. with the framework Fe 2 (BDP) 3 (BDP = benzene-1,4-dipyrazolate) (Figure 9c,f). 69 Reduction of Fe 2 (BDP) 3 with potassium naphthalenide afforded materials with the formula K x Fe 2 (BDP) 3 (0 < x ≤ 2.0). Single crystals were incorporated into FET devices by depositing platinum/carbon contacts between the MOF and prepatterned electrodes. The highest conductivity measured for one such device was 0.025 S/cm for K 0.98 Fe 2 (BDP) 3 (corresponding to about half of the Fe III centers reduced to Fe II ). The electron mobility from the same device was 0.84 cm 2 V −1 s −1 . In comparison, a device of the original Fe 2 (BDP) 3 material exhibited a conductivity of 3.5 × 10 −7 S/cm and electron mobility of 2 × 10 −3 cm 2 V −1 s −1 . The authors suspected that the true intrinsic conductivity of the material was underestimated because of contact resistance in the twoprobe configuration. Indeed, a four-probe device of Fe 2 (BDP) 3 showed a significantly higher conductivity of 9.6 × 10 −3 S/cm, suggesting that the values for the reduced material may also be underestimated to some extent.
In many of the cases presented above, electronic structure calculations have shown significant band dispersion, confirming the strong coupling between Fe d orbitals and azolate πorbitals. 68−70 This feature, however, appears to not be unique to iron-based materials. Indeed, the calculated band structures of M(1,2,3-triazolate) 2 frameworks with other metals suggest that they may also exhibit mobile charge transport pathways and relatively small band gaps. 68 If free carriers can be introduced into MOFs with related structural motifs, they will likely comprise additional platforms for conductivity. We note also that triazolate and pyrazolate frameworks generally exhibit high chemical stability because of the strong donating ability of these ligands, and therefore may be of interest for applications in which stability and conductivity are both important. 128 Pathak et al. attempted 129 to extend the through-bond approach to two dimensions by incorporating extended 2D Cu−S sheets in [Cu 2 (Hmna)(mna)][NH 4 ] (Hmna = 6mercaptonicotinic acid). Although this framework is completely nonporous and cannot be strictly called a MOF, it showed some of the highest conductivities reported to date for coordination polymer single crystals, up to ∼11 S/cm. The material behaved as a semiconductor with a low E a of ∼6 meV, noticeably lower than the calculated bandgap (1.2 eV) or observed optical bandgap (1.34 eV).

EXTENDED CONJUGATION
The discovery of free-standing graphene 130 in 2004 precipitated an avalanche of interest in the electronic properties of 2D materials. 131−136 In addition to the "top-down" methods commonly used for the preparation of these nanoscale compounds, the possibility of creating synthetic 2D materials via "bottom-up" approaches has also been explored. 134−138 Conductive 2D MOFs made via solvothermal self-assembly have emerged as 2D materials with promising transport properties. Calculated band structures suggest that they may host exotic electronic properties, including topologically protected surface states. 31,32,34−36 Members of this class exhibit the highest conductivities for porous, crystalline framework materials.
The MOFs in this section feature linkers based on conjugated organic cores and redox-noninnocent chelating groups such as ortho-diols (catechols), -diamines, and -dithiols ( Figure 10). The oxidation states for a catechol fragment, hydroquinone, semiquinone, and quinone, can also be accessed in the nitrogen-and sulfur-containing analogs. In most cases, the ligands are partially oxidized during synthesis, which favors deprotonation and subsequent framework formation. The resulting MOFs contain mixed-valent linkerspotentially leading to high carrier densitieswith some examples of charge states that can be tuned postsynthetically. Although 2D frameworks have received the most attention and constitute the vast majority of examples, we also discuss several frameworks with 3D structures showing extended metal− linker conjugation and high conductivities.
The metal−organic linkages in these MOFs are analogous to motifs known to promote charge delocalization in inorganic complexes. For example, the redox-noninnocent ligands in metal−dithiolene complexes facilitate electron delocalization between the metal and linker. 139,140 Such compounds are components in many molecular conductors (and even superconductors). 141 Ligand radical character in the bis(odiiminobenzosemiquinonate) Ni II complex is known to be stabilized by the metal binding. 142−144 Dihydroxybenzoquinone-type linkers also demonstrate strong electronic and magnetic coupling in transition metal complexes, especially in their oxidized radical forms. 145−147 We note that the structures of these materials are considered to be 2D in the sense that covalent bonds extend only within the ab plane. However, to date, mono-or few-layer samples of 2D MOFs displaying extended conjugation have not been studied in detail. Moreover, only one study on single crystals contacted along the conjugation direction has been reported. 79 Hence, interlayer effects are expected to influence the reported bulk properties. Molecular metal−dithiolene and diiminobenzosemiquinonate complexes exhibit semiconducting or metallic behavior with the highest conductivities along the stacking direction, 141,143 suggesting that transport facilitated by noncovalent c direction interactions may also contribute to the conductivities measured in powder samples of 2D MOFs. This intuition is supported by existing cross-plane conductivity measurements 79,148 and transport studies of related MOFs with no π−d conjugation. 92

Frameworks with Ditopic Redox Noninnocent Linkers
A number of materials built with tetraoxolene ligands, such as dihydroxybenzoquinonate (dhbq) and chloranilate (Cl 2 dhbq),  The easily accessible redox couples of these ligands and their resulting ability to form stable radicals form the origin of strong magnetic coupling in a number of magnetic molecular complexes. 145−147 The incorporation of these ligands into MOFs harnessed these features to produce materials with permanent porosity, high electrical conductivity, and magnetic ordering. Notably, the Fe II/III -based materials display some of the highest conductivities and magnetic ordering temperatures in MOFs.
The mechanism of conductivity in these materials is up for debate: while most works propose hopping between either redox-active metal sites or radical organic linkers to be responsible for charge transport, many of these MOFs display high degrees of electron delocalization between the linker and the metal. As a result, they show similarities to the materials in the following section, which exhibit extended conjugation between the metal and ligand.
3.1.1. Two-Dimensional Layered Frameworks with Ditopic Linkers. This structure type, which was initially reported for Mn II and Cd II chloranilates, 155 consists of hexagonal layered structures with solvent-accessible pores of ∼1.5 nm in diameter. The metal atoms are each octahedrally coordinated by three ligands (Figure 11). The frameworks are all anionic, with charge-balancing cations located either within the pore or between the layers. For simplicity, most of the following observations will discuss only the anionic framework.
The Harris group reported 149 early studies of electrical conductivity in chloranilate frameworks, focusing initially on the Fe III and Zn II materials. The materials were produced through the reaction of the corresponding divalent metal salt and chloranilic acid and are charge-balanced by dimethylam-monium ions that reside in the pores of the framework. The low barrier to oxidation of Fe II results in electron transfer from the Fe II ions to the ligands, converting two-thirds of these to the 3− radical state (Cl 2 dhbq 3−• ), and leaving the remaining third in the closed-shell Cl 2 dhbq 2− state ( Figure 12). This ligand-based mixed valency resulted in the high conductivity of 1.4(7) × 10 −2 S/cm. When desolvated, the conductivity decreased to 1.0(3) × 10 −3 S/cm, attributed to slight framework distortions. Chemical reduction of the MOF by excess Cp 2 Co led to full conversion of the ligands to the Cl 2 dhbq 3−• radical state, which caused a further decrease in conductivity to 5.1(3) × 10 −4 S/cm. The lower conductivity upon the loss of mixed valency is consistent with charge Measured at room temperature (exact temperature unspecified) or between 295 and 300 K under N 2 or Ar atmosphere. c ∥ denotes device contacted in-plane/parallel to 2D layers; ⊥ denotes device contacted cross-plane/perpendicular to 2D layers. d Derived from Arrhenius model unless otherwise indicated. For materials exhibiting more than one distinct activation regions, E a at 300 K is given. e Calculated from N 2 adsorption isotherms. f Derived from polaronic nearest-neighbor hopping. g Derived from Efros-Shklovskii variable-range hopping. The presence of strong coupling between metal centers via radical bridging ligands was further corroborated by the magnetic behavior of [Fe 2 (Cl 2 dhbq) 3 ] 2− . Solvated [Fe 2 (Cl 2 dhbq) 3 ] 2− ordered ferromagnetically at 80 K, whereas the desolvated (and less conductive) material ordered at 22 K, suggesting weaker coupling. The reduced material ordered at a higher temperature of 105 K, as would be expected for more Fe III ions being bridged by radical-containing ligands in the reduced material.
In the Zn−chloranilate framework, the oxidation of Zn II by Cl 2 dhbq 2− is not thermodynamically accessible, resulting in the full ligand population of [Zn 2 (Cl 2 dhbq) 3 ] 2− remaining in the closed-shell Cl 2 dhbq 2− states. Accordingly, the conductivity of this material was relatively low (1.5(3) × 10 −9 S/cm). The authors proposed that the closed-shell configuration of Zn II with filled d orbitals introduces a barrier to charge transport between the ligands, leading to lower conductivity. The relative influence of the d shell filling vs the ligand oxidation states could be probed via reduction of the zinc material to a similar mixed-valence state as in the Fe analog.
An isostructural Mn II -based system, [Mn 2 (Cl 2 dhbq) 3 ] 2− , showed an increase in conductivity upon reduction. 150 The conductivity value rose from 1.14(3) × 10 −13 S/cm for the fully oxidized version with Cl 2 dhbq 2− diamagnetic linkers, to 2.27(1) × 10 − 8 S/cm for the reduced version, [Mn 2 (Cl 2 dhbq) 3 ] 5− , containing Cl 2 dhbq 3−• radical linkers. To explain the difference in conductivities between the Mn II and Fe II materials, the authors proposed a poorer energetic match between the Mn atoms and the linker compared to Fe. This mismatch may have also led to noticeably higher activation energies of 0.489(8) eV for the reduced and 0.74(3) eV for the oxidized Mn frameworks. The change in conductivity was fully reversible upon chemical oxidation to the initial state [Mn 2 (Cl 2 dhbq) 3 ] 2− . Notably, like the Fe IIIbased framework, [Mn 2 (Cl 2 dhbq) 3 ] 5− displayed a ferromagnetic transition, albeit with a lower T c of 41 K.
The charge transport behavior of chloranilate frameworks with the same overall honeycomb structure and mixed-valent Fe II/III was also studied. The first report, by Shilov et al., 156 discussed the structure and electronic properties of [Fe II Fe III (Cl 2 dhbq 2− ) 2 (Cl 2 dhbq 3−• )] 2− . Unlike the aforementioned [Fe III 2 (Cl 2 dhbq) 3 ] 2− system, in which the charge balancing (Me 2 NH) + ions are located mainly inside the pores, the framework charge in the mixed-valent Fe II/III framework is balanced by hydrogen-bonded layers of [(H 3 O) 2 (phenazine)] 2+ . Shilov et al. were unable to isolate single crystals of this material, and conducted a thorough investigation of its electronic properties on polycrystalline powder samples. Mossbauer spectroscopy combined with magnetometry revealed the mixed-valent nature of the material, with the Fe atoms present both in pure 2+ and 3+ states, as well as in the mixed-charge 2.5+ state, with relative concentrations dependent on the temperature. The material showed a relatively high conductivity of 4 × 10 −5 S/cm (polycrystalline pellet) and an Arrhenius E a of 0.16 eV.
A later study reported single-crystal measurements on a m a t e r i a l w i t h t h e s a m e n o m i n a l f o r m u l a , [Fe II Fe III (Cl 2 dhbq 2− ) 3 ] − , assigned to different charge states (with all the ligands in their closed-shell Cl 2 dhbq 2− configurations, and no organic radicals). 151 Despite the lack of organic radicals, the MOF showed a high conductivity of 0.031(8) S/cm from single-crystal measurements parallel to the layers, and 1.0(9) × 10 −4 S/cm orthogonal to the layers. Similar conductivities, 3(1) × 10 −3 S/cm (parallel) and 6(2) × 10 −6 S/cm (orthogonal), were obtained for the isostructural framework containing bromanilate instead of chloranilate. The high anisotropy of the conductivity confirmed that charge transport is mainly limited within the ab plane. Though these values were from single-crystal measurements (vs powder pellet values for [Fe III 2 (Cl 2 dhbq) 3 ] 2− ), it is remarkable that the conductivities were of the same magnitude. This result demonstrates that ligand mixed valency and metal mixed valency can both enable charge transport in these systems, consistent with strong electronic coupling between metals and ligands.
A series of structurally related frameworks based on early fi r s t r o w t r a n s i t i o n m e t a l s , [ V 2 ( C l 2 d h b q ) 3 ] 2 − , [Ti 2 (Cl 2 dhbq) 3 ] 2− , and [Cr 2 (dhbq) 3 ] 1.5− showed similarly high conductivities. 152 Precursors containing the highly reducing divalent early 3d transition metal ions Ti II , V II , and Cr II were combined with Cl 2 dhbq 2− and dhbq 2− , reducing the linkers to the corresponding radical anions. The more diffuse orbitals of the early transition metals (as compared to the later Ni, Co, or Cu) were hypothesized to allow for orbital overlap with the ligand and thus more efficient charge delocalization. This approach proved successful, producing materials with conductivities of 2.7(2) × 10 −3 S/cm for [Ti 2 (Cl 2 dhbq) 3 ] 2− , 0.45(3) S/cm for [V 2 (Cl 2 dhbq) 3 ] 2− , and 1.2(1) × 10 −4 S/cm for [Cr 2 (dhbq) 3 ] 1.5− . The observed conductivities followed the same trend as the degree of electronic delocalization observed by optical spectroscopy, where the Ti and Cr-based materials can be classified as Robin−Day class II and the V-based material as class III, indicating much stronger charge delocalization in the latter.
The charge states of both the ligands and the metals in these three materials differed. From elemental analysis, optical spectroscopy, and magnetization measurements, the three m a t e r i a l s h a d t h e f o l l o w i n g a s s i g n m e n t s : [Ti IV 2 (Cl 2 dhbq 3− ) 2 (Cl 2 dhbq 4− )] 2− , [V III 2 (Cl 2 dhbq 2− )-(Cl 2 dhbq 3− ) 2 ] 2− (assigned putatively, as electronic delocalization did not allow full resolution of the metal/ligand charge states), and [Cr III 2 (dhbq 2− ) 1.5 (dhbq 3− ) 1.5 ] 1.5− . The stronger charge delocalization and significantly higher conductivity in the V-based materials are explained by a better energy match between the ligand orbitals and the metal 3d shell. Interestingly, [Ti 2 (Cl 2 dhbq) 3 ] 2− displayed strong antiferromagnetic coupling between the organic radical spins, with a Curie−Weiss temperature of −223 K, although no ordering was observed in it or the other two materials. Variable temperature conductivity measurements revealed semiconducting behavior fitting well with either polaronic nearest-neighbor These mechanisms differed from the Fe-based materials, perhaps as a result of the wider temperature range fitted (a narrower region may yield an Arrhenius-type dependence). A similar approach was later applied 157 to dense pyrazinebased square lattice coordination polymers, in which the reaction of a Cr II salt with pyrazine led to the formation of a highly conductive nonporous framework. As in the dhbq materials, the conductivity originated from the reduction of pyrazine to its radical anion form, which led to improved ligand π-system−metal d-orbital overlap, as well as introduction of mobile charge carriers. The material, CrCl 2 (pyz) 2 (pyz = pyrazine), showed a relatively high conductivity of 3.2 × 10 −2 S/cm for a pressed pellet, along with a semiconductortype temperature dependence, which could be fitted to the Mott law for two dimensions.
3.1.2. Three-Dimensional Frameworks Based on dhbq. Dihydroxybenzoquinone also forms 3D frameworks with many late transition 158 and rare-earth metals. 159 The charge transport properties of the Fe material have been evaluated in detail. 153 [Fe 2 (dhbq) 3 ] 2− is a cubic interpenetrated 3D framework with octahedral Fe III sites containing radical dhbq 3−• anions, with the as-synthesized formula (Figure 13a). The two component sublattices are enantiomeric and fully independent from each other (Figure 13b). Due to the bulky tetrabutylammonium ions compensating the charge, the material is essentially nonporous (though a high theoretical surface area of 5070 m 2 / g was calculated for the fully oxidized material with chargebalancing cations removed from the pores). The as-synthesized material showed an impressive electrical conductivity of 0.16(1) S/cm from a 2-probe pellet measurement. This is one of the highest values reported to date for a 3D material with likely isotropic charge transport. The authors proposed that the high conductivity arises from charge hopping between dhbq 3−• and dhbq 2− centers. This agrees with the lower conductivity (6.2(1) × 10 −3 S/cm) observed upon reduction of the material by an average of 0.7 e − per ligand to [Fe III 2 (dhbq) 3 ] 2.7− , which effectively lessened the extent of ligand-based mixed valency and hence the concentration of available charge hopping sites. Interestingly, both the as made and the reduced materials displayed a very low-energy absorption edge suggestive of Robin−Day type-II/III mixed valency originating from the dhbq 2−/3−• couple. This assignment implies that the charges are significantly delocalized and that description of the charge transport from a band-like perspective may be appropriate.
A Tb III -based chloranilate MOF with a three-dimensional structure has been reported to have an electrical conductivity of 6.25 × 10 −4 S/cm. 154 This lower value compared to transition metal chloranilate MOFs is likely due to the strongly ionic bonding character between the ligand and the lanthanide.

Frameworks with Flat, Multitopic, Conjugated Linkers
Layered two-dimensional frameworks based on flat, fully conjugated ligands with trigonal geometries constitute one of the most prominent classes of conductive MOFs (see Tables  3−5). To the best of our knowledge, these materials show the highest conductivities observed for ordered microporous materials. The clear and consistent strategy underlying these high conductivities, which are usually thought to originate from the strong electron delocalization within the 2D layers, also contributes to the broad appeal of these materials. The high conductivity values and straightforward design principles drive the growing number of studies on these materials.
The strong delocalization is produced by covalent bonding between the metal and the linker to form an extended πconjugated framework ( Figure 14). Three key features of the ligands enable this electronic structure. First of all, the catecholoid linkers are highly susceptible to oxidation, producing organic radicals in the MOF. Second, full conjugation within the linker, which consists of one large πsystem, facilitates the delocalization and stabilization of said radicals. Finally, good energy matching between the linker and the metal frontier orbitals allows the conjugated system to extend throughout the full layer. The overall geometry of the material is also important for the extended conjugation of the π-systems, as coplanar linkers form flat 2D layers. Chemical Reviews pubs.acs.org/CR Review Materials in this class can be divided into three broad categories, based on the linker used: semiquinoid-based materials with oxygen-based linkers, iminosemiquinoid-based materials with nitrogen-based linkers, and dithiolene-based materials with sulfur-based linkers. With the exception of several recent reports, each category consists of two families: one with larger hexa-substituted triphenylene linkers, and one with smaller hexa-substituted benzene linkers, with the former being more porous, and the latter typically being dense. Overall, increasing the bond strength between the metal and the ligand (i.e., moving from oxygen-based to nitrogen-to sulfur-based linkers) and increasing the framework density (moving from a triphenylene core to a benzene core) both appear to result in higher conductivities.
Although some trends are emerging, the limited number of materials in this class still precludes more detailed correlations among composition, structure, and properties. Although many metal−ligand pairs have been reported, comparative studies designed to isolate compositional variables (i.e., the ligand core, functional group, and metal) are necessary to disentangle their influence on conductivities. Small crystallite size or low crystallinity in many of these materials introduce uncertainties regarding the exact structures, especially related to the layer stacking arrangements. The exact oxidation and protonation states of redox noninnocent linkers are often ambiguous as well. Significant variance is also observed between different reports of the same material by different groups with respect to conductivity, crystallinity, and purity. Therefore, a more systematic approach to these materials, including standardization of synthetic and measurement techniques, will be central to the elucidation of underlying trends.
Despite these challenges for accurate, consistent structural and compositional characterization of 2D conductive MOFs, their high conductivities have attracted significant interest from applied studies in electrocatalysis, 16 148 On the basis of synchrotron PXRD and transmission electron microscopy (TEM), they found that Cu II and 2,3,6,7,10,11hexahydroxytriphenylene (H 6 HOTP) form a layered honeycomb framework with each metal coordinated by two linkers in a square planar environment, resulting in an overall formula of Cu 3 (HOTP) 2 . Co II and Ni II produced MOFs with similar honeycomb layers, but with the metal sites in octahedral environments with two terminal aqua ligands. Additionally, the framework layers in the Co and Ni MOFs are interleaved with layers of molecular [(H 2 O) 4 M] 3 HOTP clusters, which contribute to an overall formula M 9 (HOTP) 4 (discounting solvent molecules) ( Figure 15). Both the Cu and the Co/Ni structure types have a near-eclipsed arrangement of the layers, resulting in permanent microporosity, with surface areas in the range of 400−500 m 2 /g.
As implied by their distinct formulas, these two structure types also contain different ligand charge states (HOTP is assumed to be fully deprotonated in both cases). In Cu 3 (HOTP) 2 , all linkers are HOTP 3− , (i.e., oxidized by 3 electrons). In Co 9 (HOTP) 4 and Ni 9 (HOTP) 4    Although the authors of the initial report focused on the structural aspects of the materials, they also measured the electrical conductivity of Cu 3 (HOTP) 2 to be 0.1 S/cm in a single-crystal device contacted along the c axisthe highest value at the time for a porous MOF. Similar values have been reported for Cu 3 (HOTP) 2 powders and films (from 0.02 to 0.2 S/cm). 22,160−167 Although the initial report did not include conductivity data for Co 9 (HOTP) 4 and Ni 9 (HOTP) 4 , later studies 20,23,161,163,164,168 provided pressed powder pellet values in the range from 3 × 10 −3 to 0.1 S/cm for Ni 9 (HOTP) 4 and from 3 × 10 −6 S/cm to 3 × 10 −3 S/cm for Co 9 (HOTP) 4 (see Table 3). Although an Fe-based HOTP material has also been reported, showing a conductivity of 3 × 10 −3 S/cm, it appeared to be completely amorphous by PXRD. 163 Recently, Song et al. synthesized thin films of Cu 3 (HOTP) 2 with preferential orientation of the ab plane parallel to the substrate surface via a layer-by-layer method. 169 Four-probe measurements across the surface of the film (i.e., comprising predominantly in-plane contributions) showed a high conductivity of 0.29 S/cm. Variable-temperature measurements revealed an Arrhenius E a of 0.13 eV above 240 K, with Mott VRH behavior below 200 K. The thin films were investigated as potential spin-valve materials in the same study.
In a recent report, Day et al. optimized the synthesis of Cu 3 (HOTP) 2 to produce micron-sized crystals. 79 A detailed TEM and PXRD investigation showed that the layers in Cu 3 (HOTP) 2 are in fact stacked in a slightly offset fashion with a continuous shift along the c direction, creating a tilted stack, as opposed to the straight stack of Co 9 (HOTP) 4 ( Figure  16). The authors also achieved synthetic control over the morphology of the crystallites, obtaining both needle-shaped crystals with the long axis aligned with the stacking direction, as well as shorter prism-shaped crystals with nearly isotropic dimensions. Sonication-assisted liquid-phase exfoliation of the prism-shaped crystals produced thinner hexagonal flakes. The authors used electron-beam lithography to fabricate four-point probe devices measuring conductivity both perpendicular to the MOF layers ("cross-plane"; measured on the needle crystals), and within the MOF layers ("in-plane"; measured on the exfoliated flakes). Surprisingly, the measured values of 1.5 S/cm for cross-plane devices and 0.5 S/cm for in-plane devices were within the same order of magnitude. This result challenges the view that conductivity in these materials is primarily due to intralayer transport; the most likely explanation to the high cross-plane conductivity values is interlayer charge transport mediated by close stacking and interactions between the π-systems of neighboring layers.
The significance of π−π stacking with respect to the charge transport of 2D layered MOFs showing intralayer extended conjugation remains an open question. The single-crystal studies by Day et al. 79 suggest significant π−π stacking contributions, as equally high conductivities were found for inplane and cross-plane devices for Cu 3 (HOTP) 2 . More focus was put toward π−π stacking in this class of MOFs in a study by Skorupskii et al. investigating a series of Ln 1+x (HOTP) (Ln = La III , Nd III , Ho III , Yb III ) materials. 92 In contrast with the structure of Cu 3 (HOTP) 2 , the lanthanides are not located in the same plane as the linkers, but rather in-between the linker layers, connecting them into a 3D framework ( Figure 17). In this structure, the HOTP linkers form continuous stacks with extremely close π−π interactions of 3.0−3.1 Å. In addition, the precise π−π stacking distances were found to correlate directly with the sizes of the lanthanide cations. The authors found that despite the lack of conjugation in the layers, these materials displayed high conductivities of up to 0.05 S/cm (for Ho 1+x (HOTP)). Importantly, the conductivity decreased with longer stacking distances separating the linkers: the conductivity of La 1+x (HOTP) was 1 × 10 −3 S/cm, which was attributed to the larger La III pushing the layers farther apart. DFT calculations of the band structure of Ln 1+x (HOTP) showed high band spread and metallicity in the cross-plane direction, with negligible contribution of Ln III to the frontier Taken together, these findings demonstrated the importance of the cross-plane charge transport in these materials. 3.2.1.2. Other Semiquinoid Linkers. Cu 3 (HOB) 2 , a material based on the smaller hexahydroxybenzene (H 6 HOB) linker and Cu II , was also investigated for its charge transport properties. 170 The structure of the material resembles Cu 3 (HOTP) 2 but with smaller pores due to the smaller ligand core. Despite the similarities, the material showed a significantly lower conductivity of 7.3 × 10 −8 S/cm in a twoprobe pellet configuration at room temperature. The temperature dependence of the conductivity was fitted to give an Arrhenius E a of 0.45 eV, similar to many other materials in this class. A curious feature of Cu 3 (HOB) 2 is the extremely close stacking of its layers, which are separated by only 2.96 Å, which may lead to a more significant contribution of cross-plane charge transport to the electrical conductivity.
A hybrid of Cu 3 (HOTP) 2 and Cu 3 (HOB) 2 was recently reported by Yao et al. 171 In Cu 3 (HOTP)(HOB), the two linkers alternate within the layer, forming an ordered layered trigonal network. The layers stack in an eclipsed fashion, forming continuous 1D channels, similar to Cu 3 (HOTP) 2 . This structure resulted in high intrinsic porosity, with a BET surface area of 441.2 m 2 /g. Two-probe pressed pellet measurements revealed a conductivity of 2.53 × 10 −5 S/cm, considerably lower than that of Cu 3 (HOTP) 2 (up to 0.2 S/cm ), but higher than Cu 3 (HOB) 2 (7.3 × 10 −8 S/cm). The temperature dependence of the conductivity similarly showed an E a of 0.30 eVin-between those of Cu 3 (HOTP) 2 and Cu 3 (HOB) 2 . This work served as another demonstration of the tunability of MOFs.   Figure 18). The authors proposed an eclipsed stacking model for the materials with square channels of ∼15 Å in diameter. Permanent porosity was confirmed by N 2 adsorption, which revealed a BET surface area of 358 m 2 /g. The electrical conductivity of pressed pellets ranged between 10 −8 S/cm and 10 −6 S/cm, significantly lower than for HOTP-based materials. A later study reported Ni II and Cu II -based Ni(OPc) and Ni(ONPc) MOFs, which were found to be isostructural to Cu 2 [Cu(OPc)]. 25 These materials showed considerably higher conductivities, ranging from 10 −4 to 10 −2 S/cm, with the Cu IIbased materials showing higher values compared to Ni II (the authors did not propose an explanation for this difference).
Feng and co-workers also reported an all-Fe phthalocyaninebased framework with the formula K 3 Fe 2 [Fe(OPc)]. 173 X-ray photoelectron spectroscopy (XPS) indicated a mixture of Fe II and Fe III in the samples (the authors were not able to distinguish between the framework and central Pc Fe sites). The relatively low BET surface area, 206 m 2 /g, was attributed to the K + counterions in the pores. The conductivity of a van der Pauw pellet was ∼1 × 10 −5 S/cm at 300 K, with a hightemperature E a of 0.26 eV. Hall measurements revealed p-type behavior with a hole concentration of ∼2.4 × 10 −14 cm −3 and mobility of ∼0.1 cm 2 V −1 s −1 . Noncontact TRTS measurements estimated a higher mobility of 15 (2) 174 and benzenes 65 was largely inspired by the nickel bis(dithiolene) motif in molecular complexes. 175 This fragment exhibits strong charge delocalization and planar geometry, which are enabled by the redox noninnocence of the ligand. Nickel bis(dithiolene) complexes are also redox-active, exhibiting several reversibly accessible states. These features led researchers to believe that extending the motif into a coordination polymer would allow for the formation of an extended π-conjugated sheet akin to graphene.
The framework Pt 3 (THT) 2 (THT = 2,3,6,7,10,11-triphenylenehexathiolate), reported in 2014 by Xu and Cui, 174 exhibited both permanent porosity (300−400 m 2 /g) and electrical conductivity (1.4 × 10 −6 S/cm). The as-made material was anionic but could be oxidized into an almost neutral framework with I 2 . No significant impact of oxidation on the conductivity was observed. As expected from the very strong and covalent bonding between Pt and thiol ligands, the material was not highly crystalline, with only broad peaks present in the PXRD pattern. Nevertheless, the unit cell  Chemical Reviews pubs.acs.org/CR Review dimensions agreed with a structure containing the ligands connected by square-planar Pt II ions into a hexagonal net. Materials with similar structures were also isolated with Co II and Fe III . 78,176 Co 3 (THT) 2 shows somewhat improved crystallinity compared to Pt 3 (THT) 2 and was first investigated as an electrocatalyst for H 2 evolution. 16 The material was porous (370 m 2 /g) and exhibited a conductivity of 1.4 × 10 −3 S/cm at 300 K in a van der Pauw geometry, considerably higher than the Pt analog. The temperature dependence of the conductivity was complex and warrants further investigation: the authors saw an increase in resistivity with decreasing temperature (as would be expected for a semiconductor) for temperatures close to room temperature, and the opposite (as would be seen for a metal) in the low-temperature regime (below ∼180 K). We caution that similar behavior can be interpreted as phase artifacts in alternating current measurements of resistive samples. Interestingly, from the theoretical computations, the authors concluded that despite the highly covalent bonding in the sheet, the main mechanism for charge transport is expected to be along the c (stacking) direction. Also of note is the mixed valency of the metal ions: magnetometry results suggested that the correct charge assignment for the material is [Co III 2 Co II (THT) 2 ] 3− . Fe 3 (THT) 2 displayed the highest conductivity to date among THT materials, reaching 3.4 × 10 −2 S/cm at 300 K in a four-probe device. 78 This material was permanently porous, with a BET surface area of 526 m 2 /g. Although the PXRD patterns were not of sufficiently high quality to resolve the exact stacking sequence, both PXRD and high-resolution TEM measurements proved the overall layered honeycomb structure, similar to Pt 3 (THT) 2 and Co 3 (THT) 2 . The study reported Hall effect and THz spectroscopy measurements to extract a charge mobility value of ∼200 cm 2 V −1 s −1 , depending on the measurement. We note that discussions of the material as both heavily doped semiconductor (in the main text) and as an intrinsic semiconductor (in the Supplementary Information) are present in ref 78. The overall semiconducting behavior, however, was conclusively proven by variable temperature conductivity measurements. Optical absorption data suggested a direct band gap of 0.245 eV, matching well with the DFT calculated value of 0.350 eV. Interestingly, although the authors started with Fe II salts and allowed no exposure to oxidants during the reaction, the metal charge state in the final material is Fe III . As metal-based mixed valency provides an excellent way to introduce free carriers, a more rigorously air-free synthesis could allow access to a Fe II/III mixed material. A later work 177 proposed different Fe oxidation states and reported unusual maxima in variable temperature conductivity behavior (as with the report on the Co analog of this material, 176 we caution that similar behavior can be interpreted as phase artifacts). Clearly, this is a material that warrants further investigation and that could benefit from a systematic variation of synthetic conditions, as well as additional transport measurements.
The Feng group also reported a related material with Fe III , Fe 3 (PTC) (PTC = perthiolated coronene). 178 This framework was essentially nonporous: the type II N 2 adsorption isotherm revealed a (primarily external) surface area of 210 m 2 /g. The simulated structures contained acetylacetonate-like coordination modes of the thiols to the metal, unlike the typical catechol-like mode observed in Pt 3 (THT) 2 . Although the authors discussed several potential layer stacking sequences, they did not elaborate on the assignment of this particular Fe− S coordination mode. Similar to Fe 3 (THT) 2 , despite using an Fe II salt in oxidant-free conditions, the metal charge state in the material was assigned unambiguously to Fe III . The material displayed very high conductivities of up to 10 S/cm for pressed pellets in van der Pauw geometries. The conductivity increased with decreasing temperature with an Arrhenius E a of 0.2 eV, similar to most MOFs discussed in this section. Importantly, this material was also ferromagnetic below ∼20 K and is the only material in this class with reported magnetic ordering.
THT materials with Ni II and Cu II have also been reported, 179 but the structural analysis was based on PXRD patterns with essentially no discrete peaks, and thus no definite conclusions on the structures can be made.
3.2.2.2. Benzenehexathiolate MOFs. In a foundational work in 2013, Kambe et al. incorporated the nickel bis(dithiolene) motif into a π-conjugated 2D material, the Ni 3 (BHT) 2 (BHT = benzenehexathiolate) framework. 65 The as-synthesized material, which was characterized as a thin black film, showed a two-probe pressed pellet conductivity value of 0.15 S/cm and a four-probe conductivity of 2.8 S/cm (reported in a later work by the same group). 94 The higher four-probe value is likely the result of eliminating contact resistance.
These authors also probed the charge states of nickel bis(dithiolene) fragments in the MOF, and the dependence of conductivity on the redox assignments. Infrared (IR) spectroscopy and XPS studies showed that the as-synthesized material contained mixed-valence 0 and −1 charge states, with an average charge of −3/4. Both the pure 0 and −1 states were accessible through electrochemical means and chemical oxidation/reduction (in contrast with molecular species, only the 0/−1 couple appeared to be reversible). Compared to the Chemical Reviews pubs.acs.org/CR Review as-synthesized mixed-valent MOF (2.8 S/cm), the fully reduced material in the −1 state showed a lower conductivity of 6.7 × 10 −3 S/cm. The oxidized material in the 0 state showed the highest conductivity of 160 S/cm. DFT calculations and photoelectron spectroscopy (PES) suggested that both the as-made and the reduced materials were metallic in nature. However, the temperature dependence of the conductivity showed thermally activated behavior characteristic for a narrow-gap semiconductor. The authors proposed that the discrepancy could be a result of structural disorder in the real materials.
A palladium coordination polymer of BHT was later reported by the same group 180 and was proposed to have a similar structure to Ni 3 (BHT) 2 based on TEM data. However, PXRD analysis revealed no discrete peaks, and the electron diffraction data were of poor quality, prohibiting detailed structural characterization. Initial synthetic attempts led to coprecipitation of Pd 0 nanoparticles, along with the proposed MOF. Addition of an oxidizing buffer in the form of K 3 [Fe(CN) 6 ] allowed the authors to generate pure Pd 3 (BHT) 2 , as evidenced by TEM. The material exhibited a similar charge state to Pd 3 (BHT) 2 , with the palladium bis(dithiolene) fragment present in mixed-valence 0/−1 states. The average charge state was found to be −0.19, which was higher than seen in as-made Ni 3 (BHT) 2 , likely because of the oxidizing buffer added to prevent formation of Pd 0 nanoparticles. The electrical conductivity of the pure Pd 3 (BHT) 2 material was reported to be of similar magnitude to the Pd 0contaminated material (2.8 × 10 −2 S/cm), though the precise value was not specified. This conductivity is noticeably lower than that of Ni 3 (BHT) 2 , although the sample quality of Pd 3 (BHT) 2 was significantly worse as well, prohibiting any definite conclusions.
The dense coordination polymer Cu 3 (BHT) reported by Huang et al. in 2015 holds the record for the highest conductivity in MOFs with low or minimal porosity (Table  4). 76 Although the material is nonporous, its structure and composition are related to other materials in this family, meriting inclusion in this Review. In contrast to Ni 3 (BHT) 2 , in which the BHT linker acts as a tritopic ligand, in Cu 3 (BHT) the ligand bonds to six Cu II ions, thus increasing the framework density ( Figure 19). The conductivities of Cu 3 (BHT) are the highest values reported for coordination polymers. Initial four-probe measurements on thin films revealed a value of 1580 S/cm at room temperature. In later studies by the same group on more crystalline samples, 181 the conductivities reached 2500 S/cm for a similar configuration. Interestingly, while the initial work reported a 3D Mott VRH dependence of electrical conductivity, which is characteristic for semiconductors, the later study showed a metallic dependence. As the authors noted, this is likely due to the drastically improved crystallinity of the material in the later study. Field-effect transistors based on Cu 3 (BHT) films 76 displayed very high (∼100 cm 2 V −1 s −1 ) electron and hole mobilities, along with low (∼10) on/off ratios, which the authors likened to the behavior of gapless graphene. The very low on/off ratios are in good agreement with the metallic temperature dependence shown in the later study. Importantly, Chemical Reviews pubs.acs.org/CR Review Cu 3 (BHT) was also reported to transition into a superconducting state (exhibiting zero electrical resistance) below 0.25 K, marking the first such observation for a coordination polymer or MOF. 181 The components of this structure responsible for the superconductivityor even the metallic transportin this material remain to be definitively identified. More efficient stacking, higher framework density, or stronger Cu−S bonds all potentially play a role. The Zhu group also reported the selenolate analog of Cu 3 (BHT), denoted as Cu 3 (BHSe). 182 The conductivity of a pressed pellet of this dense coordination polymer was 110 S/ cm, similar to that of Cu 3 (BHT). Although thermally activated conductivity was observed, the authors proposed that Cu 3 (BHSe) may also be metallic due to weak temperature dependence and strong absorption throughout the IR region, with semiconductor-like behavior induced by grain boundaries.
Another dense phase structurally related to Cu 3 (BHT) was reported in 2008 by Turner et al. 183 This framework, Pb 3 (BHT), has the same overall placement of linkers in layers but contains larger Pb II atoms sitting in between the layers and binding them into a 3D solid. The material showed only low conductivities of 2 × 10 −6 S/cm for a two-probe pressed pellet. The conductivity nevertheless increased by 2 orders of magnitude to 0.1 × 10 −3 S/cm after doping with Tl I .
A well-characterized dense Ag I BHT polymer was also reported, 184 with a structure unrelated to Cu 3 (BHT) or Ni 3 (BHT) 2 . The material showed high conductivity values of up to 250 S/cm. Its structure contains many potential pathways for conductivity, including infinite Ag−Ag and Ag− S−Ag chains, and the origin of charge transport is likely different from that in Cu 3 (BHT) and Ni 3 (BHT) 2 . Another work reported Ag I and Au I coordination polymers based on BHT, which were inferred to be dense phases isostructural to Cu 3 (BHT). 185 However, the reported PXRD patterns bear little resemblance to those of the Cu analog, and no attempts at fitting were shown. The materials showed high conductivities of up to 363 S/cm for Ag 3 (BHT), and up to 1.12 × 10 −4 S/cm for Au 3 (BHT).

Iminosemiquinoid Linker-Based Materials.
3.2.3.1. Hexaiminotriphenylene MOFs. The highest conductivity in a porous MOF to date belongs to Ni 3 (HITP) 2 (HITP = 2,3,6,7,10,11-hexaiminotriphenylene), first reported by Sheberla et al. 75 Just like Ni 3 (BHT) 2 is inspired by metal dithiolene complexes, the design of Ni 3 (HITP) 2 is based on the Ni(isq) 2 complex, which has a fully conjugated biradical ground state. 188 As was initially proposed based on DFT calculations and later confirmed in a detailed TEM study, 79 Ni 3 (HITP) 2 forms a structure similar to Cu 3 (HOTP) 2 , where the honeycomb layers stack in a slipped but near-eclipsed fashion, creating micropores of ∼1.6 nm in diameter ( Figure  20). Unlike Cu 3 (HOTP) 2 , the offset direction in Ni 3 (HITP) 2 alternates every layer (ABAB), creating a straight rather than a tilted stack. The initial paper reported four-probe film and pressed pellet conductivities of 40 and 10 S/cm, respectively, which were the highest values shown by a porous material at the time. The temperature dependence of the conductivity was thermally activated and linear over an unusually large temperature range. Although consistent with semiconductortype behavior, this behavior did not fit to a specific transport model. The material showed intrinsic microporosity with a relatively high surface area of 630 m 2 /g. 17 A number of later studies reported conductivities ranging from 2.6(3) × 10 −3 to 150 S/cm for Ni 3 (HITP) 2 . 22,23,27,79,189 High quality thin films of this landmark material were later reported by Wu et al., who fabricated field-effect transistors 190 from Ni 3 (HITP) 2 . These devices revealed p-type semiconducting behavior with a high hole mobility of ∼40 cm 2 V −1 s −1 . This mobility is higher than for most organic  Chemical Reviews pubs.acs.org/CR Review semiconductors and is likely responsible for the high electrical conductivity of the material. The depletion-mode behavior of the FETs indicated that the as-made material possessed a sufficiently high carrier concentration. The FETs exhibited high on/off ratios of up to 1000. The excellent bulk transport properties align well with what chemical intuition would predict based on the structure of Ni 3 (HITP) 2 . Strong in-plane bonding between the iminosemiquinone HITP 3− linker and Ni II and close (3.35 Å) neareclipsed stacking of the π-conjugated layers both contribute to charge delocalization and high mobility. Further oxidation of the iminosemiquinone linkers leads to removal of electrons formation of excess holesin the material, producing a p-type semiconductor. These results, taken together with the temperature dependence, are consistent with Ni 3 (HITP) 2 being a semiconductor.
Surprisingly, in a later study, Day et al. fabricated singlecrystal four-probe devices of Ni 3 (HITP) 2 , which revealed higher conductivities (up to 100 S/cm) and, more importantly, a nonzero conductivity at temperatures approaching 0 K. 79 Day et al. proposed that this behavior can be explained by metallic charge transport in the material. 79 This apparent disagreement between the FET study and the single-crystal study can perhaps be resolved by a computational investigation by Foster et al., 63 which showed that disorder and defects in Ni 3 (HITP) 2 can lead to the opening of a band gap in the otherwise metallic material.
A Cu II -based HITP material has also been reported, with a structure apparently similar to Ni 3 (HITP) 2 . Cu 3 (HITP) 2 showed conductivities as high as 0.2 S/cm for two-probe pressed pellet measurements. 21 Poor air stability of the material precluded further investigations.
Recently, Lian et al. reported the novel Co 3 (HITP) 2 MOF, as well as mixed-metal Co x Ni 3−x (HITP) 2 MOFs. 191 However, the crystallinity of the Co 3 (HITP) 2 samples was very poor and may be at least partially responsible for the relatively low bulk conductivity of 8 × 10 −4 S/cm, several orders of magnitude lower than that of pure Ni 3 (HITP) 2 synthesized via similar methods (0.19 S/cm). The mixed-metal Co/Ni MOFs showed increasing conductivities with increasing Ni content. Because of the lack of structural data, it remains to be determined whether the lower conductivity of Co 3 (HITP) 2 compared to Ni 3 (HITP) 2 is a result of different electron counts, structural distortions, layer stacking patterns, or other factors.

Hexaiminobenzene MOFs.
A number of materials based on hexaiminobenzene (HIB), the nitrogen-containing analog of benzenehexathiol and hexahydroxybenzene, have also been made. The first report of a coordination polymer based  Lahiri et al. 192 The study showed synthesis of thin films of HIB materials based on Cu II , Ni II , and Co II . Although little experimental evidence was provided in support of structural claims, the materials showed hexagonal symmetry based on electron diffraction patterns. No conductivity values were reported, although FET devices were fabricated, demonstrating semiconducting behavior. Two later reports discussed the Cu II and Ni II -based frameworks in more detail, showing that their structures were both similar to that of Ni 3 (BHT) 2 . 29,77 In Ni 3 (HIB) 2 and Cu 3 (HIB) 2 , however, the hexagonal honeycomb layers stacked in a slipped-parallel fashion, as opposed to the eclipsed arrangement in Ni 3 (BHT) 2 . The reported conductivities varied between the different reports, ranging from 0.7 S/cm to 10 S/ cm for Ni 3 (HIB) 2 and from 0.11 S/cm to 0.7 S/cm for Cu 3 (HIB) 2 . Ultraviolet photoemission spectroscopy for both materials showed a nonzero charge density at the Fermi levelsimilar to Ni 3 (BHT) 2 indicating metallicity. However, the temperature dependence of the electrical conductivity of both Ni 3 (HIB) 2 and Cu 3 (HIB) 2 again showed a semiconductor-type dependence and was attributed to intergrain contributions dominating charge transport that is otherwise intrinsically metallic (i.e., metallic transport within a singlecrystal particle but temperature-activated transport between particles). Ni 3 (HIB) 2 and Cu 3 (HIB) 2 showed BET surface areas of 100−150 m 2 /g with type II N 2 adsorption isotherms, indicating mostly external surface adsorptionunsurprising given the extremely small particle size (∼10 nm) and small pore diameter (<1 nm).
A Co II -based material with a structure similar to that of Ni 3 (HIB) 2 was reported by Park et al., who proposed an eclipsed arrangement of the layers in the Co analog (unlike the Cu and Ni materials). 28 This stacking mode led to permanent microporosity with a BET surface area of 240 m 2 /g. The electrical conductivity of the material, as determined from twoprobe pressed pellet measurements, ranged from ∼0.1 to 1.57 S/cm, similar to the Cu II and Ni II materials.
Nanoparticulate Fe 3 (HIB) 2 and Mn 3 (HIB) 2 showed some of the highest conductivities observed in MOFs, 30 reaching 100− 360 S/cm at 300 K for pressed pellets. The authors attributed the transport properties to high crystallinity of their materials: the PXRD patterns appear to be better resolved than for other M 3 (HIB) 2 materials. PXRD and TEM suggested that the materials have an eclipsed arrangement of the 2D layers, which may also have contributed to the higher conductivities. Interestingly, the materials were obtained as multishelled hollow sphere hierarchical structures, which led to very high surface areas of over 2300 m 2 /g.  194 This framework exhibited the same honeycomb structure as Ni 3 (HITP) 2 and was calculated to favor a similar slipped-parallel stacking. The as-made material, synthesized in air-free conditions, was fully insulating, but oxidation with I 2 led to a drastic increase in conductivity to 0.01 S/cm. While technically the material could be considered a covalent organic framework beyond the scope of this Review, this work showcases the structurally tunable nature of conductive MOFs. Further work is necessary to determine the origin of the substantial differences in charge transport behavior of this material and Ni 3 (HITP) 2 .
The group of Nishihara, which originally reported Ni 3 (BHT) 2 , later reported a Ni II coordination polymer based on 1,3,5-triaminobenzene-2,4,6-trithiol, which can be thought of as a hybrid of benzenehexathiol and hexaaminobenzene. Both an oxidized 195 and a reduced form of this material have been reported. 196 The reduced form showed sharp PXRD peaks ascribed to the staggered arrangement of the hexagonal conjugated layers, similar to Ni 3 (BHT) 2 . The oxidized form presumably has a similar structure. The reduced material exhibited a low conductivity of 3 × 10 −6 S/cm, which increased to 0.3 S/cm upon oxidation, as indicated by pressed pellet measurements. This change is unsurprising given that the reduced form is unlikely to possess a significant number of free carriers, which typically originate from the oxidation of the amino moieties into their radical iminosemiquinonate form. The relative conductivities are also consistent with a DFT study of the material, which indicated the oxidized form to be metallic, and the reduced form to be a semiconductor with a band gap >1.0 eV. The two forms of the material were interconvertible by solution-phase oxidation with ferrocenium tetrafluouroborate or reduction with decamethylcobaltocene.

THROUGH-SPACE PATHWAYS
Unlike the previous two strategies, the through-space approach focuses on intermolecular interactions of organic linkers instead of covalent metal−ligand bonds. Largely inspired by organic semiconductors and charge-transfer salts, this approach utilizes the high propensity of planar conjugated organic ligands to form structures containing π−π stacking in the solid state. The intrinsic tunability of MOFs can allow for fine control over the stacking sequence and distance to a degree that is difficult to achieve in classical polymeric and molecular materials. As is the case for the other approaches, mixed valency plays a crucial role: partial oxidation of organic ligands 37,85,198 provides charge carriers and potentially even biases the system toward the formation of π−π interactions during synthesis. 199 Frameworks based on tetrathiafulvalene (TTF), an archetypal organic semiconductor building block, constitute a large portion of MOFs designed with this approach in mind. Some of the most detailed studies correlating π−π stacking configuration and conductivity were performed on materials made from TTF-based ligands. In particular, these studies demonstrate how varying parameters such as metal size 84 and solvent ratios 85 can result in structural changes that allow control over transport properties. That said, studies of these materials are not limited to TTF-based MOFs, and π−π stacking between a diverse set of other organic cores, including anthracene, 38 naphthalene, 89 and napthalenediimide, 90,91 has so far enabled efficient charge transport (Table 6). Clearly, given the vast arsenal of organic molecules that have been employed in organic semiconductors with π−π stacking in mind, we expect that this approach will become much more popular in the future.

Chemical Reviews pubs.acs.org/CR Review
In most of the existing examples, π−π stacking between ligands results in the formation of 1D conduction pathways. However, more unusual and complex arrangements, such as zigzag chains and orthogonal orientations, have also been reported. 38,88 Many 2D extended conjugation MOFs also exhibit significant π−π stacking among the covalently bonded layers, and as previously discussed, this stacking may in fact contribute as much to the conductivity as the in-plane conjugation. Therefore, charge transfer between organic components is a pervasive motif in conductive MOFs regardless of their categorization.

Tetrathiafulvalene-Based Frameworks
TTF is an organosulfur compound that is an excellent electron donor. 200 The stability of its radical cation form has made it a popular component in conducting charge-transfer salts, in which π−π stacking and close intermolecular contacts in  0.037 2-probe pellet 90 Cu(DPNDI) 2 1.2 × 10 −5 2-probe sc 91 Cd 2.39 (TPDAP) 3 1 × 10 −6 2-probe sc 210 Cd 2 (TPDAP) 2 insulating 2-probe sc 210 a See Abbreviations. b Measured at room temperature (exact temperature unspecified) or between 293 and 300 K. c Single-crystal devices are contacted parallel to the π−π stacking direction unless otherwise indicated. d Derived from Arrhenius model. For materials exhibiting more than one distinct activation regions, E a at 300 K is given. e Calculated from N 2 adsorption isotherms. f Contacted perpendicular to π−π stacking direction. Chemical Reviews pubs.acs.org/CR Review oxidized TTF stacks facilitate charge transport. 201 Functionalized TTF derivatives are also used in supramolecular chemistry because of their redox activity and participation in noncovalent interactions. 202,203 As a result of their electroactive nature and π−π stacking propensity, ligands based on TTF form conductive MOFs with a diverse range of structures. We note that conductive, nonporous coordination polymers containing TTF-based linkers, such as TTF tetracarboxylate, 204 have been reported and reviewed. 205 The isostructural series of M II 2 (TTFTB) frameworks (M = Mn, Co, Zn, and Cd; TTFTB = tetrathiafulvalene tetrabenzoate) reported by Narayan et al. 37 and Park et al. 84 contain infinite 1D helical stacks of TTF ( Figure 21). They are permanently porous, with BET surface areas between 559 and 665 m 2 /g. The single-crystal structures of these MOFs showed that the TTF moieties within the stacks have a single close S··· S contact distance of <3.8 Å. As the size of the M II cation increases, the stack effectively becomes pinched, shortening the S···S distance from 3.773(3) Å for Co II to 3.654(2) Å for Cd II .
The single-crystal conductivities measured along the stacking direction for the M 2 (TTFTB) frameworks showed a striking correlation with the S···S distances, with average values of 2.9(5) × 10 −4 S/cm for Cd 2 (TTFTB), 9(1) × 10 −5 S/cm for Mn 2 (TTFTB), 1.5(3) × 10 −5 S/cm for Co 2 (TTFTB), and 4.0(6) × 10 −6 S/cm for Zn 2 (TTFTB). DFT band structure calculations indicated that the valence band comprises S and C orbitals from the ligand TTF core, and has a large dispersion of 400 meV. Increasing the size of the metal cation and contracting the stacking distance increases the overlap of S p z orbitals and, thus, the charge mobility along the TTF stacks. Detailed investigations of the electrochemical and spectroscopic properties of these MOFs further confirmed the central role of the TTF core, showing that ligand-based IVCT mediates the charge transport. 198 The large band dispersion leads to high charge mobilities in these frameworks: FP-TRMC and TRTS studies of Zn 2 (TTFTB) revealed a mobility value 37 of 0.2 cm 2 V −1 s −1 , and photoconductivity (with 400 nm pump-pulses) of 5 × 10 −4 S/cm. 206 The large (up to 1 mm in length), high-quality single crystals that can be obtained for these M 2 (TTFTB) phases have enabled thorough characterization of their properties: for example, Cd 2 (TTFTB) was used as a model system for a comparative study of techniques to measure conductivity in MOFs. 56 Several different TTFTB MOFs based on lanthanides show diverse structures, while retaining the general trend of conductivities correlating inversely with S···S contact distances. Xie et al. reported three TTFTB MOFs with La III which crystallize with 1D stacks of TTFTB ligands and rectangular pores. 85 Notably, the H 2 O to DMF ratio used in their synthesis appeared to play a role in determining the extent of π−π stacking. The two more conductive phases showed average conductivity values of 2.5(7) × 10 −6 and 9(4) × 10 −7 S/cm, respectively, with longest S···S contacts of 3.640(4) and 4.083(3) Å. The third phase was significantly less conductive (1.0(5) × 10 −9 S/cm) and has a much longer S···S contact of 7.072(7) Å. The BET surface areas of these polymorphs ranged between 362 and 596 m 2 /g.
TTFTB-based MOFs with middle and late lanthanides also showed polymorphism dependent on the synthesis conditions ( Figure 22). Castells-Gil et al. reported 86 two structure types of TTFTB frameworks with Gd III , Tb III , Dy III , Ho III , and Er III . M 3 (TTFTB) 2 (OAc)(OH), featuring continuous π−π stacking with S···S contact distances of <4 Å, showed consistently higher conductivities, with the highest value of 1.5 × 10 −5 S/ cm measured in the Tb III material. A different family of frameworks with Tb III , Dy III , Ho III , and Er III reported by Su et al. do not have close or continuous π−π stacking and consistently exhibited relatively low conductivities between 10 −9 and 10 −8 S/cm, with marginal increases upon doping with I 2 . 87 The smallest lanthanides, Tm III , Yb III , and Lu III , form TTFTB MOFs with unusual 2D structures and showed conductivities up to 2.6 × 10 −6 S/cm. 88 In contrast with the 1D TTF stacks in most other phases with this ligand, these MOFs contain one ligand oriented orthogonally to the others.
Assuming that electron transfer between TTF cores dominates charge transport in TTFTB MOFs, the longest S···S contact in 1D ligand stacks should limit the overall rate of charge transfer in the stack. Hence, the value of the shortest S···S contact between each pair of adjacent ligands, denoted as (S···S) min , should be correlated with the conductivity. Indeed, plotting the conductivities of these materials vs the longest crystallographic (S···S) min contacts in each structure shows an inverse correlation between these parameters (Figure 23). 85 This general trend suggests that intermolecular TTF distances determine the bulk conductivities of these MOFs across different structures and compositions. Similar trends should be observed for other MOFs where the dominant transport pathway is through close ligand···ligand π-based interactions.
For instance, other TTF-containing frameworks have been synthesized with the tetrathiafulvalene tetrapyridyl (TTF-(py) 4 4 ]-(ClO 4 ). 207 They lack continuous π−π stacking pathways and have conductivities on the order of 10 −9 S/cm in their assynthesized forms. Upon treatment with I 2 , their conductivities increased to 1.3 × 10 −6 and 7.6 × 10 −5 S/cm because of the oxidation of the TTF cores to their radical cation state. Notably, the incorporation of I 3 − proceeded via a single crystalto-single crystal transformation in the latter MOF. Yu et al. reported three Mn II frameworks with TTF(py) 4 with similar conductivities as the undoped Fe II analogs (between 2.0 × 10 −10 and 6.3 × 10 −9 S/cm) because of the absence of closestacking interactions in their structures. 208 Wang et al. also combined the TTF(py) 4 ligand with TCNQ as both a second ligand and an interstitial guest in MOFs with Zn II and Cd II . 209 Though the combination of donor−acceptor systems in a single MOF framework is appealing for increasing conductivity, the conductivities of these frameworks were relatively low (∼10 −8 S/cm). After treatment with I 2 , the conductivity of the Cd II framework increased slightly to 2.16 × 10 −7 S/cm, but no evidence of additional TTF(py) 4 oxidation was apparent (the interstitial TCNQ •− was replaced with I 3 − ). The modest conductivity values were attributed to large separation between the redox-active moieties and the orthogonality of TTF and TCNQ.
In most of these reports, EPR and optical absorption spectroscopy studies on TTF-based frameworks have generally indicated that they contain some TTF •+ species in their assynthesized forms. Therefore, IVCT between partially oxidized TTF cores appears to be the general mechanism responsible for conductivity in these materials.

Other π−π Stacked Frameworks
A number of other planar organic cores have been used to make electrically conductive MOFs, including anthracene, naphthalenediimide, and naphthalene. The resulting materials have generally been studied in isolated reports, and unifying or comparative studies have yet to be carried out. Nonetheless, some of the highest conductivity values for the through-space approach have been achieved in these MOFs. Hence, more systematic investigations into non-TTF π−π stacked frameworks and their conductivities will accelerate development in this area and point out worthy synthetic targets for new linkers.
Chen et al. reported a mixed-metal MOF with the formula ZnNa 2 (AnBEB) 2 (AnBEB = 4,4′-(anthracene-9,10-diylbis-(ethyne-2,1-diyl))dibenzoate) containing unusual zigzag packing of the anthracene moieties, with an interplanar distance of approximately 3.4 Å between neighboring ligands. 38 The conductivity from single-crystal measurements was 1.3(5) × Chemical Reviews pubs.acs.org/CR Review 10 −3 S/cm. This high value was attributed to the relatively close stacking distances and the zigzag packing motif. In contrast to other MOFs in this category, there was no evidence of ligand-based mixed valency for this framework. After activation, crystallinity and porosity were lost due to structural collapse from removal of coordinated solvents. Another study reported a porous Cd II material with the N,N′-di(4-pyridyl)-1,4,5,8-naphthalenetetracarboxdiimide (DPNDI) linker. 90 The structure is a π−π stacked assembly of linear [Cd(OH 2 ) 4 (DPNDI)] polymers, which arrange into a hexagonal framework. The interlayer stacking distance is remarkably short, at only 3.18 Å (compared to typical distances of ∼3.3 Å for napthalenediimide (NDI)-based molecules). This close stacking, along with the ability of the NDI core to form stable radical anions, led to high conductivities up to 3.3 × 10 −3 S/cm measured on single crystals along the stacking axis. Measurements on pressed pellets yielded much lower values up to 7.6 × 10 −6 S/cm, perhaps due to the considerable anisotropy of the material. Unexpectedly, the desolvated form showed a higher conductivity than either of the above (3.7 × 10 −2 S/cm). However, the material exhibited a structural transition upon solvent removal, and its final structure was not fully determined.
Kuang et al. studied Cu I (DPNDI) 2 , an NDI-based framework with considerably further stacking distances than in the Cd II material, with closest contacts of ∼3.7 A. 91 The NDIs are also offset horizontally, leading to reduced overlap of their πsystems. Despite this, the MOF demonstrated a relatively high conductivity of 1.2 × 10 −5 S/cm, as determined from singlecrystal measurements.
Another Cd II -pyridyl-based framework, with the redox-active conjugated linker 2,5,8-tri(4-pyridyl)1,3-diazaphenalene, showed single-crystal conductivities of ∼1 × 10 −6 S/cm. 210 The intrinsically porous structure consists of continuous columns of the linkers, with a relatively short stacking distance of 3.3 Å. The same linker was also found to form a different framework with Cd II without continuous stacking, which was insulating. The different transport properties between these two MOFs demonstrate the importance of a continuous charge transport pathway.
Dense naphthalene-based MOFs with π−π stacking have also been reported to be conductive. Haider et al. obtained a dense Sr II framework with 1,4,5,8-naphthalenetetracarboxylate that exhibited white-light emission. 89 This structure contains 2D sheets with Sr−carboxylate chains and closely stacked naphthalene cores separated by ∼3.4 Å and showed a conductivity of ∼10 −4 S/cm. Panda and Banerjee fabricated field-effect transistors based on a series of indium isophthalate frameworks. 211 Out of the three materials investigated, only one featured close π−π stacking and showed a hole mobility of 4.6 × 10 −3 cm 2 V −1 s −1 . Electrical conductivity of the material was not reported.

REDOX HOPPING
In this section, we summarize the transport properties of MOFs that do not conform to the above-mentioned design strategies. In general, these materials do not contain welldefined crystallographic pathways that could allow for bandlike charge transport. Although many of these MOFs do contain redox-active components, they are separated by distances long enough to preclude direct orbital overlap. Therefore, charge hopping is the likely mechanism of conduction in these materials.
On average, the greater separation between hopping sites in these materials can be expected to coincide with lower charge mobilities compared to other classes of conductive MOFs. Consistent with this expectation, the highest conductivities for MOFs in this category 26,212,213 are lower than those falling under other design strategies. However, the wide range of conductivity values reported (Table 7) suggest that the factors limiting charge transport may vary widely among different materials (e.g., carrier concentration, carrier mobility, or grain boundary resistance), and can be difficult to predict from structural and compositional information alone.
In general, higher conductivities in this category appear to correlate with the presence of redox-active components in the MOFs (e.g., linkers featuring metal−dithiolene units, 26,64,214 conjugated organic cores, 213,215 and azo groups 212,216−219 ). Excitation of mobile carriers from these motifs presumably facilitates conductivity.

Metal-Based Hopping
In 2009, Takaishi et al. reported Cu[Cu(pdt) 2 ] (pdt = pyrazine-2,3-dithiolate], which exhibited a conductivity of 6 × 10 −4 S/cm. 64 The authors hypothesized that linking the known electron acceptor [Cu III (pdt)] − with donating Cu I ions could lead to the formation of an electrically conductive MOF. The authors obtained a 3D porous framework with Cu centers in a square-planar geometry ( Figure 24) upon reaction of Na[Cu-(pdt) 2 ]·2H 2 O and CuI. Analysis of the bond lengths and angles from the X-ray crystal structure indicated that electron  reported that the compound amorphized and collapsed upon desolvation. 214 They were able to synthesize the structural analog Cu[Ni(pdt) 2 ], with Ni replacing Cu in the dithiolenecoordinated site. This mixed-metal MOF retained crystallinity and microporosity after activation, exhibiting a BET surface area of 385 m 2 /g. The conductivity of as-synthesized Cu[Ni(pdt) 2 ] was reported to be 1 × 10 −8 S/cm in the original study and 2.6 × 10 −6 S/cm in a later study. 26 Upon exposure to I 2 vapor, the conductivity increased to 1 × 10 −4 S/ cm without measurable incorporation of iodine species into the framework. Analogous to the case of Cu[Cu(pdt) 2 ], redox hopping among partially oxidized nickel-dithiolene units is the likely mechanism of conduction. These results constitute the first example of a permanently microporous MOF with conductivity tunable over several orders of magnitude through doping.
Dolgopolova et al. studied the influence of exchanging Co II into Cu 3 (BTC) 2 (BTC = benzene-1,3,5-tricarboxylate) on its conductivity. 220 The particular choice of Co II was motivated by an earlier computational study, which investigated the effects of partial metal substitution in MOF-5 on its electronic structure. 221 Cu 2.4 Co 0.6 (BTC) 2 was found to be somewhat more conductive than monometallic Cu 3 (BTC) 2 , with the conductivity increasing from 2.0 × 10 −10 S/cm to 1.4 × 10 −8 S/cm with the introduction of Co II . In the same study, Dolgopolova et al. also provided a microwave conductivity investigation of the materials, showing that the effective conductivity of the exchanged material was approximately one order of magnitude higher than that of the original. This difference was attributed to the presence of Co II midgap states, which can lower the energetic barrier to charge hopping.
An analogous strategy was employed by the same group to increase the conductivity of a bimetallic actinide framework U 1.23 Th 4.77 O 4 (OH) 4 (NO 3 ) 2 (Me 2 BPDC) 5 (Me 2 BPDC 2− = 2,2′-dimethylbiphenyl-4,4′-dicarboxylate). 222 Introduction of Co II ions into the framework led to the extension of the hexanuclear cluster node to give the formula U 1.23 Th 4.77 Co 3 O 4 (OH) 4 (NO 3 ) 2 (Me 2 BPDC) 5 and resulted in enhancement of the electrical conductivity from 7.0 × 10 −10 S/ cm to 1.4 × 10 −7 S/cm. DFT calculations showed narrowing of the bandgap upon addition of Co II due to the addition of midgap Co−O-based states. This result is similar to what was observed in the previous Cu 3 (BTC) 2 study and is most likely responsible for the improved charge transport.

Linker-Based Hopping
Dolgopolova et al. also investigated ways of controlling charge transport in photoresponsive MOFs using linker isomerization. 213 The authors synthesized a set of mixed-linker frameworks with photoactive spiropyran and diarylethene fragments in the linkers. The spiropyran fragments isomerize on exposure to UV light to form merocyanine zwitterions. Similarly, diarylethene moieties form more extended con-jugated (closed) systems under UV-irradiation, which break up (open) following exposure to visible light at 590 nm ( Figure  25a). The authors used these two transformations as switches for charge transport, hypothesizing that the different charge distribution and size of the conjugated system between the nonisomerized and isomerized linkers would affect the energetic and spatial separation between hopping sites, thus allowing control over the conductivity. They found that a spiropyran-containing MOF exhibited a reversible increase in the conductance of single crystals upon UV irradiation. Reverse isomerization of merocyanine at ambient conditions complicated the study, so the authors focused on photoactive systems using diarylethene (Figure 25b). Both in situ and ex situ conductivity and diffuse reflectance measurements on three different Zn II -and Zr IV -based frameworks containing diarylethene motifs showed small but statistically significant (p < 0.05) three-to four-fold increases in conductivity upon isomerization of the linker into its "closed" form (Table 7). This study demonstrated the first realization of photoswitchable charge transport in MOFs.
Goswa mi e t al . st udied t he conduct ivit y of Zr 6 (OH) 16 (TBAPy) 2 (NU-1000; TBAPy = tetrakis(p-benzoate) pyrene) thin films at 1.6 V vs Ag/AgCl using EIS. 223 At this potential, the TBAPy linker is expected to be present in both its closed-shell TBAPy 4− and radical trianion TBAPy •3− forms. They discovered higher conductivities for films with preferential orientation along the c direction than those oriented along the ab direction, which is consistent with Chemical Reviews pubs.acs.org/CR Review closer redox hopping distances along the c direction. As expected from the generation of more charge carriers (i.e., radical ligand species), the conductivity values obtained by this method (up to 1.2 × 10 −7 S/cm) were significantly higher than those reported for NU-1000 without an applied potential (9.1 × 10 −12 S/cm). 224 Dawood et al. fabricated thin film devices of interpenetrated Zn 4 O(NDC) 3 (NDC = 2,6-naphthalenedicarboxylate) using electron beam deposition of copper electrodes. 225 Surprisingly, these devices exhibited high conductivities up to 3.98 × 10 −2 S/cm despite relatively discontinuous stacking in the structure of the MOF. The authors hypothesized that the interpenetrated arrangement of naphthalene cores allows for charge hopping.

Mixed Metal-and Linker-Based Hopping
Ghoshal and co-workers have reported several Zn II and Cd II frameworks with the 4,4′-azobispyridine linker and other carboxylate ligands, including 5-hydroxyisophthlate, 212,216 methylsuccinate, methylglutarate, and glutarate. 217 Their conductivities ranged from 1.29 × 10 −5 S/cm 217 to 1.86 S/ cm. 212 Roy and co-workers made Cd II frameworks with an extended azo-based linker, 1,4-bis(3-pyridyl)-2,3-diaza-1,3butadiene, with conductivities of 4.53 × 10 −7 S/cm 218 and 2.90 × 10 −6 S/cm. 219 All of these azo materials also exhibited Schottky diode-like behavior in thin film devices, with some increases in conductivity upon photoillumination. They were dense except for one Zn II framework with 4,4′-azobispyridine and 5-hydroxyisophthlate, which displayed a BET surface area  216 Redox-hopping among metal centers and π−π interactions among the linkers were both proposed to contribute to the observed conductivity. Huang et al. combined the triphenylenehexathiol (THT) ligand with Pb II in order to obtain a framework with the formula [Pb 6 (THT) 2 (OH)] 3− (charge-balanced by Na + in the pores). 226 While other MOFs with the THT ligands exhibit extended conjugation in 2D structures, the reaction conditions used in this study (using in situ deprotection of the linker to slow down crystallization) appeared to favor the formation of a 3D framework without close π−π interactions or continuous through-bond pathways. The conductivity of this MOF (1.1 × 10 −6 S/cm) was accordingly lower compared to other THTbased materials.
The charge transport and electrochemical properties of a porphyrin-based framework, Co 3 [Co(TCPP)] 2 (TCPP = 5,10,15,20-(4-carboxyphenyl)porphyrin), was investigated by Ahrenholtz et al. 227 The conductivity measured by EIS on a thin film grown on FTO was 3.62 × 10 −8 S/cm. Charge diffusion parameters from cyclic voltammetry and spectroelectrochemical analysis combined with the relatively low conductivity value were deemed consistent with a redox hopping mechanism of conduction between framework Co centers and the linker porphyrins.  228 Matheu et al. recently reported a 3D framework based on Co II phthalocyanine with the OPc linker and Fe III as the framework metal with the formula Mg 3 Fe 2 [Co(OPc)]. 101 Unlike other OPc frameworks, which form 2D sheets with extended metal−linker conjugation, this MOF contains no π−π stacking, with all ligands being separated by more than 13 Å. The octahedral coordination of Co II centers in the Pc units (with two axial aquo ligands) and the formation of Fe 3 (−C 2 O 2 −) 6 (OH 2 ) 2 SBUs likely both favor this more porous structure. Despite the long separation distance, the single-crystal conductivity (measured along the c direction, which contains continuous SBU−OPc linkages) was ∼10 −5 S/ cm. The relatively high value was attributed to charge hopping between redox-active Pc molecules and SBUs.
Duan et al. synthesized thin nanosheets of the dense 2D mixed-metal MOF (Ni/Fe)(NDC) for electrocatalytic water splitting. 229 The nanostructured MOF was significantly more conductive (1.0(2) × 10 −5 S/cm) than the bulk MOF (1.0(5) × 10 −8 S/cm), which was attributed to defects and vacancies induced by nanostructuring. The mixed-metal approach may also have introduced free charge carriers into the framework.
The conductivities of several dense frameworks with benzene-based carboxylate linkers without open-shell transition metals have been studied. 230−232 These values range from 10 −11 to 10 −5 S/cm. Hopping between the ligands or carboxylate-coordinated metal centers is likely to be responsible for the transport behaviors observed.

GUEST-PROMOTED TRANSPORT
Using host−guest interactions to increase conductivity is perhaps the strategy most specific to MOFs. By leveraging the intrinsic microporosity of MOFs, high loadings of electroactive molecules can be introduced into the pores to increase electrical conductivity (Table 8). In contrast with charge doping, which increases the number of charge carriers by reacting the material with (usually small quantities of) oxidizing or reducing species, guest-mediated conductivity commonly involves the introduction of stoichiometric amounts of new species into the MOF. These relatively high loadings of guests are usually necessary to form continuous charge transport pathways (and hence obtain high mobilities). Guests with extended structures, including conductive polymers and inorganic oxides, have also been introduced. In most cases, charge transfer between the framework and the guest appears to be important for increasing conductivity, presumably as a mechanism for generating free charge carriers.
The advantages of this strategy lie chiefly with the possibility of tuning the guest and framework properties in a synergistic fashion. Materials with highly tunable conductivities can be realized, 235 and conductivity modulation across 11 orders of magnitude have been demonstrated. 236 In addition, by selecting starting frameworks with high porosities and different pore sizes, the resulting guest@MOFs can retain high surface areas. 224,236−239 For frameworks that are less porous in their original state, partial loading of guests can potentially allow for the optimization of conductivity versus surface area. Matching the reduction potentials of the framework and guest, 236 as well as the sizes of the guest and pores, 240 are promising routes toward further customization of transport properties in this class of materials.
Although examples of crystallographically resolved guests exist, 224,241,242 the inclusion of guests and the formation of conductive pathways are often inferred from spectroscopic data that do not provide atomistic resolution. Finally, growth of conductive side products during guest infiltration can further complicate the attribution of increased conductivity to the guest@MOF assembly. 243 Detailed compositional and crystallographic studies, therefore, can clarify the mechanisms behind conductivity enhancement achieved through this strategy.

Iodine and Polyiodides
Although iodine loading into MOFs has been explored predominantly in the context of dopingwith the charge transport happening through pathways intrinsic to the frameworkconductivity mediated by encapsulated channels of iodide species has also been demonstrated. Zeng et al. found that a double-walled MOF Zn 3 (lac) 2 (pybz) 2 (pybz = 4-(pyridin-4-yl)benzoate, lac = DL-lactate) with 1D square channels incorporated I 2 upon soaking in a cyclohexane solution. 244 Conductivity measurements on single monoliths (loss of crystallinity was observed upon reaction with I 2 ) revealed a value of 3.42 × 10 −3 S/cm along the direction of the channels. The Co II analog incorporated a similar amount of I 2 but exhibited a lower pellet conductivity of 7 × 10 −6 S/cm. 245 The authors of these studies proposed that charge transfer between iodide species and pybz π-electrons led to these relatively high conductivities (we note that the conductivities of the guest-free frameworks were not reported). However, loss of crystallinity upon I 2 incorporation made mechanistic interpretation challenging.
Yin et al. uncovered more definitive evidence of the role of polyiodide inclusion in a related system. 241 By including I 2 in the solvothermal reaction with Cu II and pybz, they obtained Chemical Reviews pubs.acs.org/CR Review Cu 6 (pybz) 8 (OH) 2 ·I 5 − ·I 7 − with crystallographically resolved polyiodide anions incorporated into the channels. The conductivity of a thin film of the material was 8.11 × 10 −7 S/cm. Soaking in methanol removed I 2 from the polyiodides to afford Cu 6 (pybz) 8 (OH) 2 (I − ) 2 and lowered the conductivity by 2 orders of magnitude (8.04 × 10 −9 S/cm). We note that, Hall bar film 250 Cu 6 (pybz) 8  Measured at room temperature (exact temperature unspecified) or 298 K. Value given in parentheses corresponds to pristine guest-free material. c ∥ denotes device contacted parallel to guest-filled channels; ⊥ denotes device contacted perpendicular to channels. d Derived from Arrhenius model. For materials exhibiting more than one distinct activation regions, E a at 300 K is given. e Calculated from N 2 adsorption isotherms. Chemical Reviews pubs.acs.org/CR Review while these pybz materials were porous in their original forms, with BET surface areas up to 800 m 2 /g for Co 3 (DLlac) 2 (pybz) 2 , their porosities were likely not maintained after iodide inclusion. No adsorption measurements were attempted for any of these iodine-treated materials. Hu et al. combined thorough crystallographic and physical characterization to explain the enhancement of transport properties in Tb(Cu 4 I 4 )(PCA) 3 (PCA = 4-pyridinecarboxylate) upon uptake of I 2 . 246 The single-crystal conductivity of the parent material, which included framework iodide anions, was 5.72 × 10 −11 S/cm. After I 2 inclusion, the conductivity increased by 7 orders of magnitude to reach 2.16 × 10 −4 S/cm. This dramatic change was attributed to the extended {Cu 4 I 5 } layer formed by strong I − ···I 2 ···I − interactions, as seen from single-crystal X-ray diffraction ( Figure 26). Notably, this work demonstrated reversible adsorption/desorption of I 2 and associated modulation of the conductivity.
Zhang et al. found that iodine adsorption in the V 2 (OH) 2 (BPTC) framework resulted in oxidation of about 30% of the native V III centers to V IV . 247 Powder and singlecrystal X-ray diffraction analyses revealed the presence of both I 2 and I 3 − in the channels, forming a helical chain. The conductivity of the iodine-containing MOF increased to 1.2 × 10 −4 S/cm from an original value of 1.7 × 10 −10 S/cm. The authors attributed conductivity to both mixed-valent V centers and charge hopping along the iodine chains.
Other instances of iodine incorporation resulting in electrical conductivity have also been reported. Encapsulation of ∼17% iodine by weight into Eu 4 (BPT) 4 (BPT = biphenyl-3,4′,5tricarboxylate) yielded a conductivity of 8.27 × 10 −7 S/cm (the conductivity of the parent material was not reported). 248 Uptake of I 2 and I 3 − by Co 3 (NDC) 3 (NDC = 2,6naphthalenedicarboxylate) 249 resulted in a conductivity of 1.88 × 10 −6 S/cm for a thin film device. 250 Hall effect measurements were also carried out, yielding a hole mobility of 21.2 cm 2 V −1 s −1 and a carrier concentration of 5 × 10 11 cm −3 .
Hall effect measurements on films of CoBDC and CoNDC grown on TiO 2 and doped with I 2 showed similar mobility values of 21.2 and 87.6 cm 2 V −1 s −1 , respectively. 251 Finally, stoichiometric polyiodide incorporation also increased the conductivity of several tetrathiafulvalene-based MOFs, as discussed in more detail above. 87,198,207,209 In most cases, these changes were attributed to oxidation of the ligand core, although we note that transport through iodide channels may also play a role.

Organic and Organometallic Molecules
The guest-promoted approach is not limited to smaller inorganic species, as large organic molecules have also been incorporated in MOFs to increase their conductivity. Talin et al. reported that soaking thin films of Cu 3 (BTC) 2 (or HKUST-1; BTC = benzene-1,3,5-tricarboxylate) in a saturated solution of TCNQ, an electron acceptor, increased its conductivity 6 orders of magnitude from 10 −8 to 0.07 S/cm after 72 h. 235 The decrease in the BET surface area from 1844 to 214 m 2 /g indicated that TCNQ was incorporated into the pores. The conductivities increased with longer exposure times in good agreement with percolation theory, leading the authors to propose that conducting regions were formed in the MOF by TCNQ bridging the copper paddlewheel SBUs ( Figure 27). Vibrational and UV−vis spectroscopic data were consistent with partially reduced TCNQ, implying charge transfer between the framework and TCNQ.
Higher conductivities in TCNQ@Cu 3 (BTC) 2 compared to the pristine MOF were corroborated by other reports. 243,252,253 However, Schneider et al. 243 and Thurmer et al. 254 showed that partial or full conversion of Cu 3 (BTC) 2 into the dense coordination polymer Cu(TCNQ) (σ ∼ 0.1 S/cm 82,255 ) upon exposure to TCNQ proceeded under certain conditions. These works highlighted the importance of verifying the phase composition of MOFs after treatment with redox-active species and raise the question of whether the formation of a highly conductive parasitic phase contributes to the improved conductivity. Hence, additional investigation into the mechanism of conductivity enhancement in this system remains important.
TCNQ incorporation led to a 10 4 -fold improvement in conductivity in Cu 2 (TATAB) 3 (TATAB = 4,4′,4″-((1,3,5-  Chemical Reviews pubs.acs.org/CR Review triazine-2,4,6-triyl)tris(azanediyl)) tribenzoate), a denser MOF-based on the same copper paddlewheel SBU. 256 Although the overall structure and metal environment in Cu 2 (TATAB) 3 is similar to Cu 3 (BTC) 2 , the larger size of the linker leads to formation of an interpenetrated framework, with reduced porosity compared to Cu 3 (BTC) 2 . When reacted with a methanolic solution of TCNQ, Cu 2 (TATAB) 3 darkened in color, and its conductivity increased from 9.75 × 10 −12 to 2.67 × 10 −7 S/cm, as determined from two-probe pressed pellet measurements. No discussion of the exact mechanism of TCNQ interactions with the framework was provided; however, the denser arrangement of the copper paddlewheels in the material makes it difficult to envision a bridging binding mode for the TCNQ, as was proposed for Cu 3 (BTC) 2 . TCNQ incorporation was also reported to increase the conductance of thin-film MOF devices via charge transfer between TCNQ and the framework for both Co 2 (DOBDC) 257 and a 2D MOF constructed of Cu II acetate paddlewheels and a tetrapyridyl porphyrin linker. 258 The effects of ferrocene doping on the conductivity of thin films of Cu 3 (BTC) 2 have also been investigated. 259 Activated thin films of the MOF were exposed to ferrocene vapors, which led to a slight improvement of the electrical conductivity, from 1.47 × 10 −15 to 3.57 × 10 −15 S/cm. From the thickness dependence of the current density in the films, the authors concluded that the conduction mechanism was not tunneling through the films but rather hopping-type.
Guo et al. demonstrated that soaking films of a pillared paddlewheel MOF with an electron-rich pyridyl-and dipyrrolidyl-substituted naphthalene diimide (NDI) ligand in a solution of methyl viologen dication (MV 2+ ) increased the conductivity from 6 × 10 −7 to 2 × 10 −5 S/cm. 240 Soaking in solutions of the weaker π-acids dinitrotoluene and 1,5-difluoro-2,4-dinitrobenzene also improved the conductivity, though by less than one order of magnitude. In contrast, treatment with a solution of C 60 resulted in unchanged conductivity, which was attributed to size exclusion by the MOF pores. The authors proposed that donor−acceptor interactions between the π-acid guests and the electron-rich ligand introduced charge carriers and potentially also increased the charge mobility via π−π interactions.
Souto et al. encapsulated C 60 in a similarly mesoporous MOF with the formula Fe 3 O(TTFTB) 3 (MUV-2) with a redox-active TTF-based ligand. 237 Similar to the pyrene-based MOFs, a charge transfer band in the UV−vis spectroscopy data indicated donor−acceptor interactions between C 60 and the TTF core. While DFT calculations predicted favorable ligandto-guest charge transfer and a lowered bandgap in the C 60encapsulated material, the conductivity increase was modest (from 3.7 × 10 −11 S/cm for MUV-2 to 4.7 × 10 −9 S/cm for C 60 @MUV-2). The authors attributed the relatively low conductivity value to low C 60 loading (C 60 @MUV-2 remained highly porous, with a BET surface area of 1040 m 2 /g) and long distances separating the ligand TTF moieties. C 60 incorporation into thin films of surface-mounted metal− organic frameworks (SURMOFs) has also been demonstrated to increase their conductivities. Deposition of C 60 @Cu-(BPDC) (BPDC = biphenyl-4,4′-dicarboxylate) and C 60 @ Zn(TPP) (TPP = 5,15-bis(3,4,5-trimethoxyphenyl)-10,20bis(4-carboxyphenyl)porphyrin) SURMOFs was carried out by spin-coating alternating layers of solutions of the metal salt, linker, and C 60 . 260 The conductivities of the C 60 -containing frameworks (∼10 −11 S/cm) were at least 2 orders of magnitude higher than the pristine SURMOFs (<2 × 10 −13 S/cm). For C 60 @Zn(TPP), a C 60 loading of about 0.9 molecules per unit cell was assigned based on UV−vis spectroscopy and mass spectrometry. C 60 @Zn(TPP) also exhibited photoconductivity, reaching a maximum conductivity of 1.3 × 10 −9 S/cm when irradiated with 455 nm light. No surface area measurements were carried out on these materials.

Conductive Polymers and Oxides
Conductive extended solidsboth organic and inorganic have been encapsulated in MOFs. For polymer⊃MOF assemblies, conductivities higher than either of the pristine components have been reported. Enhanced transport properties in polymer⊃MOF materials have been generally attributed to charge transfer interactions between polymer chains surrounded by π-donor ligands, as well as high degrees of order and orientation in the polymers. In situ hole doping of the polymers by I 2 during the polymerization process are also likely to play a role.
Polymer chains have been incorporated into MOFs by loading monomers into the channels and then reacting the material with I 2 to induce polymerization. Wang et al. first demonstrated this approach with polypyrrole (PPy) and Zn 3 (lac) 2 (pybz) 2 , 261 likely inspired by an earlier work, 262 in which well-ordered polypyrrole was grown inside a layered coordination polymer. The resulting PPy⊃Zn 3 (lac) 2 (pybz) 2 material ( Figure 28) reached a conductivity of 1.0 × 10 −2 S/ cm, higher than both I 2 @Zn 3 (lac) 2 (pybz) 2 (3.42 × 10 −3 S/ cm), 244 238 The conductivity of the PEDOT⊃Cr-MIL-101 material with 57% polymer loading by weight was 1.1 × 10 −3 S/cm (compared to <10 −11 S/cm for Cr-MIL-101 alone). This material had a BET surface area of 803 m 2 /g, partially retaining the high porosity of the Cr-MIL-101 host. The mass fraction of PEDOT in Cr-MIL-101 was found to correlate directly with the conductivity and inversely with the surface area. Interestingly, when PEDOT was incorporated into La(BTC), a MOF with 1D channels, the composite showed a much smaller increase in conductivity than PEDOT⊃Cr-MIL-101, reaching only 2.3 × 10 −8 S/cm. The authors hypothesized that the 3D architecture of Cr-MIL-101 may enable higher conductivity by allowing for more PEDOT interchain interactions.
Motivated by the ubiquitous conductive tin oxide glasses, K u n g e t a l . i n v e s t i g a t e d t h e m o d i fi c a t i o n o f Zr 6 (OH) 16 (TBAPy) 2 (NU-1000) with postsynthetically installed tetratin(IV) oxy clusters. 239 The activated MOF was immersed in a solution of a tin precursor, which readily hydrolyzed at the hydroxyl and aqua sites of the Zr cluster. The material was then subjected to a steam treatment, and this whole process was repeated several times. The authors found that the unmodified MOF, as well as the same MOF after one or two cycles of the installation process (in which the concentration of the installed clusters was low enough to give isolated tin(IV) oxy clusters), were both insulating. However, after three cycles, the clusters formed continuous strands of tin(IV) oxide, leading to a significant improvement in conductivity, with values as high as 1.8 × 10 −7 S/cm. Despite the considerable amount of tin(IV) oxide in the material, it was still porous, with BET surface areas of 680 m 2 /g. This study demonstrated that conductive MOFs can be attained through the guest-promoted design strategy using extended inorganic solids, in addition to the more commonly utilized organic and organometallic molecules and polymers.

CONCLUSIONS AND OUTLOOK
Recent years have witnessed a broad diversification of strategies used to imbue MOFs with mobile charge carriers and pathways for charge transport. These strategies have centered upon various intrinsic structural components of MOFs, including the inorganic SBU, extended bonding between the metal and linker, and noncovalent interactions among linkers. Additionally, the porosity of MOFs has been leveraged to introduce extrinsic guests that mediate conductivity. Across these distinct approaches, we summarize unifying characteristics among the materials we have discussed: (1) The ubiquity of mixed-valence motifs. Mixed valency among metal centers or linkers can both introduce charge carriers into MOFs. For example, frameworks containing Fe II/III in chain SBUs are more conductive than those containing only divalent metals. Semiquinonate MOFs (and their imino-and thio-analogs) contain both radical and closed-shell oxidation states of the linkers in the solid state. Conductive MOFs based on tetrathiafulvalene and other conjugated organic molecules also contain a mixture of the radical cation and neutral forms of the organic cores. Some of these materials are grown with a mixture of ligand oxidation states (often with the presence of atmospheric oxygen serving as an in situ oxidant) and contain intrinsic mixed-valency in their as-synthesized forms. In other cases, mixed-valent motifs are postsynthetically introduced (e.g., deliberate oxidation of Fe II centers to Fe III , reduction of Fe III to Fe II , or partial oxidation of tetrathiafulvalene to its radical cation form using I 2 ). Materials in which the concentration of different charge states can be tuned have generally exhibited conductivities correlating with the extent of mixed-valency.
(2) The presence of continuous charge transport pathways.
Regardless of whether conduction proceeds by charge hopping or band conduction mechanisms, close proximity of the sites that charges can occupy correlates with conductivity. Moreover, continuous arrangement of these sites throughout the structure of the material is important to enable long-range charge transport. Continuous 1D, 2D, and 3D covalently bonded motifs have been demonstrated to facilitate conductivity in MOFs. In addition, in frameworks containing extensive π−π stacking, closer contact distances throughout the stack are correlated with higher conductivities. Routes to control π−π stacking motifs, as well as the connectivity and dimensionality of the frameworks, have also been demonstrated to correlate with conductivities. Finally, harnessing the inherent porosity of MOFs toward guest-promoted conductivity opens up the door to near-endless host−guest combinations with different properties. This field has seen immense growth and progress in the past five years, allowing for the relationships among structure, composition, and physical properties discussed in this Review to be established from existing studies. Nevertheless, many worthwhile directions remain to be pursued in terms of both fundamental studies and potential applications of conductive MOFs. We highlight some of the most intriguing questions and pressing challenges: (1) Disentangling the properties of charge carriers in MOFs.
Only a handful of charge mobility values have been reported for MOFs (Table 9), and many of these studies have used noncontact methods, such as FP-TRMC and TRTS. Hall effect measurements and FET devices using high-quality thin films or ideally single crystals of MOFs can determine the nature of the charge carriers (i.e., electrons or holes), as well as the carrier concentration and mobility. In general, deeper investigations into the mechanisms underlying high conductivities will yield more useful insights than a race toward record conductivity values. Additionally, computational studies (e.g., band structure calculations) will be important in corroborating experimental findings in this area. Only isolated examples of conductive MOFs with isotropic structures and crystallographically resolved transport pathways have been reported. 68,71,72,153 The problem of installing charge carriers in close proximity to one another in an isotropic, porous network presents an intriguing intellectual and synthetic challenge. On the practical side, isotropic materials should offer some advantages for integration into devices on account of their orientationindependent properties. (3) Conductivity in mixed-metal and mixed-linker systems.
Apart from mixed-valent motifs, targeting MOFs with more than one metal or ligand can be a strategy to dope these materials and increase their conductivities. Such systems could conceivably be realized through either direct synthesis or postsynthetic cation-or ligandexchange. 265,266 Preliminary examples of mixed-metal approaches 191,220,229,267 portend promising capabilities to further tune the conductivity of MOFs through compositional variation. (4) Further elucidation of transport mechanisms responsible for excellent bulk conductivities in selected 2D MOFs, especially  2 and Cu 3 (HOTP) 2 ) have indicated that the in-plane conductivities are similar in magnitude to the cross-plane conductivities. 79 These findings suggest that π−π stacking interactions between the layers may play an important role in determining the bulk transport properties of these materials. In addition to more characterization of high-quality single crystals, unambiguously determining the crystal structures, stacking motifs, and metal and linker oxidation states in these materials can address this question. Finally, studies that systematically vary the identities of the metal and ligand and examine the resulting conductivities may offer further insights. (5) Isolation and characterization of single layers of 2D MOFs.
Several theoretical works have predicted monolayers of 2D MOFs with extended conjugation to be materials with topologically nontrivial electronic states. 31−36 Although exfoliation down to single layers has been demonstrated for nonporous 2D coordination polymers that are conductive, 268 the transport properties of isolated mono-or few-layer samples of 2D MOFs with extended π−d conjugation have not been reported in detail. The realization of such materials could reveal new platforms for studying transport phenomena at the forefront of experimental physics research. On a related note, further dimensional reduction of the motifs that facilitate charge transport in MOFs to 1D and 0D structures can shed light on specific coupling mechanisms. 269 (6) Expanding the library of ligands for MOFs featuring through-space charge transport pathways. To date, most conductive MOFs with transport pathways comprising π−π stacking are made with TTF-based ligands. Though a large variety of different donor and acceptor molecules are utilized in organic electronics, most of these have not yet been incorporated into MOFs. Clever ligand designs that overcome synthetic hurdles that may be encountered in using such molecules can lead to new MOFs with through-space charge transport pathways and potentially notable photophysical properties. (7) More detailed structural investigations of host−guest interactions in MOFs. In MOFs with guest-promoted conductivities, the locations of most guest species within the pores of the MOF have not been crystallographically resolved. Determining the nature of charge-transfer interactions that increase conductivity in such systems would benefit from crystallographic investigation. Additionally, side reactivity between the guest and framework decomposition products is a potential concern in some of these systems. More in-depth structural studies (e.g., Rietveld refinement for crystalline powders, pair distribution function analysis for less crystalline materials) can also allow for more definitive assignment of enhanced transport properties to the host−guest composite versus conductive side products. (8) Inf luence of environment and atmosphere on MOF conductivity values. The successful application of conductive MOFs to chemiresistive sensing is a consequence of the sensitivity of the electronic properties of these materials to the atmosphere in which they are measured. Both reversible and irreversible changes in conductivity of MOFs may be observed upon exposure to different atmosphere (e.g., oxygen, moisture, volatile organic compounds, corrosive vapors). While such responses have been productively harnessed toward tunable conductivities and sensing applications, unmonitored atmospheric effects may also influence reported conductivity values. Hence, controlling and purposefully studying the impact of environmental effects can lead to more reproducible values in the literature and new materials for chemiresistive sensing.
(9) Best practices for materials preparation and measurement techniques. As a corollary to the previous point, depending on the material and use case, the standard (or ideal) conditions under which to measure the transport characteristics of a MOF may vary. In addition, the same material (nominally, in terms of structure and composition) prepared under different conditions or measured in different form factors may exhibit conductivity values varying over several orders of magnitude. To maintain clarity in this growing field and ensure reproducibility, we believe it is increasingly important for researchers to document their sample and device preparation procedures, as well as the details of the measurement conditions (e.g., temperature, atmosphere, illumination, nature of the electrical contacts, as well as details on the measurement apparatus). 56 In addition, reporting the temperature dependence of the conductivity (from which an activation energy and transport mechanism can be extracted) is more instructive than the room temperature value alone. We note in closing that conductivity values relevant to realworld applications have already been attained in MOFs. Inventive synthetic approaches, optimized growth techniques, and improved device fabrication can (and should) all continue to raise the ceiling of the transport properties attainable in porous, crystalline framework materials. However, equally important are the continuing refinement of design principles and more detailed mechanistic investigations into charge transport in MOFs. Future work along these lines will undoubtedly reveal novel fundamental properties and opportunities for application for this versatile class of materials.

Notes
The authors declare no competing financial interest. Mircea Dincăis a Professor in the Department of Chemistry at MIT. His group's research is broadly focused on understanding and changing materials' physical and chemical properties through means that are not typically applicable to solids. He also gets excited about making new materials that combine canonically antithetical properties, such as porosity and electrical transport, to yield something that is greater than the sum of its parts. Educated at Princeton University (A.B. in Chemistry, 2003) and UC Berkeley (PhD in Inorganic Chemistry, 2008), he aspires to one day create and control (chemical) disorder.