Design Strategies for Enhanced Conductivity in Metal-Organic Frameworks.

Metal-organic frameworks (MOFs) are a class of materials which exhibit permanent porosity, high surface area, and crystallinity. As a highly tunable middle ground between heterogeneous and homogeneous species, MOFs have the potential to suit a wide variety of applications, many of which require conductive materials. The continued development of conductive MOFs has provided an ever-growing library of materials with both intrinsic and guest-promoted conductivity, and factors which limit or enhance conductivity in MOFs have become more apparent. In this Outlook, the factors which are believed to influence the future of MOF conductivity most heavily are highlighted along with proposed methods of further developing these fields. Fundamental studies derived from these methods may provide pathways to raise conductivity across a wide range of MOF structures.

Since the first publications roughly 20 years ago, metal–organic frameworks (MOFs) have shown use in a wide range of applications such as gas storage and separation,[1−4] drug delivery,[5−8] and catalysis.[9−11] MOFs are a class of porous coordination polymers composed of multidentate organic linkers connected to metal nodes. The combination of different linkers and metals can lead to diverse structural architectures, allowing for control over dimensionality, pore shape and size, and surface area to fit the needs of the application. The high tunability, favorable properties, and typically cheap starting materials make MOFs potentially viable candidates to replace more expensive materials in many conventional applications.

A growing field of exploration is the employment of MOFs for applications involving electrical conductivity: electrocatalysis,[12,13] (photo)electrochemical energy storage,[14] batteries,[15,16] and chemiresistive sensing.[17] Due to the fact that most MOFs are insulators with conductivity values of 10–10 S/cm or lower, research has focused on ways to enhance electrical conductivity through various approaches.[18−22] It should be noted that the minimum requirements for many applications do not differ greatly from what MOFs can already achieve as device design can be made to compensate for less conductive MOFs, though some applications such as batteries and superconductors require conductivities greater than 0.1 S/cm.[23]

Several excellent reviews, perspectives, and accounts have been reported on conductive MOFs and their various applications.[18−28] In this Outlook, we focus on structure–property analysis and provide insights on outstanding fundamental questions that will help large-scale exploitations of these materials. To date, 2D and 3D MOFs have both been widely explored,[23] with hexagonal 2D MOFs based on catechol and semiquinone-type linkers showing some of the highest conductivities. Both cubic and hexagonal 3D MOFs have been reported, though hexagonal systems are much more commonly seen in the literature. With numerous structures, shapes, and sizes of MOFs in existence, there are also multiple methods of transporting charge within MOFs, namely, through-bond, through-plane, through-space, redox-hopping, and guest-promoted pathways.

Through-Bond Pathway

MOFs which grow with infinitely repeating metal-chelating atom pathways (−(M–E−)∞) may exhibit what is called through-bond conductivity. In this pathway, charge moves through the metals and bridging atoms but specifically not through the ligand backbone. The ability of the ligand to delocalize charge is irrelevant in this case, and only the ligand’s effect on the orientation of the node and binding atoms is considered. While the most prevalent examples move charge in one dimension, three-dimensional through-bond charge transport has been reported, as well.[29−31] As charge must move through both metal and coordinating atom orbitals, the use of softer binding atoms which are energetically similar to the metal node are more likely to promote this type of conductivity. MOFs composed of 2,5-dihydroxybenzene-1,4-dicarboxylic acid (DOBDC), 2,5-disulfhydrylbenzene-1,4-dicarboxylic acid (DSBDC), bis(1H-1,2,3-triazolo[4,5-b],[4′,5′-i]dibenzo[1,4] dioxin) (BTDD), or (1,2,3-triazolate)2 have been reported to show this type of behavior.[29−33]

Through-Plane (Extended Conjugation) Pathway

Functioning similarly to the through-bond pathway, through-plane pathways also include the ligand backbone (metal–chelate–linker backbone delocalization). Commonly proposed for 2D graphene analogues, this pathway makes use of π–d conjugation in the ab plane. Delocalization of charge through the ligand is necessary to be considered through-plane instead of through-bond, and therefore, ligands should promote long-range conjugation. For charge transport to occur in this manner, it is important not only that the energy levels of the metal, binding atom, and ligand backbone are similar but also that the binding atom should be in conjugation with the ligand backbone to maintain the π system. The most prevalent examples of this type of charge transport are benzene- and triphenylene-based MOFs.[34−42] Conjugation will be confined to two of the three available dimensions, but charge can still move in the c direction via the through-space pathway.

Through-Space Pathway

The 2D systems that experience through-plane conductivity can also experience through-space conductivity due to interlayer π–π stacking in the c direction. Conductivity is directly related to how close the sheets are to each other, as creating too much space between sheets will decrease the wave function overlap and charge transport efficiency. Because both through-plane and through-space pathways are dependent on extensive π conjugation in 2D systems, it can be difficult to separate the two. Often, both modes of charge transfer are seen with varied levels of contribution. The total contribution of through-space charge transport can be difficult to determine, but recent reports have suggested that this method can be the dominant transport pathway in some MOFs according to DFT calculations.[40]

Redox Hopping

For MOFs that do not exhibit good energy overlap between metal nodes and organic linkers, bonding is more ionic in nature which inhibits the previously mentioned methods of charge transport. Instead of charge being delocalized throughout the MOF, electrons can instead hop through local minima states in a manner similar to the through-space pathway. Our group was the first to present redox hopping as a possible charge transport mechanism,[24,43−45] and the number of MOFs reporting redox hopping as a charge transport mechanism continues to grow.

Guest-Promoted

The guest-promoted pathway is unique in that it can theoretically be applied to every MOF. The intrinsic porosity of MOFs is utilized to postsynthetically introduce guest species, which generate electrically conductive pathways. Published guest species show a wide range of sizes, with iodine/polyiodides, organic/organometallic species, and conductive polymers all elevating MOF conductivity.[23,24,46,47] While the versatility of this method is great, the incorporation of guest species will obviously result in a trade-off where porosity and available surface area are decreased. There are also possible issues with MOFs that have pores too small to effectively incorporate guest molecules.

With a vast number of MOF structures and charge transport methods available, there is no true best way to enhance MOF conductivity. However, certain considerations such as the chemical properties of the chelating moiety, the differences between 2D and 3D electron conduction, the role of counterion transport, and the role of grain boundaries are widely applicable starting points for fundamental studies.

Chemical Properties of the Chelating Moiety

An underexplored area in the development of conductive MOFs is the influence of the coordinating atom on charge transport. While nodes have been constructed with various different metals, the coordinating atoms have largely been limited to O, N, and S.[31−33] The relationship between the metal node and chelating atoms is important to consider as hard–soft acid–base theory suggests that hard–hard interactions will lead to ionic binding (redox hopping) while soft–soft will be more likely to form covalent interactions (through-bond, through-plane, through-space). Harder atoms such as O coordinating to comparatively soft metal ions/nodes produce energy gaps and trapped valence states that hinder conductivity.[23] Incorporating comparatively softer atoms (N, S) leads to more covalent binding, which can help optimize conductive pathways (π–d conjugation). The optimization of hardness and softness of metals and chelates will be a necessary process for enhancement of conductivity in MOFs.

MOFs based on semiquinoid-type linkers are some of the most widely studied with regards to chelate identity, with O, N, and S linkages being explored for benzene and triphenylene linkers (Figure ). Numerous theoretical approaches have been taken to understand the relationship between the metal node and chelating atom.[48−53] Cu3(HOTP)2 (HOTP = 2,3,6,7,10,11-hexahydroxytriphenylene), Ni9(HOTP)4, and Co9(HOTP)4 were reported to have maximum conductivity values of 0.2, 0.1, and 3 × 10–3 S/cm, respectively.[34−37] Changing from oxygen-based linkages to softer nitrogen-based species results in enhanced conductivity,[38,41,42] with Ni3(HITP)2 (HITP = 2,3,6,7,10,11-hexaiminotriphenylene) producing conductivity values of 0.2 and 40 S/cm for bulk pellet and film measurements, respectively. Thiol-based ligands have been a focus in recent years, though the number of published crystalline species is limited. Co3(THT)2 (THT = 2,3,6,7,10,11-triphenylenehexathiolate) (1.4 × 10–3 S/cm) and Fe3(THT)2 (3.4 × 10–2 S/cm) were reported to undergo temperature-dependent conductivity changes, with semiconducting behavior shifting to metallic behavior as the temperature is lowered.[39,40] Through the alteration of binding atoms alone, characteristics of structurally similar MOFs can be drastically changed.

Figure 1

Semiquinoid-type triphenylene and benzene-based 2D MOF structures.

MOF-74-type structures, with the formula M2(DOBDC) (DOBDC = 2,5-dioxidobenzene-1,4-dicarboxylate), have been synthesized with Mg, Mn, Fe, Co, Ni, Cu, and Zn metal nodes.[23,31] These MOFs exhibit through-bond conductivity, with charge moving through infinite M–O bonds. The conductivity of Fe2(DOBDC) was the highest at 3.2 × 10–7 S/cm while the rest exhibited poor conductivity (Mn2(DOBDC) = 3.9 × 10–13 S/cm).[32,33] Substituting thiols for hydroxyls via 2,5-disulfidobenzene-1,4-dicarboxylate (DSBDC) increases the observed conductivity by roughly an order of magnitude (Fe2(DSBDC) = 3.9 × 10–6 S/cm, Mn2(DSBDC) = 2.5 × 10–12 S/cm).[32,33] The large difference in conductivity between Fe and Mn DOBDC/DSBDC MOFs in these cases is attributed to Fe2+ having a loosely bound β-spin d band electron while the Mn2+ d band is empty, but the increase in both by an order of magnitude demonstrates the influence that metal–chelate interactions have on conductivity.

As the number of reported N- and S-based ligand systems grows, the need for greater understanding of metal–chelate interactions is becoming more apparent. Softer ligands have shown to have a positive effect on through-bond and extended conjugation-type conductivities, and the added d-orbital availability presents the opportunity for unique coordination environments. Though benzene and triphenylene systems have been widely explored and thus serve as valuable examples, these same approaches can be applied to numerous different linker architectures. For example, porphyrin- and phthalocyanine-based MOFs have been reported with O and N coordinating atoms.[54−58] The modification of chelating atoms is independent from ligand backbone complexity or pore size/shape, meaning that softer atoms can be used in multiple MOF architectures without affecting other innate structural properties which may be desirable for particular applications. In addition to O-, N-, and S-based MOF linkers, Se- and P-based linkages could serve as the future for highly conductive materials. MOFs made with Se-based linkers are already being pursued,[59] but synthetic limitations present the greatest challenge. Additionally, the strong covalent coordination can lead to mostly amorphous material with current synthetic approaches, suggesting the need for synthetic techniques to grow along with MOF ligands. However, the development of a large library of linkers with different chelating moieties is only a matter of time and necessary for the continued exploration of conductive MOFs.

2-Dimensional vs 3-Dimensional MOF Architectures

Different dimensionality can lead to different dominant modes of charge transport. The distinction between 2D and 3D MOF architectures is small as most 2D MOFs exhibit some form of interlayer stacking, often due to π–π interactions. The main mechanisms of charge transport differ from 2D to 3D, as 2D MOFs mostly display through-plane and through-space while 3D MOFs predominantly display through-bond or redox hopping. 3D MOFs can conventionally be thought of as MOFs which display connectivity in the a, b, and c directions, while 2D MOFs are only bonded in the a and b directions. While 2D MOFs display some of the highest observed conductivities, the higher surface areas of 3D MOFs are more ideal for practical applications. However, the relationship between dimensionality and charge transport in 3D MOFs is still not fully understood as, to the best of our knowledge, symmetrical conductivity in 3D has not been observed.

Ligands which can form 2D and 3D architectures may offer insight into the struggles of achieving symmetric conductivity. Dihydroxybenzoquinone (dhbq) has been used in the synthesis of 2D and 3D MOFs (Figure ),[60−66] with different charge transport mechanisms being seen for each case. The 2D MOFs display covalent bonding in the ab plane, which would suggest extended conjugation-type conductivity within the plane. However, charge transport through the c axis is expected based on previous reports, and in some cases this method of transport is significant or even dominant. By modifying the synthetic conditions, dhbq can form 3D MOFs with interpenetrated layers.[61] The interpenetrated growth and charge-balancing ions effectively eliminate porosity but allow for redox hopping through the layers via dhbq radicals. The 3D iron MOF, [Fe2(dhbq)3]2–, displays one of the highest observed conductivities at 0.16(1) S/cm, which is an order of magnitude higher than the 2D versions.[61] Similar behavior has also been seen in recent works by Bao et al.[67−69] A 3D cubic MOF was synthesized using tetrahydroxy-1,4-quinone (THQ), and FeTHQ was reported to have a conductivity of about 0.003 S/cm.[67] As with [Fe2(dhbq)3]2–, the conductivity of FeTHQ in a 3D form is higher than 2D counterparts by orders of magnitude.[68,69] While the increased conductivity when forming a 3D MOF is notable, the change in electron transport mechanism highlights the difficulty of producing 3D MOF architectures with extended conjugation. By nature, extended conjugation in all three dimensions is improbable if not impossible, leaving through-bond and redox-hopping mechanisms as the most likely scenarios of three-dimensional charge transport. Given that most intrinsically conductive 3D MOFs are hexagonal structures,[23] there will always be a shortest path for charge transport to undergo, and thus a preferred direction. Future designs must then focus on MOFs which are optimized to promote equivalent conductivity in all three dimensions.

Figure 2

2D structure of [Fe2(Cl2 dhbq)3]2– along the c (A) and a (B) axes and 3D structure of single layer (C) and interpenetrated (D) [Fe2(dhbq)3]2– with the interpenetrated layers shown in red and blue. The cubic space group of the 3D MOF means that the a, b, and c axes are identical. Reprinted with permission from refs (61 and 23). Copyright 2015 American Chemical Society. Copyright 2020 American Chemical Society.

The design of idealized 3D MOFs is a difficult task. Cubic MOFs may be the optimal architecture, but only a handful of intrinsically conductive cubic MOFs have ever been reported.[30,61,67,70] One such cubic MOF, composed of rare-earth metals (Y, Eu, La) and HOTP, possesses symmetry in the a, b, and c directions and employs a ligand often associated with high conductivity due to its ability to access mixed valence states.[70] Though the conductivity values observed for the 3D MOFs are orders of magnitude lower than 2D structures, forming new structures from existing ligands is a viable option to attempt generating symmetric conductivity in 3D MOFs. Alternatively, methods of converting 2D structures into 3D MOFs through bridging species have long been hypothesized, but these methods do not solve the issue of dimensionally limited conductivity. These approaches also hinder desirable properties in MOFs as (1) the total porosity/surface area will be decreased; (2) rigid binding will force sheets to be a set distance apart which can inhibit c direction charge transport if the bridging species is too large/larger than the ideal distance without bridging; and (3) the added coordination to metal nodes can hinder π–d conjugation and through-plane conductivity. To produce MOFs which maintain high surface area and display high, symmetric conductivity, careful consideration must be paid to MOF symmetry and the methods by which charge moves through the structure.

Counterion Transport

Early studies on charge hopping in redox-active polymer films hint that the electron motion in charge hopping processes is closely followed by counterion transport, which maintains the electroneutrality.[71] Consequently, the overall conductivity of polymeric systems is significantly affected by ion mobility. The phenomenon becomes even more prominent in nanostructured media due to several reasons. First, ion volumes are now comparable to the size of cavities, which prohibits free motion of ions through pores of materials. Second, dielectric properties of the encapsulated solvent are considerably reduced when compared to the bulk.[72] The reduced dielectric properties can contribute to the increased ion pairing strength between electroactive species and counterions, which, in turn, further incapacitates conductivity. Finally, the presence of solvent structuring in nanoconfined spaces[73] likely restricts flexibility of the medium and impacts the overall ion mobility. Indeed, diffusion-limited electron transport is commonly observed in nanoconfined redox-active systems.[74−78] Most notably, ion conductivity is one of the critical parameters for the performance of rechargeable batteries because it directly affects charging and discharging through the processes of intercalation and deintercalation.[79] Given the importance of ion transport in these processes, current efforts are focused on improving the mobility by careful electrode structuring or by incorporating secondary ion-transport channels.[80,81]

Despite numerous reports on redox hopping in MOFs, there is very little knowledge of ion transport in these materials. Our group has pioneered establishing a relationship between structural properties and charge carrier mobilities.[74,75] In the initial study, the individual diffusion coefficients for electrons (De) and ions (Di) for metallocene-decorated M-NU-1000 MOFs (M = Fe, Ru, and Os) were quantified using a theoretical model for the porous materials developed by Scholz.[74] Independent evaluation of both parameters revealed a few findings: (1) De follows the Fe-NU-1000 < Ru-NU-1000 < Os-NU-1000 trend, further validating that the electron hopping rate depends on the self-exchange rate between incorporated redox-active units (Section c). (2) Di is generally higher for smaller counterions. (3) Ion transport is the limiting process, since the ion diffusion values (10–12–10–14 cm2/s) are several magnitudes lower than electron diffusion (10–9–10–10 cm2/s).

In a separate study, we investigated the effect of architectural features of ferrocene-decorated MOFs on separate electron and ion hopping rates (Figure ).[75] For this study, MOFs were selected according to their pore sizes: MOF-808 (15 Å), NU-1000 (33 Å), and NU-1003 (43 Å). The increase in the MOF pore size resulted in a sizable improvement of ion diffusion rate (ki), whereas electron diffusion rate (ke) decreased by a factor of 3. Nonetheless, the charge hopping rate (khop) was increased by a factor of 10 with larger pore sizes despite the slower electron diffusion. It is important to note that, even in the NU-1003, the MOF with the largest pore size in this study, the electron diffusion rates were still significantly higher than ion diffusion.

Figure 3

Effect of pore size on conductivity in MOFs. Reprinted with permission from ref (75). Copyright 2020 American Chemical Society.

Results of our preliminary studies corroborated that MOFs display the same type of ion diffusion-limited electron transport observed in other nanoconfined systems. The diffusion of ions has proven to be one of the main reasons for poor conductivity in zirconium-based MOFs.[74,75] Increasing the MOF pore size certainly improves ion mobility and overall apparent conductivity, but at the same time, it negatively impacts electron diffusion. Future studies should focus on the careful design of MOF cavities that will facilitate ion transport while still maintaining high electron hopping rates, which could be accomplished with the conscientious engineering of pore and counterion shape and sizes, or by introducing functional groups within pores that would facilitate ion transport. Furthermore, gaining knowledge on solvent properties within the MOF pores may aid in achieving improved conductivities. While the results of future studies will indubitably benefit redox-hopping MOF systems, we anticipate that they can also be applied to other nanoconfined conductive materials.

Grain Boundaries and In-Crystal Defects

So far, we have discussed properties that are intrinsic to the MOF systems and electron transport processes. Apart from these properties, there is a universal characteristic for all materials developed for large-scale applications—grain boundaries and defects. Grain boundaries are commonly observed in crystalline materials and represent two-dimensional interfaces that have an atomic arrangement different from the single-crystal lattice. These boundaries are structural defects, and as such are responsible for disruption in the continuous thermal and electrical conductivity in materials. Given the importance of this in the resultant performance, grain boundaries have been extensively studied both experimentally and computationally.[82−87] It is found that grain boundaries negatively impact the conducting properties of materials, but the extent of the effect depends greatly on the given system. For example, electric conductivity in perovskite films and graphene showed a modest decrease in less than an order of magnitude across the grain boundaries (by a factor of 2 for perovskites, whereas graphene conductivity decreased roughly 3 times).[83,85] Contrary, organic semiconductors display a more pronounced (several orders of magnitude) resistance at the grain boundaries.[86,87]

While the MOF field is lacking a proper evaluation of the grain boundary effect on conductivity, there has been an initiative to obtain high-quality single crystals and thin films. Single crystals represent an ideal platform to study intrinsic conductive properties of MOFs due to the absence of structural grain boundaries. In a recent study, Dincă and co-workers demonstrated that Ni3(HITP)2 and Cu3(HHTP)2 (HHTP = 2,3,6,7,10,11-hexahydroxytriphenylene) single crystals exhibit significantly improved electric conductivities of several orders of magnitude when compared to polycrystalline films.[34,88] In addition, Ni3(HITP)2 acted as a metallic material in single crystal form, whereas the polycrystalline films demonstrated more semiconductive properties. Computational studies on Ni3(HITP)2 showed that the presence of internal interface defects introduces transport barriers via disrupted conjugation and/or decreased dispersion of electronic bands near the Fermi level, which can help explain the discrepancy between predicted metallic behavior and experimentally observed semiconductor behavior for the bulk material.[89] These studies demonstrate how the field of conductive MOFs will benefit from fabricating single crystals. Nonetheless, despite being great for studying fundamental properties, single crystals are not preferred for utilization in electronic devices given that their sizes (dimensions in nanometers and micrometers) frequently do not meet requirements for large-scale applications.

Thin films, on the other hand, are more readily applicable for large-scale applications, yet for MOFs, they are often obtained in the polycrystalline form with a large number of grain boundaries (Figure A).[90] One of the possible ways to circumvent this problem is to grow monocrystalline (single-crystal) films or highly oriented films (Figure A), where electron transport occurs in the direction of the film growth. Recent advances in thin film fabrication, particularly in layer-by-layer liquid phase epitaxy,[91,92] vapor-assisted conversion,[36,93] and chemical vapor deposition,[94] enabled production of high-quality films that are suitable for charge transport studies. Indeed, preferential orientations in Ni- and Co-CAT-1 thin films yielded electrical conductivities in the 10–3 S/cm range, comparable to the single-crystal values.[36] Similarly, a 30-fold improvement in conductivity was observed for single-crystal thin-film prepared by chemical vapor deposition.[94] Moreover, Hupp and co-workers observed a sizable increase in the redox-hopping efficiency (over 3000 times) in NU-1000 films for perpendicular transport (along the c axis in Figure B) over the parallel route (transport though ab plane), further highlighting the importance of oriented film growth on charge transport.[95] Nevertheless, obtaining high-quality single-crystal thin films remains a challenge, despite the recent advances in the field.

Figure 4

(A) Schematic representation of MOF thin films. Reprinted with permission from ref (90). Licensed under a Creative Commons Attribution (CC BY) license. (B) Charge transport in NU-1000 films through the ab plane and along the c axis. Reprinted with permission from ref (95). Copyright 2020 American Chemical Society.

In addition to grain boundaries, MOF materials suffer from in-MOF defects that can impact the overall conductivity. In-crystal defects are particularly common when a modulator is employed in order to grow large crystals.[96] The modulator competitively binds to a metal center and, depending on the strength of the metal–modulator bond, may remain in the crystal leading to missing linker or missing cluster defects. Since both types of defects can serve as a trap state and an additional barrier in charge transport, future studies should focus on the correlation between defect levels and conductivity.

The production of large enough single crystals is highly desirable for wide-ranging device applications. In reality, however, materials that are currently being produced often contain structural imperfections, such as grain boundaries and in-crystal defects. Events that take place at these sites have a negative impact in the macro-observed material charge transport properties. Therefore, it is important to be able to engineer these structural flaws, so we can achieve desired conductivities.

Conclusion

The growing interest in conductive MOFs and their high tunability has revealed a wealth of information with regards to how these frameworks undergo charge transport. Fundamental studies on these materials yielded several interesting findings: (a) The highest conductivity values are obtained for electronically compatible mixed-valence metal nodes and linkers. (b) Redox-hopping in MOFs occurs via diffusion-limited electron transport. (c) Single crystals display a significant improvement in conductivities over polycrystalline materials. Despite the great effort and appreciable advances in recent years, conductive MOFs are yet to be implemented in large-scale applications and devices. To achieve this, some important fundamental questions need to be answered. Is it possible to accomplish conductivities in 3D structures that are comparable to 2D materials? Can the redox-hopping electron transport mechanism reach the transfer rates of other bandlike systems? What is the role of the solvent in diffusion-limited charge transport? How do the grain boundaries and in-crystal defects impact materials’ charge transport properties? Not only will the answers to these fundamental questions bring us closer to the anticipated wide-ranging applications, but they will also benefit other nanoconfined conductive materials.