Report

Tunable Electrical Conductivity in Metal-Organic Framework Thin-Film Devices

See allHide authors and affiliations

Science  03 Jan 2014:
Vol. 343, Issue 6166, pp. 66-69
DOI: 10.1126/science.1246738

Guests for Conductors

Thin films of metal-organic framework (MOF) compounds are generally poor conductors because the linking organic groups are usually insulators with little π-orbital conjugation. Talin et al. (p. 66, published online 5 December) show that infiltrating films of the copper-based MOF HKUST-1 with the conjugated organic molecule 7,7,8,8-tetracyanoquinododimethane created an air-stable material with conductivities as high as 7 siemens per meter.

Abstract

We report a strategy for realizing tunable electrical conductivity in metal-organic frameworks (MOFs) in which the nanopores are infiltrated with redox-active, conjugated guest molecules. This approach is demonstrated using thin-film devices of the MOF Cu3(BTC)2 (also known as HKUST-1; BTC, benzene-1,3,5-tricarboxylic acid) infiltrated with the molecule 7,7,8,8-tetracyanoquinododimethane (TCNQ). Tunable, air-stable electrical conductivity over six orders of magnitude is achieved, with values as high as 7 siemens per meter. Spectroscopic data and first-principles modeling suggest that the conductivity arises from TCNQ guest molecules bridging the binuclear copper paddlewheels in the framework, leading to strong electronic coupling between the dimeric Cu subunits. These ohmically conducting porous MOFs could have applications in conformal electronic devices, reconfigurable electronics, and sensors.

Metal-organic frameworks [MOFs (13)] are crystalline materials with a nanoporous, supramolecular structure consisting of metal ions connected by multitopic organic ligands. These materials are typically poor electrical conductors because of the insulating character of the organic ligands and poor overlap between their π orbitals and the d orbitals of the metal ions. Combining the crystalline order of MOFs with an ability to conduct electrical charge has the potential to create a new class of materials that would enable applications such as conformal electronic devices, reconfigurable electronics, and sensors. Although strategies to engineer electrically conducting MOFs have been proposed (e.g., the use of second- or third-row transition metals, redox-active linkers, and hetero-bimetallic structures), few of these approaches have been realized (4). Until recently, only one example of an intrinsically conducting framework with permanent porosity was known: a p-type semiconducting MOF in which conductivity occurs via a redox mechanism (5). Very recently, Gándara et al. described a series of metal triazolate MOFs, one of which exhibits ohmic conductivity (6). The mechanism of conductivity in that case is not known, but it appears to be highly specific to the presence of Fe(II) in the structure, as MOFs in this series with the same structure but different divalent metals are not conducting.

An alternative approach is to use the MOF pores themselves as a venue for modulating the electrical transport properties. We reasoned that infiltrating MOFs having open metal sites with molecules capable of charge transfer that can coordinate to these sites would create a mechanism for carrier mobility. Binuclear Cu(II) paddlewheel MOF structures such as Cu3(BTC)2 (7) (also known as HKUST-1; BTC, benzene-1,3,5-tricarboxylic acid) are attractive candidates for testing this strategy. The coordination positions located on opposite ends of the Cu(II) axis of the paddlewheel—which are occupied by solvent (often water) in the assynthesized material—can be exchanged with redox-active molecules such as TCNQ. Moreover, electron paramagnetic resonance (EPR) spectra suggest a somewhat delocalized electronic structure enabled by additional spin exchange among the copper dimers, which results from the carboxylates interconnecting the paddlewheel subunits (8).

Here, we describe the realization of this strategy using a thin-film device comprising Cu3(BTC)2 (7) grown on electrodes and demonstrate control of ohmic electrical conductivity over six orders of magnitude. Silicon wafers covered with 100 nm of SiO2 were prepatterned with Pt pads 100 nm thick (dimensions 800 μm by 400 μm) and gaps of 100 μm, 150 μm, and 200 μm. Cu3(BTC)2 films with nominal thickness of 100 nm were grown on the wafers from the liquid phase, as described (9). Grazing incidence x-ray diffraction (XRD) measurements and scanning electron microscopy (SEM) imaging (Fig. 1, B and C) indicate a polycrystalline Cu3(BTC)2·xH2O film with preferred orientation along the (111) direction (10). Current-voltage (I-V) characteristics obtained on as-grown thin-film devices in air (Fig. 2A) exhibited very low conductivity (~10−6 S/m), consistent with the expected insulating nature of Cu3(BTC)2.

Fig. 1 Fabrication of conductive MOF thin-film devices and structural characterization.

(A) TCNQ molecule shown above a Cu3(BTC)2 MOF; arrow points into the pore. White, hydrogen; blue, nitrogen; cyan, carbon; red, oxygen; light brown, copper. (B) SEM image of MOF-coated device; insets are optical images of devices before and after TCNQ infiltration. (C) XRD data for powders and grazing incidence XRD for a thin film.

Fig. 2 Electronic transport characteristics of MOF thin-film devices.

(A) I-V curves before (red) and after infiltration with TCNQ (green), F4-TCNQ (gold), or H4-TCNQ (purple). (B) Channel-length dependence of conductivity for TCNQ-infiltrated devices; error bars denote SD. (C) Stability of conductivity over time for several devices. (D) I-V curve temperature dependence. (E) Arrhenius plot of the conductivity. (F) Conductivity versus exposure time for several devices. The black line is a fit to percolation theory, σ = 4 × 10–6(t – 0.5)2, where t is exposure time.

The as-grown films were infiltrated with TCNQ by first heating in vacuum at 190°C for 30 min to remove coordinated water molecules, then immediately transferring them to a saturated TCNQ/CH2Cl2 solution. I-V curves for four such devices after 72 hours of exposure to the TCNQ solution are shown in Fig. 2A (11). The infiltration led to marked increases in current, with a linear I-V curve and conductivity of 7 S/m, six orders of magnitude greater than that of the uninfiltrated device. Measurements as a function of channel length (Fig. 2B) revealed a monotonic increase in resistance with increasing electrode separation. The TCNQ-infiltrated devices were stable in ambient atmosphere up to at least 40 days (Fig. 2C). The conductivity increased with temperature (Fig. 2, D and E), following a thermally activated relation σ ~ exp(Ea/T) with a low activation energy Ea = 41 ± 1 meV, similar to values reported for high-mobility organic polymeric semiconductors such as poly-3-hexylthiophene (P3HT) (12).

In contrast to nonporous conducting coordination polymers such as CuTCNQ (1215), we could tune the device conductivity by changing the TCNQ exposure time. As seen in Fig. 2F, conductivity variation over several orders of magnitude was achievable. Furthermore, the increase in conductivity could be described by classic percolation theory (black solid line) (16), which suggests that TCNQ forms localized conducting regions rather than acting as a dopant.

Chemical and physical characterization of MOF films and powders exposed to TCNQ confirmed that TCNQ resides in the MOF pores without altering the MOF crystal structure. Powder XRD patterns of as-synthesized Cu3(BTC)2·xH2O, Cu3(BTC)2 (activated), and Cu3(BTC)2 infiltrated with TCNQ [hereafter TCNQ@Cu3(BTC)2] showed that the MOF crystalline structure (face-centered cubic) was unaffected by the infiltration process (Fig. 1C and fig. S2) (7). Rietveld refinement yielded lattice parameters of 2.617 ± 0.001 nm and 2.635 ± 0.001 nm for Cu3(BTC)2 and TCNQ@Cu3(BTC)2 powders, respectively. The peak at 2θ = 5.759° seen in the patterns of both Cu3(BTC)2·xH2O and TCNQ@Cu3(BTC)2 (but absent from that of activated MOF) is diagnostic for guest molecule binding to the open metal sites in the large pores, and indicates long-range order (17). The surface area of the activated Cu3(BTC)2 powder, obtained from N2 adsorption isotherms using the Brunauer-Emmett-Teller (BET) method, is 1844 ± 4 m2 g–1. This value is typical of high-quality Cu3(BTC)2 FmEmbedded Imagem material with little or no pore collapse or residual reactant (18). After infiltration and drying in air, the TCNQ@Cu3(BTC)2 material displayed a BET surface area of 214 ± 0.5 m2 g–1, suggesting high TCNQ loading. This result was confirmed by elemental analysis indicating a Cu3(BTC)2/TCNQ ratio of 2 on the basis of carbon, nitrogen, and hydrogen content. This corresponds to about eight TCNQ molecules per unit cell, or one TCNQ molecule per MOF pore. The presence of nitrogen in the TCNQ@Cu3(BTC)2 films was further corroborated by x-ray photoelectron spectroscopy (XPS) (fig. S3). Furthermore, visual examination of the powdered MOFs revealed the expected turquoise-blue color for the as-synthesized material and the violet-blue hue for the activated (dehydrated) MOF (fig. S1). Upon exposure to TCNQ, the color of the crystals changed to teal, indicating a perturbation of the MOF electronic structure. The color of TCNQ@Cu3(BTC)2 did not change upon exposure to air (fig. S1), which suggests that TCNQ is not displaced by atmospheric water vapor; XPS also showed a substantially lower oxygen concentration in the infiltrated specimen (fig. S3) consistent with reduced water occupation of the pores. In contrast, the color of the activated MOF before TCNQ infiltration reverted almost instantly to that of the as-synthesized (hydrated) material when exposed to atmospheric moisture.

The realization of these hybrid electronic materials raised questions concerning the nature of the TCNQ-MOF interaction and the mechanism of charge transport. We probed the TCNQ-MOF interaction in several ways. Ultraviolet-visible (UV-vis) spectra were collected from films of the uninfiltrated Cu3(BTC)2·xH2O, TCNQ@Cu3(BTC)2, and TCNQ in dilute solution. The absorption spectrum of the TCNQ@Cu3(BTC)2 film (Fig. 3A) exhibited the expected MOF peak at 340 nm, a peak at 410 nm associated with neutral TCNQ (18), and broad new absorption bands centered at ~700 nm and ~850 nm that were absent in both Cu3(BTC)2·xH2O and TCNQ in CH2Cl2. These additional bands are well-known signatures for charge transfer (13, 19). Note that reacting either copper acetate or copper sulfate with TCNQ in methanol generated no precipitates or new absorption bands, indicating that confinement in the MOF pore is essential to the formation of a charge transfer complex between Cu(II) and TCNQ.

Fig. 3 Evidence for interaction between TCNQ and the MOF.

(A) Transmission UV-vis spectra collected for a Cu3(BTC)2·xH2O film on borosilicate substrate before (red) and after adsorption with TCNQ (green) or H4-TCNQ (purple), and for TCNQ in methanol (blue). (B) Raman spectra collected for a Cu3(BTC)2·xH2O film on borosilicate substrate before (red) and after adsorption of TCNQ (green), and for TCNQ crystals deposited onto a glass slide (blue). (C) Infrared spectra collected for Cu3(BTC)2·xH2O (red), Cu3(BTC)2 (yellow), TCNQ@Cu3(BTC)2 (green), and TCNQ powder (blue). (D) Room-temperature continuous-wave EPR spectra of activated Cu3(BTC)2 (yellow), Cu3(BTC)2 stirred in methanol (red dashed), and Cu3(BTC)2 stirred in methanol containing TCNQ (green). The asterisk denotes an unidentified organic radical signal observed only in the activated Cu3(BTC)2 sample. (E) Minimum-energy configuration for TCNQ@Cu3(BTC)2 obtained from ab initio calculations. (F) Possible configuration that would provide a conductive channel through the MOF unit cell. Atom color code for (E) and (F) as in Fig. 1.

TCNQ complexes have been characterized extensively by vibrational spectroscopies, where the frequencies of C=C and C≡N stretching modes are particularly sensitive to the extent of charge transfer (13, 20, 21). Raman spectra of TCNQ@Cu3(BTC)2 (Fig. 3B) indicate that the TCNQ C=C stretching frequency shifted from 1456 cm–1 to 1437 cm–1 and new peaks appeared at 1352 cm–1 and 1296 cm–1—a strong indication that TCNQ interacts with the available coordination sites on the Cu2+ ions in the framework. A shift of 19 cm–1 for the C=C wing stretching mode suggests a partial charge transfer of ~0.3e between the framework and TCNQ (20). The nitrile stretch at 2229 cm–1 is split into two peaks at 2226 cm–1 and 2213 cm–1 (Fig. 3B, inset) indicating two non-equivalent C≡N bonding environments, in close agreement with our calculated vibration spectra (11). Infrared spectra (Fig. 3C) also show that the C≡N stretch of TCNQ is affected by adsorption into the framework, with a shift from 2223 cm–1 to 2204 cm–1 accompanied by substantial peak broadening. According to Chappell et al. (21), this shift corresponds to a charge transfer of ~0.4e between the framework and TCNQ, in reasonable agreement with the value inferred from the Raman spectra. The observation of partial charge transfer is further supported by room-temperature EPR spectra obtained from TCNQ@Cu3(BTC)2 (Fig. 3D), which display no evidence of TCNQ radical anions (22). Partial charge transfer is characteristic of many conducting TCNQ salts; however, in contrast to the well-studied CuTCNQ, in which Cu(+1) predominates, Cu in Cu3(BTC)2 remains in a +2 state after infiltration, as revealed by XPS (fig. S4).

The importance of guest-host interactions was further probed by replacing TCNQ with its fully hydrogenated counterpart, H4-TCNQ [cyclohexane-1,4-diylidene)dimalononitrile], which lacks a conjugated π electron network, and F4-TCNQ (2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane), which has a similar HOMO-LUMO (highest occupied molecular orbital–lowest unoccupied molecular orbital) gap but higher electron affinity relative to TCNQ. Elemental analysis indicates that the loading of H4-TCNQ is similar to that of TCNQ (i.e., about 1 H4-TCNQ molecule per pore). The I-V curve (Fig. 2A) for H4-TCNQ@Cu3(BTC)2 is essentially the same as that of the uninfiltrated, nonconducting MOF. The UV-vis spectrum also lacks the characteristic bands indicative of charge transfer in TCNQ@Cu3(BTC)2 (Fig. 3A). These results illustrate that the availability of guest molecule orbitals that can accept charge, as is the case in TCNQ but not H4-TCNQ, is important for achieving high conductivity. F4-TCNQ@Cu3(BTC)2 is not as conductive as TCNQ@Cu3(BTC)2. We view this result as semiquantitative, however, because F4-TCNQ is volatile, and unlike TCNQ@Cu3(BTC)2, the conductivity was not stable with time. Nonetheless, the lower conductivity measured immediately after infiltration suggests that the high electron affinity of this molecule inhibits electron mobility.

Finally, ab initio calculations suggest a possible mechanism for the appearance of conductance in TCNQ@Cu3(BTC)2 hybrids. As illustrated in Fig. 3E, the calculations predict that TCNQ binds strongly to the MOF (binding energy of 83.9 kJ/mol) and that four such molecules create a continuous path through the unit cell (Fig. 3F and movie S1). Our calculations for molecular clusters comprising two copper dimer groups (MOF secondary building units) bridged by a TCNQ molecule show that the bridging TCNQ inserts unoccupied molecular orbitals into the MOF HOMO-LUMO gap (fig. S5), producing the new charge transfer band in the visible spectrum and enabling electronic coupling between the MOF and TCNQ. Moreover, computed values of HAB, the electronic coupling matrix element for electron transfer from the MOF cluster to TCNQ [i.e., TCNQ@Cu3(BTC)2 → TCNQ@Cu3(BTC)2], combined with the value of the activation energy obtained from the temperature dependence of the conductivity, allow us to evaluate the extent of donor-acceptor coupling using the quantity 2HAB/λ (where λ is the reorganization parameter) defined by Brunschwig et al. (19). We find 2HAB/λ = 1.21 for TCNQ@Cu3(BTC)2, identifying this material as a class III system according to the Robin-Day classification scheme (19). These calculations also predict that electronic coupling in F4-TCNQ@Cu3(BTC)2 is intermediate between H4-TCNQ and TCNQ itself. The order of HAB values is H4-TCNQ < F4-TCNQ < TCNQ (0.19 eV < 1.03 eV < 2.32 eV)—a trend that is consistent with the observed conductivities.

Supplementary Materials

www.sciencemag.org/content/343/6166/66/suppl/DC1

Materials and Methods

Figs. S1 to S8

Table S1

Movie S1

References (2339)

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: We thank S. T. Meek for organic synthesis in this work and D. Ruzmetov for help in fabrication. This work was supported by the Laboratory Directed Research and Development Program at Sandia National Laboratories and the U.S. Department of Energy (DOE) SunShot Program. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin company, for the U.S. DOE National Nuclear Security Administration under contract DE-AC04-94AL85000. A.A.T. was supported by the Science of Precision Multifunctional Nanostructures for Electrical Energy Storage (NEES), an Energy Frontier Research Center funded by the U.S. DOE, Office of Science, Office of Basic Energy Sciences under award DESC0001160. F.E.G. was supported by the Office of Basic Energy Sciences, Division of Materials and Engineering Sciences, U.S. DOE, under contract DE-AC04-94AL85000. A.C., R.A.K., and H.P.Y. acknowledge support under the Cooperative Research Agreement between the University of Maryland and the National Institute of Standards and Technology Center for Nanoscale Science and Technology, award 70NANB10H193, through the University of Maryland.
View Abstract

Navigate This Article