Ningxin Jianga,
Saranya Velliyaratb,
Chen-Yu Liena,
Ha L. Nguyena,
Jan Hofmann
c,
Jie-Hao Chen
d,
Arun Ramanathan
e,
Alexander S. Filatov
a,
Henry S. La Pierre
e,
Shrayesh Patel
d,
Karena W. Chapmanc,
Jan-Niklas Boyn
b and
John S. Anderson
*a
aDepartment of Chemistry, University of Chicago, Chicago, Illinois 60637, USA. E-mail: jsanderson@uchicago.edu
bDepartment of Chemistry, University of Minnesota, Minneapolis, Minnesota 55455, USA
cDepartment of Chemistry, Stony Brook University, Stony Brook, New York, 11794, USA
dPritzker School of Molecular Engineering, University of Chicago, Chicago, Illinois 60637, USA
eSchool of Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, Georgia 30332-0400, USA
First published on 11th September 2025
Conductive coordination polymers (CPs) with sulfur-based ligands offer strong metal–ligand interactions and redox tunability, making them promising candidates for electronic applications. Tetrathiafulvalene-2,3,6,7-tetrathiolate (TTFtt) is a particularly attractive ligand. However, its strong metal–ligand covalency leads to rapid irreversible metal coordination, limiting control over structure and morphology. Here, we demonstrate structural control in Cu TTFtt CPs using a pre-synthetic redox control strategy. Two new copper-based CPs, CuTTFtt and Cu2TTFtt, have been synthesized and thoroughly characterized from differentially oxidized TTFtt synthons. CuTTFtt forms a 1D chain, while Cu2TTFtt adopts a 2D ribbon-like structure. Detailed spectroscopic studies confirm the structures of these materials as well as their ligand and metal oxidation states. Physical property measurements reveal that Cu2TTFtt exhibits higher conductivity than CuTTFtt. Furthermore, Cu2TTFtt also shows unusual diamagnetism which contrasts the paramagnetism observed in CuTTFtt and the related material NiTTFtt. Density functional theory (DFT) further elucidates the physical properties of these CPs and supports the observed conductivity trends. This study expands the structural landscape of TTFtt-based CPs and further establishes how redox-doping can tune CP structure and physical properties.
Of many possible sulfur-rich ligands, tetrathiafulvalene-2,3,6,7-tetrathiolate (TTFtt), which combines a tetrathiafulvalene (TTF) core—a well-known motif in conductive molecules—with dithiolene coordination sites, is an excellent candidate for designing highly conductive materials.23–25 Several reports have investigated the combination of this linker with transition metals, but it typically exhibits rapid reaction with metal cations. This rapid irreversible reaction makes it difficult to control CP structure or morphology, and syntheses with TTFtt often yield amorphous black powders which can be difficult to characterize despite being highly conductive.26–29 Early synthetic efforts to generate TTFtt based CPs with both Ni and Cu resulted in conductive solids, but minimal insight into their electronic and geometric structure was obtained.30 This lack of insight is largely due to challenging structural characterization which can be particularly difficult with thiolate-based systems.31–34 Hoffmann and coworkers proposed several structures for TTFtt-based materials 40 years ago this year,35 but only a 1D chain structure of NiTTFtt has been experimentally demonstrated.23 Predictions of an alternative 2D sheet structure remain experimentally unverified.36
This dearth of detail presents a significant challenge in understanding (and controlling) the structure and properties of these materials. Dimensionality (1D, 2D, and 3D) plays a critical role over physical properties including both conductivity and magnetism, as demonstrated in both carbon-based materials and reticular structures.37–39 However, studies on the dimensionality of sulfur-based frameworks are rare. This difficulty in building structure–function relationships is made even more challenging as sulfur-based ligands often feature multiple accessible oxidation states which may change concurrently with changes in dimensionality.20,21,40 Many CP syntheses occur in aerobic conditions which can lead to in situ oxidation.41–43 This redox ambiguity complicates the determination of metal oxidation states, especially with redox-active metals such as Cu, where ambiguities in oxidation states are common in thiolate-based systems.44,45 The redox activity of ligands combined with the structural challenges mentioned above, make understanding and controlling the properties of TTFtt-based materials particularly challenging.
We recently employed a transmetalation and pre-synthetic doping strategy to successfully synthesize Ni CPs of TTFtt with variable TTFtt oxidation states.40,46 Using a pre-oxidized TTFtt transmetalating synthon provides NiTTFtt with a 1D chain structure where TTFtt is in a formally doubly oxidized state. While NiTTFtt displays high conductivity despite an amorphous structure, its reduced congener Li-NiTTFtt, with an overall TTFtt4− ligand, displays intriguing photothermoelectric and thermoelectric properties.
This progress in understanding the structure and electronic properties of NiTTFtt motivates extending this synthetic control to other transition metal centers. Copper-thiolate CPs are known to exhibit electrical conductivity comparable to that of nickel-thiolate materials.45,47,48 We have therefore investigated copper coordination chemistry with TTFtt and synthesized two new materials, CuTTFtt and Cu2TTFtt. By employing similar pre-synthetic redox control of transmetalating TTFtt reagents, we can manipulate TTFtt oxidation states, with Cu2TTFtt containing TTFtt3− linkers and CuTTFtt containing oxidized TTFtt2− linkers. Thorough characterization, including X-ray absorption spectroscopy (XAS), X-ray photoelectron spectroscopy (XPS), and Raman spectroscopy, enable an accurate determination of ligand and copper oxidation states.
Structural analyses suggest that while CuTTFtt adopts a 1D chain structure similar to NiTTFtt, while Cu2TTFtt forms a 2D ribbon-like layered structure consistent with original structural models proposed by Hoffman and coworkers.35 Conductivity measurements demonstrate that Cu2TTFtt shows higher conductivity compared to CuTTFtt. In contrast to the dominant Pauli paramagnetism observed in NiTTFtt, Cu2TTFtt also shows diamagnetic behavior while CuTTFtt exhibits Curie–Weiss paramagnetism. Density functional theory (DFT) calculations were also employed to provide insight into the different physical properties of NiTTFtt, CuTTFtt, and Cu2TTFtt and validate the observed experimental trends.
These findings validate and expand the known structural types for TTFtt-based CPs and also elucidate how these structures influence charge transport properties. Moreover, the different morphologies observed for these copper-based CPs suggest that linker redox-tuning is an important strategy for controlling structure. This study motivates continued investigations into how the structure and metal identity of TTFtt-based materials dictates magnetic coupling and novel emergent properties at the interface of conductivity and magnetism.
For Cu2TTFtt, two equivalents of Cu(acacF3)2 (acacF3 = trifluoroacetylacetonate) were mixed with excess tetramethylethylenediamine (TMEDA) in tetrahydrofuran (THF), then combined with one equivalent of TTFtt(SnBu2)2 in THF to immediately generate a dark powder. It is worth noting that the addition of TMEDA to a Cu(acacF3)2 solution initially results in an immediate color change from blue to green, suggesting the formation of [(Cu(TMEDA)2)]2+. The isolated dark green (nearly black) powder was dried at 70 °C to yield Cu2TTFtt.
For Cu2TTFtt, combustion analysis yields 21.36(1)% carbon, 2.72(3)% nitrogen, and 1.75(3)% hydrogen. These results suggest some additional organic component beyond a limiting formula of Cu2C6S8. Combined with a ∼10% mass loss observed at ∼200 °C in TGA (Fig. S2), we propose a chemical formula for Cu2TTFtt as Cu2C6S8(C6H16N2)0.5 (Cu2(TTFtt)(TMEDA)0.5) with the inclusion of 0.5 TMEDA molecules per formula unit. The inclusion of TMEDA suggests a fundamentally different structure for Cu2TTFtt. The proposed chemical formulas of both CuTTFtt and Cu2TTFtt match well with the combustion analysis results shown in Table S2.
![]() | ||
Fig. 2 (A) Synchrotron (λ = 0.167 Å) PXRD patterns and of Cu2TTFtt and CuTTFtt, (B) PDF of Cu2TTFtt and CuTTFtt. Structural model of Cu2TTFtt viewed along c-axis (C) and viewed from a-axis (D). |
Pair distribution function (PDF) data (Fig. 2C and S5A) reveal local range order in these CPs. The low r region of the PDF data for both materials contains features at ∼1.4 Å, 1.8 Å, 2.3 Å, 2.7 Å and 3.0 Å that correspond to the C–C, C–S, Cu–S, C⋯S, and S⋯S distances, respectively (Fig. 2D and S5). These peaks correspond to intra-chain atomic distances within CuTTFtt and Cu2TTFtt and thus verify the presence of 1D chains in both materials. The PDF of CuTTFtt can be well-described by a single-chain model derived from NiTTFtt (Fig. S6). The difference in peak position between model and data around 2.3 Å is due to different Cu–S bond lengths compared with Ni–S distances from the NiTTFtt model.23
The PDF of Cu2TTFtt shows distinct differences in local order from the 1D materials, particularly in the intensity of the Cu–S peak at 2.3 Å and the presence of additional peaks at 3.5 Å and 3.9 Å. A similar 3.9 Å distance has been associated with a side-by-side ligand arrangement in other TTF based materials,49 and so the presence of this feature in Cu2TTFtt suggests the presence of such a side-by-side TTFtt arrangement. Combined with the increased Cu–S intensity observed in the PDF of Cu2TTFtt, we propose that these increased peak intensities correspond to additional Cu2+ ions that bind to sulfur in between negatively charged CuTTFtt2− chains in a side-by-side arrangement (Fig. 2C, D and 3). As previously mentioned, a structural model for tetrathiolate-based materials with a metal-to-ligand ratio of 2:
1 was previously proposed by Hoffmann and coworkers in 1985, but experimental validations have been lacking until the present example for Cu2TTFtt. This model has also been proposed as a potential structure for [Cux(Cu-ETT)] systems.35,36 We therefore propose a related 2D model for Cu2TTFtt with additional Cu–S bonds, which is consistent with the larger peak at 2.3 Å in the PDF data (Fig. S5B). This model consists of a 2D layered framework where copper cations connect 1D chains (Fig. 2C, D and S7). Consistent with prior data on TTFtt materials, the interlayer distance is 3.62 Å and the distance between neighboring Cu centers is 12.6 Å within a Cu–TTFtt chain. It should be noted that this construct is an idealized highly symmetric model of the material and requires a perfect 1D chain length match and alignment when propagating along the second dimension. Any mismatch in these distances/alignments would lead to disorder in the material and formation of amorphous to semi-crystalline materials as observed experimentally.
![]() | ||
Fig. 3 Demonstration of proposed structural models of CuTTFtt (middle) and Cu2TTFtt (bottom) based on PXRD, PDF and elemental analysis. |
The simulation of the PXRD pattern based on this idealized model using Pmmm space group and a unit cell of a = 6.67 Å, b = 12.61 Å, c = 3.62 Å (V = 304.3 Å3) reasonably reproduces the general features of the experimental diffraction pattern. This motivated the use of this model for a crude Rietveld refinement. This analysis is admittedly limited due to the poor crystallinity of Cu2TTFtt, but the Rietveld refinement does show a reasonable fit with the experimental data which provides some validation of the proposed structural model (Fig. S9). The refined unit cell parameters are a = 6.613(11) Å, b = 12.991(8) Å, c = 3.6136(20) Å (V = 310.4(6) Å3) with R and wR factors of 0.042 and 0.051 respectively. The only sharp Bragg peak is located at Q = 0.53 Å−1 and is assigned to a (010) reflection suggesting a more precise long-order arrangement of 1D chains in the structure compared to all other dimensions. The crystallite size (as refined over the whole powder pattern) is ∼7 nm, which is indicative of 5 to 6 Cu–TTFtt motifs in a single chain or ∼20 Cu2TTFtt layers. While this domain size is reasonable for a material with defects and disorder, it should be emphasized that this value is highly sensitive to both the sample's low crystallinity and to the limitations of our structural model. As such, the reported crystallite size should be regarded as an approximate lower bound rather than an absolute measurement.
While these X-ray analyses provide a reasonable structure for Cu2TTFtt, composition studies reveal a significant amount of TMEDA which is unaccounted for in the structural model. Attempts to incorporate TMEDA into the 2D plane or between the 2D planes do not produce physically reasonable models due to steric clashes. We instead propose that TMEDA binds exclusively to copper at the edge sites of Cu2TTFtt. TMEDA is known to act as a bidentate ligand for copper, forming complexes such as [Cu(TMEDA)Lx] or [Cu(TMEDA)2]1+/2+. When L represents sulfur-based ligands, a square planar geometry is reasonably expected in [Cu(TMEDA)L2].50 Given that Cu cations bind strongly to sulfur-based ligands, we propose that Cu(TMEDA)1+/2+ resides at the terminal positions of each Cu–TTFtt chain in Cu2TTFtt. A structural model was generated to allow AA stacking of Cu–TTFtt chains with TMEDA termination, as shown in Fig. S10. After structural optimization, the interlayer distance increased to 6.5 Å, which is longer than the expected 3.6 Å from PXRD data and suggests that Cu(TMEDA) complexes cannot stack directly on top of each other from different chains (Fig. S10). However, a structural model with TMEDA present only at terminal positions cannot account for the Cu:
TMEDA ratio of 4 unless each chain consists of only four TTFtt anions, which is inconsistent with the strong (010) peak observed at 0.80° 2θ (l = 0.167 Å). The experimentally observed Cu
:
TMEDA ratio can only be achieved if two Cu–TTFtt chains are coupled in Cu2TTFtt, as illustrated in Fig. 3 and S12. The resulting structural model, shown in Fig. 3, demonstrates a chemical formula of Cu2TTFtt(TMEDA)0.5, which aligns well with the compositional analysis and avoids TMEDA steric clashes. Thus, the combined experimental data support that Cu2TTFtt adopts a 2D ribbon-like layered structure similar to that shown in Fig. 3.
We note that, despite the stoichiometric amount of TMEDA present, no distinct peak corresponding to TMEDA can be identified in the PDF analysis of Cu2TTFtt (Fig. 2B). This absence can be attributed to two factors. Firstly, reasonable Cu–N bond lengths (∼1.9 Å) have significant overlap with the numerous C–S bonds in the material. Secondly, the number of proposed Cu–N bonds is much smaller than the other bonds represented in the PDF analysis, resulting in a lower signal intensity that cannot be directly observed.
To further characterize the structures of Cu2TTFtt and CuTTFtt, we carried out Cu K-edge XAS measurements and analyzed the EXAFS data (Fig. S13, S14, Tables S3 and S4). For CuTTFtt, the best-fit results yield a Cu–S distance of 2.28 Å, a Cu–C distance of 3.13 Å, and an average Cu coordination number of 4.0 ± 0.2. These values are consistent with previously reported Cu–S bond lengths and with our PDF analysis,51 confirming that Cu in CuTTFtt adopts a square-planar coordination environment. This finding further supports the conclusion that the structure of CuTTFtt closely resembles that of NiTTFtt, forming a one-dimensional chain-like arrangement.
For Cu2TTFtt, the EXAFS fitting gives a Cu–N distance of 1.76 Å, a Cu–S distance of 2.27 Å, a Cu–Cu distance of 2.95 Å, and a Cu–C distance of 3.14 Å, with an average Cu coordination number of 4.1 ± 0.4. The Cu–S bond length again agrees well with the PDF analysis, confirming that Cu also adopts a square-planar geometry in this compound. The relative ratio of Cu–N to Cu–S bonds (1:
8) is close to the theoretical value of 1
:
7, consistent with the presence of stoichiometric TMEDA in the structure. Together, these EXAFS results provide direct structural support that Cu2TTFtt adopts a two-dimensional ribbon-like framework.
Recent sulfur K-edge XAS studies on Ni-TTFtt molecules indicate that the first pre-edge feature in doubly oxidized TTFtt2− appears ∼0.6 eV lower in energy than the pre-edge feature in both neutral TTFtt4− and singly oxidized TTFtt3−.54 This provides a useful benchmark to examine the oxidation state of TTFtt linkers in these copper-based materials. We therefore collected sulfur K-edge XAS data on both Cu2TTFtt and CuTTFtt (Fig. 4A). Molecular analogs of copper TTFtt compounds have not yet been successfully synthesized, so we compared the observed spectroscopic features to those of analogous nickel compounds for interpretation. For CuTTFtt, the first pre-edge feature is observed at 2470.4 eV, while in Cu2TTFtt, a shoulder-like feature appears at approximately 2471.2 eV. The 0.8 eV energy difference between the two samples strongly suggests that the TTFtt motifs are in different overall redox states. By comparison, the sulfur K-edge pre-edge feature for NiTTFtt appears at 2470.7 eV, supporting that the linkers in CuTTFtt are best assigned with a formal oxidation state of TTFtt2−. However, the pre-edge positions for TTFtt3− and TTFtt4− are similar, and so the redox state of TTFtt in Cu2TTFtt cannot be determined from sulfur K-edge XAS data alone.
![]() | ||
Fig. 4 Spectroscopic studies TTFtt ligand redox states. (A) Sulfur K-edge XAS, (B) Raman spectra, and (C) S 2p XPS spectra of Cu2TTFtt and CuTTFtt. |
To further investigate the formal TTFtt redox state, Raman spectroscopy and XPS were employed. Infrared signals were difficult to interpret due to broadening, presumably from high reflectivity/shielding from potential metallic character (Fig. S15). The Raman spectra for Cu2TTFtt and CuTTFtt are more informative and are shown in Fig. 4B. Both compounds show peaks between 1400-1450 cm−1 which can be assigned to C–C vibrations.55 For Cu2TTFtt, this feature is centered at 1401 cm−1, compared to 1434 cm−1 in CuTTFtt. This shift to a lower wavenumber in Cu2TTFtt suggests differences in C–C bonding order, likely indicating longer C–C bond lengths in Cu2TTFtt and thus a more reduced formal oxidation state of the TTFtt linkers. Similar trends have been observed in NiTTFtt coordination complexes, where oxidation of the TTFtt motif decreases C–C bond length.40 In CPs such as NiTTFtt (TTFtt2−) and Li-NiTTFtt (TTFtt4−), oxidized TTFtt CPs consistently show C–C vibrations at higher wavenumbers.23,40 Thus, the higher frequency features observed in CuTTFtt vs. Cu2TTFtt further support more reduced TTFtt linkers in Cu2TTFtt.
Additional peaks are present around 900 cm−1 which can be assigned to C–S vibrations. These features are present in Cu2TTFtt and Li-NiTTFtt but are absent in CuTTFtt,55 and so we hypothesize that they may serve as a characteristic Raman signal for reduced TTFtt linkers (i.e. TTFtt4−).40 Overall, the observed Raman features strongly support the assignment of TTFtt in Cu2TTFtt as having a lower redox state than in CuTTFtt. Both samples also show a peak at 487 cm−1 which arises from Cu–S vibrations. However, Cu2TTFtt shows an extra peak at 527 cm−1 which indicates an additional sulfur ligation environment in Cu2TTFtt and further supports the proposed 2D ribbon structure.
Sulfur 2p XPS data was finally collected to further corroborate the TTFtt oxidation states (Fig. 4C). Notably, both Cu2TTFtt and CuTTFtt exhibit a main peak accompanied by two shoulder-like features. Deconvolution of the spectra into three distinct sulfur chemical states yields a reasonable fit, as shown in Fig. S17. Based on previous XPS analyses of poly[Cux(Cu-ETT)],36 the first two sets of doublets can be assigned to the reduced terminal sulfur (162.5 eV) and the oxidized terminal sulfur (164.1 eV) in the TTFtt ligand. The broad doublets at higher binding energy (164.8 eV) correspond to sulfur within the TTF core. Notably, the ratio of oxidized sulfur to reduced sulfur, determined from peak areas, increases from 1.20 in Cu2TTFtt to 2.15 in CuTTFtt, confirming a more oxidized redox state of TTFtt in CuTTFtt. Interestingly, the 162 eV feature, which intensifies with increasing oxidation of TTFtt, has also been observed in NiTTFtt (Fig. S18), poly[Cux(Cu-ETT)], and LixFe3(THT)2.21,23,36 This observation highlights XPS as a powerful tool for studying ligand redox states.
Finally, we examined the Cu K-edge XANES spectra of CuTTFtt and Cu2TTFtt (Fig. S20). The absorption edge of Cu2TTFtt is clearly shifted to lower energy relative to that of CuTTFtt. Since the Cu K-edge position decreases with decreasing oxidation state, this shift indicates the presence of a more reduced copper component in Cu2TTFtt. Together with the spectroscopic evidence for TTFtt2− ligands in CuTTFtt, this strongly supports a Cu2+ assignment in this compound, giving an overall formal redox state of (Cu2+)(TTFtt2−). For Cu2TTFtt, the ligand is expected to be either TTFtt3− or TTFtt4−; considering the lower edge position and the presence of a more reduced copper species, the most reasonable assignment is a mixed-valent state of (Cu2+)(Cu+)(TTFtt3−).51
Additional data were then collected to further understand the charge transport properties of CuTTFtt and Cu2TTFtt. Both materials show increased resistance with decreasing temperature (Fig. 5A). Fitting the temperature-dependent conductivity data from 300 K to 150 K yields activation energies of 13.4 meV for Cu2TTFtt and 31.6 meV for CuTTFtt. The smaller activation energy in Cu2TTFtt suggests either a higher carrier density or mobility in Cu2TTFtt than in CuTTFtt. The UV-vis-NIR spectra (Fig. S21) show no sharp absorbance drop for either compound, indicating a band gap smaller than 0.62 eV, consistent with the small activation energy values. Ultraviolet photoelectron spectroscopy (UPS) allows for the determination of work functions of 4.27 eV for Cu2TTFtt and 4.47 eV for CuTTFtt (Fig. S22). Notably, for both compounds, the counts per second (CPS) drop to background levels at 0 eV, indicating a low or zero density of states at the Fermi level. The valence band maxima were determined to be 0.23 eV for Cu2TTFtt and 0.63 eV for CuTTFtt. These results suggest that both compounds are small-bandgap semiconductors, differing from the glassy metallic behavior observed in NiTTFtt. This is also consistent with the lower electrical conductivity values observed for both compounds compared to NiTTFtt. The Seebeck coefficients are 10.9(2) μV K−1 for Cu2TTFtt and −2.5(5) μV K−1 for CuTTFtt. In comparison, Li-NiTTFtt exhibits a Seebeck coefficient of 10 μV K−1, while NiTTFtt has a value of −3.6 μV K−1. Notably, materials where TTFtt is in a more reduced state (Cu2TTFtt and Li-NiTTFtt) exhibit p-type behavior, whereas those where TTFtt is in a more oxidized state (CuTTFtt and NiTTFtt) exhibit n-type behavior. This trend underscores the critical role of ligand redox states in tuning the electronic structure and determining the charge-carrier type in TTFtt-containing materials.
The magnetic properties of both CuTTFtt and Cu2TTFtt were also investigated, as shown in Fig. 5B and C. In contrast to NiTTFtt, which contains square planar Ni2+ cations in a closed shell diamagnetic 3d8 electron configuration, both copper compounds presented here contain Cu2+ cations, which have a 3d9 electronic configuration and are S = 1/2, potentially leading to magnetic behavior. For CuTTFtt the temperature dependence of the magnetic susceptibility (χ) increases monotonically with decreasing temperature from 300 to 1.8 K, consistent with paramagnetic behavior (Fig. 5B). As CuTTFtt exhibits high electrical conductivity, a small band gap, and S = 1/2 spin centers, some combination of both Pauli and Curie–Weiss paramagnetism is reasonable.56 Fitting the data from 150 K to 300 K using the Curie–Weiss law yields a Curie–Weiss temperature (θCW) of −74.8(14) K, Curie constant of 0.105 emu K mol−1 and an effective magnetic moment of 0.916(3)μB/Cu2+, which is lower than the expected 1.73μB/Cu2+ for S = 1/2 spins. The χT value at room temperature is 0.08 emu K mol−1 (Fig. S25C), which is also significantly lower than the expected spin-only value of 0.375 emu K mol−1. The field dependence of magnetization at 1.8 K slowly increases nonlinearly to 0.037μB/Cu2+ up to 7T (Fig. S25D).
To differentiate the contributions of Pauli and Curie–Weiss paramagnetism in TTFtt2−-based CPs, a modified Curie–Weiss law, χ = C/(T − θCW) + χ0, incorporating a temperature-independent component (χ0), was used to fit the temperature dependence of the magnetic susceptibility for both CuTTFtt and NiTTFtt (Fig. S26) from 100 K to 300 K.56 Diamagnetic corrections were applied prior to fitting, ensuring that χ0 primarily represents the Pauli paramagnetic contribution. For NiTTFtt, C = 0.0797 emu K mol−1, χ0 = 3.28 × 10−4 emu mol−1 and θCW = 3.6 K, indicating paramagnetism dominated by Pauli contributions. In contrast, for CuTTFtt, C = 0.0739 emu K mol−1, χ0 = 3.40 × 10−5 emu mol−1 and θCW = −45.8 K, suggesting a stronger Curie–Weiss paramagnetic component due to the higher absolute θCW value and lower χ0. A possible conclusion from this fit is a lower number of carriers in CuTTFtt, leading to a smaller Pauli contribution, alongside antiferromagnetic coupling between either copper- or TTFtt-based spins. The interplay of antiferromagnetic coupling with carrier density or mobility remains an interesting arc of investigation in these materials.
For Cu2TTFtt (Fig. 5C), the magnetic susceptibility (χ) is negative and extremely small (∼10−4 emu mol−1) down to 30 K even after accounting for diamagnetic corrections. The χT value is also negative at room temperature (−0.07 emu K mol−1, Fig. S25A). The magnetization at 1.8 K and 7T is only 0.01μB, indicating that Cu2TTFtt exhibits diamagnetic behavior despite the presence of Cu2+ cations (Fig. S25B). The diamagnetic behavior of Cu2TTFtt suggests significant antiferromagnetic coupling leading to a strongly insulated singlet ground state.
To further investigate the nature of the spin centers in both compounds, we carried out X-band EPR measurements at 4 K (Fig. S27). The data were fitted using the EasySpin software57 with one and two S = 1/2 spin centers in Cu2TTFtt and CuTTFtt, respectively (Table S5, see SI for details). For Cu2TTFtt, only a single S = 1/2 resonance was observed at g = 2.011. This g-value and the observed sharp linewidth are characteristic of an organic radical and we therefore assign this signal to radical TTFtt3− linkers. SQUID susceptibility measurements show that Cu2TTFtt is diamagnetic between 100 and 300 K, but exhibits a Curie tail at low temperature, supporting a small amount of magnetic impurities. Taken together, these results suggest strong antiferromagnetic exchange between Cu2+ and TTFtt3− in Cu2TTFtt, which suppresses the Cu2+ EPR signal while leaving a small but detectable fraction of unpaired TTFtt3− spins, likely from defects or disorder. This finding also provides further support for a formal redox-state assignment of Cu2TTFtt as (Cu2+)(Cu+)(TTFtt3−).
In contrast, the EPR spectrum of CuTTFtt displays two distinct S = 1/2 resonances. A narrow and weak signal at g = 2.005 (∼1% of total intensity) can be attributed to trace radicals, likely from TTFtt3− defects. The dominant broader resonance at g = 2.036 (∼99% of the signal intensity) can be reasonably assigned to uncoupled Cu2+ spin centers based on comparison with other Cu dithiolenes.58 This assignment is also consistent with the SQUID susceptibility results, which indicate antiferromagnetic interactions between S = 1/2 Cu2+ ions.
Thus, the EPR results reveal that while CuTTFtt contains some Cu2+ spins with only a negligible amount of radical impurities, Cu2TTFtt exhibits suppressed Cu2+ signals due to strong Cu2+–TTFtt3− antiferromagnetic interactions and some residual radical signatures from the TTFtt3− linkers. These results provide additional confirmation of our proposed redox state assignments for both compounds.
First, the magnetic ground states of CuTTFtt, Cu2TTFtt, and isolated 2D sheets of Cu2TTFtt were investigated, all of which converge to a closed-shell ground state with zero magnetization. In Cu2TTFtt, the energy difference between the S = 1 and S = 0 states is significantly larger in the π-stacked system (0.337 eV) than in the isolated sheet (0.101 eV). This indicates that the π–π interactions in the stacked system make it increasingly favorable to pair electrons, promoting a closed-shell or antiferromagnetic coupling state. For CuTTFtt, the calculated energy gap between S = 1 and S = 0 is 0.203 eV, which is smaller than that of Cu2TTFtt. In the S = 1 excited state, the majority of the spin density is localized on the TTFtt linkers, with 0.35 electrons residing on the in-chain Cu atom and no significant spin density on the out-of-chain Cu. This localization likely promotes convergence to a closed-shell solution. Comparison with the vacuum-isolated sheet shows that π-stacking interactions do not significantly change the spin delocalization in the S = 1 state.
The diamagnetism observed in Cu2TTFtt aligns well with the calculated magnetic structure. The source of the paramagnetism observed in CuTTFtt is less clear. The smaller calculated energy gap between the closed-shell and S = 1 solutions suggests that a paramagnetic state is more reasonable in CuTTFtt. We also note the significantly lower high-temperature χT in CuTTFtt than would be expected for an S = ½ paramagnet which is consistent with a comparatively large energy gap. However, it is difficult to rule out paramagnetic defect sites in this amorphous material. If some copper centers are structurally distorted they may behave as isolated paramagnets. In either case, computations support that paramagnetic behavior is more reasonable in CuTTFtt, but deeper explorations of the magnetism of TTFtt CPs with paramagnetic metal centers are still warranted.
The band structure and DOS of Cu2TTFtt were also analyzed to assess conductivity along different crystallographic directions (Fig. 6C). Along the Γ–Z and Y–T directions (the TTFtt polymer chain directions, Fig. 6A), a steep TTFtt-based band crossing the Fermi level indicates metallic conductivity mediated by TTFtt. Along Γ–Z, a flatter Cu-based band above the Fermi level suggests electron localization on Cu. Along Γ–Y (the Cu chain direction, orthogonal to the TTFtt chains), a Cu-based band crossing the Fermi level implies metallic conductivity mediated by Cu atoms.
Along Γ–X and S–Y (the π-stacking direction), two steep TTFtt-based bands crossing the Fermi level indicate high inter-stack conductivity, while a flatter Cu-based band just above the Fermi level in Γ–X suggests possible Cu electron localization. The DOS at the Fermi level is dominated by S p-orbitals, with notable contributions from C p- and Cu d-orbitals, highlighting the role of TTFtt π-electrons and Cu-mediated interactions in charge transport. To examine the role of π-stacking, we also analyzed isolated 2D sheets of Cu2TTFtt (see SI Section 6). The band structure of the isolated sheets exhibits a 0.45 eV band gap located 0.3 eV above the Fermi level, with generally flatter bands. These findings confirm that π-stacking plays an important role in mediating conductivity along the polymer and Cu chains. However, we note the additional metallic directions between 1D chains in Cu2TTFtt, which may lead to the higher conductivity in this material.
To test the effect of TMEDA, we calculated the DOS for the 2D Cu2TTFtt system with TMEDA included (Fig. S28). The results confirm that TMEDA has negligible effect on the states near the Fermi level, supporting that its omission does not significantly affect the predicted electronic structure. Next, we consider the band structure and DOS of CuTTFtt to analyze the conductivity of the 1D chain along different crystallographic directions (Fig. 6D). The electronic band structure reveals multiple bands crossing the Fermi level, indicating metallic behavior. Along Γ–X (slightly off-axis to the π-stacking direction, Fig. 6B), two steep, TTFtt-based bands crossing the Fermi level suggest π-electron delocalization, while a flatter Cu-based band just above the Fermi level indicates electron localization on Cu. Along Γ–Z (the polymer chain direction), one Cu-based and one TTFtt-based band cross the Fermi level, indicating metallic character along the chain. Finally, along Γ–T (in-plane direction nearly perpendicular to the polymer chain), a less dispersive Cu-based band crossing the Fermi level suggests weaker interchain interactions. A 0.2 eV band gap appears at 1.4 eV above the Fermi level. The DOS at the Fermi level is primarily contributed by S p-orbitals, with notable C p- and Cu d-orbital contributions, again highlighting the influence of TTFtt π-electrons and Cu–TTFtt interactions in conductivity.
The band structure and DOS of NiTTFtt (Fig. S30A) are analyzed as a reference against CuTTFtt due to their different conductivities despite similar structural motifs and π-stacking interactions. Along the polymer chain and π-stacking directions, NiTTFtt exhibits metallic character, primarily driven by TTFtt with minimal Ni contributions. In contrast, a band gap appears along the in-plane direction nearly perpendicular to the polymer chain indicating hindered charge transport in this direction. CuTTFtt exhibits a higher DOS at the Fermi level, gradually decreasing to form a band gap, whereas NiTTFtt shows no such gap (Fig. S30B). Additionally, CuTTFtt shows more avoided crossings and flatter bands than NiTTFtt, suggesting more electron localization. The reduced band dispersion in CuTTFtt correlates with the experimentally observed higher conductivity in NiTTFtt relative to CuTTFtt.
Next, we compare the band structure and DOS of Cu2TTFtt and CuTTFtt to examine how the additional orthogonal Cu chains in Cu2TTFtt influence its metallic character. Along the polymer chain direction, Cu2TTFtt has a more dispersive TTFtt-based band crossing the Fermi level compared to CuTTFtt. Additionally, in this direction, Cu2TTFtt has a relatively flat Cu-based band above the Fermi level, while in CuTTFtt, the Cu-based band crosses the Fermi level. In CuTTFtt, along the in-plane direction nearly perpendicular to the polymer chain, a relatively flat Cu-based band again suggests some electron localization. In contrast, Cu2TTFtt shows a highly dispersed band along the Cu chain direction, indicating strong metallic character. The 0.2 eV band gap observed at 1.4 eV above the Fermi level in CuTTFtt is absent in Cu2TTFtt, as this gap is filled with a high density of bands from the additional Cu orbitals. Both materials display a similar total DOS at the Fermi level, but CuTTFtt shows a peak just above EF that gradually decreases until forming the 0.2 eV gap (Fig. S30C).
The significantly lower conductivity of CuTTFtt compared to NiTTFtt is noteworthy. In several conductive reticular materials, Cu-based compounds typically exhibit higher conductivity than their Ni analogs with the same ligands, such as benzenehexathiolate and hexaiminobenzene.2 This divergent behavior in TTFtt2−-based materials may be attributed to the triplet diradical nature of TTFtt2− (ref. 73) and suggests that magnetic metal centers, such as Cu2+, are detrimental to electrical conductivity when putatively magnetic linkers are present. This may plausibly arise from some degree of coupling between the paramagnetic centers and the TTFtt-based electrons which serve as carriers.
Although theoretical calculations suggest that the 2D structure of Cu2TTFtt should result in electronic conductivity comparable to that of amorphous NiTTFtt, experimental results reveal that the conductivity of Cu2TTFtt is actually one order of magnitude lower. This counterintuitive result can be primarily attributed to the presence of TMEDA, which coordinates with Cu2+ centers and acts as an insulating barrier at the grain boundaries, thereby impeding efficient charge transport between crystallites. In contrast, NiTTFtt contains no organic components beyond TTFtt itself, allowing for strong π–π interactions between chains that promote effective interchain electron transfer. Possible strategies to further enhance the conductivity of Cu2TTFtt include substituting TMEDA with smaller amines, performing post-synthetic ligand exchange, and optimizing growth/annealing conditions to enlarge crystallite sizes.
Multiple independent measurements indicate that Cu2TTFtt is mixed-valent, with coexisting Cu+ and Cu2+ centers. Mixed valency has been reported more frequently in iron-based conductive coordination polymers, where Fe2+/Fe3+ delocalization can enhance charge transport.74 By analogy, similar mechanisms may facilitate conductivity in Cu2TTFtt even though the macroscopic conductivity is likely limited by grain-boundary effects. The presence of Cu+ may also be structurally consequential: it likely contributes to stabilizing the 2D ribbon-like architecture observed for Cu2TTFtt. In particular, prior studies on poly[Cux(Cu-ETT)] plausibly feature related structural/valence motifs,36 whereas analogous 2D arrangements have not been reported for Ni-based TTFtt or ETT systems—as expected due to the much lower stability of Ni+ relative to Cu+. Collectively, these considerations highlight the importance of Cu+ in stabilizing 2D networks and suggest that mixed valency may serve as an additional feature for tuning structure and charge transport in TTFtt-based coordination polymers.
Cu2TTFtt, with its 2D structure and strong antiferromagnetic (AFM) interactions, bears some similarity to the layered cuprate materials, which exhibit high-temperature superconductivity.75 In cuprates, doping introduces charge carriers that suppress long-range AFM ordering and enable unconventional superconducting states. We speculate that a similar approach in Cu2TTFtt, specifically doping at the Cu sites or modulating the redox state of the TTFtt ligand, may result in interesting electronic and magnetic properties. This is particularly compelling given the interplay of 2D geometry, AFM interactions, and the potential for doping-induced charge delocalization in Cu2TTFtt. Experimental exploration of doping strategies and their effects on the electronic density of states, along with theoretical studies to identify accessible pairing mechanisms, are exciting future areas of study.
Comprehensive spectroscopic studies, including sulfur and copper K-edge XAS, Raman spectroscopy, and XPS, demonstrate that the oxidation states of the TTFtt ligand and Cu centers play a critical role in determining the electronic and magnetic properties of these materials. The material CuTTFtt features an oxidized TTFtt2− state, while the spectroscopic evidence supports a reduced formally (Cu2+)(Cu+)(TTFtt3−) electronic structure in Cu2TTFtt. These results further underscore the importance of precise redox state determination in sulfur-based coordination systems.
Electrical conductivity measurements show that both materials are highly conductive, with room-temperature values of 23(2) S cm−1 for CuTTFtt and 50(2) S cm−1 for Cu2TTFtt. The higher conductivity of Cu2TTFtt is due to some combination of higher carrier densities or enhanced charge mobility as based on DFT calculations. Magnetic studies reveal contrasting behaviors. 1D CuTTFtt displays paramagnetic behavior, while Cu2TTFtt is diamagnetic, likely due to strong antiferromagnetic coupling interactions. These observations provide insight into the interplay between magnetic properties, dimensionality, and electronic properties in CPs containing redox-active ligands.
In conclusion, this work demonstrates the importance of redox state control and dimensionality in tuning the structural, electronic, and magnetic properties of TTFtt-based CPs, particularly as paramagnetic ions are included into these materials. The different structures arising from differentially oxidized precursors represents a new pathway for controlling material dimensionality and crystallinity. The inclusion of TMEDA in Cu2TTFtt also raises the possibility of modulating TTFtt-based materials with additional organic components. By bridging theoretical predictions and experimental realization, this study provides important insights into the rational design of highly conductive and magnetically tunable coordination materials.
This journal is © The Royal Society of Chemistry 2025 |