Experimental and theoretical study of dimer-of-dimers-type tetrarhodium(II) complexes bridged by 1,4-benzenedicarboxylate linkers

Yusuke Kataoka *a, Kazuki Arakawa a, Hikaru Ueda a, Natsumi Yano a, Tatsuya Kawamoto b and Makoto Handa *a
aDepartment of Chemistry, Interdisciplinary Graduate School of Science and Engineering, Shimane University, 1060, Nishikawatsu, Matsue, 690-8504, Japan. E-mail: kataoka@riko.shimane-u.ac.jp; handam@riko.shimane-u.ac.jp
bDepartment of Chemistry, Faculty of Science, Kanagawa University, 2946, Tsuchiya, Hiratsuka, Kanagawa 259-1293, Japan

Received 11th August 2018 , Accepted 22nd October 2018

First published on 23rd October 2018


Two dimer-of-dimers-type tetrarhodium complexes, [Rh4(piv)6(BDC)] ([1]; piv = pivalate) and [Rh4(piv)6(F4BDC)] ([2]), in which two paddlewheel-type dirhodium units are linked by 1,4-benzenedicarboxylate (BDC) and 1,4-tetrafluorobenzenedicarboxylate (F4BDC), respectively, have been synthesized and characterized via single-crystal X-ray diffraction analyses, ESI-MS, 1H NMR, infrared spectroscopy, Raman spectroscopy, and elemental analyses. Crystal structure analyses of [1(THF)4] and [2(THF)4], which are crystallized from THF solutions of [1] and [2], respectively, revealed that dihedral angles (ϕ) between two –CO2 units and phenyl rings of the BDC linker in [1(THF)4] are almost co-planar (ϕ = 2.8°), whereas those of the F4BDC linker in [2(THF)4] are largely inclined (ϕ = 78.3°). Density functional theory calculations clarified that (i) their dihedral angles of optimized geometries of [1(THF)4] and [2(THF)4] are almost the same as their experimental geometries, and (ii) the rotation energy barriers of phenyl moieties in [1(THF)4] and [2(THF)4] estimated by potential energy surface analyses are 12.0 and 8.4 kcal mol−1, respectively, indicating that hydrogen bondings are formed between two –CO2 units and four hydrogen atoms of phenyl rings of the BDC linker in [1(THF)4], whereas two –CO2 units and four fluorine groups on the phenyl ring of the F4BDC linker in [2(THF)4] are electrostatically and sterically repulsed. Electrochemical properties and electronic structures of [1(THF)4] and [2(THF)4] are strongly influenced by the electronic states of dicarboxylate linkers, whereas absorption spectra are strongly influenced by the dihedral angles between two –CO2 units and phenyl rings of dicarboxylate linkers.


Introduction

Bottom-up design and synthesis of supra-molecular complexes such as coordination cages (CCs),1–3 coordination polymers (CPs),4,5 and metal–organic frameworks (MOFs)6–10 have been receiving much attention from coordination chemists because their complexes possess intriguing functional properties unexpected from the precursor building block complexes, which should be modified and improved to develop and progress the functionalities of the supra-molecular complexes.11–15 As is well known, the paddlewheel-type dinuclear (M2) complexes, [M2(O2C-R)4], are well utilized as the precursor building blocks for the supra-molecular complexes because of their robustness and high structural D4h symmetry.16–20 The smallest supra-molecular complexes, in which two paddlewheel-type M2 cores are connected by a dicarboxylate linker, are called the “dimer-of-dimers-type” tetranuclear (M4) complexes,21–26 which are also well recognized as the smallest model units of MOFs from theoretical chemists.27 One of the established methods for the development of dimer-of-dimers-type M4 complexes is to use the heteroleptic M2 complexes with three terminal-ligands such as diphenylformamidine derivatives, which form the coordination-inactive M–N bonds, and one carboxylate ligand, which is utilized in the ligand-exchange reactions.22–24 However, the obtained supra-molecular complexes are often less soluble for the bulkiness of the terminal ligands. Thus, investigations of their properties and utilizations as building blocks for the bottom-up synthesis of CPs and MOFs are relatively limited. In contrast, by utilizing the strong-coordination nature of metal–pivalate bonds, Chisholm et al. reported the dimer-of-dimers-type Mo4 and W4 complexes, [M4(piv)6(BDC)] and [M4(piv)6 (F4BDC)] (piv = pivalate), in which two paddlewheel-type M2 units are linked by 1,4-benzenedicarboxylate (BDC) and 1,4-tetrafluorobenzenedicarboxylate (F4BDC), respectively.28–30 For the superior solubility of the complexes, their photophysical properties were closely investigated, and it was anticipated that their complexes could be utilized as the building blocks for CPs and MOFs. However, their complexes are unstable in air and are not structurally characterized. Because the molecular structures of the complexes have not been resolved by single crystal X-ray diffraction analyses, discussions on their geometries were only for the optimized geometries calculated by conventional dispersion-free density functional theory (DFT). Obtaining singe crystal X-ray structure data and the experimental evidence of the theoretically proposed structural features are important in the aspect of understating molecular geometries and electronic structures of dimer-of-dimers-type M4 complexes as well as the development and understanding of structures and properties of CCs, CPs, and MOFs. According to this background, we have been interested in the developments of the dimer-of-dimers-type Rh4 complexes because paddlewheel-type Rh2 complexes are stable and easy to handle in air and have been already well utilized as the building blocks for robust CCs, CPs, and MOFs.31–44

In this paper, we described the synthesis, characterization, molecular structures, and electronic features of two dimer-of-dimers-type Rh4 complexes, [Rh4(piv)6(BDC)] ([1]) and [Rh4(piv)6(F4BDC)] ([2]) (see Fig. 1(c)), in which dirhodium tripivalate units are linked by a linker ligand of BDC or F4BDC. The molecular structures of [1(THF)4] and [2(THF)4], which were obtained by recrystallization from THF solutions of [1] and [2], were clearly determined by single crystal X-ray diffraction analyses and their structures are supported by dispersion corrected DFT (DFT-D).45 DFT-D calculations also showed that the energy profiles of rotation barriers of phenyl moieties in [1(THF)4] and [2(THF)4] were different due to the electrostatic interactions between phenyl and carboxylate moieties. The electronic structures, electrochemical properties, and absorption spectral features of [1(THF)4] and [2(THF)4] were also investigated and discussed in this study.


image file: c8dt03293a-f1.tif
Fig. 1 Schematic molecular structures of (a) paddlewheel-type Rh2 complex, (b) dicarboxylate linker, and (c) dimer-of-dimers-type Rh4 complex.

Results and discussion

Synthetic procedures and characterization of dimer-of-dimers-type Rh4 complexes

Complexes [1] and [2] were prepared via the ligand exchange reactions between [Rh2(piv)4] (Fig. 1(a)) and corresponding dicarboxylic acids (H2BDC for [1] and H2F4BDC for [2]; Fig. 1(b)) in EtOH under solvothermal conditions (at 423 K for 15 h) with a Teflon-lined steel autoclave. The pure products of [1] and [2] were successfully isolated via silica-gel column chromatography (eluent: 3% MeOH in CHCl3) with 23.7 and 29.4% yields, respectively. Here, [Rh2(piv)4] (recover yields: 27.4 and 27.2% in the synthesis of [1] and [2], respectively) and intermediate complexes such as [Rh2(piv)3(H-BDC)] (12.4% yield) and [Rh2(piv)3(H-F4BDC)] (1.1% yield) are also confirmed as the by-products. In addition, insoluble polymeric complexes and rhodium blacks are also confirmed on the top of the silica gel column. Both [1] and [2] are air-stable and robust toward moisture. The good solubility of [1] and [2] was also found to be similar to that of [Rh2(piv)4]. For instance, [1] and [2] are commonly dissolved in THF, acetone, MeOH, and CHCl3.

The important knowledge for the syntheses of [1] and [2] was also confirmed as follows: (i) the solvothermal reaction under an O2 atmosphere affords a large amount of rhodium black as byproducts. (ii) [1] and [2] were not obtained by the general reflux method using EtOH. (iii) The increase of reaction time for the solvothermal reaction also afforded the increase of the rhodium black, and due to its formation, the yields of desired Rh4 complexes were apparently decreased. (iv) Desired Rh4 complexes were not obtained when the reaction temperature decreased below 373 K. From these results, we concluded that the present synthetic procedures should be appropriate and effective for the syntheses of [1] and [2].

To characterize the complexes [1] and [2], ESI-MS, 1H NMR, elemental analyses, infrared spectroscopy, and Raman spectroscopy were performed. In ESI-MS with the positive ion mode, the signals observed at 1204.9829 and 1276.9423 m/z agree well with the simulated [M + Na]+ values of [1] (1204.9837 m/z) and [2] (1276.9460 m/z), respectively. As depicted in Fig. S1 and S2 in the ESI, the shapes and patterns of the isotropic signals of [1] and [2] also fit well with their simulated data. These data strongly suggested the formation of dimer-of-dimers-type Rh4 complexes. The 1H NMR spectrum of [Rh2(piv)4] in DMSO-d6 shows only one signal at 0.88 ppm due to the methyl protons of the tBu group in piv ligands. On the other hand, in [1] and [2], the methyl proton signals of the tBu group in piv ligands are divided into two signals with integral intensities of 2[thin space (1/6-em)]:[thin space (1/6-em)]1, which corresponds to the existence of cis- and trans-piv ligands relative to the dicarboxylate linkers. Compared with methyl proton signals of [1], those of [2] display relatively downfield-shifts because of the electron-withdrawing effect of the F4BDC ligands in [2]. In [1], one proton signal appears at 7.73 ppm, which is assigned to the phenyl protons of the BDC ligand. As summarized in the Experimental section of [1] and [2], no superfluous signals were found in their 1H NMR spectra, and their signals are only found within the diamagnetic region. These results indicated that obtained [1] and [2] are obviously pure compounds. The purities of [1] and [2] are also supported by CH elemental analyses. The estimated CH ratios of [1] and [2] were almost equal to the calculated CH ratios of [1] and [2], respectively (see the Experimental section). In the infrared spectra of [1] and [2], the symmetric and asymmetric vibration modes of carboxylate moieties, i.e., νsym[CO2] and νasym[CO2], are observed; the νsym[CO2] and νasym[CO2] of [1] were observed at 1415 and 1576 cm−1, respectively, while those of [2] were found at 1415 and 1592 cm−1, respectively. The estimated separation values (Δν[CO2]) between νsym[CO2] and νasym[CO2] of [1] and [2] are 161 and 177 cm−1, respectively, which are in the typical range of those observed for paddlewheel-type Rh2 complexes,38–40 indicating that valence of the Rh2 units in [1] and [2] is both Rh(II)–Rh(II) states. The vibration of the Rh–Rh bond,16i.e. ν(Rh–Rh), of the [1] and [2] was confirmed at 328 and 332 cm−1, respectively, by the Raman spectra. These values are similar to that of [Rh2(piv)4] (330 cm−1), which has a single bond between two rhodium ions.

Crystal structures of dimer-of-dimers-type Rh4 complexes

Single crystals of [1(THF)4] and [2(THF)4] suitable for single crystal X-ray diffraction analyses were obtained via the slow evaporation of THF solution of [1] and THF/MeOH solution of [2], respectively. Diffraction analyses clarified that [1(THF)4] and [2(THF)4] crystallized both in the monoclinic system (P21/c space group). Crystallographic data and selected structural parameters of [1(THF)4] and [2(THF)4] are summarized in Tables 1 and S1, 2, respectively, and the Ortep structures of [1(THF)4] and [2(THF)4] are depicted in Fig. 2 and 3, respectively.
image file: c8dt03293a-f2.tif
Fig. 2 (a) Ortep structure and (b) linker structure of [1(THF)4] (40% thermal ellipsoid probabilities). One of the disordered parts is depicted for clarity, and the full structure with disorder parts is shown in Fig. S3 in the ESI.

image file: c8dt03293a-f3.tif
Fig. 3 (a) Ortep structure and (b) linker structure of [2(THF)4] (40% thermal ellipsoid probabilities).
Table 1 Crystallographic data of [1(THF)4] and [2(THF)4]
Complexes [1(THF)4] [2(THF)4]
Formula C54H90O20Rh4 C54H86F4O20Rh4
M r (g mol−1) 1470.89 1542.86
Crystal system Monoclinic Monoclinic
Space group P21/c P21/c
a (Å) 10.4346(5) 11.5391(18)
b (Å) 33.9316(12) 17.587(2)
c (Å) 10.1175(5) 16.956(3)
α (°) 90 90
β (°) 111.302(5) 109.225(3)
γ (°) 90 90
V3) 3337.5(3) 3249.1(9)
Z 2 2
D Calc (g cm−3) 1.464 1.577
μ (mm−1) 1.036 1.077
F(000) 1508.0 1572.0
R 1 (I > 2σ(I)) 0.0392 0.0297
wR2 (I > 2σ(I)) 0.1047 0.0708
R 1 (all data) 0.0461 0.0321
wR2 (all data) 0.1093 0.0731
GOF on F2 1.109 1.062


Crystal structure analyses revealed that asymmetric units of [1(THF)4] and [2(THF)4] were both one-half of dimer-of-dimers units, which contain two Rh ions, three pivalate anions, two THF molecules, and one-half of BDC or F4BDC linkers, respectively. Therefore, the crystallographic inversion centers of [1(THF)4] and [2(THF)4] are confirmed at midpoints of BDC and F4BDC linkers, respectively. Each Rh2 unit in [1(THF)4] and [2(THF)4] is coordinated by three μ-pivalate ions and one μ-BDC or μ-F4BDC linker at equatorial positions and two THF molecules at axial positions. Thus, the Rh(II) ions are in the pseudo octahedral environment. In [1(THF)4], one pivlate ion and two THF ligands are slightly disordered as depicted in Fig. S3. The Rh–Rh bond lengths in [1(THF)4] and [2(THF)4] are estimated to be 2.3813(4) and 2.3865(4) Å, respectively, which are within the typical range of the Rh(II)–Rh(II) bond length of paddlewheel-type Rh2 complexes.16 That is, the single bonds are formed between two Rh ions in [1(THF)4] and [2(THF)4]. The averaged Rh–O(trans-piv) bonds (2.027 Å), which are located at trans-positions to the F4BDC linker, are slightly shorter than averaged Rh–O(F4BDC) bonds (2.053 Å) and four other Rh–O(cis-piv) bonds (2.034 Å) due to the effect of the electron withdrawing F4BDC linker, while the averaged Rh–O(cis-piv) (2.030 Å), Rh–O(trans-piv) (2.029 Å), and Rh–O(BDC) bonds lengths (2.037 Å) in [1(THF)4] are almost the same. Axial-coordinated THF ligands in [2(THF)4] lean onto the [2] moiety, while those in [1(THF)4] stand upright. This is assumed to be due to the difference in the crystal packing stress between [1(THF)4] and [2(THF)4]. As depicted in Fig. 2(b) and 3(b), the other important difference has been found in the orientation of the phenyl group of the linker ligand between [1(THF)4] and [2(THF)4]. The –CO2 units and the phenyl ring within the BDC linker are almost co-planar in [1(THF)4], whereas the corresponding phenyl ring of F4BDC is largely inclined to the –CO2 units in [2(THF)4]. The dihedral angles (ϕ) defined by the phenyl ring and the –CO2 plane were estimated as ϕ = 2.8(5)° (for [1(THF)4]) and 78.3(3)° (for [2(THF)4]). It was speculated that the difference could be due to the fact that [1(THF)4] forms weak hydrogen-bonding interactions between four H atoms of –C6H4 moieties and four O atoms of –CO2 moieties in the BDC linker, whereas four F atoms of –C6F4 moieties and four oxygen atoms of –CO2 moieties in the F4BDC linker in [2(THF)4] are electronically repelled. In fact, the hydrogen bonding lengths in [1(THF)4] were estimated to be 2.469 (O(7)–H(18)) and 2.478 (O(8)–H(19)) Å, respectively, which are in the typical range of the C–H⋯O bonding interactions. The evidence for the hydrogen bonding interactions existing in [1(THF)4] is theoretically proved and the calculated results are described in the next theoretical section.

Optimized geometries, electronic structures, and potential energy surfaces of dimer-of-dimers-type Rh4 complexes

In order to investigate the favourable spin states, optimized geometries, structural-depending potential energy surfaces (PESs), and electronic structures of [1(THF)4] and [2(THF)4], dispersion-corrected density functional theory (DFT-D) calculations were performed.

Initially, the single point energy calculations for the crystal structures of [1(THF)4] and [2(THF)4] at singlet, triplet, and quintet states were performed to determine their favorable spin states. As depicted in Fig. S4, the most stable spin states of [1(THF)4] and [2(THF)4] are both singlet; the total energies of [1(THF)4] and [2(THF)4] at the singlet state are 42.3 and 43.0 kcal mol−1 more stable than those at triplet states. Thus, it is considered that the thermal excitation from singlet to triplet spin states was certainly impossible. The total energies of [1(THF)4] and [2(THF)4] at quintet states are also much higher than those at singlet states. These results clearly pointed out that both the spin states of [1(THF)4] and [2(THF)4] are definitely singlet, which is consistent with the experimentally obtained results that 1H NMR spectral signals of [1(THF)4] and [2(THF)4] were observed in the diamagnetic region.

To investigate in depth the molecular geometries of [1(THF)4] and [2(THF)4], geometry optimizations of their complexes at the singlet state were performed with the solvent effect of THF. The selected (averaged) bond lengths of optimized geometries of [1(THF)4] and [2(THF)4] are summarized in Table S3. The optimized geometries of [1(THF)4] and [2(THF)4], which have the C2h symmetry, are reasonably reproduced in accordance with their geometries determined by single crystal X-ray diffraction analyses. For instance, the Rh–Rh bond lengths of optimized geometries of [1(THF)4] and [2(THF)4] are 2.413 and 2.418 Å, which are only 0.025 and 0.032 Å longer than their experimental geometries. The Rh–O(carboxylate) bonds, i.e., Rh–O(cis-piv), Rh–O(trans-piv), and Rh–O((F4)BDC) bonds of the optimized geometries of [1(THF)4] and [2(THF)4] are slightly longer (by 0.04 Å) than those of experimental structures. In the optimized geometry of [2(THF)4], the averaged bond length for Rh–O(trans-piv) (2.055 Å) is slightly shorter than those for Rh–O(F4BDC) (2.078 Å) and Rh–O(cis-piv) (2.062 Å), which is considered to be due to the effect of the electron withdrawing F4BDC linker. On the other hand, the averaged Rh–O(cis-piv) (2.065 Å), Rh–O(trans-piv) (2.066 Å), and Rh–O(BDC) bond lengths (2.071 Å) in optimized geometry of [1(THF)4] are almost the same, as shown for the experimental structure of [1(THF)4]. One of the reasons for the fact that the bond lengths of the primary coordination spheres of Rh ions in optimized geometries of [1(THF)4] and [2(THF)4] are slightly longer than those in experimental structures could be explained with the presence of crystal packing effects as previously reported in the theoretical study of the paddlewheel-type Rh2 complex.46 The averaged dihedral angles (ϕ) between two –CO2 units and phenyl rings in dicarboxylate linkers in [1(THF)4] and [2(THF)4] are 1.4 and 61.8°, respectively, which are similar to their experimental values (ϕ = 2.5(11)° and 78.3(3)° for [1(THF)4] and [2(THF)4], respectively). These results mean that the difference in the dihedral angle between [1(THF)4] and [2(THF)4] should be due to the electrostatic effects, rather than crystal packing effects.

The rotation energy barriers of phenyl rings in [1(THF)4] and [2(THF)4] were closely investigated by PES analyses for the dihedral angles between two –CO2 units and phenyl rings. As depicted in Fig. 4, the PESs of [1(THF)4] and [2(THF)4] have both one energy minima at the dihedral angles of ϕ = 0.0° and 65.0°, respectively, which are close to their optimized structures.


image file: c8dt03293a-f4.tif
Fig. 4 Potential energy surfaces (PES) for the dihedral angles between two-CO2 units and phenyl rings in [1(THF)4] (image file: c8dt03293a-u1.tif)and [2(THF)4] (image file: c8dt03293a-u2.tif).

On the other hand, an energy maximum was found at ϕ = 90.0° in the PES of [1(THF)4], although the PES of [2(THF)4] has two energy maxima at ϕ = 0.0° and 90.0°. That is, when ϕ = 0.0°, [1(THF)4] and [2(THF)4] are most energetically stable and unstable, respectively. These results support that hydrogen bonding interactions are formed between –CO2 and BDC linkers in [1(THF)4], whereas electrostatic repulsions occur between –CO2 and F4BDC linkers in [2(THF)4], as discussed in the previous section for the crystal structures of [1(THF)4] and [2(THF)4]. The existence of smaller energy maxima at ϕ = 90.0° in the PES of [2(THF)4] may come from (i) the very weak hydrogen bonding interactions between fluorine atoms in the F4BDC linker and piv ligands and (ii) unstabilization of the orbitals with interactions between π(F4BDC) and d(Rh2) orbitals.

Furthermore, the electronic structures of optimized geometries of [1(THF)4] and [2(THF)4] in THF were investigated. Fig. 5 shows the orbital energy diagrams and selected molecular orbitals (MOs) of [1(THF)4] and [2(THF)4]. The MOs from HOMO−15 to LUMO+7 of [1(THF)4] and [2(THF)4] are depicted in Fig. S5 and S6 in the ESI (here, the HOMO and LUMO are the highest occupied MO and lowest unoccupied MO, respectively.).


image file: c8dt03293a-f5.tif
Fig. 5 Orbital energy diagrams and selected MOs of [1(THF)4] and [2(THF)4].

In the occupied MO spaces, the electronic configurations of two Rh2 units in [1(THF)4] and [2(THF)4] are both best described as π8δ4σ4π*4δ*4π*44δ2σ2π*2δ*2π*2 configuration for one Rh2 unit). These results clearly support that the single bond is formed at the Rh2 units in [1(THF)4] and [2(THF)4]. As confirmed from Fig. 5, S5 and S6, the orbital populations of their MOs are equivalently distributed at two Rh2 units and orbital energies of occupied MOs of [2(THF)4] are relatively stable than those of [1(THF)4] due to the electron withdrawing effects of fluorine groups at the F4BDC linker in [2(THF)4]. Three pairs of HOMO/HOMO−1, HOMO−2/HOMO−3, and HOMO−4/HOMO−5 of [1(THF)4] and [2(THF)4] are energetically degenerated and are localized at anti-bonding π*(Rh2), δ*(Rh2), and π*(Rh2) orbitals, respectively. The most unstable bonding orbital characters of two Rh2 units are the σ(Rh2) orbitals, which form sizable anti-bonding interactions with p(O) orbitals of THF ligands and are found at HOMO−6 and HOMO−7 of [1(THF)4] and [2(THF)4]. The δ(Rh2) orbital characters are found at HOMO−8 and HOMO−9 of [1(THF)4] and [2(THF)4], in which the HOMO−8 orbital of [1(THF)4] also has large contribution of the π(BDC) orbital, while that of [2(THF)4] is relatively small.

HOMO−10 and HOMO−11 are mainly localized at the π(Rh2) orbitals, which have anti-bonding orbital interactions with p(O) orbitals of THF ligands, while π(Rh2) orbitals of HOMO−12 and HOMO−13 of [1(THF)4] and those of HOMO−13 and HOMO−14 of [1(THF)4] do not form anti-bonding orbital interactions with p(O) orbitals of THF ligands. The contributions of π(BDC) and π(F4BDC) orbitals are found at HOMO−14 of [1(THF)4] and HOMO−12 of [2(THF)4], respectively. The HOMO–LUMO gaps of [1(THF)4] and [2(THF)4] are estimated to be 3.64 and 3.84 eV, respectively. In the unoccupied MO spaces, the LUMO of [1(THF)4] is mainly localized at the BDC linker with the π*(BDC) character, while that of [2(THF)4] is delocalized at the d(Rh2) [minor contribution] and π*(F4BDC) [main contribution] orbitals. The anti-bonding σ*(Rh2) orbitals are found at LUMO+1 and LUMO+2 of [1(THF)4] and [2(THF)4]. Here, LUMO+2 of [2(THF)4] contains slight contribution of the π*(F4BDC) orbital. The unoccupied δ*(Rh2) and δ(Rh2) orbitals, which are formed by d(x2y2) characters of Rh ions, are found at LUMO+3/LUMO+4 and LUMO+5/LUMO+6, respectively, of [1(THF)4] and [2(THF)4]. LUMO+4 of [2(THF)4] also contains the small orbital contributions of the π*(F4BDC) orbital. The dominant π*(F4BDC) orbital characters, which have no orbital contributions from two Rh2 units, are found at LUMO+7 of [1(THF)4] and [2(THF)4]. From the above results, the MO interactions between the F4BDC linker and Rh2 units in [2(THF)4] are relatively stronger than those between BDC and Rh2 units in [1(THF)4].

Electrochemical properties

Electrochemical properties of [1] and [2] in THF solution were analysed via cyclic voltammetry (CV). As depicted in Fig. 6, the CVs of [1] and [2] (5.00 μmol in electrochemical cells) show one reversible redox wave at 1.20 and 1.30 V vs. SCE, respectively. Compared with the redox potentials of [Rh2(piv)4] (1.12 V vs. SCE), those of [1] and [2] are positively shifted by ca. 80 and 180 mV, respectively. It is considered that these shifts are due to the electron-withdrawing effects of phenyl moieties of BDC and F4BDC linkers in [1] and [2], respectively. It is remarkable that redox potential of [2] is further shifted to the positive side by 100 mV than that of [1] because of the electron-withdrawing effect of fluorine groups in the F4BDC linker in [2]. These results clearly suggest that the electronic states of dicarboxylate linkers in dimer-of-dimers-type Rh4 complexes strongly affect their redox potentials. The oxidation processes of [1] and [2] in THF solution were further investigated by the DFT calculations for oxidized species of [1(THF)4] and [2(THF)4] complexes, i.e., [1(THF)4]+ and [2(THF)4]+, on the assumption that [1(THF)4], [2(THF)4] and their oxidized cationic forms exist in THF solutions. As depicted in Fig. S7, LUMOs (β orbitals) of [1(THF)4]+ and [2(THF)4]+ are localized at π*(Rh2) moieties of two Rh2 units, but not at the BDC linker. Moreover, as shown in Fig. S8, a linear relationship between orbital energies of HOMOs and one-electron oxidation potentials of [1(THF)4], [2(THF)4] and [Rh2(piv)4] was observed. These indicate that the observed oxidations of [1] and [2] are metal-centred as being similar to the case of [Rh2(piv)4].
image file: c8dt03293a-f6.tif
Fig. 6 CVs of [1], [2], and [Rh2(piv)4] in THF containing 0.1 M TBAPF6.

Absorption properties

The absorption spectra of [1] and [2] as well as those of [Rh4(piv)4] are measured in THF solution, and their spectra are shown in Fig. 7. In the visible light region, the shapes of the absorption spectra of [1] and [2] are similar to that of [Rh2(piv)4]. [1] and [2] exhibit that two energetically low-lying absorption band wavelengths (λmax = 597 and 452 nm for [1]; λmax = 596 and 450 nm for [2]) are almost the same as those of [Rh2(piv)4] (λmax = 602 and 445 nm), whereas the molar absorption coefficients (ε) of the absorption maxima of [1] and [2] are ca. two times higher than those of [Rh2(piv)4] because [1] and [2] have two Rh2 units within the molecules. A higher energy absorption band (λmax = 452 nm) of [1] lies at the foot of an intense absorption band in the UV light region, thus giving a shoulder-like absorption shape. On the other hand, in the UV light region, [1] has a considerably different spectral feature from those of [2] and [Rh2(piv)4]. [2] and [Rh2(piv)4] have a shoulder band ([2]: 255 nm, [Rh2(piv)4]: 247 nm) and an intense band ([2]: 234 nm, [Rh2(piv)4]: 226 nm), whereas [1] has two intense absorption bands at λmax = 279 and 237 nm. The lower-energy band position (λmax = 279 nm) and its band shape of [1] clearly differ from those of [2], [Rh2(piv)4], and other Rh2 complexes, of which absorptions appear as shoulder bands. From the previously reported results of the theoretical calculations,47 the excitation of paddlewheel-type Rh2 complexes in the visible and UV region is mainly assigned to the d–d transitions at the Rh2 centers. There is the possibility that the band origin of the absorption at 279 nm for [1] is different from the d–d transition.
image file: c8dt03293a-f7.tif
Fig. 7 Absorption spectra of [1] (blue line), [2] (red line), and [Rh2(piv)4] (green line) in THF solution.

To investigate the absorption features of [1] and [2] in THF solution, time-dependent density functional theory (TDDFT) calculations were performed with the optimized geometries of [1(THF)4] and [2(THF)4] in THF. Calculated excitation energies, oscillator strengths, and assignments (excitation characters) of [1(THF)4] and [2(THF)4] are summarized in Tables S4 and S5. In the visible light region, two absorption bands of [1] observed at 597 and 452 nm are theoretically assigned to π*(Rh2) → σ*(Rh2) and π*(Rh2) → δ*(Rh2) transitions, respectively, while those of [2] observed at 596 and 450 nm are assigned to π*(Rh2) → σ*(Rh2) and π*(Rh2) → δ*(Rh2) transitions, respectively, with π*(Rh2) → π*(F4BDC)/d(Rh2) CT as minor excitation characters. This is considered to be one of the reasons for the large molar absorption coefficients of the lower-energy absorption band of [2] compared with that of [1]. In the UV light region, clear evidence of the CT band of [1] at 279 nm is theoretically confirmed; this CT band is assigned to δ(Rh2)/π(BDC) → π*(BDC) transition (oscillator strength: f = 0.6768). Similar intense excitation characters are not obtained from the TDDFT results of [2(THF)4]. Although the intense absorption band of [1] at 237 nm contains multiple-excitation components, the main characters are π(BDC)/d(Rh2) → π*(BDC) [dominant component] and δ*(Rh2)/π*(BDC) → π*(BDC) [minor component] transitions. The shoulder band of [2] observed at 255 nm can be assigned to the σ(Rh2) → σ*(Rh2) transition, which is consistent with previously reported theoretical and experimental results of paddlewheel-type Rh2 complexes. The intense absorption band of [2] at 234 nm contains π(F4BDC)/d(Rh2) → π*(F4BDC)/d(Rh2) and δ(Rh2) → π*(F4BDC)/d(Rh2) as major contributions with δ(Rh2) → σ*(Rh2) as minor contribution. That is, this band has CT characteristics like the case of [1(THF)4], which shows the corresponding absorption band at 237 nm.

Finally, to clarify the reason why an intense CT band is observed for [1] (at 279 nm), but not for [2] in the spectra measured in THF, TDDFT calculations were performed for [1(THF)4] and [2(THF)4], the dihedral angles being fixed at ϕ = 65° and 0°, respectively, where the ϕ values are exchanged to each other in this calculation. The intense CT band was completely eliminated around 279 nm for [1(THF)4], while an intense CT band newly appeared at a wavelength of 309 nm [δ(Rh2)/π(F4BDC) → π*(F4BDC)] for [2(THF)4]. Based on the results, it is concluded that the CT absorption band of [1(THF)4] at 279 nm should come from the large transition dipole moment due to the co-planar dihedral arrangement of ϕ = 0°, but not from the effects related to the electronic states of the dicarboxylate linker ligand.

Experimental

Materials and instrumentation

All chemical reagents and solvents used in this study were purchased from commercial sources and were used without further purifications. [Rh2(piv)4] was prepared according to procedures described in the literature.481H NMR spectra were recorded on a JEOL-500SS (500 MHz) spectrometer with DMSO-d6 as the solvent. Chemical shifts (δ/ppm) were referenced to residual DMSO (δ = 2.49 ppm). Electronspray ionization-mass spectral (ESI-MS) data were collected with a Bruker micrO-TOF II analyzer. Infrared spectra were recorded as KBr pellets with a JASCO FT-IR 660-plus spectrometer. Raman spectra were recorded on a Renishaw Raman system 2000 spectrometer equipped with a He–Ne laser (633 nm) as the excitation source. Elemental analyses were performed with a Yanaco CORDER MT-6 installed at Shimane University, Japan. UV-Visible absorption spectra in THF were recorded using a JASCO V-670 spectrometer. Cyclic voltammetry (CV) analyses were carried out in dried THF containing tetra-n-butylammonium hexafluorophosphate (TBAPF6) as an electrolyte using a BAS ALS-DY 2325 Electrochemical analyzer. A glassy carbon disk (1.5 mm radius), platinum wire, and a saturated calomel electrode (SCE) were used as the working, counter, and reference electrodes, respectively.

Synthetic procedures

Synthesis of [Rh4(piv)6(BDC)] [1]. [Rh2(piv)4] (100.0 mg, 0.164 mmol), H2BDC (13.6 mg, 0.082 mmol), and degassed EtOH solution (10.0 mL) were placed into the autoclave and sealed under an Ar atmosphere. The reaction mixture was heated at 423 K for 15 h. After cooling down to room temperature, the reaction solution was evaporated and then purified by column chromatography (silica-gel, eluent: 3% MeOH in CHCl3). The green solution was re-evaporated to dryness, and the green residue was collected and dried under vacuum at 353 K. Yield: 23.7% (24.4 mg). Anal calc. for Rh4C38H58 O16: C, 38.60; H, 4.94%. Found: C, 38.48; H, 4.66%. ESI-MS: calc. for [M + Na]+: 1204.9837 m/z; found 1204.9829 m/z. 1H NMR (500 MHz, DMSO-d6, 300 K, ppm): δ = 7.73 (s, 4H), 0.89 (s, 18H), 0.83 (s, 36H). IR data (KBr disk, cm−1): 2962 (m), 1684 (w), 1576 (s), 1486 (m), 1415 (s), 1387 (s), 1223 (m), 741 (w), 636 (w). Raman (cm−1): 328 [ν(Rh–Rh)].
Synthesis of [Rh4(piv)6(F4BDC)] [2]. [2] was synthesized by using the same procedure as that for [1], but with H2F4BDC (19.5 mg, 0.082 mmol) instead of H2BDC. Yield: 29.4% (30.2 mg). Anal calc. for Rh4C38H54O16F4: C, 36.38; H, 4.34%. Found: C, 36.56; H, 4.55%. ESI-MS: calc. for [M + Na]+ 1276.9460 m/z; found 1276.9423 m/z. 1H NMR (500 MHz, DMSO-d6, 300 K, ppm): δ = 0.90 (s, 36H), 0.89 (s, 18H). IR data (KBr disk, cm−1): 2964 (m), 1592 (s), 1485 (m), 1415 (s), 1395 (s), 1378 (m), 1223 (m), 993 (w), 761 (w), 640 (w). Raman (cm−1): 332 [ν(Rh–Rh)].

Crystallography

Single crystals of [1(THF)4] and [2(THF)4] suitable for single crystal X-ray diffraction analyses were obtained via slow evaporation of THF solutions of [1] and [2], respectively. X-ray diffraction data for [1(THF)4] were collected on a RIGAKU Saturn 724 CCD system equipped with a VariMax Mo rotating-anode X-ray generator with Mo-Kα radiation (λ = 0.71075 Å), while those of [2(THF)4] were collected at 150 K on a RIGAKU Mercury system equipped with a Mo rotating-anode X-ray generator with Mo-Kα radiation (λ = 0.71075 Å). The obtained diffraction data of [1(THF)4] and [2(THF)4] were processed with the RIGAKU CrysAlisPro and CrystalClear programs, respectively. The structures of [1(THF)4] and [2(THF)4] were solved by direct methods (SIR-2004)49 and refined using the full-matrix least-squares technique F2 with SHELXL201450 equipped in the RIGAKU CrystalStructure 4.2.5. Non-hydrogen atoms were refined using anisotropic parameters and hydrogen atoms were located in the calculated positions and refined using riding models. Crystallographic data as well as data collection and refinement of [1(THF)4] and [2(THF)4] are summarized in Table 1 and can be obtained as CIF files from the Cambridge Crystallographic Data Centre (CCDC). Deposition numbers of [1(THF)4] and [2(THF)4] are CCDC 1861521 and 1861522, respectively.

Theoretical calculation

All density functional theory (DFT) calculations used in this study were performed with the dispersion corrected B3LYP-D2 functional method based on the Gaussian 09 program package.51 LANL08(f) and cc-pVDZ basis sets were used for the rhodium and other atoms, respectively. The solvent effect of THF was modelled by using the polarizable continuum model (PCM). Molecular geometries in the ground state were optimized in THF media and were checked by vibrational frequency calculations. Vertical singlet excitation is calculated by time-dependent DFT (TDDFT). Molecular orbitals (MOs) were drawn with Gaussview 5.0.

Conclusions

In this study, two dimer-of-dimers-type Rh4 complexes, in which two paddlewheel-type Rh2 units are connected by a dicarboxylate linker, were synthesized and characterized. Single crystal X-ray diffraction analyses and DFT calculations of [1(THF)4] and [2(THF)4] clarified that two –CO2 units and the phenyl ring of the BDC linker in [1(THF)4] are almost co-planar (ϕ (dihedral angle defined by –CO2 and phenyl planes) = 2.5°) due to the hydrogen-bonding between two –CO2 units and four hydrogen atoms of the phenyl ring of the BDC linker, whereas those of F4BDC in [2(THF)4] are largely inclined (ϕ = 78.3°) due to the electrostatic and steric repulsions between two –CO2 units and four fluorine groups on the phenyl ring of the F4BDC linker. Electrochemical analyses indicated that oxidation potentials of dimer-of-dimers-type Rh4 complexes were strongly influenced by the electronic states of dicarboxylate linkers as in the cases of dimer-of-dimers-type Mo4 complexes with terminal formamidinate ligands reported by the Cotton group.52 Interestingly, the dihedral angles between two –CO2 units and the phenyl ring of the dicarboxylate linker strongly affected the absorption features of dimer-of-dimers-type Rh4 complexes. A unique CT band, which is an excitation character from δ(Rh2)/π(dicarboxylate) → π*(dicarboxylate), has been observed due to the formation of the transition dipole moment when the dihedral angle (ϕ) is nearly zero. This result suggests that the CT transition character can be controlled by the appropriate selection of the dicarboxylate linker. The further study regarding the control of electronic structures of dimer-of-dimers-type Rh4 complexes is now in progress in our laboratory.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by JSPS KAKENHI Grant Numbers 15K17897, 15H00877, 16K05722, 17J11019, and 18H05166. N. Y. acknowledges JSPS research fellowships for young scientists.

References

  1. M. Fujita, M. Tominaga, A. Hori and B. Therrien, Acc. Chem. Res., 2005, 38, 371 CrossRef PubMed .
  2. D. J. Tranchemontagne, Z. Ni, M. O'Keeffe and O. M. Yaghi, Angew. Chem., Int. Ed., 2008, 47, 5136 CrossRef CAS PubMed .
  3. S. R. Seidel and P. J. Stang, Acc. Chem. Res., 2002, 35, 972 CrossRef CAS PubMed .
  4. A. N. Khlobystov, A. J. Blake, N. R. Champness, D. A. Lemenovskii, A. G. Majouga, N. V. Zyk and M. Schröder, Coord. Chem. Rev., 2001, 222, 152 CrossRef .
  5. W. L. Leong and J. J. Vittal, Chem. Rev., 2011, 111, 688 CrossRef CAS PubMed .
  6. S. Kitagawa, R. Kitaura and S. Noro, Angew. Chem., Int. Ed., 2004, 43, 2334 CrossRef CAS PubMed .
  7. H. Furukawa, K. E. Cordova, M. O'Keeffe and O. M. Yaghi, Science, 2013, 341, 974 CrossRef CAS PubMed .
  8. L. J. Murray, M. Dincă and J. R. Long, Chem. Soc. Rev., 2009, 38, 1294 RSC .
  9. A. Corma, H. García and F. X. Llabrés i Xamera, Chem. Rev., 2010, 110, 4606 CrossRef CAS PubMed .
  10. J. Y. Lee, O. K. Farha, J. Roberts, K. A. Scheidt, S. T. Nguyen and J. T. Hupp, Chem. Soc. Rev., 2009, 38, 1450 RSC .
  11. L. Alaerts, C. E. A. Kirschhock, M. Maes, M. A. van der Veen, V. Finsy, A. Depla, J. A. Martens, G. V. Baron, P. A. Jacobs, J. F. M. Denayer and D. E. De Vos, Angew. Chem., Int. Ed., 2007, 46, 4293 CrossRef CAS PubMed .
  12. A. Shigematsu, T. Yamada and H. Kitagawa, J. Am. Chem. Soc., 2011, 133, 2034 CrossRef CAS PubMed .
  13. L. Sun, M. G. Campbell and M. Dincă, Angew. Chem., Int. Ed., 2016, 55, 3566 CrossRef CAS PubMed .
  14. T. Uemura, N. Yanai and S. Kitagawa, Chem. Soc. Rev., 2009, 38, 1228 RSC .
  15. Y. Kataoka, K. Sato, Y. Miyazaki, K. Masuda, H. Tanaka, S. Naito and W. Mori, Energy Environ. Sci., 2009, 2, 397 RSC .
  16. F. A. Cotton, C. A. Murillo and R. A. Walton, Multiple Bonds between Metal Atoms, 3rd edn, Springer, New York, 2005 Search PubMed .
  17. S. S.-Y. Chui, S. M.-F. Lo, J. P. H. Charmant, A. G. Orpen and I. D. Williams, Science, 1999, 283, 1148 CrossRef CAS PubMed .
  18. W. Mori, F. Inoue, K. Yoshida, H. Nakayama, S. Takamizawa and M. Kishita, Chem. Lett., 1997, 26, 1219 CrossRef .
  19. M. Köberl, M. Cokoja, W. A. Herrmann and F. E. Kühn, Dalton Trans., 2011, 40, 6834 RSC .
  20. M. Mikuriya, D. Yoshioka and M. Handa, Coord. Chem. Rev., 2006, 250, 2194 CrossRef CAS .
  21. F. A. Cotton, L. M. Daniels, J. P. Donahue, C. Y. Liu and C. A. Murillo, Inorg. Chem., 2002, 41, 1354 CrossRef CAS PubMed .
  22. F. A. Cotton, Z. Li, C. Y. Liu and C. A. Murillo, Inorg. Chem., 2007, 46, 7840 CrossRef CAS PubMed .
  23. P. Angaridis, J. F. Berry, F. A. Cotton, P. Lei, C. Lin, C. A. Murillo and D. Villagrán, Inorg. Chem. Commun., 2004, 7, 9 CrossRef CAS .
  24. D. A. Boyd, P. E. Fanwick and T. Ren, Inorg. Chim. Acta, 2011, 370, 198 CrossRef CAS .
  25. M. H. Chisholm and A. M. Macintosh, Chem. Rev., 2005, 105, 2949 CrossRef CAS PubMed .
  26. D. Höhne, E. Herdtweck, A. Pöthig and F. E. Kühn, Inorg. Chim. Acta, 2015, 424, 210 CrossRef .
  27. B. E. Bursten, M. H. Chisholm and J. S. D'Acchioli, Inorg. Chem., 2005, 44, 5571 CrossRef CAS PubMed .
  28. M. H. Chisholm, F. Feil, C. M. Hadad and N. J. Patmore, J. Am. Chem. Soc., 2005, 127, 18150 CrossRef CAS PubMed .
  29. B. E. Bursten, M. H. Chisholm, R. J. H. Clark, S. Firth, C. M. Hadad, P. J. Wilson, P. M. Woodward and J. M. Zaleski, J. Am. Chem. Soc., 2002, 124, 12244 CrossRef CAS PubMed .
  30. M. H. Chisholm and N. J. Patmore, Acc. Chem. Res., 2007, 40, 19 CrossRef CAS PubMed .
  31. R. P. Bonar-Law, J. F. Bickley, C. Femoni and A. Steiner, J. Chem. Soc., Dalton Trans., 2000, 4244 RSC .
  32. W. Mori, H. Hoshino, Y. Nishimoto and S. Takamizawa, Chem. Lett., 1999, 28, 331 CrossRef .
  33. S. Takamizawa, E. Nakata, H. Yokoyama, K. Mochizuki and W. Mori, Angew. Chem., Int. Ed., 2003, 42, 4331 CrossRef CAS PubMed .
  34. F. A. Cotton, E. V. Dikarev and M. A. Petrukhina, Angew. Chem., Int. Ed., 2001, 40, 1521 CrossRef CAS PubMed .
  35. M. Handa, Y. Muraki, M. Mikuriya, H. Azumi and K. Kasuga, Bull. Chem. Soc. Jpn., 2002, 75, 1755 CrossRef CAS .
  36. H. Miyasaka, C. S. Campos-Fernández, J. R. Galán-Mascarós and K. R. Dunber, Inorg. Chem., 2000, 39, 5870 CrossRef CAS PubMed .
  37. Y. Kataoka, K. Sato, Y. Miyazaki, Y. Suzuki, H. Tanaka, Y. Kitagawa, T. Kawakami, M. Okumura and W. Mori, Chem. Lett., 2010, 39, 358 CrossRef CAS .
  38. Y. Kataoka, N. Yano, T. Shimodaira, Y.-N. Yan, M. Yamasaki, H. Tanaka, K. Omata, T. Kawamoto and M. Handa, Eur. J. Inorg. Chem., 2016, 17, 2810 CrossRef .
  39. Y. Kataoka, K. S. Kataoka, H. Murata, M. Handa, W. Mori and T. Kawamoto, Inorg. Chem. Commun., 2016, 68, 37 CrossRef CAS .
  40. N. Yano, Y. Kataoka, H. Tanaka, T. Kawamoto and M. Handa, ChemistrySelect, 2016, 1, 2571 CrossRef CAS .
  41. K. Uemura and M. Ebihara, Inorg. Chem., 2011, 50, 7919 CrossRef CAS PubMed .
  42. Z. Yang, M. Ebihara and T. Kawamura, Inorg. Chim. Acta, 2006, 359, 2465 CrossRef CAS .
  43. Y. Fuma, O. Miyashita, T. Kawamura and M. Ebihara, Dalton Trans., 2012, 41, 8242 RSC .
  44. P. Amo-Ochoa, R. Jiménez-Aparicio, M. R. Torres, F. A. Urbanos, A. Gallego and C. J. Gómez-García, Eur. J. Inorg. Chem., 2010, 31, 4924 CrossRef .
  45. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys., 2010, 132, 154104 CrossRef PubMed .
  46. Y. Kataoka, Y. Kitagawa, T. Saito, Y. Nakanishi, T. Matsui, K. Sato, Y. Miyazaki, T. Kawakami, M. Okumura, W. Mori and K. Yamaguchi, Bull. Chem. Soc. Jpn., 2010, 83, 1481 CrossRef CAS .
  47. Y. Kataoka, Y. Kitagawa, T. Saito, Y. Nakanishi, K. Sato, Y. Miyazaki, T. Kawakami, M. Okumura, W. Mori and K. Yamaguchi, Supramol. Chem., 2011, 23, 329 CrossRef CAS .
  48. F. A. Cotton and T. R. Felthouse, Inorg. Chem., 1980, 19, 323 CrossRef CAS .
  49. M. C. Burla, R. Caliandro, M. Camalli, B. Carrozzini, G. L. Cascarano, L. De Caro, C. Giacovazzo, G. Polidori and R. Spagna, J. Appl. Crystallogr., 2005, 38, 381 CrossRef CAS .
  50. G. M. Sheldrick, Acta Crystallogr., Sect. C: Struct. Chem., 2015, 71, 3 Search PubMed .
  51. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, X. Li, M. Caricato, A. Marenich, J. Bloino, B. G. Janesko, R. Gomperts, B. Mennucci, H. P. Hratchian, J. V. Ortiz, A. F. Izmaylov, J. L. Sonnenberg, D. Williams-Young, F. Ding, F. Lipparini, F. Egidi, J. Goings, B. Peng, A. Petrone, T. Henderson, D. Ranasinghe, V. G. Zakrzewski, J. Gao, N. Rega, G. Zheng, W. Liang, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, K. Throssell, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, T. Keith, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, J. M. Millam, M. Klene, C. Adamo, R. Cammi, J. W. Ochterski, R. L. Martin, K. Morokuma, O. Farkas, J. B. Foresman and D. J. Fox, Gaussian 09, Revision C.02, Gaussian, Inc., Wallingford CT, 2016 Search PubMed .
  52. F. A. Cotton, J. P. Donahue, C. Lin and C. A. Murillo, Inorg. Chem., 2001, 40, 1234 CrossRef CAS PubMed .

Footnote

Electronic supplementary information (ESI) available. CCDC 1861521 and 1861522. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c8dt03293a

This journal is © The Royal Society of Chemistry 2018