Christian
Merten
and
Yunjie
Xu
*
Department of Chemistry, University of Alberta, Edmonton, Alberta T6G2G2, Canada. E-mail: yunjie.xu@ualberta.ca; Fax: +1 780 492 8231
First published on 16th May 2013
Vibrational circular dichroism (VCD) spectroscopy is a powerful tool to characterize absolute configurations and conformations of chiral organometallic systems. Such characterizations rely on the corresponding density functional theory (DFT) spectral simulations which may become very time consuming and sometimes even impossible for systems with increasing complexity. Systematic studies on small model compounds can lead to empirical structure–spectra relationships. The present work continues the systematic investigations of transition metal complexes of chiral trans-1,2-diamino-cyclohexane (chxn). VA and VCDspectra of the mixed-ligand complexes [Cu(chxn)2L]2+ with L being either ethylene diamine (en) or N,N′-dimethylethylene diamine (dmen) are measured. The comparison of the experimental spectra of the mixed-ligand complexes with those of the distorted octahedral complex [Cu(chxn)3]2+ reveals that the VA and VCD patterns below 1500 cm−1 of the three complexes are not significantly affected by the nature of the third ligand, while the VCD pattern of the NH2-bending modes in the 1500 to 1800 cm−1 region features some characteristic changes. Comparison with the corresponding DFT spectral calculations shows that these spectral differences are related to the relative abundance of the Δ- and Λ-configurations at the metal ion. In addition, the results of this study highlight that the VCD pattern of the NH2-bending modes is characteristic of the coordination number and the configuration of the metal center.
Vibrational circular dichroism (VCD) spectroscopy, the vibrational counterpart of the well-known electronic circular dichroism spectroscopy, has been successfully used to characterize structures and chirality of a range of metal complexes. VCD studies of chiral metal complexes in general,7–11 and of actual catalysts12–14 and larger MOFs have been reported.15–17 It is recognized that for larger molecules or complexes involving lanthanide ions, the corresponding theoretical spectral simulations can be extremely time consuming or even impossible at present time. Hence, it is desirable to study derivatives of smaller metal complexes systematically in order to derive some applicable structure–spectra relationships. In this context, Sato et al. investigated in detail the VCD spectra of tris(acetyl acetonate) complexes with different transition metal ions,18 of their mixed ligand complexes,19,20 and of larger multi-core complexes.21–23
Recently, we carried out a systematic study on the VCD spectra of transition metal complexes of copper(II) and nickel(II) with the chiral diamineligands, trans-1,2-diamino cyclohexane (chxn).24 It was shown that the obtained VCD spectra feature characteristic differences between the tris-complexes of Ni(II) and Cu(II), and between the tris- and bis-complexes of Cu(II). While the substitution of the metal ion caused intensity changes of specific bands, the change of the coordination number from 3 to 2 resulted in a sign inversion of the VCD couplet of the NH2 bending vibration at ∼1600 cm−1 and other changes in the VCD pattern.
In the present paper, we examine the associated VCD signatures when one chxn in [Cu(chxn)3](ClO4)2 is replaced with ethylene diamine (en) or N,N′-dimethylethylene diamine (dmen) (Fig. 1). Both new ligands are achiral and are expected to contribute little to the experimental VA and VCDspectra on their own due to their relatively weak absorbance bands. This allows one to focus on the effects of the coordination number and the preference of Δ- and Λ-configurations of the metal ion on the VCD spectra upon ligand substitution. Hence, this study presents a first generalization of the previous findings on the relation between the observed experimental VCDspectra, in particular in the NH2-bending region, and the configurations of the metal center. Interpretation of the experimental spectra is supported by density functional theory (DFT) based spectral calculations. For simplicity, the complexes with the (R,R)-enantiomer of chxn are always used when comparing experimental and calculated spectra.
![]() | ||
Fig. 1 Chemical structure of the Δ(λ)-isomer of [Cu{(R,R)-chxn}2(L)]2+. The substituted third (R,R)-chxn ligand is indicated in red dashed lines (R = H). For the mixed-ligand complexes with L = en (R = H) and L = dmen (R = CH3), the dashed parts are not present. |
The gas phase ΔE and ΔG based population weighted VA and VCDspectra are quite similar, except for the intensity variations of the NH2-bending modes (cf. Fig. S2†). The same can be said about the corresponding IEFPCM spectra. Fig. 2 shows a comparison of the experimental VA and VCDspectra of [Cu(chxn)3](ClO4)2 with the ΔEGP and ΔEPCM weighted spectra in the gas phase and with IEFPCM, respectively. First, comparing the experimental and the gas phase spectra, the main features of the experimental spectra are found to be reproduced theoretically. On the other hand, some deviations in the VCD spectral pattern in the 1375–1300 cm−1 region are observed. In addition, the medium intensity VCD band observed at 1200 cm−1 is predicted to have almost no VCD intensity at around 1178 cm−1. The corresponding VA band observed experimentally is predicted to be very weak. Similar deviations were noted with the gas phase calculations for [Ni(chxn)3]2+ in the previous study.24 Second, comparing the gas phase with the IEFPCM spectra, the VA bands and the VCD couplet of the NH2-bending modes at 1600 cm−1 are predicted to be much more intense in the IEFPCM calculations. Considering the experimental intensity ratio of the large (+) and the small (−) components of the couplet being ∼2, the IEFPCM calculation describes the VCD spectral features of the NH2-bending modes significantly better than the gas phase one. The experimental VCD pattern in the range from 1380–1300 cm−1 is also better described by the IEFPCM calculation, although the negative VCD shoulder band observed at 1380 cm−1 is predicted as a positive band at 1371 cm−1. Furthermore, both VA and VCD bands observed at 1200 cm−1 are reasonably predicted with IEFPCM. Overall, the IEFPCM based spectra agree qualitatively better with the experiment than the gas phase ones. The better agreement in the region below 1500 cm−1 is due to the somewhat different VCD patterns generated in the gas phase and in the IEFPCM solution, rather than the changes in the relative population weights of different isomers. This statement is supported by the comparison of the single-isomer spectra (Fig. S3†) generated by the gas phase and IEFPCM calculations.
![]() | ||
Fig. 2 Comparison of the experimental VA and VCDspectra of [Cu(chxn)3](ClO4)2 with the calculated spectra. Both ΔEGP (blue) and ΔEPCM (orange) population weighted spectra are shown. |
![]() | ||
Fig. 3 Comparison of the experimental VA and VCDspectra of [Cu(chxn)2(en)]2+ with the calculated spectra. Both ΔEGP (blue) and ΔEPCM (orange) population weighted spectra are shown. |
In the tris(chxn) complexes, the structure and chirality of the chxn ligand hold the NCCN torsional angle in a fixed conformation. The NCCN angle is negative for the (R,R)-enantiomer of chxn, which is also referred to as the λ-conformation. In [Cu(chxn)2(en)]2+, the en ligand is flexible and can adopt in general either the δ- or λ-conformation. This generates four possible isomers, Δ(δ-en), Δ(λ-en), Λ(δ-en), and Λ(λ-en). Furthermore, the new ligand en can bind in two different ways to the copper center: it can either occupy two equatorial positions leaving the axial positions to the nitrogen atoms of the two chxn groups, or it can place one of its nitrogen atoms on an axial position. While the first set of conformers all belong to the C2 point group, the latter set is of C1-symmetry. Hence, for the [Cu(chxn)2(en)]2+ complex, a total of 8 conformations has to be considered. The relative energy differences ΔE for the eight conformations calculated in the gas phase and using the IEFPCM of DMSO are summarized in Table 1, whereas the related ΔG values and populations can be found in Table S2 of the ESI.† The calculated relative energies indicate a clear preference for the C1-symmetric structures with the smaller and more flexible en ligand occupying an axial position. The structures of the four C1-conformers with en on the axial positions are shown in Fig. 4. The conformation of the en ligand does not significantly affect the geometry of the remaining part of the complex, as can be seen when one superimposes the calculated structures in Fig. S5 of the ESI.†
![]() | ||
Fig. 4 Optimized geometries of the four C1(en-ax) isomers of [Cu(chxn)2(en)]2+. The axial bonds are highlighted in red. The associated Cartesian coordinates are provided in the ESI.† |
Conf. | en | ΔEGP (kcal mol−1) | pop-ΔEGP (%) | ΔEPCM (kcal mol−1) | pop-ΔEPCM (%) |
---|---|---|---|---|---|
Λ(δ-en) C1 | ax | 0.00 | 31.05 | 0.27 | 17.14 |
Λ(δ-en) C2 | eq | 1.74 | 1.65 | 1.52 | 2.06 |
Δ(δ-en) C1 | ax | 0.06 | 28.25 | 0.00 | 26.86 |
Δ(δ-en) C2 | eq | 1.74 | 1.64 | 0.37 | 14.37 |
Δ(λ-en) C1 | ax | 0.30 | 18.79 | 0.15 | 20.89 |
Δ(λ-en) C2 | eq | 2.12 | 0.86 | 0.99 | 5.09 |
Λ(λ-en) C1 | ax | 0.37 | 16.63 | 0.53 | 11.05 |
Λ(λ-en) C2 | eq | 1.96 | 1.13 | 1.39 | 2.55 |
Since the ΔG (not shown) and ΔE weighted spectra are essentially identical, only the ΔEGP and ΔEPCM population weighted VA and VCDspectra are compared to the experimental spectra in Fig. 3. Both sets of simulations predict the main experimental VCD spectral features such as the couplet of the NH2-bending mode correctly. It is noted that the gas phase VCD pattern predicted for the mixed-ligand complex in the 1400–1200 cm−1 region is essentially the same as for [Cu(chxn)3]2+, although with an overall reduction in intensity. The same observation can be made about the IEFPCM spectra. It is therefore not surprising that the same discrepancies between experimental and theoretical spectra noted for the tris(chxn) complex persist for the mixed-ligand complex here.
![]() | ||
Fig. 5 Comparison of the experimental VA and VCDspectra of [Cu(chxn)2(dmen)]2+ with the calculated spectra. Both ΔEGP (blue) and ΔGGP (orange) population weighted spectra are shown. |
For the calculations on [Cu(chxn)2(dmen)]2+, the same eight diastereomeric configurations Δ(δ-dmen), Δ(λ-dmen), Λ(δ-dmen), and Λ(λ-dmen) with dmen being in an axial or equatorial position have to be considered. Additionally, the introduction of the methyl groups on each of the amino groups further increases the number of possible conformations for each of these eight configurations. By systematically substituting hydrogen atoms of the amino groups with methyl groups, four different conformations for each diastereomeric configuration can be generated. Two of them have the methyl groups bound in trans conformations, two in a cis conformation (cf. Fig. 6). When dmen binds to the metal center in the equatorial positions, only one cis conformer is obtained due to symmetry. Hence, in total, 28 conformers of [Cu(chxn)2(dmen)]2+ can be generated. Since no noticeable differences have been observed for the gas phase and the IEFPCM simulated VA and VCDspectra for the two complexes discussed above, only the gas phase calculations have been carried out for this third complex. The relative energies and Gibbs free energies, ΔEGP and ΔGGP, as well as the corresponding Boltzmann populations are summarized in Table 2.
![]() | ||
Fig. 6 Three binding configurations of the methyl groups of dmen in [Cu(chxn)2(dmen)]2+ shown along the N–Cu–N plane of the dmen-ligand exemplified for the C2-symmetric Δ(λ-dmen) isomer. The axial Cu–N bonds are highlighted in red and the nitrogen atoms of dmen are placed in the equatorial positions of the octahedral ligand sphere. The methyl groups of dmen can either be equatorial or axial with respect to the N–Cu–N plane (trans-eq and trans-ax conformer), or they are located in a cis type conformation. |
Conf. | dmen | ΔE | ΔG | pop-ΔE | pop-ΔG | |
---|---|---|---|---|---|---|
Λ(δ-dmen) | trans-eq | ax | 0.00 | 0.00 | 29.02 | 28.84 |
Λ(δ-dmen) | trans-ax | ax | 3.21 | 3.56 | 0.13 | 0.07 |
Λ(δ-dmen) | cis1 | ax | 2.26 | 2.55 | 0.63 | 0.39 |
Λ(δ-dmen) | cis2 | ax | 1.13 | 1.37 | 4.33 | 2.86 |
Λ(δ-dmen) | trans-ax | eq | 3.86 | 4.79 | 0.04 | 0.01 |
Λ(δ-dmen) | trans-eq | eq | 0.90 | 1.48 | 6.31 | 2.35 |
Λ(δ-dmen) | cis | eq | 1.84 | 1.96 | 1.29 | 1.06 |
Δ(λ-dmen) | trans-eq | ax | 0.39 | 0.05 | 14.95 | 26.30 |
Δ(λ-dmen) | trans-ax | ax | 2.63 | 2.72 | 0.34 | 0.29 |
Δ(λ-dmen) | cis1 | ax | 1.32 | 1.23 | 3.13 | 3.59 |
Δ(λ-dmen) | cis2 | ax | 1.77 | 1.61 | 1.47 | 1.91 |
Δ(λ-dmen) | trans-ax | eq | 4.20 | 4.75 | 0.02 | 0.01 |
Δ(λ-dmen) | trans-eq | eq | 0.77 | 0.79 | 7.96 | 7.56 |
Δ(λ-dmen) | cis | eq | 2.28 | 2.59 | 0.62 | 0.36 |
Δ(δ-dmen) | trans-ax | ax | 0.97 | 1.23 | 5.63 | 3.64 |
Δ(δ-dmen) | trans-eq | ax | 1.56 | 1.44 | 2.09 | 2.51 |
Δ(δ-dmen) | cis1 | ax | 1.16 | 1.11 | 4.09 | 4.44 |
Δ(δ-dmen) | cis2 | ax | 1.18 | 0.95 | 3.93 | 5.82 |
Δ(δ-dmen) | trans-ax | eq | 1.99 | 2.31 | 1.00 | 0.58 |
Δ(δ-dmen) | trans-eq | eq | 2.10 | 2.80 | 0.84 | 0.26 |
Δ(δ-dmen) | cis | eq | 2.13 | 2.65 | 0.79 | 0.33 |
Λ(λ-dmen) | trans-ax | ax | 1.60 | 1.75 | 1.94 | 1.49 |
Λ(λ-dmen) | trans-eq | ax | 1.70 | 1.75 | 1.65 | 1.51 |
Λ(λ-dmen) | cis1 | ax | 2.20 | 2.60 | 0.71 | 0.36 |
Λ(λ-dmen) | cis2 | ax | 1.48 | 1.84 | 2.37 | 1.29 |
Λ(λ-dmen) | trans-ax | eq | 1.42 | 1.71 | 2.65 | 1.61 |
Λ(λ-dmen) | trans-eq | eq | 1.92 | 2.77 | 1.13 | 0.27 |
Λ(λ-dmen) | cis | eq | 2.04 | 2.71 | 0.93 | 0.30 |
Due to the large number of conformers, it is more complicated to evaluate which conformations are preferably adopted. Nevertheless, the Boltzmann factors tabulated in Table 2 clearly show that the trans-eq forms of the Λ(δ-dmen) and Δ(λ-dmen) isomers with dmen occupying the axial position in the complex have by far the largest contribution to the conformational equilibrium of [Cu(chxn)2(dmen)]2+. The preference of these two configurations could be understood by considering steric hindrance caused by the two methyl groups. Such repulsive energy is at the lowest in the trans-eq conformation when more space is available for the methyl groups (cf. Fig. 6).
The theoretical spectra shown in Fig. 5 are obtained by using the population factors listed in Table 2. No significant differences between the ΔEGP and ΔGGP weighted spectra of [Cu(chxn)2(dmen)]2+ are observed. Despite the existence of a large number of relevant conformers, reasonably good agreement between the simulated and experimental spectra has been achieved for this complex, as for the other two complexes discussed above. It is worth pointing out that the significantly lower intensity of the negative part of the couplet of the NH2-bending modes observed experimentally is well captured by the simulation. The considerable intensity decrease observed for the VCD bands below 1300 cm−1 is also well reproduced by the current simulation.
![]() | ||
Fig. 7 Comparison of the single isomer spectra of Cu(chxn)32+ (top trace) and the en-ax (black) and en-eq (orange) isomers of Cu(chxn)2(en)2+. |
The NH2-bending modes centred at 1600 cm−1 are highly sensitive to the Δ- and Λ-configurations of the metal ion. The third ligand substitution introduces additional sensitivity to the δ- and λ-configurations of the en ligand, especially when the metal ion is in the Λ-configuration and, to a lesser degree, to whether the third ligand occupies the axial or equatorial positions. Specifically, the Δ-isomers all show a −/+ pattern while the Λ-isomers show an all-positive pattern in the case of the Λ(λ-en) isomer and a +/− pattern for the Λ(δ-en) isomer. The next region from 1500 cm−1 to 1200 cm−1 shows only sensitivity to whether the third ligand, en, is in the δ- or λ-configuration. This means that the VCD features in this region remain essentially unchanged for Δ(λ-chxn) and Λ(λ-chxn), and for Δ(λ-en) and Λ(λ-en), with en in either the axial or equatorial position, whereas those associated with (δ-en) appear differently. In the region below 1300 cm−1, which is covered by strong bands from the solvent and the anion in the experiment, the VCD pattern is sensitive to the configuration of the metal ion and the conformation of the third ligand, as well as the binding site occupied by the third ligand. Closer examination in this region reveals another characteristic feature for the Δ-isomers with a more intense negative band at 1150 cm−1 than the Λ-isomers. Similar conclusions can also be drawn from the gas phase spectra of the above complexes (Fig. S6†).
Since these metal complexes contain multiple relevant conformers, the relative abundance of them is also important for the comparison between theoretical and experimental spectra. Table 3 summarizes the calculated populations of the Δ- and Λ-isomers of the three investigated complexes based on the ΔE and ΔG values. While the ΔE based gas phase populations suggest that the Δ- and Λ-forms of the complexes studied are of more or less the same stability, the ΔGGP populations, however, show a switch of preference from the Δ- to the Λ-form upon third ligand substitution. The ΔE and ΔG values obtained with the IEFPCM calculations show a clear preference for the Δ- over the Λ-form for the two complexes calculated. This switch of conformational preferences depending on the choice of ΔE or ΔG for its calculation has been observed before.29,30
pop-ΔEGP | pop-ΔGGP | pop-ΔEPCM | pop-ΔGPCM | |||||
---|---|---|---|---|---|---|---|---|
Σ(Δ) | Σ(Λ) | Σ(Δ) | Σ(Λ) | Σ(Δ) | Σ(Λ) | Σ(Δ) | Σ(Λ) | |
a IEFPCM calculations were not carried out for [Cu(chxn)2dmen]2+. See the main text for the discussion. | ||||||||
[CuL3]2+ | 44.7 | 55.3 | 33.4 | 66.6 | 68.5 | 31.5 | 88.7 | 11.3 |
[CuL2en]2+ | 49.5 | 50.5 | 62.0 | 38.0 | 67.2 | 32.8 | 84.6 | 15.4 |
[CuL2dmen]2+ | 46.9 | 53.1 | 57.6 | 42.4 | —a | — | — | — |
As pointed out before, the spectral patterns of the experimental VA and VCDspectra of all three complexes appear fairly similar. The main differences observed are the overall VCD intensity for each complex and the relative intensity of the couplet of the NH2 bending modes (see Fig. S4†). In the case of the [Cu(chxn)2(dmen)]2+ complex, changes are also noted for the VA intensities of some bands. The first and the last observations can be directly related to the number of chiral chxn ligands in the complex and to the substitution of two NH2-groups by NHCH3-groups in dmen, respectively. The interpretation for the relative intensity of the couplet of the NH2 bending modes is somewhat more involved. As discussed before, while Δ-conformers show −/+ couplets with similar strong intensity, Λ-conformers exhibit roughly +/− couplets with lower intensity. Hence, the intensity ratio of the two components of the couplet is a characteristic for the relative population of the Δ- and Λ-conformers. For example, by varying the ratio of the Δ- and Λ-forms of the presented theoretical spectra, the NH2-bending modes of [Cu(chxn)3]2+ would become all positive when one assumes a conformational distribution of 25% Δ- and 75% Λ-conformers. Generally, the Δ-form can be regarded as the dominating form in all three investigated complexes since the −/+ pattern is preserved.
As reported previously, the change of the coordination number from 3 to 2 resulted in a sign inversion of the VCD couplet of the NH2 bending vibration modes at ∼1600 cm−1 from the tris- to bis-complexes of [Cu(chxn)2,3]2+.24 The current comparative study with both the homoleptic tris-complex [Cu(chxn)3]2+ and the mixed-ligand complexes, [Cu(chxn)2(en)]2+ and [Cu(chxn)2(dmen)]2+, further confirms that such sign change is characteristic of the coordination number of all ligands, rather than the total number of chiral ligands.
Considering the results of the calculations, several conclusions can be drawn. First of all, the calculated spectra agree reasonably well with the corresponding experimental data. Second, the signs of the VCD couplet of the NH2-bending modes are characteristic of the coordination number of the ligands around the metal ion for the homo- and heteroleptic complexes studied. Third, the Δ- and Λ-configurations each feature a distinct VCD pattern at the NH2-bending modes, and the ratio of the (−) and (+) components of the doublet is related to the relative abundance of the Δ- and Λ-forms. In all three investigated complexes, the dominant configuration of the metal center is Δ. Fourth, the cause of the less satisfying agreement between experiment and theory in the 1400–1300 cm−1 region is unclear. Further studies of the related deuterated species and the mixed-ligand complexes of [Cu(chxn)2]2+ with larger achiral and also chiral diamineligands are planned in order to shed light on this disagreement. A good understanding of the cause is crucial for a reliable VCD spectral interpretation of chiral metal complexes and even larger metal organic networks.
Footnote |
† Electronic supplementary information (ESI) available: Selected structural parameters of [Cu(chxn)3]2+, various comparative plots, all calculated relative energy differences and single-conformer spectra, as well as the Cartesian coordinates for [Cu(chxn)3]2+. See DOI: 10.1039/c3dt50459j |
This journal is © The Royal Society of Chemistry 2013 |