Yusen
Qiao
ab,
Haolin
Yin
a,
Liane M.
Moreau
b,
Rulin
Feng
c,
Robert F.
Higgins
a,
Brian C.
Manor
a,
Patrick J.
Carroll
a,
Corwin H.
Booth
b,
Jochen
Autschbach
c and
Eric J.
Schelter
*a
aP. Roy and Diana T. Vagelos Laboratories, Department of Chemistry, University of Pennsylvania, 231 South 34 Street, Philadelphia, Pennsylvania 19104, USA. E-mail: schelter@sas.upenn.edu
bChemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA
cDepartment of Chemistry, University at Buffalo, State University of New York, Buffalo, New York 14260, USA
First published on 9th December 2020
A series of cerium(IV) mixed-ligand guanidinate–amide complexes, {[(Me3Si)2NC(NiPr)2]xCeIV[N(SiMe3)2]3−x}+ (x = 0–3), was prepared by chemical oxidation of the corresponding cerium(III) complexes, where x = 1 and 2 represent novel complexes. The Ce(IV) complexes exhibited a range of intense colors, including red, black, cyan, and green. Notably, increasing the number of the guanidinate ligands from zero to three resulted in significant redshift of the absorption bands from 503 nm (2.48 eV) to 785 nm (1.58 eV) in THF. X-ray absorption near edge structure (XANES) spectra indicated increasing f occupancy (nf) with more guanidinate ligands, and revealed the multiconfigurational ground states for all Ce(IV) complexes. Cyclic voltammetry experiments demonstrated less stabilization of the Ce(IV) oxidation state with more guanidinate ligands. Moreover, the Ce(IV) tris(guanidinate) complex exhibited temperature independent paramagnetism (TIP) arising from the small energy gap between the ground- and excited states with considerable magnetic moments. Computational analysis suggested that the origin of the low energy absorption bands was a charge transfer between guanidinate π orbitals that were close in energy to the unoccupied Ce 4f orbitals. However, the incorporation of sterically hindered guanidinate ligands inhibited optimal overlaps between Ce 5d and ligand N 2p orbitals. As a result, there was an overall decrease of ligand-to-metal donation and a less stabilized Ce(IV) oxidation state, while at the same time, more of the donated electron density ended up in the 4f shell. The results indicate that incorporating guanidinate ligands into Ce(IV) complexes gives rise to intense charge transfer bands and noteworthy electronic structures, providing insights into the stabilization of tetravalent lanthanide oxidation states.
Unlike other lanthanides, the trivalent and tetravalent oxidation states are readily accessible in molecular cerium complexes.5 The 4f → 4f transitions for the Ce3+ ion fall in the infrared region, hence the colors (and the emission colors) of cerium(III) complexes are affected by 4f → 5d absorptions.1,5–12 The 4f electron, upon absorption of light, becomes chemically accessible through promotion to 5d orbital sets. As a result, unprecedented luminescent properties as well as photoredox catalysis involving cerium(III) complexes as photosensitizers can be achieved by judicious ligand selection.8,12–19
With respect to cerium(IV) complexes, their intense colors are primarily due to parity-allowed ligand-to-metal charge transfer (LMCT) transitions of the Ce4+ cation.20–24 The electronic structures of the cerium(IV) complexes are sensitive to the ligand environment.25 Since the electron is transferred into a metal-based orbital, the LMCT can result in the reduction of the metal cation. For example, in the presence of excess NEt4Cl, irradiation at the absorption region of [CeIVCl6]2− leads to the formation of the hexachlorocerate(III) anion ([CeIIICl6]3−).26 Taking advantage of the strong excited-state reduction power, fast electron-transfer kinetics of [CeIIICl6]2− and photoreduction of [CeIVCl6]2−, photoinduced dehalogenation and borylation of aryl chlorides by [CeIIICl6]3− have been accomplished.16,27
In addition to the reactivity of such LMCT excited states, we hypothesize that LMCT transitions can be used to achieve unprecedented electronic structures of cerium(IV) complexes. Most of the known cerium(IV) complexes display absorption bands in two major regions of the spectrum. The first one is ∼300–400 nm, where Ce[NP(pip)3]4 (pip = piperidinyl, 335 nm),23 Ce[η2-ON(tBu)(2-OMe-5-tBu-C6H3)]4 (359 nm),28 Ce(tmhd)4 (tmhd = 2,2,6,6-tetramethyl-3,5-heptanedionato, 372 nm),6,29 [NEt4]2[CeCl6] (375 nm),26,30,31 and Ce(IV) oxo complexes [Li(2,2,2-cryptand)]{OCe[N(SiMe3)2]3} (306 nm),15 and [Rb(OCe(TriNOx))] (TriNOx3− = {(2-tBuNO)C6H4CH2}3N]3−, 380 nm),32 fall into this category. The second one is ∼500–650 nm, which has Ce{N[CH2CH2N(SiMe2tBu)]3}I (ca. 500 nm),33 CeX[N(SiMe3)2]3 (X = F, Cl, Br, I, ca. 500 nm),22 Ce(IV) imido complexes [M(solv)x][CeN(3,5-(CF3)2C6H3)(TriNOx)] (M = Li, K, Rb, Cs, solv = TMEDA, THF, Et2O, or DME, ca. 510 nm),32 Ce[2-(tBuNO)py]4 ([2-(tBuNO)py]− = N-tert-butyl-N-2-pyridylnitroxide, 524 nm),34 Ce(COT)2 (COT = cyclooctatetraenyl, 569 nm),35 and Ce(Odpp)4 (dpp = 2,6-(C6H5)2-C6H3, 624 nm)36 in this category. For the complexes in the second category, noteworthy electronic structures have been observed. For example, the LMCT band at 569 nm of Ce[2-(tBuNO)py]4 indicates mixing of vacant Ce 4f and 5d orbitals with filled ligand-based orbitals, that provides a strong stabilization of the CeIV oxidation state in this complex.34 Moreover, Ce(COT)2 has a multiconfigurational ground-state and displays temperature-independent paramagnetism.37,38 We stress that it has been shown that the degree of multiconfigurational character determined from a calculation depends on rotations of the natural orbitals in the active space without any effect on the calculated energy.39 This was used to indicate partial equivalence39 between multiconfigurational character and single configurational character with significant covalence in Ce(COT)2,40 although in that system a purely single configurational calculation was not obtainable. With that fact in mind, metal–ligand covalency in the cerium(IV) imido complexes is more significant than that of the thorium(IV) analogs.32,41,42 In terms of other lanthanide compounds with low energy LMCT bands, Cp*2Yb(bipy) (Cp* = pentamethylcyclopentadienyl, bipy = 2,2′-bipyridine), which has a low-lying LMCT band at ca. 2105 nm, shows significant metal–ligand orbital overlap and strong exchange coupling (2J = −0.11 eV).43 A molecular Tb(IV) compound, Tb(NP(1,2-bis-tBu-diamidoethane)(NEt2))4, with a LMCT band at 575 nm demonstrates a small HOMO–LUMO gap (1.28 eV), covalent metal–ligand bonds, and a multiconfigurational ground state.44 Remarkable lanthanide-ligand orbital mixing has also been spectroscopically determined in tetravalent lanthanide oxides, LnO2 (Ln = Ce, Pr, and Tb),45 and Pr(IV) and Tb(IV) siloxide compounds.46–48 Inspired by these findings, we are interested in developing and studying cerium(IV) complexes that show red-shifted absorption bands (>650 nm). We postulate that cerium(IV) complexes with such low energy LMCT bands would have strong metal–ligand interactions, leading to unusual electronic structures. Herein, we report that applying the guanidinate ligands to the Ce(IV) cation gives rise to intense colors and anomalous electronic structures, characteristic of 4f and 5d covalency. These results promote the understanding of bonding and electronic structures for lanthanide complexes and inform design strategies for manifesting and exploring multi-configurational electronic structures.
Scheme 1 (a) Syntheses of 1-Cl, 1-Cl−, 2-Cl, and 2-Cl−. (b) Syntheses of [3+][BArF4] and [4+][BArF4]. The colors for each of the compound labels were selected accurately represent their actual colors (see Fig. 2 inset, vide infra). |
Treatment of the bis(guanidinate) cerium(III) complex [(Me3Si)2NC(NiPr)2]2CeIII[N(SiMe3)2] (3) with [Cp2Fe][BArF4] in CH2Cl2 followed by recrystallization from CH2Cl2/toluene layering, afforded dark cyan crystals of complex {[(Me3Si)2NC(NiPr)2]2Ce[N(SiMe3)2]}[BArF4] ([3+][BArF4]) in 51% yield. Single crystal X-ray diffraction confirmed the bis(guanidinate) structural motif for [3+][BArF4] (Fig. 1). The closed shell nature of complex [3+][BArF4] was evident from 1H NMR, 13C NMR, and 19F NMR spectra collected in CD2Cl2 (Fig. S1–S3†). In contrast, our attempts to prepare the corresponding cationic cerium(IV) complex of 1 and 2 using oxidants with non-coordinating anions (e.g., [Cp2Fe][BArF4]) all resulted in intractable mixtures. Therefore, the previously reported CeIV–Cl complexes, CeIVCl[N(SiMe3)2]3 (1-Cl)22,49 and [(Me3Si)2NC(NiPr)2]CeIVCl[N(SiMe3)2]2 (2-Cl)13 were used for comparison in this study. Complexes 1-Cl and 2-Cl were prepared by oxidation of 1 and 2 with Ph3CCl in 80% and 72% yields, respectively. The corresponding one electron reduced products of 1-Cl and 2-Cl, 1-Cl− (ref. 14) and 2-Cl−, respectively, were also prepared by reacting 1 or 2 with NEt4Cl in dichloromethane. Attmepts to synthesize 3-Cl− and 4-Cl− by reacting 3 and 4 with NEt4Cl in CH2Cl2, respectively were unsuccessful as evidenced by 1H NMR spectroscopy in CD2Cl2. The average Ce–Cl bond length of 2-Cl− (2.808(3) Å) was longer than that of 1-Cl− (2.7611(9) Å), suggesting more steric bulk around the Ce3+ cation in 2-Cl− than that in 1-Cl− due to only one guanidinate ligand in the coordination sphere. However, when we attempted to synthesize the corresponding CeIV–Cl complexes for 3 and 4, no reaction was observed when 3 and 4 were treated with Ph3CCl. This result can be attributed to the large steric bulk of ligands around the Ce(III) cation in the bis- and tris(guanidinate) complexes (3 and 4), that prevent oxidation of the Ce(III) cation by an inner-sphere electron transfer pathway.14 Even using a less sterically demanding inner-sphere oxidant, C2Cl6, resulted in no color change when added to 3 and 4 where only starting material was observed by 1H NMR spectroscopy in C6D6.
Complex | E (eV) |
---|---|
Ce[NP(pip)3]4 | 3.70 |
Ce[N(SiHMe2)2]4 | 2.80 |
Ce{N[CH2CH2N(SiMe2tBu)]3}I | 2.50 |
1-Cl | 2.48 |
[Li(TMEDA)][CeN(3,5-(CF3)2C6H3)(TriNOx)] | 2.38 |
Ce[2-(tBuNO)py]4 | 2.37 |
CeI[N(SiMe3)2]3 | 2.34 |
Ce(NiPr2)4 | 2.20 |
Ce(COT)2 | 2.18 |
2-Cl | 1.86 |
[3+][BArF4] | 1.81 |
[4+][BArF4] | 1.58 |
Fig. 3 Normalized X-ray absorption spectra at the Ce L3 absorption edge of 1-Cl (red), 2-Cl (black), [3+][BArF4] (cyan), and [4+][BArF4] (green). |
Fig. 4 Correlation of calculated LUMO energy versus experimental E1/2 of 1-Cl, 2-Cl, [3+][BArF4], and [4+][BArF4] (red circles). The previous LUMO energy correlation (blue dashed line) was performed on a series of Ce(IV) complexes with a range of ligand fields and coordination geometries (blue circles).58 Data for multiconfigurational Ce(IV) porphyrinate and phthalocyanate complexes are shown as light green circles. |
The χT products were measured from 300 to 2 K at an applied field of 0.5 T for 4 and [4+][BArF4] (Fig. 5). At room temperature, the χT value of 4 was 0.68 emu K mol−1, consistent with a Ce(III) cation in a 2F5/2 ground state (Fig. 5, left).38,61–64 Upon cooling, the χT product decreased steadily to 0.33 emu K mol−1 at 10 K. This decrease in moments at lower temperature was attributed to the depopulation of crystal-field energy levels created by perturbations of the J = 5/2 manifold. Complex [4+][BArF4] showed a different magnetic susceptibility response, where a linear decrease from high (300 K) to low temperature (∼50 K) in the χT product was observed. After removing diamagnetic contribution using Pascal's constants,65 the paramagnetic contribution from a residual impurity of 4 (1.2%, J = 5/2) was removed through least squares fitting over a temperature range of 50–300 K (see the ESI† for details). The resulting susceptibility plots for [4+][BArF4] are shown in Fig. 5, right, affording χTIP = 3.2(3) × 10−4 emu mol−1. The resulting χTIP value of [4+][BArF4] was similar in magnitude to those of Ce(COT)2 (χTIP = 1.4(2) × 10−4 emu mol−1),53 Ce(acac)4 (acac = acetylacetonate, χTIP = 2.1(2) × 10−4 emu mol−1),52 Ce(tmtaa)2 (tmtaaH2 = tetramethyldibenzotetraaza[14]annulene, χTIP = 2.33(6) × 10−4 emu mol−1),52 and Ce[(1,4-SiiPr3)2C8H4]2 (χTIP = 4.5(3) × 10−4 emu mol−1).60
Overall, our experimental data confirm the Ce(IV) oxidation states and elucidate structure–property relationships for the Ce(IV) guanidinate-amide complexes. Compounds with more guanidinate ligands tend to show larger steric encumbrance about the Ce(IV) cations, lower energy absorption bands, larger nf values, and more positive Ce(IV/III) reduction potentials. We next carried out quantum mechanical calculations to understand the role of guanidinate ligands further, especially the impact on LMCT transitions, 4f and 5d covalency, and stabilization of the Ce(IV) oxidation states.
Fig. 7 Calculated HOMOs for 1-Cl, 2-Cl, [3+], and [4+]. Significant contributions from π orbitals of guanidinate ligands were evident for 2-Cl, [3+], and [4+]. |
The presence of low energy absorption bands in CeIV guanidinate complexes was also evident from the computed HOMO–LUMO gaps. Computed energy gaps of 2.42, 2.36 and 2.20 eV were found between the HOMO and LUMO orbitals for 2-Cl, [3+], and [4+]. These HOMO–LUMO gaps were smaller than that of 1-Cl at 3.07 eV. While LUMO orbitals had primarily cerium 4f character (>89%) and only minor 5d character (<2%) in these CeIV complexes (Table 2), the atomic orbital contributions to the HOMOs were distinctly different between 1-Cl and other CeIV complexes with guanidinate ligands (Fig. 7). The computed HOMO of 1-Cl was found to be mainly composed of amide p character (Table S18†). In contrast, significant AO contributions from guanidinate p character were identified for the computed HOMOs of 2-Cl, [3+], and [4+] (Table 2). An increase in guanidinate AO character of HOMOs through 2-Cl, [3+], to [4+] was observed. Only minor Ce 4f (<6%) and 5d (<4%) AO character was found in HOMO (Table S18†). Hence, the unusual colors and low-lying absorption bands of CeIV guanidinate complexes are attributed to the presence of guanidinate π orbitals which are ∼2 eV below the empty cerium 4f orbitals.
Complex | 1-Cl | 2-Cl | [3+] | [4+] |
---|---|---|---|---|
Guanidinate p in HOMO | — | 82.61 | 85.26 | 89.28 |
Ce 4f in LUMO | 92.84 | 93.57 | 90.76 | 89.83 |
The computed HOMO of [4+] also indicated the correct symmetry for π-bonding interactions between the linear combination of guanidinate π orbitals and fy(3x2−y2) atomic orbital of Ce (Fig. 8). To illustrate possible bonding interactions between the ligands and cerium, we depicted a schematic molecular orbital diagram of [4+] with an ideal C3v local symmetry. Besides the HOMO with a2 symmetry resulting from interaction between the ligand π orbitals and the cerium 4f orbital, the computed degenerate HOMO−1 and HOMO−2 with e symmetry correspond to π interactions between ligand orbitals and cerium fxyz and fz(x2−y2) atomic orbital, respectively. As a result, we postulated that despite the small overlap of the core-like 4f orbitals with guanidinate π orbitals, these orbital arrangements with matched symmetries provided stabilization to the Ce(IV) oxidation state in complex [4+][BArF4].
Fig. 8 Schematic molecular orbital diagram of [4+] with ideal C3v local symmetry and computed a2 and e MOs. The HOMO–LUMO energy gap is 2.20 eV. |
An assessment of the covalency of cerium–ligand bonds can be made based on the electron populations of the natural atomic orbitals (NAOs), which are listed in Table 3, along with the charges of the Ce centers determined by the natural bond orbital (NBO) analyses. The extent of covalency is another indicator of the degree of LMCT in the ground state wavefunctions. The data in Table 3 showed that the Ce charge increased from 1-Cl to [4+] no matter which computational method was used. [4+] with the most positive reduction potential had the largest Ce charge. The 5d population, calculated by all methods, consistently decreased from 1-Cl to [4+]. However, the 4f population, calculated by DFT methods, increased from 1-Cl (0.91 based on PBE-DFT) to [4+] (0.94 based on PBE-DFT). The trend was opposite to that calculated by complete active space self-consistent field (CASSCF) method. The CASSCF calculations are affected by limitations of the active spaces, due to the sheer sizes of the systems, in particular [3+] and [4+]. The 4f populations obtained from the hybrid B3LYP-DFT and, especially, the non-hybrid Kohn–Sham (KS)-DFT (PBE-DFT) were significantly higher. The 4f populations obtained with KS-DFT and non-hybrid functionals such as PBE were shown previously to be somewhat inflated by the KS delocalization error.72 However, the static correlation error73 (i.e., the error that arises from treating multi-configurational states) was numerically probed to be the smallest with non-hybrid functionals.72,74 Therefore, the PBE calculations would seem to be a good compromise for describing the complexes. Moreover, the PBE calculations reproduce the trend in the spectroscopic nf values (Fig. S73†). Since the 4f population increased to a lesser extent compared to the decrease of 5d population, there is an overall decrease of the ligand-to-metal electron donation and increase of the Ce charge from 1-Cl to [4+].
Complex | 1-Cl | 2-Cl | [3+] | [4+] | |
---|---|---|---|---|---|
CASSCF | Ce charge | 2.34 | 2.37 | 2.65 | 2.78 |
4f | 0.41 | 0.40 | 0.36 | 0.34 | |
5d | 1.03 | 1.00 | 0.83 | 0.73 | |
B3LYP | Ce charge | 2.18 | 2.26 | 2.44 | 2.51 |
4f | 0.71 | 0.69 | 0.72 | 0.74 | |
5d | 0.84 | 0.80 | 0.61 | 0.53 | |
PBE | Ce charge | 1.98 | 2.05 | 2.25 | 2.31 |
4f | 0.91 | 0.90 | 0.93 | 0.94 | |
5d | 0.84 | 0.79 | 0.60 | 0.53 |
To further understand the differences in 4f and 5d population and metal–ligand covalency in the Ce(IV) guanidinate complexes, 2-Cl, [3+], and [4+], natural localized molecular orbitals (NLMOs)75 were obtained from the NBO analyses. Fig. 9 shows the NLMOs for different types of Ce–N bonds in 2-Cl. 2-Cl (Fig. 9a) was used as an example to establish the bonding model, as it demonstrated the most stabilized Ce(IV) oxidation state. Only the PBE-DFT method was used here because it afforded results consistent with the experimental data. Notably, different bonding models for Ce–amide bonds and Ce–guanidinate bonds were illustrated (Fig. 9b–f). Each amide ligand interacted with the Ce atom through one Ce–N σ bond (Fig. 9b) and one Ce–N π bond (Fig. 9c). However, the two N atoms, N(3) and N(4), of the guanidinate ligand interacted with the Ce atom unequally, resulted from the steric congestion of the guanidinate ligand. The bond distance of Ce–N(3) was slightly shorter than that of Ce–N(4) by 0.03 Å. The shorter Ce–N(3) bond had one Ce–N σ bond (Fig. 9d) and one Ce–N–C π bond (Fig. 9e), while the Ce–N(4) bond had only σ bond weight (Fig. 9f). Similar bonding situations were observed for [3+] and [4+]. Moreover, the NLMOs indicated that the Ce–N–C π bond for the guanidinate ligand had larger Ce 4f weight (84%) than that of the Ce–N π bond formed for the amide ligand (66%). The larger Ce 4f weight was consistent with a better energy match between Ce 4f orbitals and guanidinate π orbitals than that between Ce 4f orbitals and amide nitrogen p orbitals. However, the Ce–N–C π bond had the least Ce 5d character (16%) among all bonding types. The Ce–N bond distances for the guanidinate ligand were about 0.2 Å longer than that for the amide ligand. Therefore, increasing the number of sterically hindered guanidinate ligands from 2-Cl to [4+] resulted in less spatial overlap between Ce 5d orbitals and ligand π orbitals. Overall, the consecutive replacements of the amide ligand by the rigid guanidinate ligand decrease the ligand donation into the metal 5d shell, due to less optimal overlap when going from 2-Cl to [4+], while simultaneously increasing the 4f participation in the metal–ligand bonding orbitals. Related scenarios have been reported for Ce(IV)/Th(IV) imido complexes,42 lanthanide dioxides and sesquioxides,45,76 [LnCl6]x− (x = 2, 3; Ln = CeIV, CeIII, NdIII, SmIII, EuIII, GdIII),30 and [AnCl6]2− (An(IV) = Th, U, Np, Pu).77 The NLMOs show one Ce–Cl σ bond and two Ce–Cl π bonds, which have similar bonding characteristics to the computed Ce-amide bonds. This is also reflected on the positive charges of 1-Cl and 2-Cl compared to [3+] and [4+] for the Ce atom (Table 3).
Fig. 9 Crystal structure and Ce–N natural localized molecular orbitals (NLMOs) (±0.03 a.u. isosurfaces) and weight-% of atomic shell contributions to the NLMOs of 2-Cl. (a) Thermal ellipsoid plot of 2-Cl at the 30% probability level.13 Selected bond lengths (Å): Ce(1)–N(1) 2.384(4), Ce(1)–N(2) 2.422(4), Ce(1)–N(3) 2.529(3), Ce(1)–N(4) 2.545(4). (b) σ(Ce–N(1)) for the amide ligand. (c) π(Ce–N(1)) for the amide ligand. (d) σ(Ce–N(3)) for the guanidinate ligand. (e) π(Ce–N(3)–C) for the guanidinate ligand. (f) σ(Ce–N(4)) for the guanidinate ligand. |
Overall, our computational analyses elucidate the impacts of guanidinate ligand frameworks on the experimental observables. First, the nature of unusual colors and low-lying absorption bands for Ce(IV) guanidinate complexes, 2-Cl, [3+][BArF4], and [4+][BArF4], is ligand-to-metal charge transfer (LMCT) transitions from guanidinate π orbitals to the cerium 4f shell. Moreover, a better energy match between the guanidinate π orbitals and the cerium 4f orbitals increases the participation of Ce 4f orbitals in the metal–ligand bonds, consistent with the increase of nf values extracted from the Ce L3-edge XANES spectra. Incorporation of sterically bulky guanidinate ligands decreases the overlap between cerium 5d orbitals and ligand π orbitals, resulting in a reduction of overall ligand-to-metal electron donation and thereby larger Ce charge, more positive Ce(IV/III) reduction potential, and less stabilized Ce(IV) oxidation state. Similarly, tuning the Ce(IV/III) reduction potentials for Ce(III) bis(guanidinate) mono(amide) photocatalysts by a modification of ligand steric profiles has been achieved by some of us.17
Footnote |
† Electronic supplementary information (ESI) available. CCDC 1965035–1965036. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0sc05193d |
This journal is © The Royal Society of Chemistry 2021 |