Nikolay A.
Pushkarevsky
*ab,
Mikhail A.
Ogienko
a,
Anton I.
Smolentsev
a,
Igor N.
Novozhilov
a,
Alexander
Witt
c,
Marat M.
Khusniyarov
c,
Vladimir K.
Cherkasov
de and
Sergey N.
Konchenko
ab
aNikolaev Institute of Inorganic Chemistry, Siberian Division of RAS, Akad. Lavrentieva str. 3, 630090 Novosibirsk, Russia. E-mail: nikolay@niic.nsc.ru
bDepartment of Natural Sciences, Novosibirsk State University, Pirogova Street 2, 630090 Novosibirsk, Russia
cDepartment of Chemistry and Pharmacy, Friedrich-Alexander-University of Erlangen-Nuremberg, Egerlandstr. 1, 91058 Erlangen, Germany
dG. A. Razuvaev Institute of Organometallic Chemistry of RAS, Tropinina St. 49, 603950 Nizhny Novgorod, Russia
eN. I. Lobachevsky Nizhny Novgorod State University, Gagarin Ave., 23, 603950, Nizhny Novgorod, Russia
First published on 1st December 2015
The first examples of samarium, europium, and ytterbium complexes with 3,6-di-tert-butyl-o-benzoquinone (3,6-dbbq) in the form of catecholate have been obtained by reactions of the quinone with the corresponding lanthanocenes, (n = 1 or 2) in solution. In the course of the reactions lanthanide ions lose one or two Cp* ligands, which take part in reduction of a quinone molecule into a catecholate anion (dbcat, 2−). As a result of the reactions, Sm and Yb clearly yield dimeric complexes [(LnCp*)2(dbcat)2], where each Ln ion loses one Cp* ligand. Eu forms a trimeric complex [(EuCp*)(Eu·thf)2(dbcat)3], in which one Eu ion is coordinated by one Cp* ligand, while two Eu ions have lost all Cp* ligands and are coordinated by THF molecules instead. Magnetic properties corroborate the assignment of oxidation states made on the basis of single-crystal X-ray diffraction: all the quinone ligands are present in the catecholate state; both Sm/Yb ions in the dimers are in the +3 oxidation state, whereas the Eu trimer contains two Eu(II) and one Eu(III) ions. Cyclovoltammetry studies show the presence of two reversible oxidation waves for all complexes, presumably concerned with the redox transitions of the dbcat ligands.
In the context of the synthesis and investigation of new lanthanide complexes with possibly high exchange constants between a lanthanide ion and a ligand, the lanthanide complexes with o-benzoquinone-based (bq) ligands are promising. Indeed, the o-quinone based ligands can exist in three charged states (quinone, semiquinone (sq, 1−), and catecholate (cat, 2−)), so the redox transitions can vary in a wide range.12,13 Besides, the ligands can be functionalized in a number of ways, including conversion to iminoquinones,14 thus providing variable donor and steric behaviours and a possibility to tune metal–ligand interactions. Despite this prominent ligand diversity, only a limited number of quinone-based Ln(III) complexes have been described and magnetically studied until now.4–6,15,16 A few complexes of Ln(II) with sq and cat ligands are known; to the best of our knowledge, the only examples are homoleptic [Ln(dbcat)] (Ln = Sm, Eu) and heteroleptic [EuLi4(dbsq)2(dbcat)2(LiI)2(thf)6] complexes obtained by Bochkarev and co-workers in the reaction of EuI2 or Ln(N(SiMe3)2 with 3,5-di-tert-butyl-o-benzoquinone (3,5-dbbq).17 Lately, a ligand containing 3,6-di-tert-butyl-o-benzoquinone (3,6-dbbq) coupled with tetrathiafulvalene was employed for syntheses of bridged dimeric Ln complexes (Ln = Dy, Er, Yb); in the case of dysprosium, each Dy3+ ion was seen to behave as an independent single-ion magnet.18,19 The possibility of tuning the redox properties of a ligand by changing its substituents or by using iminoquinone derivatives14 can be used to enhance the magnetic interactions between the Ln ion and the ligand in its radical-ion form.
Substituted cyclopentadienyl (Cpx) and indenyl (Ind) complexes of lanthanides(II) have been widely used for syntheses of complexes with radical ligands.20–29 In some cases, not only the Ln cation, but also Cpx and Ind ligands take part in the reduction process, and their role as a reducing agent is sufficiently induced by their bulkiness, in addition to the redox potential of the metal centre.
We were interested in finding new ways of producing the quinone complexes of lanthanides, for which easily accessible Cpx complexes can be good starting compounds, while using the 3,6-dbbq-based ligands could provide necessary solubility of the complexes and sterical protection owing to bulky butyl groups.
The reactions of lanthanocenes(II) (Ln = Sm, n = 2; Ln = Yb, n = 1) with 3,6-dbbq were carried out in hexane solutions starting from deep-frozen mixtures. Upon warming to room temperature, the lanthanocene readily dissolves to give a green (Sm) or sky-blue (Yb) solution; upon successive dissolution of the quinone the reaction slowly proceeds to result in a yellow (Sm) or light-brown (Yb) clear solution and crystalline precipitate of the corresponding compound. The products [(SmCp*)2(dbcat)2] and [(YbCp*)2(dbcat)2] (1 and 2, correspondingly, Scheme 1) are well soluble in hexane and were crystallized from small volumes of this solvent; a by-product, , is retained in the solution.
Crystal structures were determined for both complexes 1 and 2, which possess close molecular structures and isostructural crystal lattices (Fig. 1). The binuclear complexes contain a crystallographic centre of inversion in the middle of the Ln–Ln vector with one quinone and one Cp* ligand per Ln atom. The quinone ligands are bound in the chelate-bridging mode: the two O atoms are coordinated to a Ln atom (2.3127(18) and 2.2269(17) Å for Sm, 2.2264(19) and 2.1285(18) Å for Yb), while one of them is additionally bound to a second Ln atom at a longer distance (by ca. 0.09 Å for Sm, 0.04 Å for Yb; see all distances in the captions to Fig. 1). It is noteworthy that the carbon atom C1 next to the μ-O atom is also located within a bonding distance to the Ln atom and the angles Ln–O1–C1 are slightly less than 90°. The mean-squared plane of the conjugated quinone ligand is nearly perpendicular to the plane formed by the Ln′, C1, and O1 atoms (84.4° for Sm, 82.1° for Yb), as well as to the flat Ln2O2 ring (84.8° for Sm, 89.5° for Yb). Out of more than a hundred structurally characterized complexes of transition metals (and several of lanthanides) with the 3,6-dbbq-based ligand, no analogous chelate-bridging mode has been documented to date (based on data from Cambridge Structural Database (CSD)33). The characteristic lengths of C–O and C–C bonds suggest that the quinone ligand acquires the catecholate (2−) state.17,34,35 The IR spectra of 1 and 2 are nearly identical, in accordance with the same coordination environment of the complexes (Fig. S1, ESI†). Consequently, both Ln ions in 1 and 2 are in the 3+ oxidation state, which is consistent with the rather light colour of the compounds. To obtain this composition, the 2e reduction of the neutral quinone requires 1e from the Ln ion and 1e from the Cp*− ligand, which corresponds to the 1:1 ratio of the reagents:
The participation of the Cp* and related indenyl ligands in reduction processes of Ln complexes is well documented in cases of “sterically induced reduction” (SIR), that were described for the complexes with a sterically encumbered coordination sphere of a Ln ion.27,29 In our case, the addition of one quinone ligand to is likely to proceed fast, concurrent with the oxidation of Ln2+ to Ln3+ and reduction of the dbbq to the semiquinone (1−) or catecholate (2−) state. Then, upon the formation of a dimer by coordination of the third O atom to the Ln3+ ion, the steric crowding becomes strong enough to facilitate the cleavage of the Cp* fragment accompanied (or followed) by further reduction of the quinone ligand to the catecholate state. This reaction sequence is especially noticeable during the synthesis of 1 or 2, where the characteristic colour of lanthanocene(II) does not appear in a cold solution, even before all of the quinone is dissolved; on the contrary, the blue colour of the reaction mixture is ascribed to the intermediate species [LnIII(Cp*)2(dbsq)], which then transforms into the brown product 2 upon stirring at room temperature, or, faster, upon slight warming. This sequence, i.e. a fast redox process connected with the oxidation of a metal centre followed by a slower process of sterically induced reduction, has been proven previously by Evans.27,36
The reaction of the europocene [Eu(Cp*)2(thf)] with 3,6-dbbq (Scheme 2) was carried out identically with Sm and Yb analogues, however, the colour of the resulting solution turned out to be much darker. The product [(EuCp*)(Eu·thf)2(dbcat)3] (3) is quite soluble in hexane, thus the yield of crystals was somewhat lower.
The crystal structure of 3 reveals a trinuclear complex (Fig. 2), where the three metal atoms are coordinated by three bridging quinone ligands. It can be conceived as a trigonal prism of six O atoms with all the square faces capped by Eu(L) vertices. Remarkably, the Eu centres have different ligand environments: all metal ions are bound to two bridging quinone ligands, while in the terminal position one Eu atom retains a Cp* ligand and the other two bear a thf molecule. There are two types of ligand bridging: the two dbbq ligands lying out of the Eu3 plane are of the μ3,η4 type, while the in-plane dbbq ligand acquires the μ,η4-bridging mode. The out-of-plane quinone ligands are located closer to the Eu1 atom: the Eu1–O bonds (2.280(2)–2.310(2) Å) are noticeably shorter than the bonds between the same O and the corresponding Eu2 or Eu3 atoms (2.487(2)–2.506(2) Å; see bond lengths in the Fig. 2 caption). Nevertheless, these ligands are not noticeably tilted towards Eu2 or Eu3; as opposed to the structures of 1 and 2, the angles between their mean-squared planes and the Eu3 plane are 79.2 and 77.4°, and all the distances Ln⋯C are much longer than corresponding Ln–O bonds. The in-plane quinone ligand is bound to the metal atoms nearly symmetrically with all the Eu–O bonds lying in the range 2.429(3)–2.457(3) Å and the mean-squared plane of the ligand being perpendicular (89.4°) to the Eu3 plane. All the C–O (1.356(4)–1.373(5) Å) and interjacent C–C bonds (1.419(5)–1.430(6) Å) in the quinone ligands point to their dianionic catecholate state. The IR spectrum of 3 indicates the same characteristic absorptions as those for 1 or 2 (Fig. S1, ESI†). Thus, Eu in 3 is present in two oxidation states, one cation of Eu3+ and two cations of Eu2+, which correspond to Eu–O bond distances and the darker colour, and are well reflected by magnetic properties of the complex (vide infra). The oxidation of Eu in the course of the reactions does not proceed to the full extent; instead, the Cp* ligands from two thirds of the starting europocene(II) are oxidized and lost. Again, the reagent stoichiometry is 1:1 as follows:
The presence of unoxidized Eu2+ in the complex can be explained by its significantly lower reduction potential as compared to Sm and Yb low-valent metallocenes.27 It is noteworthy that in case the latter reaction is carried out in hexane, as in the case of compounds 1 and 2, small amounts of a pale-grey microcrystalline precipitate remain insoluble after recrystallization of the product 3. Changing the solvent from hexane to toluene results in a much cleaner reaction and the pale-grey by-product is formed in vanishingly small amounts. The IR spectrum of this by-product shows a very similar pattern to that of magnesium catecholate 4 (see below) and, according to elemental analysis, its composition is close to the formula [Eu(dbcat)(thf)]n (attempts to grow crystals suitable for X-ray structural analysis were unsuccessful; see the ESI† for further details). Presumably, the latter complex can be formed in the reaction of europocene and dbbq if only the Cp*− ligands take part in the reduction and Eu remains in the 2+ state.
Considering different reaction behaviours of lanthanocenes, it is not fully clear, which factors govern the partial or total loss of Cp* ligands from the Ln atoms, as well as Ln2+ → Ln3+ redox processes. Presumably, both the redox potentials of corresponding lanthanocenes and the steric bulkiness of the Cp* ligand in the coordination sphere of a given Ln ion have an influence. To make the reduction properties of the Cp*− anions more evident, one must exclude the possibility of a metal cation reducing the quinone. This can be achieved by involving in the reaction an analogous complex with a redox-innocent metal cation (2+), for which the alkaline-earth elements are a good choice. In the reaction of magnesocene with dbbq (1:1) in thf carried out analogously to those of lanthanocenes, the solution appeared blue-green after mixing of the reagents and retained the same colour for several hours indicating the initial formation of semiquinone complexes. Considering that only Cp*− can act as a reductant, these complexes are supposedly [MgCp*(dbsq)(thf)n] species. After prolonged stirring the solution turned yellow and finally the already known complex [(Mg(thf)2)2(dbcat)2] (4) was isolated. This compound was initially obtained by Piskunov and co-workers in the reaction of amalgamated Mg with a quinone in THF.37 Substantial delay in the second step of the reduction process observed during the preparation of 4 is similar to that of the reactions with the lanthanocenes, and may be caused by the slow electron transfer between the Cp*− anion and the dbsq ligand in the same complex. However, to resolve whether this reductive action of Cp*− depends on the steric crowding in the Mg coordination sphere would require additional studies.
A dinuclear molecule of complex 4 (Fig. 3) resembles those of complexes 1 and 2 with regard to the coordination mode of the quinone ligand. The molecule is located in a common position of the crystal structure, so the bond lengths in both halves are not equal, but differ by no more than 0.01 Å (see the distances in the caption to Fig. 3). Unlike compounds 1 and 2, the Mg atoms are not bearing Cp* ligands, so the quinone ligands are present in the catecholate state, in agreement with the C–O and C1–C2 bond lengths,17,34,35 the IR spectral data,37 as well as its colourless appearance. The Mg2O2 ring is not flat (the angle between two Mg2O planes is 133.65(8)°), and both dbcat ligands are turned to the same side of the Mg2O2 butterfly, as opposed to the dimeric molecules of 1 and 2. Interestingly, the crystal structure of the similar dimeric pyridine complex [Mg2(3,6-dbbq)2(py)4]37 differs in the manner of coordination of quinone ligands: they acquire the same chelate-bridging mode as in 4, but both dbcat ligands chelate the same Mg ion and bind the second Mg ion with bridging O atoms to give 6- and 4-fold coordinated Mg centres in the same molecule. Apparently, both types of arrangements of catecholate ligands in dimeric complexes are close in energy and depend on the steric demand or rigidity of the second ligand (THF or pyridine). Hence, both Cp*− ligands of the initial Mg complex participate in the reduction of dbbq to dbcat ligands, and are replaced by THF molecules in the course of the reaction. The differences in the reaction behaviour between Mg and Ln complexes correlate with the ionic radii of the metal atoms involved: supposedly, the noticeably smaller Mg2+ (0.66 Å for a 5-fold coordination) constrains the Mg2O2 unit to acquire a bent geometry and does not provide enough space near the metal for a bulky Cp* ligand, unlike the much larger Sm3+ and Yb3+ ions (0.96 and 0.87 Å for a 6-fold coordination, correspondingly38). To corroborate this, experiments with larger alkaline-earth metals (Ca, Sr, Ba) will be useful, which could be considered for further work.
Fig. 4 Variable-temperature effective magnetic moments measured at external magnetic field 0.1 T: 1 – black circles, 2 – blue squares, 3 – red triangles. |
The magnetic moment of Yb complex 2 measured at RT is 5.48μB. Upon lowering the temperature, the moment decreases gradually to a value of 4.69μB at 40 K followed by an abrupt decrease to 0.74μB recorded at 2 K (Fig. 4). The latter points to antiferromagnetic interactions present in the solid 2. A mononuclear Yb(III) complex is expected to have a RT moment in the range 4.3–4.9μB.39–41 Thus, for a dinuclear complex we expect to obtain μeff = 21/2μeff(YbIII) = 6.1–6.9μB. The experimental value is slightly lower than the predicted one. This is likely due to some diamagnetic or weakly paramagnetic impurities present in the solid 2. Alternatively, a quantum admixture of Yb(II) and Yb(III) states cannot be fully excluded.10,11,20 No saturation of magnetization was observed within 0–5 T at 2 K (ESI†).
Magnetic susceptibility measurements performed on a trinuclear Eu complex 3 reveal a RT magnetic moment of 10.98μB. This moment is nearly constant in the temperature range 30–300 K (Fig. 4). Upon cooling below 30 K the magnetic moment increases gradually reaching a value of 13.21μB at 2 K. This behaviour points to the presence of weak ferromagnetic interactions in the solid 3. However, the ferromagnetic behaviour can be suppressed at high magnetic fields (H = 5 T, see the ESI†). Taking into account the common RT magnetic moments for mononuclear complexes of Eu(II) (7.6–8.0μB) and Eu(III) (3.7–4.2μB),42 the moment for 3 is calculated as (2μeff2(EuII) + μeff2(EuIII))1/2 = 11.4–12.1μB. The experimentally observed value (10.98μB) is slightly lower than that predicted. This is likely due to the presence of a small amount of diamagnetic or weakly paramagnetic impurities in the highly oxygen-sensitive sample 3. Thus, the magnetic data confirm the presence of one Eu(III) and two Eu(II) ions in 3. Note that a much higher RT magnetic moment (13.3–14.0μB) would be expected for an uncoupled system with three Eu(II) ions and a ligand-radical. At very low temperatures magnetization saturates showing a plateau at 11–12NaμB in the variable-field series (ESI†). Due to the detected ferromagnetic interactions, 3 might exhibit SMM properties, which was investigated by ac susceptibility measurements. Unfortunately, no ac signal was observed, which precludes slow magnetic relaxation and thus SMM behaviour.
Complex | E a, V | E 1/2, V | |
---|---|---|---|
Cp*− − e− = Cp* | dbcat2− − e− = dbsq−˙ | dbsq−˙ − e− = dbbq | |
1 | +0.25 | +0.52 | +0.89 |
2 | +0.23 | +0.50 | +0.88 |
3 | — | +0.51 | +0.88 |
4 | — | −0.31 | +0.65 |
Complexes 1 and 2 possess additional irreversible oxidation peaks (Eaca. +0.23 V) which probably correspond to the oxidation of the Cp*− ligand. Cp*− is expected to be oxidized at lower potentials than Cat2− and leaves as , in accordance with the chemical behaviour of these groups during the synthesis of 1–4. There are two irreversible peaks of smaller intensity in the negative region for 1 and 2 (−0.45 and −0.60 V vs. Fc+/Fc), which can correspond to the reduction of the species obtained upon oxidation of Cp* ligands.53 These irreversible oxidation and reduction peaks are absent in the case of 4, and, rather unexpectedly, in the case of 3; the latter fact cannot be ascribed to the insufficient concentration of Cp* ligands, since the concentration of complexes was maintained at the same level. Potentially, the reaction with the remaining water could lead to the partial removal of Cp* ligands owing to hydrolysis, but the preservation of the initial colour of the solutions during the CV experiments, and careful preparation of the solvent and solutions (similarly to the other complexes) does not support this supposition. More uncommonly, the Eu curves do not contain any waves attributable to the Eu2+/Eu3+ transition; presumably, this wave is outside of the measurement window (−2.6 to +1.4 V vs. Fc+/Fc).
It is evident that the full explanation of the redox processes in the Ln–catecholate systems requires more systematical study.43,51 Chemical oxidation (e.g. with AgI or FcPF6) may be useful to obtain similar complexes with semiquinolate radicals; in the case of the compounds with Cp* ligands it may lead to the oxidation and substitution of the latter, before the oxidation of the dbcat ligands, which can serve as a method for introducing other ligands in the Ln coordination sphere.
1 | 2 | 3 | 4 | |
---|---|---|---|---|
Empirical formula | C48H70O4Sm2 | C48H70O4Yb2 | C60H91Eu3O8 | C48H80Mg2O9 |
Formula weight | 1011.74 | 1057.12 | 1396.21 | 849.74 |
Temperature (K) | 150(2) | 150(2) | 150(2) | 150(2) |
Crystal size (mm3) | 0.22 × 0.14 × 0.05 | 0.26 × 0.24 × 0.14 | 0.22 × 0.15 × 0.08 | 0.28 × 0.22 × 0.18 |
Crystal system | Triclinic | Triclinic | Monoclinic | Monoclinic |
Space group | P | P | P21/n | P21/c |
Z | 1 | 1 | 4 | 4 |
a (Å) | 10.3712(2) | 10.3705(4) | 12.5855(3) | 21.0570(8) |
b (Å) | 10.6440(2) | 10.5729(4) | 20.3033(6) | 12.0413(5) |
c (Å) | 12.1789(3) | 12.1066(5) | 24.2945(7) | 19.0031(6) |
α (°) | 70.4500(10) | 69.6750(10) | ||
β (°) | 66.9300(10) | 65.6400(10) | 104.6510(10) | 95.7500(10) |
γ (°) | 69.3720(10) | 70.2300(10) | ||
V (Å3) | 1127.05(4) | 1104.04(8) | 6006.0(3) | 4794.1(3) |
D calcd (g cm−3) | 1.491 | 1.590 | 1.544 | 1.177 |
μ (Mo Kα) (mm−1) | 2.620 | 4.250 | 3.144 | 0.102 |
θ range (°) | 2.10–27.53 | 1.90–27.59 | 1.73–27.69 | 1.94–27.51 |
h, k, l indices range | −13 ≤ h ≤ 12; | −13 ≤ h ≤ 13; | −7 ≤ h ≤ 16; | −27 ≤ h ≤ 26; |
−13 ≤ k ≤ 13; | −8 ≤ k ≤ 13; | −26 ≤ k ≤ 25; | −15 ≤ k ≤ 15; | |
−15 ≤ l ≤ 15 | −11 ≤ l ≤ 15 | −31 ≤ l ≤ 29 | −12 ≤ l ≤ 24 | |
F(000) | 514 | 530 | 2816 | 1856 |
Reflections collected | 10517 | 8694 | 37990 | 36354 |
Unique reflections | 5163 (Rint = 0.0291) | 5075 (Rint = 0.0140) | 13733 (Rint = 0.0251) | 10979 (Rint = 0.0522) |
Observed reflections [I > 2σ(I)] | 4788 | 4771 | 10958 | 6770 |
Parameters refined | 255 | 255 | 699 | 544 |
R[F2 > 2σ(F2)] | R 1 = 0.0215 | R 1 = 0.0182 | R 1 = 0.0301 | R 1 = 0.0541 |
wR2 = 0.0518 | wR2 = 0.0457 | wR2 = 0.0552 | wR2 = 0.1303 | |
R(F2) (all data) | R 1 = 0.0247 | R 1 = 0.0205 | R 1 = 0.0482 | R 1 = 0.1039 |
wR2 = 0.0530 | wR2 = 0.0466 | wR2 = 0.0595 | wR2 = 0.1438 | |
GOOF on F2 | 1.051 | 1.048 | 1.025 | 1.046 |
Δρmax, Δρmin (e Å−3) | 1.173, −0.764 | 1.305, −0.800 | 1.770, −0.858 | 0.564, −0.421 |
Footnote |
† Electronic supplementary information (ESI) available: Additional magnetic data, IR spectra, and X-ray crystallographic data. CCDC 1409389–1409392. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5dt03573b |
This journal is © The Royal Society of Chemistry 2016 |