Asymmetric Assembly of Chiral Lanthanide(III) Tetranuclear Cluster Complexes Using Achiral Mixed Ligands: Single-molecule Magnet Behavior and Magnetic Entropy Change.

It is challenging to use achiral ligands to spontaneously construct chiral molecular magnets. In this work, two new Ln4 cluster complexes based on N,N'-(1,3-propanediyl)bis[N-[1,1-bis(hydroxymethyl)-2-hydroxyethyl]amine] (H6L) have been assembled, which are crystallized in a chiral space group due to the asymmetric distribution of acetate (OAc-) groups and hexafluoroacetylacetonate (F6acac-) groups on both sides of the parallelogram-like Ln4 core. Complex 1, [Dy4(H3L)2(OAc)3(F6acac)3]·5MeOH·2H2O, exhibits single-molecule magnet properties at the zero field with the U eff/k value of 48.4 K; notably, besides the Orbach process, the Raman process is also prominent for the magnetic relaxation of 1. Complex 2, [Gd4(H3L)2(OAc)3(F6acac)3]·4MeOH·2.5H2O, displays a large magnetocaloric effect, whose largest -ΔS m value is 21.88 J kg-1 K-1 (when T = 2 K and ΔH = 5 T); it thus can be utilized as a good magnetic refrigeration molecular material.

Introduction

Recently, lanthanide(III) cluster complexes have attracted great attention in the field of single-molecule magnets (SMMs)[1] and magnetic refrigeration molecular materials,[2,3] which is closely related to the large spin value of lanthanide(III) ions. If the specified lanthanide(III) ions such as dysprosium(III) ions have strong magnetic anisotropy, they are suitable for the construction of SMMs, and when the specified lanthanide(III) ions are gadolinium(III) ions, they are suitable for the assembly of magnetic refrigeration molecule materials, owing to the largest ground state spin value (SGd = 7/2) and no magnetic anisotropy.[4,5] Because the magnetic axis and the symmetry of each dysprosium(III) ion in the cluster complex are difficult to control,[6] the research progress of dysprosium(III) cluster complexes in the SMM field is relatively slow. However, gadolinium(III) cluster complexes have obvious advantages in the study of magnetic refrigeration molecular materials.[7−12]

On the other hand, if chirality is introduced into molecular magnets, it will bring valuable physical properties such as nonlinear optics,[13−16] ferroelectricity,[17−20] and magneto-optical effects,[21−27] making them attractive multifunctional molecular materials. The general case is to obtain chiral structured molecule-based magnets by chiral ligand coordination[28−37] or even by cocrystallizing with chiral organic molecules.[38] Another more challenging case is to use achiral ligands to spontaneously construct chiral structured molecule-based magnets, which involve helical chirality,[39] Δ/Λ octahedral coordination configuration of transition metal ions,[40] and cooperative orientation of anions and cations in the axial direction.[41] It is well known that the chirality occurs when different functional groups are attached to the carbon atom in organic chemistry, however, the chirality of cluster SMMs through the asymmetric distribution of different achiral ligands on both sides of the cluster core has never been reported.

Lanthanide(III) tetranuclear clusters with various core structures, which include chain, square, butterfly or rhombus, and cube, might be the most studied clusters in SMMs.[42−56] They can not only exhibit large energy barrier values but also can be used to study multistep magnetic relaxation behavior. Recently, we constructed a parallelogram-like Dy(III) tetrametallic SMM, [Dy4(H3L)2(OAc)6]·2EtOH,[57] using acetate ligands and the bis-tris propane ligand, that is N,N′-(1,3-propanediyl)bis[N-[1,1-bis(hydroxymethyl)-2-hydroxyethyl]amine] (H6L, Scheme ). Interestingly, we further found that if another ligand, hexafluoroacetylacetonate anion (F6acac–), was added, we could obtain Ln(III) tetranuclear clusters (Ln = Dy and Gd) crystallized in a chiral space group. Herein we report the room-temperature syntheses, X-ray crystal structures, SMM properties, and magnetic entropy changes of two chiral Ln(III) tetranuclear clusters with the N,N′-(1,3-propanediyl)bis[N-[1,1-bis(hydroxymethyl)-2-hydroxyethyl]amine] ligand, the acetate ligand, and the hexafluoroacetylacetonate ligand. [Dy4(H3L)2(OAc)3(F6acac)3]·5MeOH·2H2O (1) and [Gd4(H3L)2(OAc)3(F6acac)3]·4MeOH·2.5H2O (2), both have a parallelogram-like Ln4 cluster core: complex 1 shows SMM properties at the zero field, while complex 2 displays a large magnetocaloric effect.

Scheme 1

Molecule Structure of the Bis-tris Propane Ligand (H6L)

Experimental Procedures

Dy(F6acac)2Ac·2H2O and Gd(F6acac)2Ac·2H2O were presynthesized by the method described in the literature.[58−60]

Preparation of 1

H6L (0.125 mol), Dy(F6acac)2Ac·2H2O (0.25 mmol), LiOH·H2O (0.50 mmol), and 20 mL of methanol were added to a 50 mL Erlenmeyer flask and stirred for 5 h to obtain a colorless solution. After filtration, the filtrate was left to slowly evaporate the solvent. Colorless small block single crystals of 1 were harvested in about ten days, which were washed with 10 mL of water and 10 mL of methanol in turn, and were then air-dried naturally. Yield (calculated based on Dy): 35%. Anal. calcd for C48H80Dy4F18N4O31 (1): C, 26.19; H, 3.66; N, 2.55%. Found: C, 26.23; H, 3.69; N, 2.52%. IR (KBr, cm–1): 3674 (w), 3397 (br s), 2936 (w), 2884 (w), 1661 (s), 1559 (s), 1529 (m), 1502 (m), 1450 (m), 1347 (w), 1255 (s), 1206 (s), 1144 (s), 1100 (w), 1027 (m), 946 (w), 855 (w), 797 (w), 757 (w), 662 (m), 618 (w), 584 (w), 560 (w), 527 (w), 495 (w), 420 (w).

Preparation of 2

H6L (0.125 mol), Gd(F6acac)2Ac·2H2O (0.25 mmol), LiOH·H2O (0.50 mmol), and 40 mL of MeOH/CH2Cl2 (v/v = 1) were added to a 100 mL Erlenmeyer flask and stirred for 5 h to obtain a colorless solution. After filtration, the filtrate was left to slowly evaporate the solvent. Colourless small block single crystals of 2 were harvested in about ten days, which were washed with 10 mL of water and 10 mL of methanol in turn, and were then air-dried naturally. Yield (calculated based on Gd): 40%. Anal. calcd for C47H79F18Gd4N4O30.5 (2): C, 26.15; H, 3.69; N, 2.59%. Found: C, 26.11; H, 3.73; N, 2.54%. IR (KBr, cm–1): 3676 (w), 3397 (br s), 2940 (w), 2887 (w), 1661 (s), 1559 (s), 1527 (m), 1502 (m), 1448 (m), 1348 (w), 1255 (s), 1206 (s), 1144 (s), 1096 (w), 1026 (m), 950 (w), 854 (w), 797 (w), 741 (w),728 (w), 662 (m), 618 (w), 584 (w), 557 (w), 527 (w), 489 (w), 412 (w).

Result and Discussion

Synthesis

The bis-tris propane ligand is a common polydentate ligand containing not only the N coordination site but also the O coordination site, which had been successfully used to assemble some 3d cluster complexes,[61,62] 3d-3d heteronuclear cluster complexes,[63,64] and 3d–4f heteronuclear cluster complexes.[65−67] We also recently used it to solvothermally react dysprosium acetate and lithium hydroxide in EtOH at 100 °C, yielding a dysprosium(III) tetranuclear cluster SMM, [Dy4(H3L)2(OAc)6]·2EtOH, which shows double magnetic relaxation behavior at the zero field.[57] In this study, we used the bis-tris propane ligand to react with Ln(F6acac)2Ac·2H2O (Ln = Dy and Gd) and lithium hydroxide in MeOH or (MeOH + CH2Cl2) for several hours at room temperature, and then slowly evaporated the solvent to obtain lanthanide(III) tetranuclear cluster compounds 1 and 2. As expected, the hexafluoroacetylacetonate anion from the Ln(F6acac)2Ac·2H2O starting materials participates in the coordination acting as a terminal ligand, being assembled into the structures of complexes 1 and 2, just like the acetate anion does. However, the molecules of complexes 1 and 2 contain three acetate groups and three hexafluoroacetylacetonate groups, and these different ligands have to be asymmetrically distributed on both sides of the Ln4 cluster core, resulting in the loss of centrosymmetry of the entire molecule. Therefore, they are eventually crystallized in the chiral space group P1 (Table ). Unfortunately, it is known from the process of single-crystal structure analysis that the two chiral isomers in 1 or 2 are combined in the form of twin crystals, so they cannot be separated manually under a polarizing microscope. In contrast, [Dy4(H3L)2(OAc)6]·2EtOH is a centrosymmetric molecule crystallized in the centrosymmetric space group, Pbca,[57] because its six acetate anions are symmetrically distributed on both sides of the Dy4 cluster core. This research suggests a new strategy to construct the chiral SMM by the asymmetric distribution of different achiral ligands on both sides of the cluster core.

Table 1

Crystal Data and Structural Refinement Parameters for 1 and 2

	1	2
formula	C48H80Dy4F18N4O31	C47H79F18Gd4N4O30.5
Fw	2201.16	2159.14
crystal system	triclinic	triclinic
space group	P1	P1
a [Å]	11.0727(2)	11.1069(2)
b [Å]	12.9870(2)	12.9886(2)
c [Å]	13.6934(2)	13.7301(2)
α [deg]	69.304(2)	68.973(2)
β [deg]	89.4750(10)	89.2670(10)
γ [deg]	89.9360(10)	89.2490(10)
V [Å3]	1841.97(6)	1848.62(6)
Z	1	1
ρcalc [g cm–3]	1.984	1.939
μ [mm–1]	4.138	3.667
T [K]	170	170
λ (Mo Kα) [Å]	0.71073	0.71073
reflections collected	46,806	47,985
unique reflections	14,449	13,077
observed reflections	13,257	11,850
parameters	955	954
GoF [I ≥ 2σ(I)]	1.040	1.030
R1 [I ≥ 2σ(I)]	0.0262	0.0348
WR2 [I ≥ 2σ(I)]	0.0601	0.0867
Flack parameter (twin)	0.413(19)	0.32(3)

Crystal Structures

The structures of cluster compounds 1 and 2 are very similar (Figure ), and we, thus, focus on describing the crystal structure of 1. As shown in Figure a, complex 1 has an approximately parallelogram-shaped Dy4 core, with the Dy···Dy side lengths of 3.534, 3.822, 3.532, and 3.818 Å, respectively. Two bis-tris propane ligands each provide two η3-CH2O– oxygen atoms to bridge the four planar Dy3+ ions from above and below, using the η3:η3:η1:η1:η1:η2:μ4 coordination mode observed in the SMM [Dy4(H3L)2(OAc)6]·2EtOH.[57] Such a [Dy4(OCH2−)4] core is very similar to that in [Dy4(H3L)2(OAc)6]·2EtOH.[57] The Dy2 atom and the Dy3 atom are both nine-coordinated, which are coordinated by one bis-tris propane ligand providing four oxygen atoms and two nitrogen atoms, another bis-tris propane ligand offering two η3-CH2O– oxygen atoms, and one η2-AcO– ligand supplying one oxygen atom. By SHAPE software[68] analysis, it can be seen that the coordination configurations of these two Dy(III) ions are both spherical capped square antiprisms, and the deviation values from the C4 symmetry are 1.029 for the Dy2 atom (Table S1) and 0.952 for the Dy3 atom (Table S2). However, both the Dy1 and Dy4 atoms are eight-coordinated: the Dy1 atom is bonded by one oxygen atom of the η3-CH2O– group and one oxygen atom of the η2-CH2O– group from one H3L3– ligand, one oxygen atom of the η3-CH2O– group from another H3L3– ligand, one oxygen atom from the η2-AcO– bridging ligand, two oxygen atoms from one AcO– terminal ligand, and two oxygen atoms provided by one F6acac– terminal ligand; meanwhile, the Dy4 atom is bonded by one oxygen atom of the η3-CH2O– group, one oxygen atom of the η2-CH2O– group from one H3L3– ligand, one oxygen atom of the η3-CH2O– group from another H3L3– ligand, one oxygen atom of one η2-AcO– bridging ligand, and four oxygen atoms supplied by two F6acac– terminal ligands. After SHAPE software[68] analysis, the coordination geometries of the two Dy(III) cations were determined to be the square antiprism, with the deviations from the D4 symmetry of 1.530 for the Dy1 atom (Table S3) and 1.418 for the Dy4 atom (Table S4). The Dy–N bond distances (mean 2.535 Å, Table ) of 1 are a little shorter than those in [Dy4(H3L)2(OAc)6]·2EtOH (average 2.584 Å),[57] and the Dy–O bond distances (mean 2.390 Å, Table ) of 1 are comparable with those in [Dy4(H3L)2(OAc)6]·2EtOH (average 2.397 Å).[57]

Figure 1

Crystal structures of 1 (a) and 2 (b). All H atoms and lattice H2O and MeOH molecules are not shown for clarity.

Table 2

Selected Bond Lengths (Å) and Angles (Deg) for 1 and 2

complex 1
Dy1–O1	2.442(12)	Dy1–O2	2.392(14)
Dy1–O3	2.473(13)	Dy1–O4	2.395(8)
Dy1–O5	2.391(14)	Dy1–O7	2.324(12)
Dy1–O13	2.268(12)	Dy1–O16	2.345(12)
Dy2–O6	2.378(12)	Dy2–O7	2.457(11)
Dy2–O8	2.442(12)	Dy2–O10	2.367(12)
Dy2–O11	2.295(11)	Dy2–O16	2.418(12)
Dy2–O17	2.534(11)	Dy2–N1	2.473(14)
Dy2–N2	2.585(15)	Dy3–O7	2.512(13)
Dy3–O8	2.411(11)	Dy3–O13	2.285(11)
Dy3–O14	2.446(12)	Dy3–O16	2.421(11)
Dy3–O17	2.455(12)	Dy3–O23	2.318(13)
Dy3–N3	2.572(13)	Dy3–N4	2.509(14)
Dy4–O8	2.365(11)	Dy4–O11	2.295(11)
Dy4–O17	2.261(12)	Dy4–O19	2.410(14)
Dy4–O20	2.433(10)	Dy4–O21	2.405(14)
Dy4–O22	2.374(14)	Dy4–O24	2.399(12)
N1–Dy2–N2	73.3(4)	N4–Dy3–N3	74.6(4)
Dy1–O7–Dy2	106.0(5)	Dy1–O7–Dy3	93.7(4)
Dy2–O7–Dy3	84.7(4)	Dy3–O8–Dy2	87.3(4)
Dy4–O8–Dy2	94.6(4)	Dy4–O8–Dy3	106.3(5)
Dy4–O11–Dy2	101.3(4)	Dy1–O13–Dy3	101.7(4)
Dy1–O16–Dy2	106.6(5)	Dy1–O16–Dy3	95.6(4)
Dy2–O16–Dy3	87.6(4)	Dy3–O17–Dy2	84.3(4)
Dy4–O17–Dy2	94.8(4)	Dy4–O17–Dy3	108.2(4)
complex 2
Gd1–O1	2.469(16)	Gd1–O2	2.380(16)
Gd1–O3	2.458(18)	Gd1–O4	2.412(14)
Gd1–O5	2.462(14)	Gd1–O7	2.343(14)
Gd1–O13	2.300(15)	Gd1–O16	2.359(14)
Gd2–O6	2.402(16)	Gd2–O7	2.508(14)
Gd2–O8	2.465(14)	Gd2–O10	2.405(16)
Gd2–O11	2.344(14)	Gd2–O16	2.475(14)
Gd2–O17	2.505(14)	Gd2–N1	2.509(17)
Gd2–N2	2.593(17)	Gd3–O7	2.538(15)
Gd3–O8	2.416(14)	Gd3–O13	2.289(14)
Gd3–O14	2.458(14)	Gd3–O16	2.430(14)
Gd3–O17	2.464(14)	Gd3–O23	2.351(16)
Gd3–N3	2.596(16)	Gd3–N4	2.529(17)
Gd4–O8	2.388(14)	Gd4–O11	2.289(15)
Gd4–O17	2.325(14)	Gd4–O19	2.468(18)
Gd4–O20	2.456(14)	Gd4–O21	2.429(18)
Gd4–O22	2.416(18)	Gd4–O24	2.392(16)
Gd1–O7–Gd2	105.3(6)	Gd1–O7–Gd3	93.7(5)
Gd2–O7–Gd3	84.1(4)	Gd3–O8–Gd2	87.7(4)
Gd4–O8–Gd2	94.6(5)	Gd4–O8–Gd3	107.2(5)
Gd4–O11–Gd2	100.7(5)	Gd3–O13–Gd1	101.9(5)
Gd1–O16–Gd2	105.8(5)	Gd1–O16–Gd3	96.2(5)
Gd3–O16–Gd2	87.1(4)	Gd3–O17–Gd2	85.7(4)
Gd4–O17–Gd2	95.1(5)	Gd4–O17–Gd3	107.6(5)

Moreover, there exist extensive hydrogen bonds between lattice methanol molecules, between lattice water molecules, and between the lattice molecule (MeOH or H2O) and the cluster molecule (Figure S1), these weak intermolecular interactions act to stabilize the crystal structure.

The structure of complex 2 (Figure b) is very similar to that of complex 1 (Figure a). However, the Gd–N bond lengths for 2 (average 2.557 Å, Table ) are slightly longer than the Dy–N bond distances for 1 (average 2.536 Å, Table ), and the Gd–O bond lengths (mean 2.413 Å, Table ) for 2 are also larger than the Dy–O bond distances for 1 (average 2.390 Å, Table ) because of the lanthanide contraction effect.

Notably, there are three acetate ligands and three hexafluoroacetylacetonate ligands in the molecules of complexes 1 and 2; these different ligands cannot be symmetrically distributed on both sides of the Ln4 cluster core, which thus causes the whole molecule to lose its centrosymmetry. Complexes 1 and 2 represent the first chiral Ln(III) cluster complexes constructed by the asymmetric distribution of different achiral ligands on both sides of the cluster core.

Magnetic Properties

Variable temperature dc magnetic susceptibilities of both complexes (Figure ) showed that the χT values at 300 K are 56.61 cm3 kmol–1 for 1 and 31.56 cm3 kmol–1 for 2, which are in good agreement with the calculated values of 56.68 cm3 kmol–1 for the four uncoupled Dy3+ ions and 31.75 cm3 kmol–1 for the four isolated Gd3+ ions, respectively. As the temperature decreases, their χT values start to decrease slowly at first, then decrease sharply at about 25 K, and the values at 2 K are 23.94 cm3 kmol–1 for 1 and 14.19 cm3 kmol–1 for 2. The 1/χ–T data of 2 conform to the Curie–Weiss law (Figure S2). After being fitted, the C value of 31.66 cm3 kmol–1 and the θ value of −2.15 K were given for 2. This negative and small θ value indicates that there exists weak antiferromagnetic coupling among the Gd3+ ions. The 1/χ–T data of 1 can also be fitted using the Curie–Weiss law to obtain the θ value of −4.01 K and the C value of 57.01 cm3 kmol–1 (Figure S3). The negative and small θ value of 1 implies that besides the thermal depopulation of mj levels of the Dy3+ ion, there may also be weak antiferromagnetic coupling among the Ln3+ ions in 1, similar to 2.

Figure 2

Plots of χT vs T of 1 and 2.

The field-dependent magnetization of both complexes at different temperatures was measured too. For 1, the M–H/T plots do not overlap at 2–6 K (Figure S4), indicating that 1 has magnetic anisotropy, which is generally beneficial for SMM properties.

Figure a shows the M–H plots at 2–10 K of 2, which can be used to evaluate the magnitude of the magnetocaloric effect. According to the Maxwell formula, ΔSm(T)Δ = ∫[∂M(T,H)/∂T]dH,[69] we could figure out the corresponding magnetic entropy change values of 2 at different temperatures under varied magnetic field differences (ΔH). Figure b reveals that the −ΔSm value gradually increases both with increasing ΔH and with decreasing temperature. The largest −ΔSm value of 21.88 J kg–1 K–1 occurs at 2 K when ΔH is 5 T, this value is smaller than the value of 31.43 J kg–1 K–1 calculated using the formula, −ΔSm = nR ln(2S + 1) (n = 4, S = 7/2 and the R value is 8.314 J mol–1 K–1), owing to the antiferromagnetic interaction among Gd3+ ions in 2. This value is larger in magnetic refrigeration molecule materials, and comparable with those of other Gd4 cluster complexes when ΔH is 5 T.[70−75]

Figure 3

Magnetization vs field plots of 2 at 2–10 K (a); plots of −ΔSm vs T of 2 (b).

As to the magnetic dynamics of 1, we first measured the ac magnetic susceptibility of 1 under zero dc field. The temperature-dependent ac magnetic susceptibility indicates the χ″–T plots display frequency dependence and the trend of double magnetic relaxation, but the peak shape in the high-temperature region is not obvious (Figure S5). In order to investigate whether the existence of quantum tunneling effects prevents such peaks in the high-temperature region from appearing, we measured the magnetic field-dependent ac magnetic susceptibility at 13 K and 997 Hz (Figure S6). However, the optimal magnetic field has not been found. Moreover, we tried to measure the ac magnetic susceptibility at 997 Hz under 1400 and 2000 Oe, and they also cannot form peaks around 13 K, similar to the ac magnetic susceptibility at 0 Oe (Figure S7), the quantum tunneling effect is thus excluded.

However, the frequency-dependent ac magnetic susceptibility at the 0 Oe field reveals that the χ″−ν plots can show a temperature-dependent peak in a wide range of 2–14 K (Figure a), although there is also a trend of double magnetic relaxation. Using these data to plot the ln(τ) versus 1/T curve (Figure b), it can be seen that the plot deviates significantly from the straight line at lower temperatures, indicating that besides the Orbach process, the magnetic relaxation also has the two-phonon Raman process. Therefore, we fitted this curve with the equation τ–1 = τ0–1 exp(−Ueff/kT) + CT including both the Raman process and the Orbach process, yielding n = 1.67, C = 12.39 s–1 K–1.67, τ0 = 4.9 × 10–6 s, and Ueff/k = 48.4 K. The large C value of 1 suggests that the Raman process is more prominent in the magnetic relaxation.[76] Furthermore, the Raman process can also be seen in the χ″−ν plots (Figure a), which show not only the broad peak shape at low temperature but also the high-frequency plateau shape with increasing temperature.[77−79] The τ0 value of 4.9 × 10–6 s is a normal value for SMMs.[1] The Ueff/k value (48.4 K) is comparable with that of the fast Orbach process (44.0 K) but smaller than that of the slow Orbach process in [Dy4(H3L)2(OAc)6]·2EtOH (107.0 K).[57]

Figure 4

Frequency dependence of χ″ for 1 at zero dc field (a); ln(τ) vs 1/T plot for 1, the solid line represents the best fitting (b).

In addition, the Cole–Cole curves in the χ″−χ′ plots show partial characteristics of the double magnetic relaxation (Figure S8), which are not surprising because 1 contains Dy3+ ions in two coordination configurations, and the Cole–Cole plots at 2–14 K could be fitted by the formula containing two Debye functions.[57,80,81] The fitting results showed that its α1 value range is 0.15–0.41 and its α2 value range is 0.33–0.56 (Table S5). The larger values of α1 and α2 may be closely related to the obvious Raman mechanism in the magnetic relaxation process.[38,76] There is no open hysteresis loop for 1 at 1.9 K (Figure S9).

Conclusions

In summary, two chiral Ln4 cluster complexes derived from N,N′-(1,3-propanediyl)bis[N-[1,1-bis(hydroxymethyl)-2-hydroxyethyl]amine] were synthesized successfully although no chiral ligands were used. It is the asymmetric distribution of acetate ligands and hexafluoroacetylacetonate ligands on both sides of the parallelogram-like Ln4 core, which leads to the chirality of the whole molecule. The Dy complex exhibits SMM properties at the zero field, including not only the Orbach process but also the Raman process. While the Gd complex has large magnetic entropy changes, it is a potential molecular material for magnetic refrigeration. This work demonstrates that the chiral Ln(III) cluster complexes can be constructed from achiral ligands through the asymmetric distribution of different ligands on both sides of the cluster core. More such chiral cluster molecular materials can be assembled by changing the ligands at both ends of the metal cluster core or changing the metal cluster core itself.