Improving the performance of β-diketonate-based Dy^III single-molecule magnets displaying luminescence thermometry†

Airton Germano Bispo-Jr ^ab, Laurence Yeh ^a, Dylan Errulat ^a, Diogo Alves Gálico ^a, Fernando Aparecido Sigoli ^b and Muralee Murugesu *^a
aDepartment of Chemistry and Biomolecular Sciences, University of Ottawa, Ottawa, Ontario K1N 6N5, Canada. E-mail: M.Murugesu@uottawa.ca
^bDepartment of Inorganic Chemistry, Institute of Chemistry, University of Campinas, Campinas 13083-970, Brazil

First published on 16th June 2023

---

---

Inspired by the fast-paced digital era, single-molecule magnets (SMMs) have emerged over the past thirty years as promising candidates for ultra-compact information storage in next-generation molecular devices.¹ Dy^III coordination compounds stand out as SMMs due to the unquenched angular momentum in the ⁶H_15/2 ground state, strong spin–orbit coupling, and an unparalleled uniaxial anisotropy.² In this regard, mononuclear Dy^III complexes with β-diketonate ligands have been widely investigated as a result of the highly tuneable structure and bidentate chelating coordination mode that enables the formation of stable complexes.³ These are often paired with nitrogen-containing heterocycles in order to impart new structural and electronic properties.⁴ Moreover, the triplet state energy of β-diketonates such as acetylacetonate (acac⁻) and N-heterocycles like 2,2′-bipyrimidine (bpm) may enable energy transfer (ET) to the ⁴F_9/2 excited level of Dy^III, favouring a bright luminescence through the antenna effect.⁵ In such multifunctional luminescent SMMs, the Dy^III luminescence may also enable the probing of in situ temperature by luminescence thermometry which takes advantage of the thermal population of Stark sublevels (M_J) arising from the Dy^{III 4}F_9/2 emitting level.⁵

Several β-diketonate Dy^III mononuclear SMMs have been reported (Table S1, ESI†) over the years, which typically exhibit an effective energy barrier (U_eff) that ranges between 50 and 300 K.³ On the other hand, for cases where luminescence thermometry is investigated as a feature of Dy^III SMMs, the U_eff is generally lower than 150 K (Table S2, ESI†), showing that boosting their magnetic and luminescent performance still remains an open challenge. That is because in order to maximize the magnetic anisotropy of Dy^III, a high pseudo-symmetry of the Dy^III polyhedron is targeted as a means to decrease the equatorial crystal field (CF) contribution and increase the axial CF terms.² Conversely, the luminescence thermometry based on Stark sublevels could be boosted by distorted pseudo-symmetries of the Dy^III polyhedron, which should decrease the energetic separation of the M_J sublevels arising from the ⁴F_9/2 emitting level. Searching for suitable candidates for luminescent mononuclear SMMs, we encountered the crystal structures of the complexes [Ln(acac)₃bpm] (Ln = Eu, Tb), which were absent of synthetic or characterization details (CCDC number = 906668, 906664, 906667, 906666). Not only are these complexes promising as multifunctional luminescent SMMs, they also have the potential to act as building blocks for extended polynuclear structures. Yet, only a few examples of mononuclear Ln^III complexes with an “open pocket” on bpm have been reported, presumably due to the tendency of Lewis acidic Ln^III to bind to neutral N-donors. The only examples to date are [Ln(L¹)₃bpm] (L = N,N′-dipyrrolidine-N′′-trichloracetylphosphortriamide; Ln = La, Nd, Eu, Gd, Er, Y)⁶ and [Ln(NO₃)₃(bpm)(MeOH)₂]·MeOH (Ln = Ce, Nd, Sm).⁷ A few examples feature two bpm per Ln^III, but this configuration limits their tunability.^7,8 In light of this guidance, herein, we present the synthesis and characterization of [Dy(acac)₃bpm] (1) and we demonstrate how the coordination of bpm can promote high U_eff in multifunctional β-diketonate mononuclear Dy^III SMMs featuring luminescence thermometry.

The synthesis of 1 was achieved by mixing equimolar amounts of [Dy(acac)₃(H₂O)₂] and bpm in methanol, as described in the supplementary information. The complex crystallizes in the triclinic crystalline system – P space group – (Fig. 1), along with one methanol molecule within the lattice; other crystallographic data are displayed in Table S3 (ESI†). The primary coordination environment of Dy^III (coordination number = 8) is filled by six oxygen atoms coming from acac⁻ and two nitrogen atoms from the bpm with bond distances of 2.6211(20) Å (Dy-N1) and 2.5842(20) Å (Dy-N3), and Dy–O distances in the range 2.3081(20)–2.3360(19) Å (Table S4, ESI†). The DyO₆N₂ polyhedron is found to have a D_4d pseudo-symmetry, which was determined by continuous shape measurement (CShM = 0.893, Fig. 1a and Table S5, ESI†).⁹ The principal magnetic axis was calculated using the Magellan software (Fig. 1).¹⁰ The orientation of the magnetic anisotropy was found to bisect two acac⁻ ligands in the axial positions, with the bpm ligand and one acac⁻ (O3 and O4) coordinated transverse to the principal axis. The Dy–N bond lengths in 1 are generally longer than in [Dy(acac)₃X]¹¹ (X = a neutral bidentate ligand forming two Dy–N bonds) congener SMMs displaying similar disposition of the neutral terminal ligand around the magnetic axis, as addressed in Table S4 and Fig. S3 (ESI†). The longer Dy–N bonds in 1 are a result of the non-coordinated nitrogens that weaken the bonds to Dy^III through electron-withdrawing effects. Finally, the shortest Dy⋯Dy intermolecular distance in 1 is 8.4139(6) Å, as revealed by the packing arrangement, Fig. S4 (ESI†).

Before further investigating the role of the bpm ligand on the slow relaxation of magnetization, photoluminescence spectra were measured at 9 K (Fig. S5, ESI†) to estimate the splitting of the Dy^III ground and emitting levels that arise due to the CF. In the excitation spectrum (Fig. S5a, ESI†), the S_n ← S₀ electronic transitions of acac⁻ as well as bpm are observed. Upon ligand excitation (Fig. S5a, ESI†), the typical Dy^III emission coming from the ⁴F_9/2 → ⁶H_15/2,_13/2,_11/2 transitions are detected while the ⁴F_9/2 emitting-level lifetime is 18.53 ± 0.09 μs (Fig. S6, ESI†). In the excitation spectrum monitored over the ⁴F_9/2 ← ⁶H_15/2 transition range (Fig. S7, ESI†), five components are observed. This aligns with expectations, given that the ⁴F_9/2 excited state has a total angular momentum quantum number (J) of 9/2 and that Dy^III is a Kramers ion. Consequently, the maximum CF splitting of a level into Kramers doublets (KDs, equivalent to the M_J sublevels) should be J + 1/2. On the other hand, by monitoring the same transition in the emission spectra (Fig. S7, ESI†), at least 15 components are detected, which are more than the 8 KDs expected for the ⁶H_15/2 level. This observation leads us to conclude that at 9 K, the upper energy KD2 arising from the ⁴F_9/2 excited level is populated, giving rise to the so-called “hot-bands”. This behaviour happens due to ET from the ligands to the ⁴F_9/2 emitting level of Dy^III. The position of the ⁴F_9/2 ↔ ⁶H_15/2 transition in the excitation and emission spectra were used to calculate the relative energy of each KD coming from these levels (Fig. 1b and Table S6, ESI†), revealing an energetic separation between KD1 and KD2 of 187 cm⁻¹ (⁶H_15/2 ground level) and 74 cm⁻¹ (⁴F_9/2 emitting level).

To evaluate the influences of bpm acting as a terminal ligand over the magnetic dynamics of 1, direct-current (dc) and alternating-current (ac) magnetic susceptibility studies were undertaken. For the dc magnetic susceptibilities measured under an applied field (H) of 1000 Oe (Fig. S8, ESI†), the χ_MT product reaches a value of 14.14 cm³ mol⁻¹ K at 300 K, which is in agreement with the theoretical free-ion value for a single Dy^III ion (S = 5/2, L = 5, ⁶H_15/2, g = 4/3, C = 14.17 cm³ mol⁻¹ K). At lower temperatures, there is a gradual decrease of χ_MT until 3.7 K, followed by a steep downturn to 10.94 cm³ mol⁻¹ K (1.8 K). This behaviour can arise from combined thermal depopulation of Stark sublevels, significant magnetic anisotropy, as well as non-negligible antiferromagnetic intermolecular interactions between Dy^III.⁵ In the magnetization plots (Fig. S10, ESI†), the non-overlap of reduced magnetization curves measured under different temperatures (Fig. S10, ESI†) confirms the presence of non-negligible magnetic anisotropy, whose s-shape in the isotherm line at 1.9 K also indicates the presence of magnetic blocking. This is corroborated by zero-field-cooled/field-cooled (ZFC/FC) measurements (Fig. S9, ESI†), revealing a divergence of the two data sets at 3.0 K due to the pinning of the magnetic moment below this temperature region. Consequently, magnetic hysteresis measurements from 70 to −70 kOe (average sweep rate of 25 Oe s⁻¹, Fig. S11, ESI†), reveal magnetic blocking up to a temperature of 3.5 K, which is in agreement with the ZFC/FC measurement. Rapid demagnetization is observed at zero field for all temperatures studied, resulting in butterfly-shaped loops. This results from ground-state quantum tunnelling of the magnetization (QTM) in the absence of a magnetic field, whose presence is discussed subsequently.

To confirm the presence of slow relaxation of the magnetization, the properties of 1 were studied via ac susceptibility measurements. To first investigate the temperature dependence of the relaxation dynamics, variable-temperature ac susceptibility was undertaken in the absence of a magnetic field (Fig. S12, ESI† and Fig. 3a). Under these conditions, a single frequency-dependent ac susceptibility signal (Fig. 2a), is observed up to 24 K. The out-of-phase ac susceptibility (χ′′) was fitted using the generalized Debye model (eqn S1, Table S7, ESI†)¹² to extract the temperature-dependent relaxation times (τ, Table S7, ESI†). To examine the relaxation mechanisms responsible for the ac susceptibility behaviour, the inverse of the relaxation times (τ⁻¹) were fitted with eqn S3 (ESI†) (Fig. 2c). A fit equation with contributions from QTM, Raman, and Orbach relaxation mechanisms provided the best match to the experimental data. The fit reveals that QTM is the main relaxation pathway at temperatures below 8 K, resulting in a plateau of the relaxation times (τ_QTM = 4.5 × 10⁻³ s). This is consistent with the hysteresis measurements where magnetization is spontaneously lost at H = 0 Oe. Within the temperature range 12–18 K, the relaxation times follow a power law which is indicative of Raman being the main contributor within this temperature range (n = 3.9 and C = 1.4 × 10⁻³ s⁻¹ K⁻ⁿ). Above 18 K, the Orbach process becomes dominant, with a U_eff = 309 K (214.7 cm⁻¹) and τ₀ = 3.2 10⁻¹¹ s (Table S8, ESI†).

Ac susceptibility measurements under variable field were also undertaken to investigate the dependence of the magnetic relaxation on the applied static field (H = 0–5000 Oe, 6 K) (Fig. S13, ESI†). From 0 to 1000 Oe, a frequency-dependent χ′′ signal is observed in the higher-frequency (HF) region while from 200 to 5000 Oe, another frequency-dependent signal of χ′′ takes place in the lower-frequency (LF) region. The χ′′ curves were fitted by applying the generalized single (eqn S1) or double (eqn S2) Debye model fits (Table S9, ESI†) to calculate the field-dependent τ values.¹² The dependence of the τ⁻¹ on the applied magnetic field for both LF and HF processes (Fig. S14, ESI†) reveals that an applied magnetic field of 1200 Oe should be sufficient to simultaneously favour only the LF process and minimize the QTM contribution through the removal of the degeneracy between the Dy^III KDs.¹³

As expected, the QTM contribution is eliminated under an external magnetic field of 1200 Oe, which is attested by the temperature-dependent χ′′ signals over the entire studied temperature range (5–24 K, Fig. 2b). The τ⁻¹ calculated from the fitting of the χ′′ using the generalized Debye model,¹² (Table S10, ESI†) were plotted as a function of the temperature (Fig. 2d), and fitted with eqn S3 (ESI†). The best fit was obtained from an equation considering only the Raman and Orbach mechanism contributions, confirming that the QTM is effectively supressed (fitting parameters displayed in Table S8, ESI†). In this case, the Raman mechanism (n = 6.1, C = 3.1 10⁻⁵ s⁻¹ K⁻ⁿ) is the main contributor to the relaxation dynamics up to 19 K, while at higher temperatures, the Orbach mechanism dominates (U_eff = 345 K (239.7 cm⁻¹), τ₀ = 6.1 10⁻¹² s). This enhancement of the U_eff from 214.7 cm⁻¹ (H = 0 Oe) to 239.7 cm⁻¹ (H = 1200 Oe) is due to the decreased QTM contribution under the applied magnetic field.¹³ Moreover, since the U_eff obtained under an applied magnetic field of 1200 Oe is close to the energy of KD4 (269 cm⁻¹) arising from the Dy^{III 6}H_15/2 ground state, it is likely that the magnetic relaxation occurs mainly via the KD4 sublevel. A small different between the experimental U_eff compared to the KD4 energy obtained from the photoluminescence spectrum is expected given the contribution of Raman and QTM to the relaxation dynamics.

Comparing the magnetic properties of 1 to the state-of-the-art SMMs based on mononuclear Dy^III β-diketonate complexes³ (Table S1, ESI†), the U_eff value obtained from 1 is among the largest, confirming that bpm as a terminal ligand plays a positive role on the magnetic dynamics of Dy^III β-diketonate complexes. That is because the slightly-distorted D_4d pseudo-symmetry of the Dy^III polyhedron enables a considerable separation of the KDs arising from the Dy^{III 6}H_15/2 ground state, as highlighted by luminescence, which should minimize the transverse CF terms, minimizing the QTM relaxation.¹⁴ Furthermore, the long Dy–N bonds in 1 should also contribute to decrease the transverse CF terms, enhancing the U_eff.² This effect is a consequence of the nearby non-coordinated nitrogen atoms imparting electron-withdrawing effects on the coordinated nitrogen atoms, which lengthens the Dy–N bonds.

Apart from the slow relaxation of magnetization in [Dy(acac)₃bpm], the integration of self-temperature monitoring is attractive since temperature is a paramount parameter affecting the SMM behaviour. To study this feature, temperature-dependent emission spectra were recorded for the emission band assigned to the ⁴F_9/2 → ⁶H_13/2 transition (Fig. 3a). As expected, changes in the relative intensities of the Stark components of the ⁴F_9/2 → ⁶H_13/2 band due to the temperature take place, which arises from the thermal population of KD2 from KD1, as represented in Fig. 1b. The ratio between the integrated intensities within the 555–570.8 nm (I₁) and 577.0–577.6 nm (I₂) ranges was used to calculate a thermometric parameter (Δ = I₂/I₁, Fig. 3b), and its dependency on temperature was fitted to a logistic function (S6, ESI†). The relative sensitivity (S_r) is larger than 1% K⁻¹ in the temperature range 30–140 K, and the maximum relative sensitivity (S_m = 1.51% K⁻¹) is achieved at 70 K (Fig. 3c), which is comparable to other multifunctional SMMs based on Dy^III operating in the same temperature range (Table S2, ESI†).⁵ On the other hand, within the SMM operating temperature range (up to 24 K), the thermal sensitivity is within the range 0.2–0.7% K⁻¹, which is lower than other Dy^III SMMs,⁵ albeit displaying a comparatively lower U_eff than [Dy(acac)₃bpm]. The temperature uncertainty lies within the range 0.12–0.02 K, which is typical of assays performed with highly-sensitive photomultipliers.

Overall, this study highlights the versatility of bpm to act as terminal ligand to build multifunctional mononuclear complexes with magnetic and luminescent properties. We have synthesized the mononuclear complex [Dy(acac)₃bpm], yielding an effective barrier of 309 K (at 0 applied dc field) and waist-restricted hysteresis up to 3.5 K. Complex 1 also displays the highest U_eff reported so far for a multifunctional SMM presenting luminescence thermometry. We highlight that the D_4d pseudo-symmetry of the Dy^III polyhedron favours a large CF splitting of KDs arising from the Dy^III ground state, while the longer Dy–N bonds should minimize the equatorial CF terms that decrease the QTM relaxation. The thermal population of KD2 arising from the Dy^{III 4}F_9/2 emitting level enabled the development of a luminescent temperature probe featuring a maximum sensitivity of 1.51% K⁻¹ (70 K). Aside from the magnetic and luminescent features, we have also developed the first formalized synthetic procedure to access mononuclear lanthanide β-diketonate complexes with an accessible coordination site on the bpm. As such, we anticipate this complex could serve as a starting point for multifunctional materials with a new level of tunability in the future. Therefore, the lessons learned from this study help to understand how astute molecular design combined with the ligand-to-Dy^III energy transfer can perform synergistically to enhance the performance of luminescent mononuclear Dy^III SMMs.

M. M. acknowledges the financial support provided by the University of Ottawa, the Canadian Foundation for Innovation (CFI), and the Natural Sciences and Engineering Research Council of Canada (NSERC). A. G. B. J. acknowledges the São Paulo Research Foundation FAPESP (2021/09755-0 and 2019/23763-6). F. A. S. acknowledges FAPESP: 2021/06326-1 and 2021/08111-2, INCT/INOMAT- (CNPq: 465452/2014–0 and FAPESP: 50906–9/2014).

Conflicts of interest

There are no conflicts to declare.

References

1. K. L. M. Harriman, D. Errulat and M. Murugesu, Trends Chem., 2019, 1, 425 CrossRef CAS .
2. J. D. Rinehart and J. R. Long, Chem. Sci, 2011, 2, 2078 RSC .
3. (a) P.-P. Cen, S. Zhang, X.-Y. Liu, W.-M. Song, Y.-Q. Zhang, G. Xie and S.-P. Chen, Inorg. Chem., 2017, 56, 3644 CrossRef CAS PubMed ; (b) M. Kong, X. Feng, J. Li, Z.-B. Hu, J. Wang, X.-J. Song, Z.-Y. Jing and Y.-Q. Zhang, Dalton Trans., 2020, 49, 14931 RSC ; (c) J. Xi, X. Ma, P. Cen, Y. Wu, Y.-Q. Zhang, Y. Guo, J. Yang, L. Chen and X. Liu, Dalton Trans, 2021, 50, 2102 RSC .
4. A. Øwre, M. Vinum, M. Kern, J. van Slageren, J. Bendix and M. Perfetti, Inorganics, 2018, 6, 72 CrossRef .
5. D. Errulat, R. Marin, D. A. Gálico, K. L. M. Harriman, A. Pialat, B. Gabidullin, F. Iikawa, O. D. D. Couto Jr, J. O. Moilanen, E. Hemmer, F. A. Sigoli and M. Murugesu, ACS Cent. Sci., 2019, 5, 1187 CrossRef CAS PubMed .
6. K. O. Znovjyak, O. V. Moroz, V. A. Ovchynnikov, T. Y. Sliva, S. V. Shishkina and V. M. Amirkhanov, Polyhedron, 2009, 28, 3731 CrossRef CAS .
7. G. Zucchi, O. Maury, P. Thuéry, F. Gumy, J.-C. G. Bünzli and M. Ephritikhine, Chem. – Eur. J., 2009, 15, 9686 CrossRef CAS PubMed .
8. A. Fratini, G. Richards, E. Larder and S. Swavey, Inorg. Chem., 2008, 47, 1031 CrossRef .
9. M. Pinsky and D. Avnir, Continuous Symmetry Measures. 5. The Classical Polyhedra, Inorg. Chem., 1998, 37, 5575 CrossRef CAS PubMed .
10. N. F. Chilton, D. Collison, E. J. L. McInnes, R. E. P. Winpenny and A. Soncini, Nat. Commun., 2013, 4, 2551 CrossRef PubMed .
11. (a) G.-J. Chen, Y.-N. Guo, J.-L. Tian, J. Tang, W. Gu, X. Liu, S.-P. Yan, P. Cheng and D.-Z. Liao, Chem. – Eur. J., 2012, 18, 2484 CrossRef CAS PubMed ; (b) Y.-Z. Tong, C. Gao, Q.-L. Wang, B.-W. Wang, S. Gao, P. Cheng and D.-Z. Liao, Dalton Trans, 2015, 44, 9020 RSC ; (c) M. Guo, J. Wu, O. Cador, J. Lu, B. Yin, B. Le Guennic and J. Tang, Inorg. Chem., 2018, 57, 4534 CrossRef CAS PubMed ; (d) G.-J. Chen, C.-Y. Gao, J.-L. Tian, J. Tang, W. Gu, X. Liu, S.-P. Yan, D.-Z. Liao and P. Cheng, Dalton Trans, 2011, 40, 5579 RSC ; (e) G.-J. Chen, Y. Zhou, G.-X. Jin and Y.-B. Dong, Dalton Trans, 2014, 43, 16659 RSC ; (f) E. V. Gorshkov, D. V. Korchagin, E. A. Yureva, G. V. Shilov, M. V. Zhidkov, A. I. Dmitriev, N. N. Efimov, A. V. Palii and S. M. Aldoshin, Magnetochemistry, 2022, 8, 151 CrossRef CAS .
12. S. M. J. Aubin, Z. Sun, L. Pardi, J. Krzystek, K. Folting, L.-C. Brunel, A. L. Rheingold, G. Christou and D. N. Hendrickson, Inorg. Chem., 1999, 38, 5329 CrossRef CAS .
13. J.-L. Liu, Y.-C. Chen and M.-L. Tong, Chem. Soc. Rev., 2018, 47, 2431 RSC .
14. A. Swain, T. Sharma and G. Rajaraman, Chem. Commun., 2023, 59, 3206 RSC .

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 2258317. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3cc02148c

This journal is © The Royal Society of Chemistry 2023