Photoresponsive luminescent single-molecule magnets based on dysprosium–anthracene complexes: regulating the de-dimerization temperature of the photocycloaddition product by co-ligand

Abstract

Lanthanide–anthracene complexes that can undergo reversible photocycloaddition reaction are attractive for the development of photoresponsive luminescent single-molecule magnets (SMMs). However, how to regulate the de-dimerization temperature of in situ formed dianthracenes remains an open question. Here, we report two binuclear dysprosium–anthracene complexes [Dy₂(SCN)₄(L)₂(depma)₂(DEPP)₂] (1) and [Dy₂(SCN)₄(L^Me)₂(depma)₂(DEPP)₂] (2), where L is 2,6-dimethoxyphenol, L^Me is 4-methyl-2,6-dimethoxyphenol, depma is 9-diethyl-phosphonomethylanthracene, and DEPP is diethylpropylphosphonate. Both undergo single-crystal-to-single-crystal (SC–SC) photocycloaddition reactions to form the 1D coordination polymers [Dy₂(SCN)₄L₂(depma₂)(DEPP)₂]_n (1UV) and [Dy₂(SCN)₄(L^Me)₂(depma₂)(DEPP)₂]_n (2UV), where depma₂ is photo-dimerized depma, concomitant with changes in photoluminescence and magnetic properties. Interestingly, the de-dimerization temperatures of 1UV (80 °C) and 2UV (71 °C) are much lower than those of the known lanthanide–dianthracene complexes (≥100 °C). We found that the stability of in situ formed dianthracene depends largely on the spacing and slip angle of the anthracene pair in the original complex, with spacing being more important, and these parameters can be modulated by choosing suitable co-ligands. In addition, we investigated the kinetics of the photocycloaddition reaction of 1 at different temperatures and found that the rate of the reaction reached a maximum at the temperature of complete de-dimerization.

---

Introduction

Lanthanide-based single-molecule magnets (Ln-SMMs) have attracted considerable attention for their potential applications in information storage, molecular devices, and sensors.^1–5 Photo-responsive Ln-SMMs are of particular interest because their magnetic and optical properties can be manipulated by light,^6–8 akin to the other emissive magnetic systems such as spin crossover and metal–cyanide compounds.⁹ Previous studies have demonstrated that the incorporation of photoactive components is an effective method for realizing photoresponsive Ln-SMMs with tunable magnetic and optical properties,^10–15 but reports showing synergistic modulation of luminescence and SMM properties are still limited.^16,17 Among the known photoactive components, anthracene and its derivatives are well recognized for their rich photophysical properties and the ability to undergo [4 + 4] photocycloaddition reaction under UV light irradiation.^18–22 Over the past few years, we and others have constructed a series of lanthanide–anthracene complexes that exhibit light-triggered significant changes in luminescence and SMM properties.^23–26 This change is caused by the structural transformation induced by photocycloaddition of the face-to-face π–π interacting anthracene pairs. More interestingly, the process is reversible, with complete recovery to the initial un-dimerized state upon heating for a few minutes.^23,25 The temperature at which the in situ formed dianthracene undergoes de-dimerization is about 100 °C or more. This fact raises the question: can we regulate the de-dimerization temperature of in situ formed Ln-dianthracene complexes to lower temperatures for practical applications?

Our previous work has shown that the Ln-dianthracene compounds formed using pre-photodimerized dianthracene ligands have much higher de-dimerization temperatures (140–190 °C)^27–31 than those formed in situ (ca. 100–140 °C).^23,25,32,33 This result is reasonable because structural transformation induced by photocycloaddition involves rearrangement of the positions of the metal ions and ligands, which is not trivial in the solid state.^23,25,26 As a consequence, dianthracenes formed by in situ solid-state photochemical reactions are usually not in their lowest energy state and are therefore less stable than those formed by pre-photodimerization reaction.³⁴ We envision that the stability of dianthracene formed in situ may be related to the stacking pattern of the anthracene pairs prior to light irradiation. The larger the slip angle and plane-to-plane distance of the two anthracene moieties, the less stable the dianthracene is, and hence the lower the temperature of de-dimerization. Thus, in order to regulate the de-dimerization temperature of lanthanide–anthracene complexes after photocycloaddition, it is necessary to change the stacking pattern of the anthracene moieties, which can be achieved by selecting a suitable auxiliary ligand or through ligand modification.^35,36

To demonstrate the above concept, we report here two Dy-anthracene complexes, [Dy₂(SCN)₄(L)₂(depma)₂(DEPP)₂] (1) and [Dy₂(SCN)₄(L^Me)₂(depma)₂(DEPP)₂] (2), where L is 2,6-dimethoxyphenol, L^Me is 4-methyl-2,6-dimethoxyphenol, depma is 9-diethyl-phosphonomethylanthracene, and DEPP is diethylpropylphosphonate [CH₃CH₂CH₂PO(OCH₂CH₃)₂] (Scheme 1). Both can undergo photocycloaddition reaction under UV light irradiation in a single-crystal-to-single-crystal (SC–SC) fashion to form the one-dimensional (1D) coordination polymers [Dy₂(SCN)₄(L)₂(depma₂)(DEPP)₂]_n (1UV) and [Dy₂(SCN)₄(L^Me)₂(depma₂)(DEPP)₂]_n (2UV), in which depma₂ is the photodimerized depma. Notably, compounds 1 and 2 have very similar binuclear structures except that the bridging 2,6-dimethoxyphenol ligand in 2 has an additional methyl group. Such a small difference leads to distinct changes in the stacking pattern of anthracene pairs. Moreover, the structure of 1 is similar to the known compound [Dy₂(SCN)₄(L)₂(depma)₂(H₂O)₂] (3) except for the auxiliary ligand DEPP.³² Compared to H₂O, DEPP has large alkyl groups which may cause steric hindrance and affect the stacking of anthracene groups. As a result, the slip angle (θ) and plane-to-plane distance (d_pp) decrease in the following order: 2 (24.6°, 3.503 Å) > 1 (19.7°, 3.443 Å) > 3 (16.9°, 3.410 Å). Accordingly, the de-dimerization temperature of the photocycloaddition products increases in the sequence: 2UV (71 °C) < 1UV (80 °C) < 3UV (100 °C). As far as we are aware, 2UV shows the lowest de-dimerization temperature among the known metal–dianthracene complexes. We also studied the kinetics of photocycloaddition reaction of compound 1 at different temperatures and found that the reaction rate reached a maximum at the temperature of complete de-dimerization. This work not only provides a feasible route to regulate the de-dimerization temperature of metal–anthracene complexes after photocycloaddition, but also may shed light on the development of anthracene-based photoresponsive materials for practical applications.

Scheme 1 Structural transformation of an anthracene pair after UV light irradiation.

Results and discussion

Crystal structures

Single-crystal structure analyses revealed that compounds 1 and 2 crystallized in the triclinic P1̄ space group (Tables 1 and S1). As shown in Fig. 1a, compound 1 has a centrosymmetric binuclear structure. The asymmetric unit contains one Dy^III ion, two SCN⁻ ions, one DEPP molecule, one depma, and one L ligand. Each Dy^III ion is coordinated with four O atoms (O4, O5, O5A, and O6A) from the two L ligands, one O atom (O7) from DEPP, one O atom (O1) from depma and two N atoms (N1, N2) from SCN⁻ to form an eight-coordinated [DyO₆N₂] configuration. The Dy–O bond lengths are 2.295(3)-2.657(4) Å, and the Dy–N bond lengths are 2.379(4)–2.392(5) Å (Table S2). According to continuous shape measure (CShM) analysis,³⁷ the geometry of the [DyO₆N₂] polyhedron in 1 is best described as distorted Biaugmented trigonal prism J50 (CShM = 2.588, C2v, Table S3). The two equivalent Dy centres are bridged by two oxygen atoms O5 and O5A from two tridentate ligands of L⁻, forming a binuclear unit of Dy₂O₂ with a Dy1-O5-Dy1A angle of 114.9° and Dy1⋯Dy1A distance of 3.871 Å. The two depma ligands locate on the two sides of the Dy₂O₂ dimer and each is π–π interacted with the anthracene groups of the adjacent dimer. Thus, an infinite supramolecular chain is constructed through intermolecular π–π interactions of the anthracene groups (d_C2–C9A = 3.771 Å, centre-to-centre distance d_cc = 3.773 Å, plane-to-plane distance d_pp = 3.443 Å) (Fig. 1b). The slip angle of the anthracene pair, defined as the angle between the centroid–centroid line and the vertical line, is 19.7°. The supramolecular chains are further connected by extensive hydrogen bond interactions through C9–H9⋯S2, C33–H33B⋯S2 and C25–H25⋯S1 contacts (Table S4 and Fig. S1a, S2a, S3a), forming a 3D supramolecular network. The shortest intermolecular Dy⋯Dy distance is 8.566 Å.

Table 1 The cell and structural parameters of 1, 1UV, 2 and 2UV

	1	1UV	2	2UV
T (K)	193	193	193	193
Crystal system	Triclinic	Triclinic	Triclinic	Triclinic
Space group	P1̄	P1̄	P1̄	P̄1
a (Å)	11.510(3)	11.584(1)	11.684(5)	11.753(1)
b (Å)	12.095(4)	12.160(1)	11.772(6)	11.913(1)
c (Å)	15.978(8)	15.597(1)	15.675(7)	15.751(1)
α (°)	75.0(1)	76.6(1)	79.6(1)	77.928(1)
β (°)	77.5(1)	78.8(1)	88.7(1)	87.062(1)
γ (°)	74.4(1)	72.2(1)	72.5(1)	73.664(1)
V (Å^3)	2044.0(1)	2016.7(2)	2021.1(2)	2069.5(2)
Dy1-O5, Dy1-O5A/Å	2.295(3), 2.297(3)	2.311(3), 2.290(3)	2.250(5), 2.284(5)	2.285(4), 2.303(4)
Dy1-O1, Dy1-O7/Å	2.319(3), 2.295(4)	2.314(4), 2.289(4)	2.330(5), 2.271(6)	2.344(4), 2.301(5)
Dy1-O4, Dy1-O6A/Å	2.657(4), 2.596(4)	2.643(4), 2.601(4)	2.635(6), 2.555(6)	2.601(4), 2.649(4)
Dy1-N1, Dy1-N2/Å	2.392(5), 2.379(4)	2.388(5), 2.398(5)	2.388(7), 2.360(7)	2.380(6), 2.389(6)
Dy1⋯Dy1A/Å	3.871(1)	3.888(1)	3.833(1)	3.872(1)
Dy1-O5-Dy1A/°	114.9(1)	115.4(2)	115.4(2)	115.1(1)
Slip angle θ/°	19.7	—	24.6	—
d cc/Å	3.773	—	3.868	—
d pp/Å	3.443	—	3.503	—
d C2–C9A/Å	3.771	1.669	3.860	1.689

---

Fig. 1 (a) The molecule structure and (b) 1D supramolecular chain of 1. (c) The building unit and (d) 1D chain in structure 1UV. (e) The molecule structure and (f) 1D supramolecular chain of 2. (g) The building unit and (h) 1D chain in structure 2UV. The DEPP ligands are highlighted by red colour.

Compound 2 has a similar structure to 1 except that the L ligand is replaced by its methyl derivative, L^Me (Fig. 1e). This replacement has an effect on the coordination environment of the Dy centre, as evidenced by the changes in bond lengths and angles (Tables 1 and S5). Accordingly, the [DyO₆N₂] polyhedron in 2 deviates more from the biaugmented trigonal prism (CShM = 2.669 vs. 2.588 in 1). Moreover, the Dy1⋯Dy1A distance over the phenolic oxygen bridges slightly reduces from 3.872 Å in 1 to 3.833 Å in 2, but the shortest intermolecular Dy⋯Dy distance of 2 is slightly increased (8.605 vs. 8. 566 Å in 1). It is worth noting that the minor change in the bridging ligand poses distinct influences on the stacking pattern of anthracene groups. The centre-to-centre (d_cc) and plane-to-plane (d_pp) distances in 2 are larger than those in 1 (3.868 and 3.503 Å vs. 3.773 and 3.443 Å in 1) (Table 1 and Fig. 1f). The slip angle of the π–π interacting anthracene groups in 2 is also larger than that in 1 (24.6° vs. 19.7° in 1). We expect that these structural differences will affect the photophysical and magnetic properties of the two complexes (vide infra). Nevertheless, the face-to-face π–π interacting anthracene units in both 1 and 2 satisfy the Schmidt's rule for photocycloaddition reaction.³⁸

Photophysical and photochemical properties of 1 and 2

To investigate the photophysical properties of compounds 1 and 2, we first measured their UV-vis diffuse reflectance spectra in the solid state. Both exhibit strong and broad bands in the 200–600 nm range (Fig. S4), attributed to the π* ← π transition of the anthracene ligand. Additionally, a weak peak was observed at 754 nm for both complexes, which is assigned to the f–f transition from the ground state ⁶H_15/2 to the excited states ⁶F_3/2 and ⁶F_5/2 of the Dy³⁺ ion.³⁹

We then recorded the photoluminescence (PL) spectra of 1 and 2. Both exhibited bright yellow luminescence. Upon excitation at 365 nm, compound 1 shows a broad and strong emission band peaking at 555 nm with a lifetime of 55.2 ns and a quantum yield of 5.09% (Fig. 2a). This band was originated from the excimer emission due to the presence of the face-to-face π–π interacting anthracene rings.^35,40 Interestingly, although compound 2 displays a similar emission profile, the emission peak is markedly blue-shifted to 520 nm (Fig. 2b). This blue-shift should be related to its structural difference compared to 1. As mentioned above, the plane-to-plane spacing and slip angle of the anthracene pairs in 2 are larger than those in 1 (3.503 Å and 24.6° in 2vs. 3.443 Å and 19.7° in 1). Therefore, we expect a weaker electronic coupling and hence a higher energy of the excimer state,⁴¹ a shorter emission lifetime (14.4 ns) and a lower quantum yield (4.96%) for 2 compared to those for 1.⁴²

Fig. 2 PL spectra excited at 365 nm for 1, 1UV and 1UV-R (a), and 2, 2UV and 2UV-R (b).

Since the stacking pattern of the anthracene pairs in 1 and 2 is consistent with the Schmidt's rule for photocycloaddition reactions, we next investigated their photochemical properties. Fig. 3a shows the PL spectra of 1 irradiated under 395 nm UV light (LED, IUVOT 50 W) for different times. It is obvious that the intensity of excimer emission at 555 nm decreases with the increase of irradiation time. Meanwhile, after 2 min of irradiation, new peaks at 424 nm and 445 nm are clearly seen, and their intensities increase with irradiation time. These two peaks are assigned to the π* → π transitions of dianthracene,³⁵ implying the occurrence of photocycloaddition for compound 1. The emission profile becomes constant after 15 min of irradiation. Impressively, the photochemical reaction is associated with a pronounced emission colour change from yellow-green (CIE 1931 coordinates (0.40, 0.53)) to blue-white (CIE 1931 coordinates (0.27, 0.35)) (Fig. 3b).

Fig. 3 (a and c) PL spectra for 1 (a) and 2 (c) with irradiation time upon 395 nm UV-light irradiation (λ_ex = 365 nm); (b and d) the CIE graphs of 1, 1UV (b) and 2, 2UV (d).

The photocycloaddition reaction is also observed for compound 2. However, in this case, the intensity decrease of the excimer emission is less pronounced than that of compound 1. More interestingly, the peak wavelength shows a redshift upon light irradiation from 520 nm before irradiation to 535 nm after irradiation for 20 min (Fig. 3c). According to the structural description, the anthracene pair in 2 has a larger slip angle and plane-to-plane spacing than compound 1. We hypothesize that when exposed to UV light, the photochemical reaction first occurs at the surface and then rapidly advances to the interior of the crystal.⁴⁰ The formation of dianthracene at the beginning of the reaction may affect the way the unreacted anthracene pairs are stacked, causing a slight decrease in their slip angle, which in turn leads to a redshift of the excimer emission peak.⁴³ In addition, the newly emerged peaks at 422 and 445 nm are relatively weak. As a result, the emission colour change for compound 2 before and after UV irradiation is not as significant as compound 1, as evidenced by the CIE 1931 coordinates from (0.29, 0.52) to (0.31, 0.50) (Fig. 3d).

To obtain pure phases of the photocycloaddition reaction products for physical property measurements, we irradiated 20 mg of the crystal samples of 1 and 2 separately with 395 nm UV light for 12 h. The products are named as 1UV and 2UV, respectively. Their PXRD patterns matched well with those simulated from the single crystal data of 1UV and 2UV (Fig. S5 and S6). We measured the infrared (IR) spectra of 1UV and 2UV in comparison with those of 1 and 2. As shown in Fig. S7 and S8, the vibrational absorption peaks of depma in 1 (810 cm⁻¹) and 2 (732 cm⁻¹) disappeared, while those of depma₂ emerged in 1UV (1225, 876, and 686 cm⁻¹) and 2UV (1281, 885, 693 cm⁻¹). To examine quantitatively the yields of photocycloaddition reactions of 1 and 2, we further measured ¹H NMR spectra of 1UV and 2UV. The peaks at 7.5–8.6 ppm and 6.7–7.3 ppm are assigned to depma and depma₂,^25,44 respectively (Fig. S9 and S10). By integrating the peaks of anthracene and dianthracene, we can estimate the photochemical reaction yields of 1 and 2 to be 99.9% and 93.2%, respectively, indicating that these photocycloaddition reactions are almost complete for the two compounds. Fig. 2 shows the PL spectra of 1UV and 2UV. For 2UV, we still observe a significant broad band around 550 nm, attributed to the presence of a small amount of un-dimerized anthracene pair.⁴⁵ To determine whether the reaction can be reversed under illumination, we took 1UV as an example and irradiated it with 254 nm light for 12 h. The ¹H NMR result showed that only 2.7% of 1UV was reversed to 1 (Fig. S11). Obviously, the complete de-dimerization of 1UV induced by light is not feasible.

To gain insight of the structures of photocycloaddition reaction products, we irradiated single crystals of 1 and 2 under 395 nm light (183 mW cm⁻²) for 30 min. Both exhibited SC–SC structural transformation to form 1D coordination polymers [Dy₂(SCN)₄L₂(depma₂)(DEPP)₂]_n (1UV) (Fig. 1c and d) and [Dy₂(SCN)₄(L^Me)₂(depma₂)(DEPP)₂]_n (2UV), respectively (Fig. 1g and h). Structural analyses revealed that 1UV and 2UV crystallized in the triclinic P1̄ space group, similar to those of 1 and 2 (Table 1). As expected, the light-induced structural transformation leads to changes in Dy–O(N) bond lengths and O(N)-Dy-O(N) angles (Tables S6 and S7). However, these changes are quite small compared to the other known Dy-anthracene complexes. The coordination geometries of Dy^III ions in 1UV and 2UV can still be best described as the Biaugmented trigonal prism J50, but the deviations are slightly different from their original ones (CShM = 2.700 for 1UVvs. 2.588 for 1, 2.565 for 2UVvs. 2.669 for 2) (Table S3). The most significant change is found for the face-to-face stacking anthracene pairs. The central C2⋯C9A distances are remarkably shortened from 3.771 Å to 1.669 Å for 1, and from 3.860 Å to 1.689 Å for 2. Interestingly, the C2–C9A bond length in 1UV (1.669 Å) is shorter than that in 2UV (1.689 Å), indicating that the in situ formed dianthracene in 1UV could be more stable than that in 2UV. The different stability of the in situ formed dianthracene in 1UV and 2UV is also reflected by their de-dimerization temperatures.

Thermally induced de-dimerization of dianthracene in 1UV and 2UV

Thermogravimetric analysis (TGA) revealed that compounds 1UV and 2UV did not show weight loss below 180 °C (Fig. 4a and c). However, the DSC curves showed broad exothermic peaks in the temperature ranges of 55–95 °C and 45–95 °C for 1UV and 2UV, respectively (Fig. 4b and d), corresponding to the de-dimerization of dianthracene ligands. The maximum of the exothermic peak, defined as the de-dimerization temperature, appeared at 80 °C for 1UV and 71 °C for 2UV. The enthalpy changes (ΔH) are estimated to be −49.8 kJ mol⁻¹ for 1UV and −53.7 kJ mol⁻¹ for 2UV. To obtain fully de-dimerized products, we kept 1UV at 80 °C for 10 min or 2UV at 71 °C for 10 min, the resulting products are named as 1UV-R and 2UV-R. The PXRD, PL, IR, and UV-vis diffuse reflectance spectra of 1UV-R and 2UV-R are identical to those of 1 and 2, respectively (Fig. 2 and S4–S8), indicating that the original compounds 1 and 2 were fully recovered by heating 1UV and 2UV. Apparently, the photocycloaddition reactions of 1 and 2 were reversible, which was also supported by single crystal structural analysis. To determine whether the SC–SC structural transformation can be repeated, we conducted five cycles of irradiation/heating on 1 using the same crystal. We found cracks in the obtained crystal, but the crystal fragments can be used for single-crystal structural analysis (Fig. S12). For compound 2, we repeated the irradiation/heating process for three cycles and found that the crystal did not fracture, still suitable for single-crystal structural analysis (Fig. S13).

Fig. 4 (a and c) The TG curves of compound 1, 1UV (a) and 2, 2UV (c); (b and d) the DSC curves of compound 1UV (b) and 2UV (d).

We noticed that the de-dimerization temperatures of compounds 1UV and 2UV are significantly lower than those for the other related compounds [Dy₂(SCN)₄L₂(depma₂)(H₂O)₂]_n (3UV, 100 °C),³² [Dy₂(SCN)₄L₂(dmpma₂)(H₂O)₂]_n (4UV, 100 °C),³² and [Dy₂(SCN)₄L₂(dmpma₂)(dmpma)₂]_n (5UV, 124 °C),³³ where dmpma represents 9-dimethyl-phosphonomethylanthracene and dmpma₂ is the photo-dimerized dmpma. Compounds 3UV, 4UV and 5UV are photocycloaddition products of [Dy₂(SCN)₄L₂(depma)₂(H₂O)₂] (3), [Dy₂(SCN)₄L₂(dmpma)₂(H₂O)₂] (4), and [Dy₂(SCN)₄L₂(dmpma)₄] (5), respectively. Considering that all these compounds have similar binuclear core structures, the difference in de-dimerization temperatures of 1UV–5UV must be related to the stability of the dianthracene units formed in situ in the solid state. We hypothesize that the stability of dianthracene is highly dependent on the stacking pattern of anthracene pairs in its original compounds 1–5 (Fig. 1 and S14). When the two anthracene groups are arranged in parallel, the stacking pattern can be characterized by two key parameters, i.e., the slip angle (θ) and plane-to-plane distance (d_pp). Anthracene pairs with a larger slip angle and d_pp value are expected to result in dianthracene units with less stability.

Table 2 summarizes some of the structural parameters of the anthracene pairs in 1–5 and their corresponding de-dimerization temperatures for the photocycloaddition products 1UV–5UV. Clearly, the de-dimerization temperatures follow the order: 2UV (71 °C) < 1UV (80 °C) < 3UV (100 °C), 4UV (100 °C) < 5UV (124 °C). Among these five compounds, 2UV has the lowest de-dimerization temperature (71 °C), while its corresponding anthracene pair in the original compound 2 has the largest slip angle (24.6°) and d_pp value (3.503 Å). In contrast, compound 5UV has the highest de-dimerization temperature (124 °C), while its corresponding anthracene pair in the original compound 5 has the smallest slip angle (16.9°) and d_pp value (3.355 Å). For compounds 3UV and 4UV, although the slip angle of the anthracene pair in 3 (16.9°) is smaller than that in 4 (19.9°), the d_pp value in 3 (3.410 Å) is close to that in 4 (3.407 Å). This may explain why the de-dimerization temperature is similar for 3UV and 4UV (100 °C). It is noteworthy that the slip angle in compound 1 is close to that in 4 (19.7° in 1vs. 19.9° in 4), but the d_pp value in 1 (3.485 Å) is larger than that in 4 (3.407 Å). As a result, the de-dimerization temperature of 1UV (80 °C) is much lower than that of 4UV (100 °C). All these results validate the above hypothesis that the stability of the in situ formed dianthracene depends strongly on the slip angle and d_pp values of the anthracene pair, while the plane-to-plane distance (d_pp) seems to be more important.

Table 2 Comparison of the de-dimerization temperature and related structural parameters of dysprosium anthracene-based complexes

Compound	d CC /Å	d pp /Å	d C2–C9A /Å	Slip angle^d/°	d Dy⋯Dy	∠P1–C1–C2/°	T d /°C	Ref.
a The centre-to-centre distance. b Plane-to-plane distance. c The central C2–C9A distance of the π–π interacting anthracene pair. d The angle between the centroid–centroid line and the vertical line in anthracene. e The Dy⋯Dy distance over the anthracene pair of dianthracene bridge. f The de-dimerization temperature. The values in the parentheses are the temperature ranges of the exothermic peaks determined by DSC curves; n.a. means not available.
1	3.771	3.485	3.773	19.68	15.617	114.5		This work
1UV			1.669		15.613	125.9	80 (55–95)	This work
2	3.868	3.503	3.860	24.60	15.607	115.1		This work
2UV			1.689		15.657	124.7	71 (45–95)	This work
3	3.690	3.410	3.700	16.88	14.671	114.6		21
3UV			1.689		14.775	121.6	100 (92–105)	21
4	3.652	3.407	3.654	19.91	14.592	114.6		21
4UV			n.a.		n.a.	n.a.	100 (90–105)	21
5	3.688	3.355	3.687	16.91	16.783	115.3		22
5UV			1.635		15.733	124.9	124 (112–133)	22

---

When photocycloaddition of π–π interacting anthracene pair in Dy-anthracene complexes occurs, we expected that the Dy⋯Dy distance over the dianthracene bridge would change significantly. This is indeed the case for compounds 3 and 5 whose distances are shortened by 0.104 Å and 1.050 Å, respectively. Interestingly, the change in the Dy⋯Dy distance in compound 2 (0.050 Å) is much less significant, whereas this change is hardly visible in compound 1 (0.004 Å). The fact that the metal ion spacing remains essentially unchanged before and after photocycloaddition favours, on the one hand, the ability of the structural transformation to proceed in a single-crystal-to-single-crystal fashion, while on the other hand, it leads to a decrease in the stability of the resulting dianthracene product. As a result, the ∠P1–C1–C2 angles after photocycloaddition are significantly increased compared to those in the initial samples (Table 2), deviating markedly from the characteristic angle of sp³-hybridized carbon atoms (109.5°). However, we note that the ∠P1–C1–C2 angle for 1UV (125.9°) is larger than that for 2UV (124.7°), but the former has a higher de-dimerization temperature. Clearly, the extent to which this ∠P1–C1–C2 angle deviates from 109.5° is less important in contributing to dianthracene stability than the anthracene stacking pattern discussed above.

The fact that the dianthracene in 1UV and 2UV can dissociate at temperatures well below 100 °C demonstrates the possibility to regulate the de-dimerization temperature of the in situ formed dianthracene by introducing a suitable co-ligand, which is DEPP in the present cases. Moreover, compounds 1UV and 2UV have very similar structures except for the additional methyl group in the ligand 2,6-dimethoxyphenol in 2UV, but their de-dimerization temperatures are different by 9 °C. This result implies that the de-dimerization temperature of in situ formed Ln-dianthracene compounds can also be regulated by ligand modification. To the best of our knowledge, the de-dimerization temperature of 2UV (71 °C) is the lowest among the known metal–dianthracene complexes.

Photocycloaddition reaction of 1 at different temperatures

The fact that de-dimerization of 1UV and 2UV can be observed at temperatures below 80 °C opens the possibility of rapid switching of the PL properties of compounds 1 and 2 above room temperature. On the one hand, elevated temperature may make excited state molecules more susceptible to structural relaxation, thus accelerating the photocycloaddition reaction rate. On the other hand, elevated temperatures also increase the de-dimerization rate of the photocycloaddition products formed in situ. Thus, the PL spectra as a function of irradiation time at elevated temperatures is the result of competing photo-dimerization and de-dimerization processes. To investigate the effect of temperature on photochemical reactions, compound 1 was chosen as a representative for detailed study because it showed significant spectral changes before and after photocycloaddition reaction.

We selected single crystals of 1 and studied the PL spectra after irradiation for different times at temperatures 20–120 °C using a home-built fluorescence microscope. Due to the limited light sources available for this instrument, we used 375 nm laser for UV irradiation and excitation with initial power of 21.75 μW and power density of 3000 mW cm⁻². All measurements were conducted at the same position on the crystal to ensure in situ testing conditions.

Fig. 5 shows the real-time PL spectral change and the normalized selected PL spectra at different times of compound 1 at 20–120 °C. It is clear that in all cases the peak intensity at 550 nm, which corresponds to the excimer emission of anthracene pair, progressively decreases. While the peak intensity at 425 nm, which corresponds to the emission of dianthracene, increases with prolonged illumination time. The results indicate that photodimerization reaction still occurs at 80 °C and above. It is worth noting that the peak of the excimer emission occurs at 550 nm instead of 555 nm due to the different instruments used to measure the PL spectra of single crystals and bulk samples. In addition, we observed the emergence of weak peaks in the near infrared (NIR) region of 800–850 nm, with the maximum at 833 nm, which increases with irradiation time. This peak is attributed to the ⁴F_9/2 → ⁶F_9/2, ⁶H_7/2 transitions of the Dy³⁺ ions.^39,46 Obviously, the dianthracene formed in situ is able to transfer energy to and sensitize the Dy³⁺ ions,^23,47 thereby inducing the NIR luminescence of the Dy³⁺ ions. The f–f transitions in the visible region cannot be identified.

Fig. 5 The real-time PL spectral change (up) and the normalized selected PL spectra at different times (down) of compound 1 at 20 °C (a), 40 °C (b), 60 °C (c), 80 °C (d), 100 °C (e), and 120 °C (f). The 375 nm laser was used for UV irradiation and excitation.

We further attempted to determine the rate constants of the compound 1. We note that few kinetic studies of anthracene photodimerization reactions have been reported in the literature.^48–53 Infrared spectroscopy is generally used to follow the progress of the reaction in the solid phase,^35,52 or absorption spectroscopy is used to study the reactions in the solution phase.^48,49 As far as we are aware, there is only one reported case of using fluorescence spectroscopy to study the kinetics of anthracene photodimerization reaction at room temperature.⁵³

We first monitored the intensity change of the peak at 425 nm. The conversion ratio was calculated using the equation y = |I − I₀|/ΔI, where I and I₀ represent the real-time and initial PL intensity, respectively, and ΔI represents the difference between the initial and final intensity when the PL profile becomes constant. We observed that ln(1 − y) exhibited an approximately linear relationship with irradiation time under different temperatures, which is consistent with first-order reaction kinetics (Fig. S15). Similar kinetic behaviour has been observed in a few other anthracene-based compounds.^54–56 The ln(1 − y) vs. time plots were well fitted by using the first-order logarithmic equation ln(1 − y) = −kt, where k is the rate constant, to obtain the reaction rate constant k at temperatures in range of 20–120 °C (Table 3). It is clear that the k value increases with increasing temperature until 100 °C, above which the k value decreases with temperature (Fig. 6a). The same conclusion can be obtained by fitting the data using the intensity change of the peak at 550 nm (Fig. 6a, S16 and Table S8). According to the Arrhenius equation k = A exp(−E_a/RT), the activation energy can be derived by linear fitting of the ln k vs. 1/T plot in range of 20 to 100 °C, which is 22.13 ± 3.02 kJ mol⁻¹ (based on the 420 nm peak) or 25.72 ± 4.19 kJ mol⁻¹ (based on the 550 nm peak) (Fig. 6b). As far as we know, the activation energy of anthracene photocycloaddition reaction has not been reported before. However, Chen and co-workers recently conducted theoretical studies on the photophysical and photochemical properties of mononuclear compounds Ln(depma)(hmpa)₂(NO₃)₃ (1-Ln, Ln = Eu, Tb).⁴⁷ They found that the calculated barriers for the ligand-centred intersystem crossing (ISC) ¹ππ* → ³ππ* transition, which led to the [4 + 4] photocycloaddition reaction, were 8.8 kcal mol⁻¹ (36.78 kJ mol⁻¹) for 1-Eu and 9.1 kcal mol⁻¹ (38.04 kJ mol⁻¹) for 1-Tb. These calculated values are relatively close to those observed for compound 1.

Table 3 The rate constant obtained at different temperatures by fitting the intensity data at 425 nm (k₁) and 550 nm (k₂)

T/°C	20	40	60	80	100	120
k 1/s^−1	0.023(1)	0.039(1)	0.046(2)	0.095(2)	0.172(9)	0.131(8)
k 2/s^−1	0.024(1)	0.050(6)	0.057(5)	0.111(16)	0.287(47)	0.258(28)

---

Fig. 6 (a) The k vs. T plots for compound 1. The solid lines are eye-guided. (b) The ln k vs. 1/T plots for 1. The solid lines are best fits.

As already mentioned, the PL spectra obtained after irradiation at higher temperatures are the result of competing photo-dimerization and de-dimerization processes. The latter should be solely due to temperature effects, since irradiation of anthracene with a 375 nm laser leads only to dimerization without the reverse process of de-dimerization.⁵⁷ The DSC curve of 1UV revealed that the de-dimerization process started at about 55 °C, reached a maximum at 80 °C and ended at 95 °C. Kinetics studies have shown that the rate of photocycloaddition reaction of 1 reaches a maximum at 100 °C and decreases above this temperature. The results indicate that it is still possible for a photocycloaddition reaction of anthracene to occur when the temperature for complete de-dimerization is attained. To verify this conclusion, compound 1 was placed on a hot plate and irradiated with 395 nm UV light (100 mW cm⁻²) for 2 h at constant temperatures of 60 °C, 80 °C, and 100 °C. The NMR results showed that the reaction conversion rate decreased with increasing temperature, but dimerized products were still formed even at 100 °C (Fig. S17–S19). To the best of our knowledge, although a few studies of the kinetics of the anthracene photocycloaddition reaction at or below room temperature have been reported,^35,49,57,58 similar kinetic studies above room temperature have never been documented in the literature. This finding is important for the selection of a suitable temperature for rapid structural and photoluminescence switching of anthracene-based systems.

Light-induced changes in magnetic properties of 1 and 2

Structural analyses have demonstrated that photocycloaddition reaction causes slight changes in the bond lengths and angles of the Dy³⁺ ions in 1 and 2. These changes in coordination environments should affect their magnetic properties. Fig. S20 shows the temperature dependent magnetic susceptibilities of the two compounds before and after photocycloaddition. At 300 K, the χ_MT values are 28.50 and 28.27 cm³ K mol⁻¹ for 1 and 2, respectively, which are consistent with the expected value of two isolated Dy^III ions (28.34 cm³ K mol⁻¹, ⁶H_15/2, S = 5/2, L = 5, g_J = 4/3). As the temperature decreases, the Χ_MT values remain nearly constant until 100 K, below which they decrease slowly to reach the minimum values of 27.40 and 28.06 cm³ K mol⁻¹ at 45 K for 1 and 2, respectively, attributed to the depopulation of Stark sublevels. The rapid increase of the χ_MT values below 45 K suggest the existence of ferromagnetic interaction between the Dy^III ions. Similar phenomenon was observed in some other binuclear Dy₂O₂ complexes with phenoxy bridges.^59–61 The field dependent magnetization curves were also measured at different temperatures. The magnetization (M) values at 2 K and 70 kOe are 10.81 Nβ for 1 and 10.55 Nβ for 2 (Fig. S21 and S22), which are lower than the theoretical saturation values of 20 Nβ for two Dy^III ions. This may be due to the presence of significant magnetic anisotropy and/or low-lying excited states of the Dy^III ions in compounds 1 and 2.⁶² Interestingly, the χ_MT vs. T and M vs. H curves of 1UV and 2UV almost overlap the curves for their counterparts prior to UV irradiation (Fig. S20 and S23).

Significant changes are found in the magnetic dynamics. For both compounds 1 and 2, we observed frequency-dependent in-phase and out of phase magnetic susceptibilities under zero dc field (Fig. 7 and S24), which is characteristic of single-molecule magnet (SMM) behaviour. The Cole–Cole plots can be fitted by the generalized Debye model to extract the relaxation time (τ) (Fig. S25).⁶³ The distribution coefficient (α) values are found to be in the ranges of 0.27–0.47 for 1 and 0.19–0.41 for 2, respectively (Tables S8 and S9), indicating the existence of a broad relaxation time distribution. Fig. 7c and f show the ln τ vs. T⁻¹ plots for both compounds. For 1, the linear relationship between 4 K and 6 K indicates that the relaxation is dominated by the Orbach process, while below 4 K, a nonlinear relationship is observed, indicating the domination of relaxation by Raman processes. Similar profile is observed for compound 2. Therefore, the plot of ln τ vs. T⁻¹ can be fitted by using the eqn (1) containing Raman and Orbach processes, where U_eff is the effective energy barrier.

τ⁻¹ = CTⁿ + τ₀⁻¹ exp(−U_eff/kT) (1)

Fig. 7 (a, b, d and e) The out-of-phase ac susceptibilities for 1 (a), 1UV (b), 2 (d) and 2UV (e); (c and f) the plot of ln τ vs. T⁻¹ for 1, 1UV (c) and 2, 2UV (f).

The best fits gave parameters U_eff = 57(2) K, τ₀ = 10^−7.8(2) s, n = 2.78(9), and C = 2.64(23) K^−2.78 s⁻¹ for compound 1, and U_eff = 78(2) K, τ₀ = 10^−8.6(1) s, n = 2.52(8), and C = 0.42(4) K^−2.52 s⁻¹ for compound 2 (Table 4).

Table 4 Magnetic parameters for compounds 1, 1UV, 2 and 2UV derived by fitting the ln τ⁻¹vs. T plots to eqn (1)

	U eff/K	τ 0/s	C/K^−n s^−1	n
1	57	1.74 × 10^−8	2.64	2.78
1UV	94	3.30 × 10^−8	0.49	2.99
2	78	2.60 × 10^−9	0.42	2.52
2UV	84	6.06 × 10^−9	0.27	2.80

---

Notably, the effective energy barrier of 2 (78 K) is higher than that of 1 (57 K). According to the structural description, the main difference between 1 and 2 is that the latter has an additional methyl group in the bridging 2,6-dimethoxyphenol ligand. This difference causes slight changes in the coordination environment of the Dy^III ion. Compared to those in 1, the Dy1-O5 and Dy1-O5A bond lengths and Dy1⋯Dy1A distance in 2 are shorter [2.250(5), 2.284(5) and 3.833(1) Å vs. 2.295(3), 2.297(3) and 3.871(1) Å in 1], but the Dy1-O5-Dy1A angle in 2 is larger [115.4(2)° vs. 114.9(1)° in 1]. Previous studies on the related compound [Dy₂(SCN)₄L₂(dmpma)₄] (5) have demonstrated that the magnetic axes lie in the Dy₂O₂ plane along the Dy1–N1 bond (Fig. 8).³³ Given that the Dy1–N1 bond length is nearly the same in the two structures and most of the equatorial Dy1–O(N) bond lengths in 2 are slightly shorter than those in 1 (Table 1), it is difficult to judge which structural factor plays the key role in enhancing the SMM performance of 2. Nevertheless, the effective energy barriers of 1 and 2 are comparable to those of compounds 5 (55.9 K)³³ and [Dy₂(L¹)₂(L²)₂(CH₃OH)₂] (L¹ = [((2-hydroxyphenyl)imino)methyl]phenol, L² = 2,6-dimethoxyphenol) (69.0 K),⁶⁴ but smaller than those of compounds 3 (110.8 K),³²4 (117.7 K),³² [Dy₂(NO₃)₄(H₂O)₂(L)₂]₂·CH₃CN (220.1 K),⁶⁵ and [Dy₂(DMOAP)₂(DBM)₄]₂·CHCl₃ (DMOAP = syringaldehyde, DBM = 1,3-diphenylpropane-1,3-dione) (175.2 K)⁶⁶ which contain similar Dy₂O₂ dimer cores.

Fig. 8 The coordination environment of the dysprosium ion in 1. The green arrow represents the orientation of the local main magnetic axis.

After light irradiation, the product 1UV shows magnetic dynamics different from that of 1, with the peaks in the vs. ν curves appearing at higher frequencies. However, the vs. ν profiles for 2UV are very similar to those for 2. The ln τ vs. T⁻¹ curves can again be fitted by using the eqn (1), resulting in parameters U_eff = 94(5) K, τ₀ = 10^−8.5(5) s, n = 2.99(18), and C = 0.49(11) K^−2.99 s⁻¹ for 1UV, and U_eff = 84(3) K, τ₀ = 10^−8.2(2) s, n = 2.80(12), and C = 0.27(4) K^−2.80 s⁻¹ for 2UV, respectively (Table 4). Compared to 1, the effective energy barrier of 1UV is increased by 37 K. We notice that the Dy1–O distances in 1 are either slightly shortened or elongated after light irradiation (Table 1). The most significant change is found for the Dy1–N bond lengths. The axial Dy1–N1 bond length is similar [2.388(5) Å in 1UVvs. 2.392(5) Å in 1] while the equatorial Dy1–N2 distance is elongated [2.398(5) Å in 1UVvs. 2.379(4) Å in 1], which may contribute to the increase of the U_eff value of 1UV. Surprisingly, the energy barrier of 2 remains almost the same before and after photocycloaddition. Structural analysis revealed that all Dy1–O(N) bond lengths in 2UV are slightly expanded except for the axial Dy1–N1 which is slightly shortened [2.380(6) Å in 2UVvs. 2.388(7) Å in 2] (Table 1). As a result, the overall magnetic anisotropy of 2UV could be similar to 2 and, therefore, the similar U_eff values.

Another interesting observation is the change in the Raman index value. According to the Debye model theory, the theoretical value of Raman index for Kramers ion system should be n = 9.⁶⁷ However, the presence of low-energy phonon vibrations in the system can cause the Debye model to deviate, and the Raman index usually falls within the range of 2–6.⁶⁸ This is indeed the case for compounds 1 and 2 as well as their photocycloaddition products which have the Raman index of 2.52–2.99. The n values of 1UV (n = 2.99) and 2UV (n = 2.80) are slightly higher than those of 1 (n = 2.78) and 2 (n = 2.52), attributed to the formation of 1D coordination polymers.

By fitting the ln τ⁻¹vs. T plots to Raman process in the low-temperature region using equation τ ∼ [exp(ℏω/k_BT)],⁶⁹ we obtained the vibration energies (ℏω) of 3.88 cm⁻¹ for 1, 4.76 cm⁻¹ for 1UV, 4.25 cm⁻¹ for 2, and 5.21 cm⁻¹ for 2UV (Fig. S26). If the Orbach process is combined, we can use eqn (2) to fit the entire dataset and obtained the energy barriers (U_eff) as well as the vibration energies (ℏω) (Table S12 and Fig. S27). The U_eff values are close to those obtained using eqn (1). While the vibration energies are 4.88 cm⁻¹ for 1, 6.02 cm⁻¹ for 1UV, 5.03 cm⁻¹ for 2, and 6.00 cm⁻¹ for 2UV. Notably, compounds 1 and 2 exhibit a significant increase in vibrational energy after photocycloaddition, indicating that the photocycloaddition reaction effectively suppresses low-energy phonon vibrations.

Theoretical calculations of magnetic properties

Theoretical calculations can reveal the relationship between structure and magnetic properties. Based on the geometries determined by X-ray crystallography, we performed complete-active-space self-consistent field (CASSCF) calculations on individual Dy^III fragments in the complexes 1, 1UV, 2, and 2UV using OpenMolcas⁷⁰ and SINGLE_ANISO^71–73 programs (see SI for details). Table S13 lists the calculated energy levels, g (g_x, g_y, g_z) tensors and the dominant m_J values of the lowest eight Kramers doublets (KDs) of each Dy^III fragment. The energy gap between the two lowest KDs of 1_Dy1 and 1UV_Dy1 is close, with a similar situation observed for 2_Dy1 and 2UV_Dy1. In all cases, the ground state is primarily composed of m_J = ±15/2 states (Table S14), resulting in a small transversal magnetic moment. However, the first excited state KD is significantly populated by multiple m_J states, leading to a large transversal magnetic moment in the first excited state KDs of 1–2UV.

Fig. S29 illustrates the corresponding magnetisation blocking barriers for the Dy^III fragments in 1–2UV. The transversal magnetic moments in the ground state KDs are all less than 0.5 × 10⁻¹μ_B, thus suppressing the quantum tunnelling of magnetisation (QTM) in the ground states at low temperatures. Notably, the transversal magnetic moments of the four Dy^III fragments in the first excited state are 0.79, 0.12 × 10¹, 0.31 and 0.29μ_B, respectively, all exceeding 0.5 × 10⁻¹μ_B. Consequently, rapid QTM is permitted in their first excited state KD. The calculated energy barriers for 1_Dy1, 1UV_Dy1, 2_Dy1 and 2UV_Dy1 are 177.3, 185.4, 162.7 and 173.7 cm⁻¹, respectively, which are quite close to each other. The barriers for 1UV_Dy1 and 2UV_Dy1 are slightly higher than those for 1_Dy1 and 2_Dy1, consistent with experimental findings. Considering the detrimental effects of anharmonic phonons, Raman magnetic relaxation, QTM, et al. on energy barriers, experimentally measured energy barriers (U_eff) are typically lower than the calculated values.^74–79

Although the magnetic anisotropy in 1–2UV primarily originates from individual Dy^III ions, Dy^III-Dy^III interactions still exert a certain influence on its slow magnetic relaxation process. By comparing the calculated and experimentally measured magnetic susceptibility of 1–2UV, we utilised the POLY_ANISO program^71–73 to fit the exchange coupling constant J̃_exch and intermolecular interaction zJ′. The parameters in Table 5 were calculated based on Dy^III ions with a pseudospin of S̃_Dy = 1/2. The magnetic susceptibility fitting includes total coupling parameters J̃_total (dipolar and exchange). Fig. S30 displays the calculated and experimental χ_MT vs. T curves for complexes 1–2UV, with all fitting results showing good agreement with experimental data across the entire temperature range.

Table 5 The fitted exchange coupling constant (J̃_exch), calculated dipole–dipole interaction (J̃_dip) and total constant (J̃_total) between magnetic centre ions in 1–2UV. The fitted intermolecular interactions (zJ′) in compounds 1, 1UV, 2 and 2UV are −0.005, −0.01, −0.001 and −0.003 cm⁻¹, respectively

	1	1UV	2	2UV
J̃ exch	2.5	2.5	2.5	2.5
J̃ dip	5.9	5.8	5.8	5.8
J̃ total	8.4	8.3	8.3	8.3

---

The positive J̃_total obtained from Lines model⁸⁰ indicate that the Dy^III–Dy^III interactions within the 1–2UV complexes are ferromagnetic, which agrees with the experimental results. Surprisingly, however, the total and exchange coupling constants for these four compounds are almost identical. This may be attributed to the minimal variations in structural parameters among them, which collectively exerted negligible influence on the intramolecular magnetic interactions within the dimer. We present the exchange energies, the energy differences between each exchange doublet Δ_t and the main values of the g_z for the lowest two exchange doublets of 1–2UV. As shown in Table S15, the g_z values for the ground exchange states of compounds 1, 1UV, 2 and 2UV are 39.671, 39.679, 39.395 and 39.711, respectively, further confirming that the Dy^III–Dy^III interactions in 1–2UV are all ferromagnetic. Fig. S31 illustrates the principle magnetic axes of Dy^III ions in 1–2UV, where the angles between the magnetic axes and the vectors connecting the two Dy^III ions in 1, 1UV, 2 and 2UV are 2.3°, 2.0°, 2.9° and 2.6°, respectively. These magnetic axes are parallel to each other and closely aligned with the direction connecting the two Dy^III ions. This results in an extremely weak transversal component of the induced dipolar field between dysprosium ions, thus significantly reducing the efficiency of the quantum tunnelling mechanism, a process typically highly active in high concentration lanthanide-based SMMs.

Conclusions

We report two compounds [Dy₂(SCN)₄(L)₂(depma)₂(DEPP)₂] (1) and [Dy₂(SCN)₄(L^Me)₂(depma)₂(DEPP)₂] (2). Both can undergo a SC–SC photocycloaddition reaction under UV irradiation, to form chain compounds [Dy₂(SCN)₄L₂(depma₂)(DEPP)₂]_n (1UV) and [Dy₂(SCN)₄(L^Me)₂(depma₂)(DEPP)₂]_n (2UV). This structural transformation leads to significant alterations in the PL properties and magnetic dynamics of compound 1. But for compound 2, the changes in PL and magnetic properties were not pronounced. The introduction of the auxiliary ligand DEPP and the modification of the bridging ligand L result in changes in the stacking pattern of anthracene pair. Consequently, the de-dimerization temperatures of 1UV (80 °C) and 2UV (71 °C) are remarkably lower than those of known dianthracene-based compounds, with 2UV showing the lowest de-dimerization temperatures. We found that the stability of in situ formed dianthracene depends largely on the plane-to-plane distance and slip angle of the anthracene pair in the original complex, with the former appearing to be more important. We also investigated the temperature effect on the photocycloaddition reaction of 1 using fluorescence spectroscopy. The reaction kinetics followed the first-order reaction characteristics, with the rate increasing with temperature and peaking at 100 °C. Interestingly, the photocycloaddition reaction of anthracene may still occur even at temperatures reaching complete de-dimerization. This finding enables temperature-controlled regulation of photodimerization/de-dimerization reactions of anthracene, offering potential applications in molecular photonics and devices. In addition, we observed that compounds with their photodimerized products showing lower de-dimerization temperatures may undergo less structural alterations before and after illumination, resulting in smaller changes in luminescence and magnetic properties. Therefore, how to achieve rapid and reversible photodimerization/de-dimerization reactions at relatively low temperatures while eliciting significant luminescence and/or magnetic changes remains a critical challenge for lanthanide anthracene-based materials.

Experimental

Materials and methods

Materials and physical measurements. The ligands 9-diethyl-phosphonomethylanthracene (depma) were synthesized according to the literature.⁸¹ All other starting reagents and solvents were obtained from commercial sources and used directly. The elemental analysis for C, H and N was performed using a PE 240C analyzer. The infrared (IR) spectra were measured in the range 4000–400 cm⁻¹ using KBr pellets on Bruker Tensor 27 spectrometer. The thermogravimetric (TG) analyses were carried out on a Mettler Toledo TGA/DSC instrument in the range of 30–600 °C under N₂ atmosphere with a heating rate of 5 °C min⁻¹. Powder X-ray diffraction (PXRD) data were recorded on a Bruker D8 advance diffractometer with Cu-Kα radiation in a range of 5–50° at room temperature. The differential scanning calorimetry (DSC) measurement was conducted on a Mettler DSC823e instrument at a heating rate of 5 K min⁻¹. The UV/Vis spectra were measured on a PerkinElmer Lambda 950 UV/VIS/NIR spectrometer using powder samples. The steady fluorescence spectra were recorded using Spectrofluorimeter PE LS55. Time resolved fluorescence and quantum yield measurements were performed on an Edinburgh FLS 980 at room temperature. Single crystal fluorescence tests at different temperatures were performed on a home-built widefield microscope with 375-nm CW diode laser as the light source. The PL was collected by an oil immersion objective lens (Olympus UPlanFLN 60×, NA = 1.25) and imaged by an EMCCD camera (iXon Ultra 888, Andor), after passing through a 400-nm longpass filter (ET400LP, Chroma). The ¹H NMR spectra were recorded on BRUKER AVANCE III 400 MHz spectrometer. The dc and ac magnetic susceptibility data were collected on polycrystalline samples by Quantum Design vibrating sample magnetometer (VSM). A drop of Paraffin less than 2 mg was added to avoid the movement and reorientation of samples during measurement. The dc susceptibilities were corrected for diamagnetic contributions of holder, paraffin and sample.⁸²

Single crystal X-ray crystallography

Single crystals were used for data collections on Bruker APEX duo diffractometers using graphite-monochromated Mo-Kα radiation (λ = 0.71073 Å). The data were integrated using the Siemens SAINT program.⁸³ Multiscan absorption corrections were applied. The structures were solved by intrinsic phasing methods and refined on F² by full-matrix least squares using SHELXTL.⁸⁴ All the non-hydrogen atoms were refined anisotropically. All H atoms were placed in theoretical positions and refined isotropically. The residual electron densities were of no chemical significance.

Synthesis of [Dy₂(SCN)₄(L)₂(depma)₂(DEPP)₂] (1). To 5 mL of MeCN, 2,6-dimethoxyphenol (HL, 13.4 mg, 0.1 mmol) and Et₃N (14 μL, 0.1 mmol) were sequentially added, followed by 1 mL of the 0.1 mol L⁻¹ Dy(SCN)₃ solution.²⁴ The mixture was stirred at room temperature for 5 min. Depma (30.0 mg, 0.1 mmol) and DEPP (15.2 mg, 0.1 mmol) were then added. After 5 min of stirring, the reaction mixture was filtered to afford a clear yellow solution. Slow evaporation of the filtrate at ambient temperature over 24 h afforded yellow block crystals of compound 1. Yield: 42.9 mg (45.6% based on Dy). Elemental Anal. Calcd (%): C, 45.98; H, 5.04; N, 2.98. Found (%): C, 45.39; H, 5.11; N, 3.07.

Synthesis of [Dy₂(SCN)₄(L^Me)₂(depma)₂(DEPP)₂] (2). Compound 2 was synthesized using a method similar to that of compound 1, except that the ligand HL was replaced with 4-methyl-2,6-dimethoxyphenol (HL^Me, 16.8 mg, 0.1 mmol). Yield: 41.1 mg (43.1% based on Dy). Elemental Anal. Calcd (%): C, 46.57; H, 5.18; N, 2.94. Found (%): C, 46.14; H, 5.17; N, 3.07.

Author contributions

Y.-H. Q., X.-F. M. and X.-D. H. conducted experiments and analysed the data. Y.-H. Q. and X.-F. M. wrote the manuscript. X. H. and Y. T. performed single crystal fluorescence measurements at different temperatures and the related data analysis. S.-S. B. assisted in the analysis of the single-crystal data. Y.-Q. Z. performed theoretical calculations of magnetic properties and the related data analysis. L.-M. Z. conceived the project and revised the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

CCDC 2455154 (2), 2456341 (1UV), 2456346 (2UV) and 2456349 (1) contain the supplementary crystallographic data for this paper.^85a–d

The data supporting this article have been included as part of the SI. Supplementary information: additional crystallographic details, and other characterizations (PXRD, IR, UV-Vis and photoluminescence spectra, thermal analysis, photographs). See DOI: https://doi.org/10.1039/d5sc04192a.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (22273037, 92356303, 21731003).

References

1. S. J. Loeb, Chem. Soc. Rev., 2007, 36, 226–235.
2. J. J. Zakrzewski, M. Liberka, J. Wang, S. Chorazy and S. Ohkoshi, Chem. Rev., 2024, 124, 5930–6050.
3. N. Deorukhkar, C. Egger, L. Guénée, C. Besnard and C. Piguet, J. Am. Chem. Soc., 2024, 146, 308–318.
4. T. G. Ashebr, H. Li, X. Ying, X.-L. Li, C. Zhao, S. Liu and J. Tang, ACS Mater. Lett., 2022, 4, 307–319.
5. A. G. Bispo Jr, Coord. Chem. Rev., 2025, 537, 216685.
6. (a) J.-T. Chen, T.-D. Zhou and W.-B. Sun, Dalton Trans., 2023, 52, 4643–4657; (b) P.-Y. Liao, Y. Liu, Z.-Y. Ruan, H.-L. Wang, C.-G. Shi, W. Deng, S.-G. Wu, J.-H. Jia and M.-L. Tong, Inorg. Chem., 2023, 62, 1075–1085.
7. K. Rogacz, M. Brzozowska, S. Baś, K. Kurpiewska and D. Pinkowicz, Inorg. Chem., 2022, 61, 16295–16306.
8. X.-J. Luo, Y. Liu, H.-D. Pan, S. Peng, Z.-B. Hu, Q.-H. Teng, Y. Li, F.-P. Liang and K. Wang, Cryst. Growth Des., 2023, 23, 4825–4835.
9. (a) M. Estrader, J. S. Uber, L. A. Barrios, J. Garcia, P. Lloyd-Williams, O. Roubeau, S. J. Teat and G. Aromí, Angew. Chem., Int. Ed., 2017, 56, 15622–15627; (b) J. J. Zakrzewski, R. Jankowski, M. Romanowska, J. Wang, D. Pinkowicz, B. Sieklucka, S.-i. Ohkoshi and S. Chorazy, Angew. Chem., Int. Ed., 2025, 64, e202424651.
10. Y.-J. Ma, J.-X. Hu, S.-D. Han, J. Pan, J.-H. Li and G.-M. Wang, J. Am. Chem. Soc., 2020, 142, 2682–2689.
11. X.-Q. Wang, Y.-W. Geng, Z. Wang, C. Xie, T. Han and P. Cheng, J. Am. Chem. Soc., 2025, 147, 18044–18053.
12. L.-F. Wang, J.-Z. Qiu, Y.-C. Chen, J.-L. Liu, Q.-W. Li, J.-H. Jia and M.-L. Tong, Inorg. Chem. Front., 2017, 4, 1311–1318.
13. L.-Z. Cai, Q.-S. Chen, C.-J. Zhang, P.-X. Li, M.-S. Wang and G.-C. Guo, J. Am. Chem. Soc., 2015, 137, 10882–10885.
14. M. Hojorat, H. Al Sabea, L. Norel, K. Bernot, T. Roisnel, F. Gendron, B. L. Guennic, E. Trzop, E. Collet, J. R. Long and S. Rigaut, J. Am. Chem. Soc., 2020, 142, 931–936.
15. P. Selvanathan, G. Huang, T. Guizouarn, T. Roisnel, G. Fernandez-Garcia, F. Totti, B. Le Guennic, G. Calvez, K. Bernot, L. Norel and S. Rigaut, Chem.–Eur. J., 2016, 22, 15222–15226.
16. Y. Xin, J. Wang, M. Zychowicz, J. J. Zakrzewski, K. Nakabayashi, B. Sieklucka, S. Chorazy and S. Ohkoshi, J. Am. Chem. Soc., 2019, 141, 18211–18220.
17. J. Wang, J. J. Zakrzewski, M. Zychowicz, V. Vieru, L. F. Chibotaru, K. Nakabayashi, S. Chorazy and S. Ohkoshi, Chem. Sci., 2021, 12, 730–741.
18. P. Giri, A. Panda and M. K. Panda, Chem.–Eur. J., 2024, 30, e202303836.
19. S. Bai, L.-F. Wang, Z.-W. Wu, T. Feng and Y.-F. Han, Dalton Trans., 2022, 51, 8743–8748.
20. N. Huang, X. Ding, J. Kim, H. Ihee and D. Jiang, Angew. Chem., Int. Ed., 2015, 54, 8704–8707.
21. M. Servalli, N. Trapp and A. D. Schlüter, Chem.–Eur. J., 2018, 24, 15003–15012.
22. K. Yuhara and K. Tanaka, Angew. Chem., Int. Ed., 2024, 63, e202319712.
23. X.-D. Huang, Y. Xu, K. Fan, S.-S. Bao, M. Kurmoo and L.-M. Zheng, Angew. Chem., Int. Ed., 2018, 57, 8577–8581.
24. X.-D. Huang, G.-H. Wen, S.-S. Bao, J.-G. Jia and L.-M. Zheng, Chem. Sci., 2021, 12, 929–937.
25. X.-D. Huang, X. Ma and L.-M. Zheng, Angew. Chem., Int. Ed., 2023, 62, e202300088.
26. Y.-H. Wang, Z.-N. Gao, S. Liang, J. Li, W.-J. Wei, S.-D. Han, Y.-Q. Zhang, J.-X. Hu and G.-M. Wang, Research, 2024, 7, 0411.
27. J.-C. Liu, X.-D. Huang, Q. Zou, S.-S. Bao, X.-Z. Wang, J.-Y. Ma and L.-M. Zheng, J. Mater. Chem. C, 2020, 8, 7369–7377.
28. Q. Zou, T. Shang, X.-D. Huang, Q.-Q. Guo, J.-G. Jia, S.-S. Bao, Y.-Q. Zhang and L.-M. Zheng, J. Mater. Chem. C, 2021, 9, 10749–10758.
29. Q. Zou, J.-C. Liu, X.-D. Huang, S.-S. Bao and L.-M. Zheng, Chin. Chem. Lett., 2021, 32, 1519–1522.
30. X.-D. Huang, B.-K. Hong, G.-H. Wen, S.-H. Li and L.-M. Zheng, Chem. Sci., 2023, 14, 1852–1860.
31. Q.-Q. Su, S.-S. Bao, X.-D. Huang, Q. Teng, X.-F. Ma, Y.-H. Qin and L.-M. Zheng, Cryst. Growth Des., 2024, 24, 10314–10325.
32. X.-F. Ma, X.-D. Huang and L.-M. Zheng, Cryst. Growth Des., 2023, 23, 1095–1103.
33. X.-D. Huang, X.-F. Ma, T. Shang, Y.-Q. Zhang and L.-M. Zheng, Inorg. Chem., 2023, 62, 1864–1874.
34. L. Zhu, A. Agarwal, J. Lai, R. O. Al-Kaysi, F. S. Tham, T. Ghaddar, L. Mueller and C. J. Bardeen, J. Mater. Chem., 2011, 21, 6258.
35. X.-D. Huang, J.-G. Jia, M. Kurmoo, S.-S. Bao and L.-M. Zheng, Dalton Trans., 2019, 48, 13769–13779.
36. T.-D. Nguyen, M.-T. Lau, K.-L. Hoang, T.-H. Dinh, H.-H. Nguyen and M.-H. Nguyen, Inorg. Chim. Acta, 2022, 541, 121066.
37. D. Casanova, M. Llunell, P. Alemany and S. Alvarez, Chem.–Eur. J., 2005, 11, 1479–1494.
38. G. M. J. Schmidt, Pure Appl. Chem., 1971, 27, 647–678.
39. W. T. Carnall, P. R. Fields and K. Rajnak, J. Chem. Phys., 1968, 49, 4424–4442.
40. X.-F. Ma, Y. Guo, X.-D. Huang, G.-H. Wen, S.-S. Bao, Y.-Q. Zhang and L.-M. Zheng, Dalton Trans., 2022, 51, 12026–12030.
41. Y. Shen, H. Liu, J. Cao, S. Zhang, W. Li and B. Yang, Phys. Chem. Chem. Phys., 2019, 21, 14511–14515.
42. S. Wang, H. Liu, S. Zhao, Q. Wu, Z. Yang, D. Yang, Y. Lv, Q. Su, S.-T. Zhang and B. Yang, Chem. Sci., 2025, 16, 3275–3284.
43. L.-Y. Hsu, S. Maity, Y. Matsunaga, Y.-F. Hsu, Y.-H. Liu, S.-M. Peng, T. Shinmyozu and J.-S. Yang, Chem. Sci., 2018, 9, 8990–9001.
44. M. Tu, H. Reinsch, S. Rodríguez-Hermida, R. Verbeke, T. Stassin, W. Egger, M. Dickmann, B. Dieu, J. Hofkens, I. F. J. Vankelecom, N. Stock and R. Ameloot, Angew. Chem., Int. Ed., 2019, 58, 2423–2427.
45. (a) R. Gao, Q. Zou, Q.-Q. Su, X.-F. Ma, Y.-H. Qin, R. Liao, S.-S. Bao and L.-M. Zheng, Chin. Chem. Lett., 2025, 36, 110404; (b) R. Gao, S.-S. Bao, Q.-Q. Su, X.-F. Ma and L.-M. Zheng, Inorg. Chem., 2025, 64, 9303–9313.
46. N. Kariaka, D. Panasiuk, V. Trush, S. Smola, N. Rusakova, V. Dyakonenko, S. Shishkina, A. Lipa, A. Bienko, J. Nasalska, P. Gawryszewska and V. Amirkhanov, Molecules, 2025, 30, 1245.
47. L. Wu, X.-D. Huang, W. Li, X. Cao, W.-H. Fang, L.-M. Zheng, M. Dolg and X. Chen, JACS Au, 2024, 4, 3606–3618.
48. Z. Chernia and D. Gill, Chem. Phys. Lett., 1993, 212, 57–63.
49. A. Kislyak, H. Frisch, M. Gernhardt, P. H. M. Van Steenberge, D. R. D'hooge and C. Barner-Kowollik, Chem.–Eur. J., 2020, 26, 478–484.
50. T. J. Gately, W. Sontising, C. J. Easley, I. Islam, R. O. Al-Kaysi, G. J. O. Beran and C. J. Bardeen, CrystEngComm, 2021, 23, 5931–5943.
51. A. Das, A. Danao, S. Banerjee, A. M. Raj, G. Sharma, R. Prabhakar, V. Srinivasan, V. Ramamurthy and P. Sen, J. Am. Chem. Soc., 2021, 143, 2025–2036.
52. R. Moré, M. Scholz, G. Busse, L. Busse, C. Paulmann, M. Tolkiehn and S. Techert, Phys. Chem. Chem. Phys., 2012, 14, 10187.
53. W. Li, T. J. Gately, D. Kitagawa, R. O. Al-Kaysi and C. J. Bardeen, J. Am. Chem. Soc., 2024, 146, 32757–32765.
54. H. Bouas-Laurent, A. Castellan, J.-P. Desvergne and R. Lapouyade, Chem. Soc. Rev., 2000, 29, 43–55.
55. J. Gemen, M. J. Białek, M. Kazes, L. J. W. Shimon, M. Feller, S. N. Semenov, Y. Diskin-Posner, D. Oron and R. Klajn, Chem, 2022, 8, 2362–2379.
56. S. Kataoka, D. Kitagawa, H. Sotome, S. Ito, H. Miyasaka, C. J. Bardeen and S. Kobatake, Chem. Sci., 2024, 15, 13421–13428.
57. A. Kislyak, D. Kodura, H. Frisch, F. Feist, P. H. M. Van Steenberge, C. Barner-Kowollik and D. R. D'hooge, Chem. Eng. J., 2020, 402, 126259.
58. X.-F. Ma, X.-L. Hou, Y.-H. Qin, Q. Teng, S.-S. Bao, Y.-X. Tian and L.-M. Zheng, Chem. Sci., 2025, 16, 10340–10348.
59. P. Lin, T. J. Burchell, R. Clérac and M. Murugesu, Angew. Chem., Int. Ed., 2008, 47, 8848–8851.
60. L. Zhang, J. Jung, P. Zhang, M. Guo, L. Zhao, J. Tang and B. Le Guennic, Chem.–Eur. J., 2016, 22, 1392–1398.
61. M. Kong, X. Feng, J. Wang, Y.-Q. Zhang and Y. Song, Dalton Trans., 2021, 50, 568–577.
62. H.-R. Tu, W.-B. Sun, H.-F. Li, P. Chen, Y.-M. Tian, W.-Y. Zhang, Y.-Q. Zhang and P.-F. Yan, Inorg. Chem. Front., 2017, 4, 499–508.
63. K. S. Cole and R. H. Cole, J. Chem. Phys., 1941, 9, 341–351.
64. S. Mukherjee, J. Lu, G. Velmurugan, S. Singh, G. Rajaraman, J. Tang and S. K. Ghosh, Inorg. Chem., 2016, 55, 11283–11298.
65. P. Hu, L. Cao, A. Liu, Y. Zhang, T. Zhang and B. Li, Dalton Trans., 2021, 50, 12814–12820.
66. W. Zhang, P. Chen, H. Li, Y. Zhang, P. Yan and W. Sun, Chem.–Asian J., 2018, 13, 1725–1734.
67. Y. Ma, Y. Zhai, Q. Luo, Y. Ding and Y. Zheng, Angew. Chem., Int. Ed., 2022, 61, e202206022.
68. L. Gu and R. Wu, Phys. Rev. B, 2021, 103, 014401.
69. M. Briganti, F. Santanni, L. Tesi, F. Totti, R. Sessoli and A. Lunghi, J. Am. Chem. Soc., 2021, 143, 13633–13645.
70. I. F. Galván, M. Vacher, A. Alavi, C. Angeli, F. Aquilante, J. Autschbach, J. J. Bao, S. I. Bokarev, N. A. Bogdanov, R. K. Carlson, L. F. Chibotaru, J. Creutzberg, N. Dattani, M. G. Delcey, S. S. Dong, A. Dreuw, L. Freitag, L. M. Frutos, L. Gagliardi, F. Gendron, A. Giussani, L. González, G. Grell, M. Y. Guo, C. E. Hoyer, M. Johansson, S. Keller, S. Knecht, G. Kovacevic, E. Källman, G. L. Manni, M. Lundberg, Y. J. Ma, S. Mai, J. P. Malhado, P. Å. Malmqvist, P. Marquetand, S. A. Mewes, J. Norell, M. Olivucci, M. Oppel, Q. M. Phung, K. Pierloot, F. Plasser, M. Reiher, A. M. Sand, I. Schapiro, P. Sharma, C. J. Stein, L. K. Sørensen, D. G. Truhlar, M. Ugandi, L. Ungur, A. Valentini, S. Vancoillie, V. Veryazov, O. Weser, T. A. Wesołowski, P.-O. Widmark, S. Wouters, A. Zech, J. P. Zobel and R. Lindh, J. Chem. Theory Comput., 2019, 15, 925–5964.
71. L. F. Chibotaru, L. Ungur and A. Soncini, Angew. Chem., Int. Ed., 2008, 47, 4126–4129.
72. L. Ungur, W. Van den Heuvel and L. F. Chibotaru, New J. Chem., 2009, 33, 1224–1230.
73. L. F. Chibotaru, L. Ungur, C. Aronica, H. Elmoll, G. Pilet and D. Luneau, J. Am. Chem. Soc., 2008, 130, 12445–12455.
74. A. Lunghi, F. Totti, R. Sessoli and S. Sanvito, Nat. Commun., 2017, 8, 14620–14626.
75. F. Lu, M. M. Ding, J.-X. Li, B.-L. Wang and Y.-Q. Zhang, Dalton Trans., 2020, 49, 14576–14583.
76. F. Lu, W.-X. Guo and Y.-Q. Zhang, Inorg. Chem., 2022, 61, 295–301.
77. M.-M. Ding, T. Shang, R. Hu and Y.-Q. Zhang, Dalton Trans., 2022, 51, 3295–3303.
78. T. Shang, F. Lu, J. Tao and Y.-Q. Zhang, J. Phys. Chem. A, 2023, 127, 3088–3095.
79. D.-M. Zhu, F. Lu, X.-L. Zhang and Y.-Q. Zhang, Inorg. Chem., 2025, 64, 4209–4212.
80. M. E. Lines, J. Chem. Phys., 1971, 55, 2977–2984.
81. D.-K. Cao, Y.-W. Gu, J.-Q. Feng, Z.-S. Cai and M. D. Ward, Dalton Trans., 2013, 42, 11436–11444.
82. O. Kahn, Molecular Magnetism, Wiley-VCH, New York, 1993.
83. SAINT, Version 8.40A Program for Data Extraction and Reduction, Bruker Nano. Inc., 2019.
84. SHELXT 2014/5, Sheldrick, 2014; SHELXL 2018/3, Sheldrick. 2018 .
85. (a) Y.-H. Qin, X.-F. Ma, X. Hou, X.-D. Huang, S.-S. Bao, Y. Tian, Y.-Q. Zhang and L.-M. Zheng, CCDC 2455154: Experimental Crystal Structure Determination, 2025,  DOI:10.5517/ccdc.csd.cc2ndsjd; (b) Y.-H. Qin, X.-F. Ma, X. Hou, X.-D. Huang, S.-S. Bao, Y. Tian, Y.-Q. Zhang and L.-M. Zheng, CCDC 2456341: Experimental Crystal Structure Determination, 2025,  DOI:10.5517/ccdc.csd.cc2ng0tz; (c) Y.-H. Qin, X.-F. Ma, X. Hou, X.-D. Huang, S.-S. Bao, Y. Tian, Y.-Q. Zhang and L.-M. Zheng, CCDC 2456346: Experimental Crystal Structure Determination, 2025,  DOI:10.5517/ccdc.csd.cc2ng0z4; (d) Y.-H. Qin, X.-F. Ma, X. Hou, X.-D. Huang, S.-S. Bao, Y. Tian, Y.-Q. Zhang and L.-M. Zheng, CCDC 2456349: Experimental Crystal Structure Determination, 2025,  DOI:10.5517/ccdc.csd.cc2ng128.

---

Footnote

† These authors contributed equally to this work.

---