Insights into the Effect of Trans-to-Cis Photoisomerization of a Co-coordinated Stilbene Derivative on the Luminescence of Di-β-diketonate Lanthanide Complexes.

Five lanthanide complexes constructed from a stilbene derivative, (E)-N',N'-bis(pyridin-2-ylmethyl)-4-styrylbenzoyl hydrazide (HL), and two β-diketonates (2-thenoyltrifluoroacetonate, tta), with or without a trifluoroacetate anion (CF3CO2 -), namely, [Ln(tta)2(HL) (CF3CO2)] [LnC45H32F9N4O7S2, Ln = La (1), Nd (2), Eu (3), or Gd (4)] and [Yb(tta)2(L)] (YbC43H31F6N4O5S2 (5), L = deprotonated HL), were synthesized and characterized. Crystals of these five complexes were obtained and analyzed by single-crystal X-ray diffraction. These complexes all belonged to the monoclinic P21/c space group. For La3+, Nd3+, Eu3+, and Gd3+, the central lanthanide ion was nine-coordinate with a monocapped twisted square antiprism polyhedron geometry. The central Yb3+ ion of complex 5 was eight-coordinate with a distorted double-capped triangular prism polyhedron geometry. Among the five complexes, trans-to-cis photoisomerization of the stilbene group in gadolinium complex 4 showed the largest quantum yield. Complexes 2, 3, and 4 showed dual luminescence and photoisomerization functions. The luminescence change of complex 3 was reversible upon the trans-to-cis photoisomerization process. The sensitization efficiencies of luminescent europium complex 3 in acetonitrile solutions and in the solid state were 49.9 and 42.6%, respectively. These medium sensitization efficiencies led to the observation of simultaneous photoisomerization and luminescence, which further confirmed our previous report that photoisomerization of the stilbene group within complexes was related to the lanthanide ion energy level and whether a ligand-to-metal center or ligand-to-ligand charge-transfer process was present. In complexes 1-5, in addition to the intramolecular absorption transition of the ligand itself (IL, πHL-πHL * and πtta-πtta*), the presence of a ligand-to-ligand charge-transfer transition between tta and HL (LLCT, πtta-πHL * or πHL-πtta *) indicated whether the triplet-state energy of HL was able to transfer to the excited energy level of the lanthanide ions, leading to different extents of HL photoisomerization. These results provide an important route for the design of new dual-function lanthanide-based optical switching materials.

Introduction

In recent years, smart materials have been studied extensively for their important applications in molecular switches,[1−6] white light-emitting diodes,[7−9] information storage,[10,11] anti-counterfeiting,[12,13] separation and detection,[14−16] drug delivery,[17,18] and other fields.[19−24] Smart material molecules can reversibly change their physical and chemical properties under different external stimuli.[25,26] Among various external stimuli, including light energy, electrical energy, and chemical energy, light is the most suitable and convenient source of clean energy and has advantages of precise control as a stimulus source and energy-saving and environmental protection characteristics. Previously reported light stimulation of smart molecules includes the trans-to-cis isomerization of azobenzene and stilbene derivatives,[27−38] photocyclization of diarylethylenes,[39−41] [2 + 2] cycloaddition of alkenes,[42−44] and [4 + 4] photodimerization of anthracene.[45−48] Photoinduced trans-to-cis isomerization has attracted much attention in inorganic and organic chemistry, materials science, and biology. Photoinduced trans-to-cis isomerization is the reversible conversion of trans isomers to cis isomers under a certain light wavelength.[27,49] Isomers with different molecular structures have different physical and chemical properties (including magnetic, optical, and electrical properties), which can be controlled by illumination. This feature is widely used in biochemistry, pharmacology, and supramolecular chemistry.[50−56] Much research has been conducted on the photoinduced trans-to-cis isomerization of various azobenzene and stilbene derivatives. In addition, we have studied a series of lanthanide β-diketonate complexes functionalized with azobenzene, finding that azobenzene groups attached to lanthanide β-diketonate complexes were stable in both solution and the solid state, exhibiting the reversible trans-to-cis photoisomerization properties of azobenzene and good fatigue resistance.[30,34,57,58] The introduction of electron donor substituents is not conducive to improving the photoisomerization quantum yield and photoisomerization rate.[34] Although some ternary complexes of europium and ytterbium β-diketonate complexes functionalized with azobenzene exhibit both luminescence and photoisomerization properties, energy transfer from an azobenzene derivative to lanthanide ions is difficult owing to the first triplet-state energy level of azobenzene derivatives being lower than the excited-state energy level of most lanthanide ions.[57] Therefore, most azobenzene-functionalized lanthanide β-diketonate complexes exhibit photoisomerization but not luminescence emission. Azobenzene-functionalized lanthanide β-diketonate complexes with good dual-function properties are difficult to achieve. To obtain lanthanide complex materials with both luminescent and trans-to-cis photoisomerization properties, we focused on designing stilbene derivatives, which are known for their rich optical properties, such as trans-to-cis photoisomerization, photocyclization, photodimerization, and fluorescence, for introduction into lanthanide β-diketonate complexes.[35] The as-designed stilbene derivative ligand, (E)-N′,N′-bis(pyridin-2-ylmethyl)-4-styrylbenzoyl hydrazide (HL), is a tetradentate chelating ligand that can strongly coordinate with lanthanide ions. HL was combined with benzoyl trifluoroacetone (Htfd), a classic β-diketone ligand that sensitizes lanthanide ions, to obtain lanthanide β-diketonate complexes co-coordinated with a stilbene derivative. Stilbene derivative HL co-coordinated in the tfd–lanthanide β-diketonate complex maintained its trans-to-cis photoisomerization properties. The extent of stilbene photoisomerization within complexes was related to the lanthanide ion energy level and whether a ligand-to-metal center or ligand-to-ligand charge-transfer process was present. Trans-to-cis photoisomerization of the stilbene group in the gadolinium complex afforded an approximate fivefold enhancement in the quantum yield compared with HL alone. The Yb and Nd complexes had dual characteristics of infrared luminescence and trans-to-cis photoisomerization. However, in the europium complex, highly efficient ligand energy transfer to the Eu3+ ion led to completely suppressed stilbene group photoisomerization. The energy transfer efficiency among ligands HL and tfd and the lanthanide ions affected the lifetime of the lanthanide ion excited-state and photoisomerization behavior of the stilbene group. Therefore, we planned to design and synthesize dual-function materials with photoisomerization properties and excellent optical properties. In this context, we selected 2-thenoyl trifluoroacetone (Htta), a ligand with a better “antenna effect”,[59,60] for co-coordination with lanthanide ions to study its luminescence and photo-induced trans-to-cis isomerization and further clarify the relationship between luminescence and isomerization within these complexes.

Results and Discussion

X-ray Crystallographic Analysis

Crystals of the five complexes were obtained by slow evaporation from a methanol solution over 1 week. Crystal analysis showed that the five complexes had similar structures and all belonged to the monoclinic P21/c space group. For complexes 1–4, the central lanthanide ion was coordinated with six oxygen atoms and three nitrogen atoms to form a stable nine-coordinate structure with monocapped twisted square antiprism polyhedron geometry. HL is a tetradentate ligand that provides one oxygen atom and three nitrogen atoms as chelating atoms. The other four chelating oxygen atoms were provided by two 2-thenoyltrifluoroacetonate (tta) ligands, while the sixth oxygen was derived from the trifluoroacetate anion hydrolyzed by the ligand Htta. In complexes 1–4, intermolecular hydrogen bonding existed between the carbonyl oxygen of trifluoroacetate and hydrogen on the hydrazine nitrogen in HL. The coordination mode of complex 5 was slightly different from those of complexes 1–4, with the central Yb(III) ion having a coordination number of 8 with a distorted double-capped triangular prism polyhedron geometry. No trifluoroacetate anion was involved in coordination in 5 owing to removal of a proton from the hydrazine nitrogen in HL to balance the charge. This was similar to our previously reported Yb3+ complexes functionalized by azobenzene groups.[58] The crystal structures of complexes 4 and 5 are shown in Figure , while those of complexes 1–3 are shown in Figure S1 in the Supporting Information. The crystallographic parameters of complexes 1–5 are provided in Table , while selected bond lengths and angles are shown in Tables S1 and S2 (Supporting Information). The lanthanide ion–oxygen bond length ranges in complexes 1–5 were 2.468–2.537, 2.357–2.451, 2.371–2.425, 2.377–2.465, and 2.206–2.475 Å, respectively. The corresponding lanthanide ion–nitrogen bond length ranges were 2.734–2.873, 2.644–2.755, 2.639–2.752, 2.613–2.706, and 2.475–2.546 Å, respectively. These results were within the ranges of reported lanthanide β–diketonate complexes.[30,34,58] The measured XRD powder diffraction spectra of complexes 1–5 were in agreement with the simulated values (Figure S2 in the Supporting Information), confirming that the obtained five complexes had a pure phase.

Figure 1

Crystal structures of complexes 4 (a) and 5 (c) (all hydrogen atoms are omitted for clarity) and coordination polyhedron geometry of central La(III) (b for complex 4, d for complex 5).

Table 1

Crystallographic Data for Complexes 1–5

crystal data	1	2	3	4	5
formula	LaC45H32F9N4O7S2	NdC45H32F9N4O7S2	EuC45H32F9N4O7S2	GdC45H32F9N4O7S2	YbC43H31F6N4O5S2
FW	1114.77	1122.13	1127.82	1133.11	1034.88
crystal system	monoclinic	monoclinic	monoclinic	monoclinic	monoclinic
space group	P21/c	P21/c	P21/c	P21/c	P21/c
a (Å)	9.891(5)	24.269(4)	24.1658(4)	9.762(4)	8.308 (10)
b (Å)	17.372(10)	9.788(2)	9.7895(2)	17.811(6)	26.613(4)
c (Å)	30.071(2)	19.295(4)	19.2341(4)	32.585(17)	18.586(3)
α (°)	90	90	90	90	90
β (°)	93.666(6)	101.887(2)	101.285(2)	111.049(5)	95.7(10)
Γ (°)	90	90	90	90	90
V (Å)	5157.0(5)	4485.4(2)	4462.26(15)	5288.0(4)	4089.5(10)
Z	4	4	4	4	4
D (g cm–3)	1.436	1.659	1.679	1.423	1.681
θ range (°)	5.87 to 133.07	7.44 to 133.20	7.46 to 131.338	5.75 to 129.99	6.64 to 154.78
refl. col.	29078	28401	31102	29121	28488
data/res./para.	8837/1078/733	7912/1029/787	7585/924/761	8958/733/698	8299/282/624
GOF	1.044	1.054	1.162	1.057	1.027
final R	R1 = 0.0889	R1 = 0.0922	R1 = 0.0518	R1 = 0.1001	R1 = 0.0351
R (all data)	R1 = 0.1585, wR2 = 0.1373	R1 = 0.1082, wR2 = 0.2544	R1 = 0.0627, wR2 = 0.1401	R1 = 0.1229, wR2 = 0.3024	R1 = 0.0444, wR2 = 0.0903

Thermal Analysis of Complexes 1–5

Thermogravimetric analyses (TGAs) of complexes 1–5 were conducted using powder samples in the temperature range of 30–800 °C under a nitrogen atmosphere. The TGA plots showed that complexes 1–4 had similar thermal decomposition processes, with complex 5 also showing the same trend (Figure S3 in the Supporting Information). Complexes 1–4 underwent the first stage of decomposition in the temperature range of 255–380 °C with mass losses of 27.05, 27.92, 27.65, and 27.67%, respectively. These results were similar to theoretically calculated percentages for the loss of one trifluoroacetate anion and one tta dionate from the complexes (29.98, 29.84, 29.63, and 29.47%, respectively). Meanwhile, the TGA curves of complex 5 showed a first-stage mass loss of 24.23% in the temperature range of 270–370 °C, indicating a mass loss of one tta dionate from the complex (calculated, 23.43%). All complexes further decomposed between 380 and 800 °C. These thermal stability results showed that the trifluoroacetate anion and tta dionate were more readily removed from the complexes, indicating that HL was a strongly coordinating tetradentate ligand for chelating lanthanide ions.

Magnetic Properties of Complex 4

The magnetic properties of complex 4 were also studied. Variable temperature molar magnetic susceptibility (χM) in the range of 2–300 K was measured under an applied field of 1000 Oe (Figure S4 in the Supporting Information). The room-temperature χMT value was 7.42 cm3 K mol–1, which was slightly lower than the theoretical value (7.87 for one isolated Gd3+ ion) owing to quenching of the orbital magnetic moment contribution to the center ion. The values of C (cm3 K mol–1) and θ (K) obtained from the best fit χMT data (χM–1 vs T) using the Curie–Weiss equation were 7.14 and −0.77, respectively, showing the presence of a weak antiferromagnetic interaction between Gd3+ ions.[61,62]

Photoisomerization Behavior of Complexes 1–5

UV–vis absorption spectra of Htta, HL, and their complexes 1–5 in acetonitrile at room temperature were obtained, with the corresponding spectral data listed in Table . The spectral shapes of complexes 1–5 showed iteration of the free ligand Htta and HL absorption spectra (Figure ), with a red shift of about 20 nm at the long wavelength and a small shift at the short wavelength compared with free ligands. The molar absorption coefficients of complexes 1–5 at the maximum wavelength were in the same order as those of the free ligands. However, the results for complex 1 were higher than those of the free ligands, perhaps owing to the rigid structure resulting from the coordination effect of the lanthanum ion with no f electrons. Density functional theory (DFT)/time-dependent (TD)–DFT calculations (Table ) showed that complex 1 has the largest oscillator strengths for the intramolecular charge transition of the ligands (IL, πHL–πHL*), ligand-to-ligand charge transfer (LLCT, πHL–πtta*), and ligand-to-metal certer charge transfer (LHLMCT) transitions among the five complexes. This might be the reason that complex 1 shows a different behavior.

Table 2

UV–Vis Absorption Data of Ligands and Complexes 1–5 in Acetonitrile Solutions

compounds	λmax[nm] (εmax[104 L·mol–1·cm–1])
Htta	292(1.95)	262(2.64)
HL	317(5.16)	262(1.96) 268(2.13)
1	335(10.50)	269(4.14)
2	337(5.14)	269(2.04)
3	335(6.52)	269(2.59)
4	336(4.94)	270(1.91)
5	337(5.44)	269(2.45)

Figure 2

UV–vis absorption spectra of Htta, HL, and complexes 1–5 in acetonitrile solutions.

Table 5

Solvent-Corrected (Acetonitrile) Major Transition Energies (eV) to Excited States of HL and Complexes 1–5 with Contributing Excitation (%), Oscillator Strengths (f), Associated wavelengths (λ), and Dominant Excitation Characteristics

compounds	transition	f	E/ev	λ/nm	character
complex 1	HOMO → LUMO (18%), HOMO → L + 1 (79%)	0.594	3.45	359	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	HOMO → L + 2 (95%)	0.678	3.54	350	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	H – 1 → L + 1 (92%)	0.152	3.64	341	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	H – 2 → L + 1 (87%)	0.301	3.71	334	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	H – 1 → L + 2 (74%), H – 2 → L + 1 (12%)	0.434	3.78	328	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
complex 2	HOMO(A) → LUMO(A) (64%)	0.185	3.35	370	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	H – 1(A) → L + 2(A) (13%)	0.118	3.43	361	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	HOMO(A) → L + 2(A) (20%)	0.141	3.49	355	IL(πHL–πHL*)LLCT(πHL–πtta*)LttaMCT
	HOMO(A) → L + 2(A) (25%)	0.485	3.53	351	IL(πHL–πHL*)LLCT(πHL–πtta*)LttaMCT
	HOMO(A) → L + 4(A) (22%),	0.187	3.74	331	IL(πHL–πtta*)LHLMCT
	H – 2(A) → L + 1(A) (17%)				IL(πHL–πtta*)LLCT(πtta–πHL*)
	H – 1(A) → L + 4(A) (24%)	0.254	4.29	289	IL(πHL–πtta*)LttaMCT
complex 3	HOMO → LUMO (18%), H – 2 → LUMO (11%)	0.245	3.40	364	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	H – 2 → L + 2 (22%), H – 2 → L + 3 (13%)				IL(πHL–πtta*)LLCT(πtta–πHL*)
	HOMO → LUMO (28%), H – 2 → L + 2 (13%)	0.222	3.42	362	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	H – 1 → L + 3 (12%)				IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	HOMO → L + 2 (50%), H – 1 → L + 6 (12%)	0.292	3.51	353	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	H – 1 → L + 5 (17%), H – 1 → L + 6 (37%),	0.139	3.53	351	IL(πtta–πtta*)LttaMCT
	HOMO → L + 2 (16%)				IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	HOMO → L + 3 (93%)	0.309	3.59	345	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	H – 2 → L + 1 (32%), H – 2 → L + 5 (11%),	0.191	3.80	326	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	H – 2 → L + 6(22%)
	H – 1 → L + 2 (18%), H – 1 → L + 4 (16%),	0.264	3.84	323	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	H – 1 → L + 5 (11%)	0.402	3.55	349
complex 4	HOMO (A) → L + 1 (A) (27%)	0.349	3.57	347	IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	HOMO (A) → LUMO (A) (13%)				IL(πHL–πHL*)LLCT(πHL–πtta*)LHLMC′
	HOMO (A) → L + 1 (A) (21%), HOMO (A) → L + 2 (A)
	H – 1 (A) → LUMO (A) (54%),	0.101	3.63	341	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	H – 2 (A) → LUMO (A) (40%),	0.314	3.70	335	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
	H – 1 (A) → L + 1 (A) (33%)
	H – 1 (A) → L + 2 (A) (23%)	0.215	3.80	326	IL(πtta–πtta*)LLCT(πtta–πHL*)LttaMCT
complex 5	H – 1 (A) → LUMO (A) (42%)	0.132	3.51	354	IL(πtta–πtta*)LLCT(πhydrazide group–πtta*)
					LttaMCT
	H – 1 (A) → L + 1 (A) (27%)	0.366	3.57	347	IL(πtta–πtta*)LLCT(πhydrazide group–πtta*)
	H – 3 (A) → LUMO (A) (15%),	0.792	3.59	345	IL(πtta–πtta*)LLCT(πHL–πtta*)LttaMCT
	HOMO (A) → L + 2 (A) (26%)				IL(πHL–πHL*)
	H – 2 (A) → LUMO (A) (24%)	0.106	3.60	344	IL(πtta–πtta*)LLCT(πHL–πtta*)LttaMCT
	H – 2 (A) → LUMO (A) (29%)	0.158	3.61	343	IL(πtta–πtta*)LLCT(πHL–πtta*)LttaMCT
	H – 4 (A) → L + 1 (A) (25%)	0.208	3.80	326	IL(πtta–πtta*)LLCT(πhydrazide group–πtta*)
					LMCT
	H – 3 (A) → L + 1 (A) (15%)	0.362	3.85	322	IL(πtta–πtta*)LLCT(πhydrazide group–πtta*)
					LMCT
	H – 4 (A) → L – + – 1 (A) (20%)	0.193	3.89	319	IL(πtta–πtta*)LLCT(πhydrazide group–πtta*)
					LMCT

The five complexes were similar to our previously reported lanthanide complexes, with only the two trifluoro-1-phenyl-butanedionate (tfd) ligands changed to tta ligands, but they showed very different photoisomerization properties. The five complexes showed similar photoisomerization behavior in acetonitrile solutions but different photoisomerization rates and quantum yields. For example, the UV–vis absorption spectra of complex 3 in acetonitrile solutions showed a similar variation to that of free ligand HL. In acetonitrile solutions, under 312-nm UV light irradiation, the characteristic absorption peak intensity of the π–π* transition at 335 nm gradually decreased with increasing irradiation time. Simultaneously, the characteristic absorption peak intensity of the π–π* transition at 269 nm gradually increased, with an isosbestic point appearing at 287 nm (Figure ). Complex 3 reached a photostable state after irradiation for 300 s. The solution was then subjected to 254 nm UV light irradiation, which caused the characteristic absorption peak intensity at 335 nm to increase with increasing irradiation time, with a constant intensity maintained after 600 s. These changes indicated that the stilbene group in the complex underwent a trans–cis–trans photoisomerization process.

Figure 3

UV–vis spectral change of complex 3 in acetonitrile solutions (2.0 × 10–5 mol/L) upon irradiation at UV-312 nm and recoverable irradiation at UV-254 nm as a function of time.

Subsequently, the reversibility and recyclability of the trans–cis–trans photoisomerization behavior of complex 3 in acetonitrile solutions were tested. UV–vis absorption spectra of the solution were recorded alternately using 312 nm UV light irradiation for 3 min followed by 254 nm UV light for 5 min. As shown in Figure , the trans-to-cis photoisomerization reaction of complex 3 was reversible and recyclable. Although both the pure ligand and complex 3 underwent photolysis, the number of cycles for complex 3 under an alternating UV light irradiation was higher than that of the pure HL ligand owing to the enhanced photostability of the stilbene group after complex formation.

Figure 4

Cycles of maximum absorption intensity at 335 nm of complex 3 and 317 nm of HL in acetonitrile solutions are modulated by the irradiation of UV light at 312 nm for 3 min and 254 nm for 5 min, alternatively.

The photoisomerization properties of the four other complexes were also explored, showing similar photoisomerization behavior in acetonitrile solutions but with different photoisomerization quantum yields and rate constants. From variations in the UV absorption spectra with increasing irradiation time, the trans-to-cis photoisomerization quantum yields and first-order kinetic rate constants were determined. The UV–vis spectra of the other four complexes in acetonitrile solutions under alternating irradiation with UV light at 312 and 254 nm are shown in Figures S5 and S6 (Supporting Information). Table shows the trans-to-cis quantum yields (Φt–c) under irradiation at 312 nm and photoisomerization rate constants (k, s–1) of HL and complexes 1–5 in acetonitrile (UV light power, 266 μW cm–2 at 312 nm). Although nuclear magnetic resonance (NMR) spectroscopy can be used to obtain an accurate quantum yield,[63,64] the quantum yields herein, calculated using only the UV–vis spectrum variations, were “apparent”. These calculations were performed using the same method as in our previous report.[34] The photoisomerization quantum yields of the complexes were all higher than that of HL, while their isomerization rate constants (kiso) were positively correlated with the photoisomerization quantum yields. The photoisomerization rate constant and quantum yield of complex 4 were largest among the five complexes. This was similar to our previously reported 4,4,4-trifluoro-1-phenylbutane-1,3-diketonate gadolinium complex co-coordinated with a stilbene derivative ([Gd(tfd)2(HL) (CF3COO)]).[35] However, the photoisomerization rate constants (kiso) and quantum yields (Φt–c) of the five complexes and HL were in the order of 4 > 3 > 5 > 2 > 1 > HL. Complex 3 was different from the reported 4,4,4-trifluoro-1-phenylbutane-1,3-diketone europium complex co-coordinated with a stilbene derivative ([Eu(tfd)2(HL) (CF3COO)]), which exhibited no detectable photoisomerization. This was due to the difference in energy transfer efficiency between the ligand and lanthanide ion, which will be discussed below in the studies on photoluminescence properties and theoretical calculations.

Table 3

Photoisomerization Rate Constants Kiso (s–1) and Quantum Yields (Φt–c) of HL and Complexes 1–5 in Acetonitrile Solutions

compounds	103kiso	102Φt→c	R1a	R2a	R1a,c	R2a,c
HL	4.8 ± 0.5	1.9 ± 0.1	1.0	1.0	1.0	1.0
1	12.9 ± 0.2	1.9 ± 0.1	2.7	1.0	0.1	1.0
2	15.1 ± 0.5	3.3 ± 0.3	3.1	1.7	3.4	2.0
3	23.2 ± 1.8	10.4 ± 0.3	4.8	5.5	b	b
4	33.7 ± 2.4	10.9 ± 0.3	7.0	5.7	4.8	4.5
5	17.3 ± 0.3	5.7 ± 0.3	3.6	3.0	1.3	1.2

R1 is the ratio of the Kiso value of a complex to HL and R2 is the ratio of the Φt→c value of a complex to HL.

No detectable value for complex 3 in ref (35).

Data calculated from ref (35).

Photoluminescence Properties of Complexes 2, 3, and 5

Nd3+, Eu3+, and Yb3+ β-diketonate complexes are known to emit strong red or near-infrared (NIR) luminescence if efficient energy transfer from the β-diketonate and auxiliary ligands to the central lanthanide(III) ions is present. First, the luminescence properties of complex 3 at room temperature were investigated. Photoluminescence spectra of complex 3 in acetonitrile solutions and in the solid state are shown in Figure . Complex 3 displayed pure red photoluminescence and its CIE coordinate was x = 0.67 and y = 0.33 (Figure S7 in the Supporting Information). Upon excitation at 355 nm, complex 3 exhibited characteristic Eu(III) ion emission bands of 5D0 → 7FJ (J = 0–4) transitions. The 5D0 → 7F2 transition at 614 nm from the induced electric dipole hypersensitive transition was strongest and dominated the whole spectrum. The 5D0 → 7F0,1,3 transitions at 578, 590, and 650 nm, respectively, were relatively weak. Only one peak for the 5D0 → 7F0 (0–0) transition and three Stark components for the magnetic dipole 5D0 → 7F1 transition, in particular the splitting pattern of 5D0 → 7F2 in the emission spectra, suggested the major presence of a single chemical environment around the Eu(III) ion and lower complex symmetry. The overall luminescence quantum yields (Φtot) of complex 3 were determined experimentally to be 0.21 and 0.39 in acetonitrile solutions and in the solid state, respectively, at room temperature, which were slightly higher than those of complex [Eu(tfd)2(HL) (CF3COO)] with a similar structure (Φtot = 0.19 in acetonitrile and 0.22 in solid). Furthermore, the 5D0 emission decay curve excited at 355 nm showed a single exponential behavior monitored within the 5D0 → 7F2 transition (Figure S8 in the Supporting Information), yielding lifetimes (τobs) of 616 and 765 μs in acetonitrile solutions and in the solid state, respectively. The radiation life time (τrad) of Eu3+ ions can be calculated using eq (65)where AMD,0 is the spontaneous emission probability of the 5D0 → 7F1 transition in vacuo (14.65 s–1), Itot is the total integral of the whole emission spectrum, IMD is the integral of the 5D0 → 7F1 transition emission peak, and n is the refractive index of the solvent (1.5 in the solid state).

Figure 5

Luminescence spectra of complex 3 in acetonitrile solutions (2.0 × 10–5 mol/L) and in the solid state (λex = 355 nm). The inset shows images of a bright red color luminescence in the solid state under a 365 nm UV lamp.

Therefore, using the equation ΦLn = τobs/τrad,[65] the intrinsic quantum yields (ΦLn) in acetonitrile solutions and in the solid state were calculated to be 42.4 and 91.5%, with sensitization efficiencies (ηsen = Φtot/ΦLn) of 49.9 and 42.6%, respectively (Table ). These lower sensitization efficiencies indicated that the ligand energy transfer to the Eu3+ ion was still less efficient in this complex, with sufficient energy remaining for HL to undergo trans-to-cis photoisomerization. This explained why photoisomerization of the stilbene group was observed in complex 3 but suppressed complex [Eu(tfd)2(HL) (CF3COO)] with a similar structure, which exhibited a higher sensitization efficiency of 97.9% in acetonitrile solutions and 96.7% in the solid state, resulting in almost no energy remaining for HL to undergo trans-to-cis photoisomerization.

Table 4

Luminescence Physical Parameters of Complexes 2, 3, and 5

complex	environment	τobs (μs)	τrad (μs)	Φtot (%)	ΦLn (%)	ηsens (%)
2	solution Solid	3.2			1.3
	solid	6.1			2.4
3	solution	616	1452	21.10	42.4	49.9
	solid	765	835	38.94	91.5	42.6
5	solution	12.1			0.6
	solid	14.3			0.7

Furthermore, the emission of complex 3 in the photostationary cis form in the solid state was checked (Figure ). Under irradiation at 312 nm, the luminescence intensity of 5D0 → 7FJ (J = 0–4) transitions of complex 3 all increased with increasing irradiation time. The photoisomerization reaction reached a photostable state after 5 min. The powder was then irradiated with UV light at 254 nm, which gradually decreased the luminescence intensity with increasing irradiation time, finally reaching a new photostable state after 15 min. The changes in luminescence intensity with trans-to-cis photoisomerization of complex 3 were also reversible and recyclable and observed for more than 20 cycles with small photolysis (Figure S9 in the Supporting Information). A similar luminescence change was also observed for complex 3 in acetonitrile solutions (Figure S10 in the Supporting Information). This luminescence variation might result from a new crystal packing within the cis form, introducing the hypersensitive transition enhancement in complex 3. These results show that complex 3 is a new lanthanide-based optical switching material.

Figure 6

Luminescence spectral change of complex 3 in the solid state upon irradiation at UV-312 nm (a) and recoverable irradiation at UV-254 nm (b) as a function of time.

The NIR luminescence of complexes 2 and 5 in acetonitrile solutions and in the solid state was also examined. With excitation wavelengths set at 358 and 365 nm, respectively, complexes 2 and 5 showed luminescence peaks characteristic of Yb3+ and Nd3+ ions, respectively (Figure ). The strong luminescence emission peak of complex 2 was located at about 1062 nm, which corresponded with the 4F3/2 → 4I11/2 transition of Nd3+ ions. The weak and broad emission peaks at 887 and 1335 nm corresponded to 4F3/2 → 4I9/2 and 4F3/2 → 4I13/2 transitions, respectively. Complex 5 also showed three luminescence emission peaks, among which the emission at 980 nm was strongest, attributed to the 2F5/2 → 2F7/2 transition of Yb3+ ions, while the broad emission peaks at around 1015 and 1050 nm were attributed to the Stark splitting of the 2F7/2 multiplex.

Figure 7

NIR luminescence spectra of complexes 2 (a, λex = 365 nm) and 5 (b, λex = 358 nm) in acetonitrile solutions (1.0 × 10–4 mol/L) and in the solid state.

The NIR intrinsic quantum yield of the lanthanide ions can be estimated using the equation ΦLn = τobs/τ0, where τ0 is the natural lifetime (0.25 ms for Nd3+ and 2.0 ms for Yb3+).[66,67] The luminescence lifetime (τobs) and intrinsic quantum yield (ΦLn) of complexes 2 and 5 are shown in Table . The decay curves of the strongest peaks of complexes 2 and 5 as a function of time, which fitted a single exponential behavior, were used to calculate τobs, as shown in Figure S11 (Supporting Information). The NIR luminescence properties of the two complexes were similar to our reported Nd3+ and Yb3+ complexes with similar structures, which also had observed lifetime values in the order of 1–10 μs.[35] This indicated that the energy transfer efficiency of the two series of complexes was similar. Therefore, the photoisomerization rates and quantum yields of the two series of Nd and Yb complexes were also similar. The triplet excited-state energy levels (3ππ*) of ligands can be determined from the phosphorescence spectra of their corresponding Gd complexes.[68] HL did not show phosphorescence at room temperature. The phosphorescence of Htta, Htfd, and their Gd complexes at room temperature was measured, with ligands and their complexes showing the same maximum emission wavelength of around 20,202 cm–1 for Htta and 21,413 cm–1 for Htfd (Figure S12), respectively, in agreement with the literature.[69] The triplet energy of HL estimated by TD–DFT was 18,083 cm–1. The energy levels of ligands HL, Htta, and Htfd and lanthanide ions Eu3+, Nd3+, Yb3+, and Gd3+ are shown in Scheme . All triplet energy levels of ligands were higher than the Eu3+ 5D0 (17,241 cm–1), Nd3+ 4F3/2 (11,500 cm–1), and Yb3+ 2F5/2 (10,260 cm–1) energy levels. HL might directly transfer energy to lanthanide ions through the triplet excited state. This explains why the stilbene-attached lanthanide β-diketonate complexes showed stronger luminescence than azobenzene-functionalized lanthanide complexes.

Scheme 2

Schematic Representation of Energy Levels of the First Excited States of Ligands HL, Htta, and Htfd and the Related Energy Levels of Eu3+, Nd3+, Yb3+, and Gd3+

Electronic Absorption Analysis

Complexes 1–5 in the acetonitrile solvent were successfully optimized using DFT, with the optimized structural geometric parameters shown in Tables S3 and S4. The optimized partial bond lengths and angles were compared with those obtained by X-ray single-crystal diffraction (measured crystal data shown in Tables S1 and S2, Supporting Information), showing some differences. The optimized lanthanide ion–oxygen bond length ranges in complexes 1–5 were 2.463–2.560, 2.386–2.445, 2.337–2.467, 2.322–2.470, and 2.229–2.263 Å, respectively. The corresponding lanthanide ion–nitrogen bond length ranges were 2.813–2.998, 2.729–2.779, 2.712–2.901, 2.719–2.918, and 2.522–2.585 Å, respectively. They are close to the crystal data (in crystal, Ln–O bond length ranges in complexes 1–5 were 2.468–2.537, 2.357–2.451, 2.371–2.425, 2.377–2.465, and 2.206–2.475 Å, respectively. The Ln–N bond length ranges were 2.734–2.873, 2.644–2.755, 2.639–2.752, 2.613–2.706, and 2.475–2.546 Å, respectively). These differences were within a reasonable range and acceptable for the simulation process conducted on the acetonitrile solution, while experimental data were obtained in the crystal state. The calculated bond lengths and angles were essentially consistent with the measured bond lengths and angles, confirming the rationality of the large-core effective core potentials (ECPs).

Electron density distributions of the highest occupied molecular orbitals (HOMOs) and lowest unoccupied molecular orbital (LUMO) of complexes 1–5 in the acetonitrile solvent were obtained, as shown in Figure . Among them, complexes 2, 4, and 5 were optimized using the UB3LYP functional and only their α-orbitals were shown. The HOMOs of the five complexes had similar orbital distributions and were mainly distributed on the stilbene group of ligand HL, while the LUMOs showed slight differences. Complexes 2, 3, and 4 showed similar LUMO distributions, with LUMOs mainly distributed on the two tta ligands, the stilbene group of ligand HL, and partially on the central lanthanide ion. Meanwhile, the LUMO of complex 1 was mainly distributed on the two tta ligands, partially distributed on the central lanthanide ion, and slightly distributed on the stilbene group of ligand HL. The LUMO of complex 5 was mainly distributed on the two tta ligands and partially distributed on the central Yb3+ ion, with no electron density on the stilbene group of ligand HL. The LUMOs +1 and +2 of complexes 1–4 all had similar electron density distributions on the two tta ligands, stilbene group of HL, and was partially on the central metal. Furthermore, the HOMOs −1 and −2 of complexes 1–4 all pull electron density away from the HL group toward the two tta ligands and partially toward the central metal. Although the LUMOs showed slight differences among the five complexes, the HOMO and LUMO energy levels were close to each other, with only complex 5 showing a small gap.

Figure 8

HOMO and LUMO energy levels and orbital distributions of complexes 1–5. Most relevant MOs of complexes 1–5 associated with vertical excitation are shown in Table .

Based on the optimized geometric structure, electron absorption spectra calculated by TD–DFT were compared with those obtained experimentally. The simulated electronic absorption spectra are shown in Figure S13 (Supporting Information). The strongest absorption peaks of the five complexes in the acetonitrile solution obtained experimentally were all observed at around 335 nm, while the absorption peaks calculated by the polarized continuum model (PCM) simulation in the acetonitrile solvent were all around 345 nm. The simulated electronic absorption spectra were close to the experimental data. The solvent-corrected main transition energies (eV), with contributing excitation (%), oscillator strengths (f, only showing f > 0.1), associated wavelengths (λ), and dominant excitation character, are shown in Table . Complexes 1–4 showed similar mixed transitional characteristics. TD–DFT/PCM calculations showed that the absorption band in the range of 230–400 nm measured experimentally mainly originated from the IL (πHL → πHL* and πtta → πtta*), LLCT (πHL → πtta* or πtta → πHL*), and a small charge-transfer transition from ligands to the central Ln3+ ion (LMCT). As shown in Table , the photoisomerization rate constants and quantum yields of complexes 1–4 evidently increased with increasing atomic number from La to Gd. Among the four complexes, complex 3 was rather different from our previously reported [Eu(tfd)2(HL) (CF3COO)] complex,[35] in which photoisomerization of the stilbene group was suppressed owing to the presence of LLCT πHL → πtfd* and the absence of backward LLCT πtfd → πHL*. This resulted in complex [Eu(tfd)2(HL) (CF3COO)] having a higher sensitization efficiency of 97.9% in the acetonitrile solution and 96.7% in the solid state, making the isomerization reaction difficult to observe and the quantum yield too low to be detected. This study shows that complex 3 has good dual photoisomerization and photoluminescence functions owing to its lower sensitization efficiency.

Complex 4 showed the largest contribution from the LLCT (tta to HL) transition, which favored the energy transfer from tta to HL and promoted photoisomerization. Furthermore, energy transfer from the ligands to the gadolinium ion was inhibited, resulting in complex 4 showing the largest photoisomerization quantum yield. The transition characteristic of complex 5 was slightly different from those of complexes 1–4 but similar to that of the tfd-coordinated complex [Yb(tfd)2(HL)]. The LLCT transition between ligands mainly originated from the hydrazide group in HL to the tta ligand, while the LLCT from tta to HL (πtta → πHL*) was absent. The LMCT transition mainly originated from the tta ligand to the central Yb3+ ion. Therefore, the luminescence and trans-to-cis photoisomerization properties of complex 5 were similar to those of complex [Yb(tfd)2(HL)].

Conclusions

Five new di-β-diketonate lanthanide complexes co-coordinated with a stilbene derivative were synthesized and characterized. The crystals of the five complexes belonged to the monoclinic P21/c space group. For complexes 1–4, the central lanthanide ion was nine-coordinate with a monocapped twisted square antiprism polyhedron geometry. The central Yb(III) ion of complex 5 was eight-coordinate with a distorted double-capped triangular prism polyhedron geometry. The photoinduced trans-to-cis isomerization properties of the five complexes in acetonitrile solutions showed that their photoisomerization rate constants (kiso) and quantum yields (Φt–c) and those of the free ligand HL were in the order of 4 > 3 > 5 > 2 > 1 > HL. Complex 4 showed the best photoisomerization properties accompanied by magnetic properties. Complex 3 showed reversible luminescence change upon the trans-to-cis photoisomerization process. Such complexes are promising substitutes for pure ligands in molecular switch applications. The magnitude of the trans-to-cis isomerization yield and rate constant of these complexes and the sensitization efficiency of ligand energy transfer to the lanthanide ion were attributed to the extent of the ligand-to-ligand charge-transfer transition from HL to tta (πHL → πtta*) and ligand-to-metal energy transfer.

Experimental Section

Materials and Methods

LnCl3·6H2O (Ln = La, Nd, Eu, Gd, and Yb) was purchased from Beijing Huawei Ruike Co. Htta and 2-(chloromethyl)pyridine hydrochloride were obtained from Alfa Aesar, and all other chemicals were purchased from Sinopharm Chemicals Group Co. The melting point was determined using an X-4 micromelting point apparatus (Shanghai Precision Scientific Instrument Co. Ltd.) without correction. Electrospray ionization (ESI)–mass spectrometry (MS) was performed using a Bruker ESQUIRE-3000 Plus liquid chromatography–MS/MS spectrometer. Elemental analysis was performed using a Vario EL III elemental analyzer. TGA was conducted using an SDT Q600 thermogravimetric analyzer under a nitrogen atmosphere with a heating rate of 10 °C/min. Solution 1H-NMR spectra were acquired using a Bruker ADVANCE 400 spectrometer at 400 MHz and 25 °C with tetramethylsilane as the internal reference. IR spectra (KBr pellets) were obtained using a Nicolet AVATAR FT-IR 330 spectrometer. Absorption spectra were recorded using a Shimadzu UV224012PC absorption spectrometer. Luminescence spectra were obtained using a Hitachi F7000 fluorescence spectrophotometer at room temperature with excitation and emission slits set at 2.5 and 5.0 nm, respectively. The overall quantum yields of the compounds were determined in the organic solvent at room temperature using the emission quantum yield (Φ = 0.028) of Ru(bpy)2Cl2 using an air-saturated aqueous solution as the standard.[70] Solid sample luminescence quantum yields and lifetimes were determined using an Edinburgh instruments FLS980 fluorescence spectrometer equipped with a Hamamatsu R5509 NIR photomultiplier tube, and excitation and emission slits were set to 2.0 and 5.0 nm, respectively. Magnetic measurements were conducted at a sweep rate of 1.5 K min–1 on a Quantum Design SQUID XL7 magnetometer over a temperature range of 2–300 K. When measuring photoisomerization quantum yields, UV light (254 and 312 nm) was acquired from a Shanghai Yihui ZF-2 UV analyzer. This type of analyzer can emit radiation at wavelengths of 254, 312, and 365 nm. The light intensity was measured using a UV-A radiation meter (Optical Electrical Apparatus Factory, Beijing Normal University). Photoirradiation experiments were performed in acetonitrile solutions using a UV analyzer. A quartz cell with a light path length of 10 mm was used for photoisomerization measurements. The concentration of all complexes was 2.0 × 10–5 mol L–1. The photoisomerization quantum yield (φiso) and trans-to-cis first-order kinetics isomerization rate constant (kiso) were calculated using previously reported methods.[34] The quantum yields and rate constants were averages of three measurements and data errors were within 10%.

Syntheses of the Co-Coordinated Stilbene Ligand (HL) and Lanthanide(III) Complexes 1–5 (Scheme )

HL was prepared in exactly the same way as the previous synthesized method.[35] The lanthanide complexes were synthesized as follows: 0.42 g of ligand HL (1.0 mmol) and 0.30 g of Htta (2.5 mmol) were mixed and dissolved in 20 mL of tetrahydrofuran. After the mixed compounds were completely dissolved, 1 M sodium hydroxide solution was added dropwise slowly to the solution to adjust the pH to about 7. It should be pointed out that some of the Htta ligand is hydrolyzed to produce the CF3COO– anion in the system. After that, the solution was heated to 65 °C, and 1.0 mmol LnCl3·6 H2O in 5 mL of methanol solution was added and allowed to stand for refluxing for 10 h. Then, the solution was filtered while hot, and the resulting solution was rotated to obtain the solids. The solid powder was washed and dried to obtain pure lanthanide complexes. The pure solid powder was dissolved in methanol and then filtered, and the solution was allowed to stand in the dark to volatilize. Crystals were obtained after 1 week.

[La(tta)2(HL) (CF3COO)] (1)

Light yellow crystal, yield: 70% (0.76 g); ESI(+)-MS: m/z (CH3OH) = 1001.0 [M+]; 1H NMR (400 MHz, DMSO-d6): δ (ppm) 9.68 (s, 1 H), 8.49 (s, 2 H), 7.84 (dd, J = 7.4, 8 H),7.62 (m, 5H), 7.38–7.16 (m, 10 H), 6.28 (s, 2 H), 4.30 (s, 4 H); FT-IR (cm–1): 3214, 3062 (C=C–H, m), 1680 (C=O, s), 1610 (C=C, s), 1538 (N–H, m), 1344, 1285(C–N, s), 1178, 1127 (C–F, s), 943 (C–S–C, s) 783, 711, 641 (Ph–H, w); elemental analysis calculated for LaC45H32F9N4O7S2 (%): N, 5.03; C, 48.48; H, 2.89; S, 5.75; found (%): N, 5.01; C, 48.39; H, 3.06; S, 5.70.

[Nd(tta)2(HL) (CF3COO)] (2)

Light lavender crystal, yield:72% (0.80 g); ESI(+)-MS: m/z (CH3OH) = 1006.0 [M+]; FT-IR (cm–1): 3209 (C=C–H, m), 1680 (C=O, s), 1614 (C=C, s), 1539 (N–H, m), 1414, 1311 (C–N, s), 1178, 1135 (C–F, s), 933 (C–S–C, s), 757, 683, 641 (Ph–H, w): elemental analysis calculated for NdC45H32F9N4O7S2 (%): N, 5.01; C, 48.25; H, 2.88; S, 5.73; found (%): N, 5.07; C, 47.52; H, 2.93; S, 5.72.

[Eu(tta)2(HL) (CF3COO)] (3)

Yellow crystal, yield: 65% (0.78 g); ESI(+)-MS: m/z (CH3OH) = 1015.0 [M+]; 1HNMR (400 MHz, DMSO-d6): δ (ppm) 9.66 (s, 1 H), 8.47 (s, 2 H), 7.75–7.62 (m, 9H), 7.37–7.25 (m, 7H), 6.47–6.37 (s, 4H), 4.52–4.29 (m, 6 H); FT-IR(cm–1): 3177 (C=C–H, m), 1689 (C=O, s), 1621 (C=C, s), 1538 (N–H, m), 1412, 1303 (C–N, s), 1185, 1137 (C–F, s), 933 (C–S–C), 785, 706, 641 (Ph–H, w); elemental analysis calculated for EuC45H32F9N4O7S2 (%): N, 4.97; C, 47.92; H, 2.86; S, 5.69; found (%): N, 4.95; C, 47.33; H, 2.77; S, 5.63.

[Gd(tta)2(HL) (CF3COO)] (4)

Transparent crystal, yield: 75% (0.83 g); ESI(+)-MS: m/z (CH3OH) = 1020.0 [M+]; FT-IR (cm–1): 3208, 3025 (C=C–H, m), 1688(C=O, s), 1613 (C=C, s), 1538(N–H, m), 1412, 1297 (C–N, s), 1178, 1138 (C–F, s), 934(C–S–C, s) 783, 711, 641 (Ph–H, w); elemental analysis calculated for GdC45H32F9N4O7S2 (%): N, 4.94; C, 47.70; H, 2.85; S, 5.66; found (%): N, 4.95; C, 47.50; H, 2.62; S, 5.58

[Yb(tta)2(L)] (5)

Transparent crystal, yield: 55% (0.56 g); ESI(+)-MS: m/z (CH3OH) = 1036.0 [M+]; FT-IR (cm–1): 3032, 2915 (C=C–H, m), 1686 (C=O, s), 1631 (C=C, s), 1545 (N–H, m), 1412, 1304 (C–N, S), 1187, 1143 (C–F, s), 933(C–S–C, s), 766, 700, 643 (Ph–H, w); elemental analysis calculated for YbC43H31F6N4O5S2 (%): N, 5.41; C, 49.90; H, 3.02; S, 6.20; found (%): N, 5.42; C, 49.37; H, 2.93; S, 6.26.

Structural Determination of Complexes 1–5 by X-ray Diffraction

Structural determination of complexes 1–5 was performed using X-ray diffraction. Single-crystal X-ray diffraction data of the five complexes were collected using a Rigaku Oxford Diffraction XtaLAB Synergy diffractometer with micro-focus sealed X-ray Cu Kα radiation (λ = 1.54184 Å) at 100 K. Crystallography data reduction and absorption correction were performed using CrysAlisPro 171.41_64.110a software (Rigaku Oxford Diffraction, 2015). The structures were solved by intrinsic phasing (SHEXT) and refined by full-matrix least-squares calculations based on F2 using the SHELXTL-2014 software package.[71,72] All non-hydrogen atoms were refined anisotropically. The disordered solvent in the structures was removed using the PLATON SQUEEZE program.[73] Crystallographic details for complexes 1–5 are summarized in Table .

Theoretical calculations

Ground-state structures of the five complexes were optimized in the acetonitrile solvent using DFT. The solvent was simulated using the PCM. In our previous report on HL-related lanthanide complexes, the Cam-B3lYP functional combined with default MWD28 specified for the SDD [Stuttgart Dresden triple zeta ECPs] basis set was found to correlate well with the experimental absorption spectra.[35] In the present study, the SDD basis set was again selected, and both the B3lYP and CAM-B3LYP functionals were used to model the electronic structures of complexes. However, the B3LYP functional was found to yield results more consistent with the observed properties of the complexes. Therefore, the B3LYP functional was used, the 6–31(g) basis set was selected for elements C, H, O, N, S, and F, and relativistic ECP and segmented basis sets were used for lanthanide ions. Optimized structures were verified by frequency calculations to check that the energy minimum was reached. The optimized ground-state structure was then used to calculate the singlet and triplet excitation energies of the five complexes, and their electronic spectra were calculated using TD–DFT. The G16 package was used for all calculations. A free GaussSum program was selected to analyze the molecular orbitals.[74]