Highly efficient light harvesting of a Eu(iii) complex in a host–guest film by triplet sensitization

Trivalent lanthanide complexes are attractive light emitters owing to their ideal high color purity. Sensitization using ligands with high absorption efficiency is a powerful approach to enhancing photoluminescence intensity. However, the development of antenna ligands that can be used for sensitization is limited due to difficulties in controlling the coordination structures of lanthanides. When compared to conventional luminescent Eu(iii) complexes, a system composed of triazine-based host molecules and Eu(hfa)3(TPPO)2 (hfa: hexafluoroacetylacetonato and TPPO: triphenylphosphine oxide) significantly increased total photoluminescence intensity. Energy transfer from the host molecules to the Eu(iii) ion occurs via triplet states over several molecules, according to time-resolved spectroscopic studies, with nearly 100% efficiency. Our discovery paves the way for efficient light harvesting of Eu(iii) complexes with simple fabrication using a solution process.

Highly effi cient light harvesting of a Eu(III) complex in a host-guest fi lm by triplet sensitization We have lighted up a trivalent europium (Eu(III)) complex intensively in a host-guest thin fi lm, which exhibits a very narrow-band red emission. Luminescent Eu(III) complexes are generally made using organic ligands that absorb light strongly, but it is diffi cult to design such ideal ligands. We have fabricated a simple thin fi lm composed of a host molecule that absorbs light strongly and a guest Eu(III) complex, and achieved a photoluminescence intensity 400 times higher than that of the Eu(III) complex. We have also revealed the detailed light harvesting mechanism using timeresolved spectroscopies.

rsc.li/chemical-science Introduction
Molecule-based light-emission technologies have been intensively developed over the last few decades owing to their various applications, such as display panels, bioimaging sensors, optical telecommunications, and laser diodes. [1][2][3] Light-emitting materials must emit narrow-band light to achieve high color purity in their applications. However, general organic molecular emitters exhibit broadband emissions with a full width at halfmaximum (FWHM) of 70-100 nm. Trivalent lanthanide (Ln(III)) complexes exhibit narrow-band emissions with FWHMs of 10-20 nm caused by transitions between f-orbitals in Ln(III) that are shielded by electrons in the occupied 5s and 5p orbitals; 4 however, direct photoexcitation of the Laporte-forbidden f-f transition in Ln(III) with a small absorption coefficient (3 < 10 M −1 cm −1 ) is difficult. 5 Many efforts have been made to overcome this difficulty by synthesizing organic ligands with large absorption coefficients and appropriate energy levels, which play an important role in efficient photosensitizers and intra-molecular ligand-to-Ln(III) energy transfer in complexes. [6][7][8] The overall photoluminescence (PL) intensity (I PL ) 9,10 of an Ln(III) complex in a dilute solution or thin lm is expressed as where 3 ligands and f tot represent the sum of all ligand absorption coefficients and the overall luminescence quantum yield of Ln(III) from ligand photoexcitation via intra-molecular energy transfer, respectively. This equation indicates that high PL intensity requires high light-absorption ability and luminescence quantum yield. Eu(III)(hfa) 3 (DPPTO) 2 (hfa: hexa-uoroacetylacetonato and DPPTO: 2-diphenyl phosphoryl triphenylene Fig. S1 †) is one of the Ln(III) complexes with the highest luminescence intensity in solution, I PL = 90 000 M −1 cm −1 . The following three factors contribute to this high I PL : (i) high 3 ligands = 170 000 M −1 cm −1 of the two DPPTO ligands owing to their triphenylene chromophores, 11 (ii) high f tot = 0.53 owing to suppression of nonradiative decay due to low vibrational frequencies of the phosphine-oxide linker in the DPPTO ligand and CF bonds in the hfa ligand, [12][13][14][15][16] and (iii) enhancement of transition intensities in Eu(III) due to the asymmetric structure 17-21 formed by the hfa and DPPTO ligands. However, further improvement of luminescence intensity by designing new ligands is limited due to the difficulty of synthesizing ligands that simultaneously contain multiple chromophores with a large absorption coefficient and stable coordination to Ln(III) compared to general transition metal ions.
We propose that a host-guest system, composed of pconjugated molecules and Ln(III) complex emitters, is an ideal system for drastically increasing the PL intensity of Ln(III) complexes while requiring minimal fabrication. Multiple pconjugated molecules with high absorption coefficients serve as antennae for photosensitizing Ln(III) complexes in this system; therefore, a much higher absorption coefficient, denoted by 3 hosts in this case, is expected in eqn (1) when compared to the molecular Ln(III) complex. Moreover, no linkers were required to coordinate Ln(III). However, to achieve high f tot , such a hostguest system must overcome additional challenges; not only intra-molecular energy transfer but also inter-molecular energy transfer processes from host molecules to the Ln(III) complex are involved, and each process must be highly efficient. To achieve a very high I PL for the host-gust system, it is essential to understand the mechanisms of the entire energy transfer process and design lossless energy transfer processes. To address these concerns, we chose a simple Eu(III) complex, Eu(III)(hfa) 3 (TPPO) 2 (TPPO: triphenylphosphine oxide), 22,23 in which intra-molecular energy transfer occurs from the hfa ligands. We discovered that a host-guest system composed of a 2,4,6-tris(biphenyl-3-yl)-1,3,5-triazine (mT2T) host and the Eu(III) complex achieves an I PL that is three orders of magnitude greater than the I PL of the Eu(III) complex itself and that ∼40 host molecules work for light harvesting of one Eu(III) complex via lossless triplet-triplet inter-molecular energy transfer.
In general, host-guest systems of Ln(III) complexes have been fabricated as emitting layers in organic light-emitting diodes (OLEDs). [24][25][26] Host molecules are known to affect the emission properties of Ln(III) complexes. Pietraszkiewicz et al. fabricated a 5 wt% Eu(nta) 3 SFXPO-complex-doped host-guest lm (nta: 1-(2-naphthoyl)-3,3,3-triuoroacetonate and SFXPO: spiro-uorene-xanthene diphosphine oxide) with a higher PL quantum yield (PLQY) of 0.86 than 0.64 in a neat lm. 27 This shows that host-guest systems may improve the efficiency of energy transfer processes when compared to molecular Ln(III) complexes. Buczko 4 ]complexes, Ln = Gd(III) or Lu(III), and found 347 times larger emission intensity compared to the neat lm due to intermolecular energy transfer from ligands of (N(C 2 H 5 ) 4 )[Ln(hfa) 4 ] and diminishment of concentration quenching. 28 Nonetheless, there are no design strategies apart from matching the energy of their lowest excited states, [29][30][31][32][33] and there are few direct observations of inter-and intra-molecular energy transfer processes in host and guest systems. [34][35][36][37][38] We fabricated 10 wt% Eu(hfa) 3 (TPPO) 2 -doped lms with the ve host molecules demonstrated below and a polymethyl methacrylate (PMMA) polymer with no photosensitization ability. We measured their I PL and discovered that the mT2T host molecule exhibited the highest I PL = 3 600 000 M −1 cm −1 , which was approximately four hundred times greater than that of the PMMA host (Fig. 1). To investigate the source of the relatively high I PL , we used time-resolved PL spectroscopy (TR-PL) and femtosecond transient absorption spectroscopy (fs-TAS) in a multiscale temporal range from sub-picoseconds to hundreds of microseconds to investigate the emission mechanisms of the host-guest lm. Beginning with the initial excitation of the host molecules and ending with the emission of the Eu(III) complex, we elucidated the mechanisms of all processes in the lm: (1) the intersystem crossing (ISC) in the host molecule, (2) the inter-molecular energy transfer process from the host molecules to the ligands of the guest Eu(III) complex, (3) the intra-molecular energy transfer process from the ligands to the Eu(III) ion, and (4) the emission processes of f-f transitions in the Eu(III) ion. Furthermore, we discovered that the yields of all energy transfer processes, (1)-(3), were nearly unity and that the yield of the Eu(III) ion emission process (4) determined the overall quantum yield of the lm. This highly efficient PL is attributed to ideal triplet sensitization processes: rapid and efficient ISC in mT2T results in efficient triplet-triplet intermolecular energy transfers with no losses.
To investigate the sensitization ability of the host molecules, we compared the PL properties of the host-guest lms to those of the Eu-PMMA lm (Fig. 3A). Aer photoexciting the Eu-PMMA lm with 315 nm light, emission bands from the Eu(III) ion were observed at 581, 594, 615, 654, and 701 nm (Fig. S4 †) and are assigned to the transitions 5 D 0 / 7 F J and J = 0, 1, 2, 3, and 4, respectively. 4 The Eu(III) ion sensitization by the hfa ligands in the Eu-PMMA lm was conrmed because excitation at 315 nm selectively excites the hfa ligands ( Fig. S3A †). Also, aer photoexcitation with 260-267 nm light, the same emission bands in the Eu(III) ion were observed in all of the host-guest lms ( Fig. 3B and S5 †). Given that the absorption coefficients of the host molecules in this wavelength range are much larger than those of the guest complex ( Fig. S3 †), this indicates that inter-molecular energy transfer from the host molecules to the Eu(III) complex occurs in all host-guest lms.
The sensitization efficiencies of the host-guest lms are discussed qualitatively based on their emission spectra. A broad emission band located at around 400 nm was observed for Eu(hfa) 3 (TPPO) 2 -doped SF3TRZ (Eu-SF3TRZ), mCBP (Eu-mCBP), CBP (Eu-CBP), and T2T (Eu-T2T) lms, in addition to the emission from the Eu(III) ion ( Fig. S5 †). These bands were assigned to the uorescence from the lowest singlet excited state (S 1 ) of each host molecule ( Fig. S6 †), indicating imperfect energy transfer to the Eu(III) complex. In contrast, no emission band of the host molecule was observed in the Eu(hfa) 3 (TPPO) 2doped mT2T (Eu-mT2T) lm (Fig. 3B). This nding suggests that inter-molecular energy transfer from the host mT2T molecules to the Eu(III) complex occurs extremely efficiently in the lm. In fact, f tot of Eu-mT2T (f tot = 0.84) was signicantly higher than that of Eu-PMMA (f tot = 0.60) ( Table 1). The excitation spectra probed at the Eu(III) ion emission, which coincided with the absorption spectra of these lms, further supported efficient sensitization ( Fig. 3B and C).  (1) and (2). The PLQYs (f tot ) were measured at each photoexcitation wavelength (l ex ) (3 ligands , 3 mol host , n host , I PL ; see the text for details)

Guest
Host Eu conc./wt%  We measured the photophysical properties of Eu-mT2T lms with different doping ratios of the Eu(III) complex to mT2T to estimate the number of host molecules that contributed to one Eu(III) ion emission (Table 1). Considering this result, I PL of host-guest lms can be expressed as where 3 mol host and n host represent the molar absorption coefficient of the host molecule and the number of host molecules that contributed to one Eu(III) ion emission which is calculated from the doped molar ratio between the host molecule and the Eu(III) complex, respectively. The f tot was the highest when the mixing ratio was 10%, implying that ∼22 host molecules contributed to the emission of one Eu(III) complex via energy transfer between the host and guest molecules. The I PL is plotted as a function of n host in Fig. 4. The I PL for the Eu(hfa) 3 (TPPO) 2 -doped lms increases monotonically even when n host = 47, indicating that more than 47 host molecules work as photosensitizers for one Eu(III) complex. Note that the maximum I PL is more than two orders of magnitude greater than that of the conventional highly luminescent complex Eu(hfa) 3 (DPPTO) 2 . 11 Initial process aer photoexcitation: rapid and efficient ISC in mT2T Understanding the mechanisms of intra-and inter-molecular energy transfer processes requires understanding the photophysical processes occurring in host molecules. To investigate the processes in the time domain, we measured and compared the fs-TAS spectra of the Eu-mT2T lm (Fig. 5A) and the mT2T neat lm (Fig. S7A †). The absorbance change (DAbs.) of the Eu-mT2T lm increased in the <600 nm range, whereas it decreased in the >600 nm range, with an isosbestic point at 600 nm. We note that the isosbestic point does not appear as an intersection at one point due to a slight spectral shi over time. This is presumably due to the dynamical effect uniquely seen in a solid state, for example, dielectric relaxation. We performed a global analysis of the fs-TAS spectra, assuming a sequential model with two components because the isosbestic point indicates an exclusive transition between two states. 41 With a time constant of 71.2 ± 0.6 ps, the rst component was converted to the second component (Fig. 5B). The evolution associated spectra (EAS) and concentration kinetics of the two components are shown in Fig. 5C and D, respectively. We analyzed the fs-TAS of the mT2T neat lm in the same way to assign the observed species. The global analysis also resulted in similar EASs (Fig. S7C †), with a time constant of 44.9 ± 0.4 ps (Fig. S7D †). The two EAS components were reasonably assigned to S 1 and the lowest triplet excited state (T 1 ) in the case of the neat lm, and the time constant represented the ISC rate. Therefore, the two components observed in the Eu-mT2T lm were assigned to S 1 and T 1 of the host mT2T molecule. These time constants are much faster than those of the ISC process for aromatic organic molecules in general, which can be explained by the presence of lone pairs in the triazine moieties in mT2T  accelerating the ISC rate. 42,43 We conclude that mT2T undergoes a rapid and nearly unity ISC before exciton diffusion and energy transfer in the host-guest lm.

Quantum yields of each energy transfer process
Given that rapid ISC occurs rst in the host lm, Fig. 6 shows the predicted energy transfer processes in the Eu-mT2T lm aer photoexcitation. We rst estimated the luminescence quantum yield of the Eu(III) ion (f Eu ) in the host-guest lms to estimate the quantum yield of each process. The natural radiative rate constant of the Eu(III) ion (k Eu r ) can be calculated from the ratio of the total Eu(III) ion transition to the magnetic dipole transition ( 5 D 0 / 7 F 1 ) in the observed emission spectrum of the Eu(III) ion (Fig. 3B, eqn (S1); see the ESI text for details †). 44 f Eu is also determined using the ratio of k Eu r to the observed decay rate constant of the 5 D 0 state (k Eu obs ) (eqn (S2) and (S3) †). The estimated f Eu , rate constants, and parameters are summarized in Table S1. † We calculated h sens because f tot represents the product of f Eu and overall photosensitization efficiency (h sens ; eqn (S4); Table S1 †). It is worth noting that h sens in the 10 wt% Eu-mT2T lm is nearly unity, with h Eu-mT2T sens = 0.97. This indicates approximately perfect sensitization of photoexcited mT2T to the Eu(III) complex in the Eu-mT2T lm, which is more efficient than sensitization of hfa ligands in the Eu-PMMA lm, h Eu-PMMA sens = 0.71. To identify the factor that improves energy transfer efficiency in the Eu-mT2T lm, we compared the sensitization efficiencies of intra-molecular energy transfer in the Eu-PMMA lm (h Eu-PMMA sens ) and the inter-and intra-molecular energy transfer in the Eu-mT2T lm (h Eu-mT2T sens ). h Eu-PMMA sens is expressed as follows: where f ligand ISC and k ET represent the ISC yield of the ligand and the efficiency of intra-molecular energy transfer, respectively. In contrast, h Eu-mT2T sens is expressed as where f host ISC and z ET represent the yield of ISC of the host and the efficiency of inter-molecular energy transfer, respectively. Fig. 6B shows the symbols for the Eu-mT2T lm. We conducted time-resolved measurements in Eu(hfa) 3 (TPPO) 2 -doped neat (Eu-neat) and Gd(hfa) 3 (TPPO) 2 -doped neat (Gd-neat) lms to estimate the k ET of the Eu(III) complex ( Fig. S8 and S9 †). We can observe intrinsic emission from the hfa ligands in the Gd-neat lm because there is no intra-molecular energy transfer to Gd(III) due to the energy level mismatch (Fig. S9A-C †). We compared the time constants of the fs-TAS and TR-PL measurements to analyze each time constant in the energy transfer process (Fig. S10 †). The k ET of the Eu(III) complex was >0.99 when the lifetimes of the T 1 ligands between the Eu-neat lm and the Gd-neat lm were compared (Table S2, eqn (S5) †). We conclude that the quantum yield of the ISC at the ligands, f ligand ISC , in the Eu-PMMA lm is the dominant factor in the relatively low h Eu-PMMA sens . Sensitization processes in the case of the Eu-mT2T lm, in contrast, occur via inter-molecular energy transfer via T 1 states between the mT2T and hfa ligands. Because the quantum yield of ISC in mT2T is nearly unity, the Eu-T2T lm achieves more efficient sensitization.

Time-domain view of the whole sensitization processes
Multiscale TR-PL measurements were performed to quantify energy transfer processes in the time domain. The pseudo-2D color plot of the TR-PL of the Eu-mT2T lm aer photoexcitation at 267 nm is shown in Fig. 7A-C. Following photoexcitation, a broad emission band in the 350-500 nm range was observed ( Fig. 7A and D). This band shape is approximately identical to that of the uorescence observed in the mT2T neat lm (Fig. S11 †); this emission is attributed to the S 1 emission in mT2T. Note that no emission from mT2T was observed in the steady-state emission spectrum (Fig. 3B) because its timeintegrated intensity was much smaller than that of the Eu(III) ion. The decay time constant of mT2T emission was estimated to be <100 ps (Fig. 7G), which agrees with the time constant of ISC in mT2T (∼70 ps) estimated from fs-TAS spectra.
Following the rapid decay of uorescence from mT2T, narrow-band emissions in the nanosecond to microsecond time range were observed (Fig. 7B, C, E and F). According to Dicke's diagram, all of these emission bands were assigned to the f-f transitions in the Eu(III) ion. 45 The 535, 555, and 585 nm bands in the nanosecond region ( Fig. 7A and E) were assigned to the transitions 5 D 1 / 7 F 1 , 5 D 1 / 7 F 2 , and 5 D 1 / 7 F 3 , respectively. The bands at 590 and 615 nm in the microsecond region ( Fig. 7B and F) were assigned to 5 D 0 / 7 F 1 and 5 D 0 / 7 F 2 , respectively. The rise time constant of the 5 D 1 state was estimated to be 41.0 ± 0.8 ns (Fig. 7H), which is much longer than the time constant of the ISC in mT2T (<100 ps). The direct energy transfer from the T 1 state in mT2T to the excited states in the Eu(III) ion is ruled out because this energy transfer occurs by the Dexter mechanism with electron exchange and the distance between mT2T and the Eu(III) ion in the lm is too long for such an energy transfer to occur. Therefore, this result indicates that the energy transfer to the Eu(III) complex is ∼40 ns and is mediated via the T 1 states. The rise time is determined by multiple processes: (1) the ISC of mT2T (<100 ps), (2) the triplet exciton diffusion in the host matrix, (3) the triplet-triplet energy transfer between mT2T and the hfa ligand, and (4) the intramolecular energy transfer from the hfa ligand to the Eu(III) ion.

Mechanisms of efficient triplet-state mediated energy transfer and sensitization
Here, we discuss the origins of efficient inter-molecular energy transfer in the Eu-mT2T lm in terms of T 1 energy matching. We compared the T 1 energies of the mT2T molecule to those of the hfa ligands of the Eu(III) complex. The T 1 energies for mT2T and hfa estimated from the phosphorescence spectra (Fig. S12 †) were found to be 2.66 and 2.70 eV, respectively. Due to the close proximity of the energies, an efficient energy transfer from mT2T to the Eu(III) complex is anticipated. We also compared the T 1 energies of the other host molecules to those of hfa, conrming that energetically resonant conditions in the T 1 energies of the host and hfa are important for the higher PLQYs in the host-guest lms (Fig. S12G †). This indicates that intermolecular energy transfer from host molecules to the Eu(III) ion occurs via the T 1 of hfa. Furthermore, the phosphorescence of hfa ligands was observed following energy transfer from T 1 in mT2T ( Fig. S12A and S13 †) in the Gd(hfa) 3 (TPPO) 2 -doped mT2T (Gd-mT2T) lm, consistent with efficient host-to-hfa energy transfer based on triplet-triplet energy transfer. We concluded that the energy resonance in T 1 between mT2T and hfa causes highly efficient inter-molecular energy transfer sensitization (z ET ∼ 1.0).
It is worth mentioning that our triplet-based emission enhancement strategy offers several advantages over the co-uorescence effect, in which a Gd or Lu complex works as a sensitizer, as described in ref. 28. Firstly, we achieved a signicantly greater enhancement in emission compared to the previous work. Secondly, we utilized organic compounds as host molecules, resulting in a substantial improvement in absorption ability. Thirdly, the host-guest lms are fabricated through a simple solution process.
There are two possible assignments to the slow rise time. One is the energy transfer from the T 1 of mT2T to the T 1 of the ligand and the other is the triplet exciton diffusion in the host matrix. Finally, we discuss triplet exciton diffusion in the host matrix by comparing energy transfer to the Eu(III) complex and PLQYs in Eu-mT2T lms with different mT2T and Eu(III) complex ratios (Fig. S14, Table S3 †). As the concentration of the Eu(III) complex increased, the rise time constant of 5 D 1 got shorter (Fig. S14 †). The rise time constant of the low ratio of the mT2T (Eu(III) complex concentration of 50 wt%) lm exhibits a fastest rise time constant (∼29 ns). Because the lifetime of T 1 in mT2T is likely to be much longer than the diffusion time scale, sensitization processes mediated by triplet-triplet energy transfer are effective in realizing ideal sensitization for Eu(III) ion emission (Fig. S15 †).

Conclusions
We demonstrated highly efficient light harvesting of the Eu(III) ion in complex-doped host-guest lms with Eu(hfa) 3 (TPPO) 2 . When we used mT2T, a triazine derivative that works well as an energy-harvesting antenna, we observed a signicant increase in the luminescence intensity of the Eu(III) ion. From the photoexcitation of host molecules to the emission of Eu(III) ion, we estimated the quantum yields of all energy transfer processes and discovered that all energy transfer processes occur with nearly unity quantum yield. We conclude from the TR-PL and fs-TAS measurements that efficient energy transfer occurs via resonant energy transfer from T 1 of mT2T to T 1 of hfa following rapid and highly efficient ISC in mT2T. This mechanism can avoid energy loss in the ISC process in the hfa ligands of an Eu(III) complex and overcome the intrinsic limitations of conventional direct sensitization by the ligands. Based on these results, we propose a novel light harvesting method for Ln(III) with simple fabrication: a host-guest lm composed of host molecules with efficient ISC, which works as an efficient photosensitizer, and a guest Ln(III) complex with ligands having a T 1 state whose energy matches that of T 1 in the host, which works as an efficient energy accepter and emitter.

Fabrication of thin lms
Thin lms for optical measurements were fabricated by spincoating on quartz substrates. The quartz substrates were washed by ultrasonic cleaning with acetone and isopropanol. For the preparation of neat lms, the emitter compounds were dissolved in chloroform (10 wt%). To prepare the host-guest lms, a weight ratio of 1 : 9 of the guest molecule and the host molecule was dissolved in chloroform to obtain an overall concentration of 10 wt%. Before use, the solution was ltered through a 0.2 mm lter, and the quartz substrates were heated to 80°C. The solution was spin-coated onto quartz substrates for 60 s at 1000 rpm and then annealed at 70°C for 10 min. Table  S4 † shows the thicknesses of the lms. Thin lms for refractive index measurements were fabricated on silicon substrates using vacuum vapor deposition at a pressure of less than 10 −3 Pa. We used a xed deposition rate of 0.5 nm s −1 and a thickness of 100 nm.

General methods
UV-vis absorption spectra were measured using a PerkinElmer LAMBDA 950 spectrophotometer. Excitation and PL spectra of the Eu(hfa) 3 (TPPO) 2 -doped PMMA lm (excitation wavelength for PL spectra: l ex = 315 nm and probe wavelength for excitation spectra: l em = 615 nm), Eu(hfa) 3  The photoluminescence quantum yields were measured using a Hamamatsu Photonics Quantaurus-QY instrument equipped with an integrating sphere. A time-correlated singlephoton counting lifetime spectroscopy system (HAMAMATSU Quantaurus-Tau C11367-21, C11567-02, and M12977-01) was used to measure PL lifetimes.
The refractive indexes and thicknesses of the lms were measured using variable-angle spectroscopic ellipsometry (M-2000U, J. A. Woollam Co., Inc., United States).

Time-resolved photoluminescence (TR-PL)
TR-PL measurements were performed using a streak camera system (Hamamatsu C4780, time resolution < 30 ps) synchronized with a Ti:sapphire regenerative amplier (Spectra-Physics, Spitre Ace, pulse duration = 120 fs, repetition rate = 1 kHz, pulse energy = 4 mJ per pulse, central wavelength = 800 nm). 34 The samples were excited by the third harmonic (267 nm) of the fundamental pulse from the amplier. Before measuring, all lms were encapsulated. The excitation energy was kept to less than 0.8 mJ cm −2 .

Femtosecond transient absorption spectroscopy (fs-TAS)
Transient absorption (TA) measurements were conducted using the pump-probe method. 46 The light source was a Ti:sapphire regenerative amplier system (Spectra-Physics, Spitre Ace, pulse duration = 120 fs, repetition rate = 1 kHz, pulse energy = 4 mJ per pulse, central wavelength = 800 nm) seeded using a Ti:sapphire femtosecond mode-locked laser (Spectra-Physics, Tsunami). The output of the amplier was divided into two pulses for the pump and probe. The samples were pumped by the third harmonic of the fundamental pulse from an amplier (267 nm). The broadband probe pulse (450-750 nm) was generated using a sapphire crystal of 1 mm thickness. The pump and probe pulse beam sizes at the sample position were <0.7 mmf and <0.5 mmf, respectively. A PC-controlled mechanical delay state was used to adjust the delay time between the pump and probe pulses. The probe pulse that passed through the sample lms was dispersed using a polychromator (JASCO, CT-10, 300 grooves/500 nm), and the spectra were captured using a multichannel detection system with a CMOS sensor (UNISOKU, USP-PSMM-NP). To avoid damage, all the lms were encapsulated before being measured and were mechanically moved continuously. The excitation energy was kept to less than 0.4 mJ cm −2 . The recorded data were analyzed using a Python-based homemade program. Note that the Eu-mT2T lm shows much stronger emission compared to general Eu(III) complexes, so that no TA data were available at the positions of the Eu(III) ion emission.

Author contributions
SM prepared the lms. YK synthesized the complexes. KG and SM measured the optical properties of the sample. SM conducted the time-resolved spectroscopic measurements. KG and SM analyzed all the data. KO, KM, and SM draed the original manuscript. KO and KM supervised the study. All authors contributed to the review and editing of the manuscript and critically commented on the project.

Conflicts of interest
There are no conicts to declare.