Heterometallic lanthanide complexes with site-specific binding that enable simultaneous visible and NIR-emission

Macrocyclic lanthanide complexes have become widely developed due to their distinctive luminescence characteristics and wide range of applications in biological imaging. However, systems with sufficient brightness and metal selectivity can be difficult to produce on a molecular scale. Presented herein is the stepwise introduction of differing lanthanide ions in a bis-DO3A/DTPA scaffold to afford three trinuclear bimetallic [Ln2Ln’] lanthanide complexes with site-specific, controlled binding [(Yb2Tb), (Eu2Tb), (Yb2Eu)]. The complexes display simultaneous emission from all LnIII centers across the visible (TbIII, EuIII) and near infra-red (YbIII) spectrum when excited via phenyl ligand sensitization at a wide range of temperatures and are consequently of interest for exploiting imaging in the near infra-red II biological window. Analysis of lifetime data over a range of excitation regimes reveals intermetallic communication between TbIII and EuIII centers and further develops the understanding of multimetallic lanthanide complexes.


Introduction
The photophysical properties of lanthanide ions in the common +III oxidation state (Ln III ) has become a wide and complex field of research across multiple disciplines and applications, particularly Ln III luminescence for bioimaging (Bünzli and Piguet, 2005;Bünzli, 2010;Monteiro, 2020). Previous works have exploited "windows" of attenuation in the absorption profiles of biological tissues at near-infrared (NIR) wavelengths, allowing deeper penetration of such wavelengths and therefore enhanced imaging (Hemmer et al., 2016;Fan and Zhang, 2019). The Ln III ions are of particular interest due to their characteristic line-like emission across the visible and NIR range, low autofluorescence and photobleaching and high signal-to-noise ratios in comparison to classic organic fluorophores. A caveat to their use is intrinsically poor extinction coefficient as a result of partially-forbidden f-f transitions, which is mitigated by sensitization strategies via organic chromophores (the antenna effect) or d-and f-block metal complexes (Faulkner and Pope, 2003;Natrajan et al., 2009). Furthermore, X-H (X = O, N, C) oscillators are well documented in their ability to vibrationally quench Ln III excited states due to significant overlap between vibrational overtones of these bonds and emissive Ln III energy levels (Doffek et al., 2012).
This work further develops a previously designed ligand based on a multi-macrocyclic architecture of 1,4,7,10-tetraazacyclododecane-1,4,7-triacetic acid (DO3A) and diethylenetriaminepentaacetic acid (DTPA) (Faulkner and Pope, 2003). Polyamino carboxylate ligands are well known to securely bind lanthanides in a range of environments and have displayed the ability to act as trinuclear heterometallic binding ligands in a site-selective [TbYbTb] arrangement. Herein, we develop the scaffold to include two [LnTbLn] complexes (Ln III = Yb III , Eu III ) and a complementary [YbEuYb] species ( Figure 1) and further investigate the photophysical capabilities of the ligand across various temperature windows and excitation regimes. The metal selection results in clear Ln III -centred emission across the visible (Tb III , Eu III ) and NIR (Yb III ) spectrum, which overlaps with both NIR-I (650-950 nm) and NIR-II (1000-1350 nm) imaging windows and therefore invites application in imaging technologies (Foucault-Collet et al., 2013;Jin et al., 2022). The simultaneous response from both Ln III centres has further application as a dual-modal device; the unique magnetic and photophysical behaviour of these metals facilitates use as both MRI contrast agents and optical probes (Rivas et al., 2013;Xu et al., 2013).
Low temperature, solid-state and deuterated media measurements are prioritized to mitigate non-emissive quenching mechanisms. These experiments were also designed to facilitate observation of energy transfer (ET) between metal centers; the relative excited state energies of the chosen metal combinations have been exploited for ET processes in numerous multimetallic Ln III systems (Bispo-Jr et al., 2018;Abad Galán et al., 2021). This work contributes toward the growing library of bimetallic trinuclear Ln III systems that have been studied to elucidate the characteristics of intermetallic communication in discrete molecular complexes (Aboshyan-Sorgho et al., 2012;Zaïm et al., 2014;Tropiano et al., 2015;Maniaki et al., 2023). Finally, the presence of ET in molecular Ln III systems is also useful for imaging applications such as those employing two-photon processes like upconversion (Aboshyan-Sorgho et al., 2011;Nonat et al., 2019).

Synthesis and characterization
The overall synthesis of target complexes [Yb 2 Tb], [Eu 2 Tb], and [Yb 2 Eu] was adapted and modified from a previously reported literature procedure (Natrajan et al., 2009). Experimental procedures and characterization are detailed on pages S2-S41.
1,4,7,10-tetraazacyclododecane (cyclen) was purchased from CheMatech and used without further purification. All other Frontiers in Chemistry frontiersin.org 02 reagents and solvents were purchased from Sigma-Aldrich, Fluorochem Ltd. or Apollo Scientific Ltd. and used without further purification. Electrospray +/− (ES-MS) spectra were recorded on a Thermo Orbitrap Exactive Plus mass spectrometer. MALDI-TOF spectra were recorded on a Shimadzu Biotech Axima Confidence mass spectrometer. FT-IR spectra were recorded on a Bruker ALPHA I FT-IR spectrometer. Elemental analysis data were recorder using a Thermo Scientific FlashSmart Elemental Analyzer.
NMR spectra were recorded on a Bruker AVIII HD 500 MHz spectrometer (BBFO inverse probe) in deuterated chloroform, deuterium oxide or deuterated methanol and analyzed using MestReNova 14.1.0. Chemical shifts in parts per million (ppm-δ) are reported relative to residual proton resonances and an internal tetramethylsilane reference. Splitting abbreviations: s: singlet, br. s: broad singlet, d: doublet, dd: doublet of doublets, t: triplet, dt: doublet of triplets, m: multiplet. Blank sections of spectra or those containing solvent resonances are omitted in certain spectra for clarity. Due to complex isomerism between the square and twisted square antiprismatic (SAP ↔ TSAP) forms of multimacrocyclic cyclen compounds, 1 H NMR assignments were often achieved via correlation with 2D COSY, HSQC and HMBC data where possible (Miller et al., 2010;Tircso et al., 2011). 1 H NMR data for compounds containing paramagnetic atoms (Yb III , Tb III , Eu III ) were processed using a line broadening/apodization factor of 1.5-5 Hz and baseline corrected using a multipoint baseline correction with a Whittaker, cubic spline or segment algorithm. The chemical shift values of the 1 H resonances in all the lanthanide (III) complexes are reported without assignment due to the complex nature of the paramagnetic NMR assignment of polyaminocarboxylate lanthanide compounds (Sørensen et al., 2017). 13 C data could not be collected for the same paramagnetic compounds.
Energy minimization of crystal structures was carried out using Avogadro 1.2.0. Compound structures were downloaded as mol2 files from the Cambridge Crystallographic Data Centre (CCDC) and Ln . . . C distances measured directly from crystallographic data without any further modification to the structure. The superimposed Yb III -Gd III structure was achieved via manual manipulation of each structure to minimize steric clash and bonding of the two structures. Bond angles and lengths were maintained during any new bond formation. The auto optimization tool was then used to minimize the energy of the new structure (UFF force field, steepest descent algorithm). Intermetallic measurements were then conducted on this new, minimzed structure.

Luminescence spectroscopy
Luminescence spectra were recorded on an Edinburgh Instruments FLS1000 Photoluminescence Spectrometer. Solidstate and low temperature spectra were recorded using a cryostat attachment with the sample deposited on a fused silica slide and were the default method of measurement unless stated otherwise. Solution measurements were recorded using a Hellma quartz glass 3.5 mL cuvette with a 1 cm path length and a sample absorbance of 0.1. UV-VIS spectra were recorded using a Mettler-Toledo UV5Bio spectrometer. Samples were excited using a 450 W Xe lamp with a long-pass filter on the detection arm and emission captured by PMT-900 (visible) and PMT-1700 (NIR) detectors. Any direct comparison of spectra used identical settings; excitation/emission monochromator slit widths and post-collection processing were identical for all. Lifetime measurements were collected using a microsecond flash lamp operating at 40 Hz (Tb III , Eu III data) or 100 Hz (Yb III data). Plotting, fitting and analysis of data was carried out using Origin 2019b. All data were fitted with exponential decay models starting with the fewest terms (mono-, bi-exponential) until sufficiently good fit residuals were achieved. In particular, 2component fits were always compared against tri-exponential alternatives and found to better fit the data via residual, visual, Chi 2 and R 2 analysis.
Shorter-lived lifetimes from Yb III emission require consideration of the instrument response function (IRF) of the excitation source, both of which are on a microsecond timescale. An IRF trace was recorded at 100 Hz in line with the procedure detailed by the instrument manufacturer (Edinburgh Instruments) by matching excitation and detection wavelengths (λ = 280 nm) and recording the decay. This was repeated across different temperatures, both with the sample present and separately using milk powder to provide scatter (in both solid and solution-state). The variation in IRF trace between different variables is minimal. Detector response of the (NIR) PMT-1700 was also considered by recording a pseudo-IRF detecting scattered 2λ light from visible excitation (λ ex = 600 nm, λ em = 1200 nm), to provide a similar result. The convolution tool in Origin was used to generate a decay trace with the appropriate IRF taken into consideration, which was then fitted to an exponential decay profile in the same manner as all other lifetime fits. Fitting of Tb III and Eu III signals used the IRF as a benchmark to ensure fitting parameters were only applied after the instrument response had decayed to background and therefore could not contribute to the resulting lifetime.

Photophysical properties of the near infra-red-visible emitting complex [Yb 2 Tb]
Initial excitation via the phenyl linker (λ ex = 280 nm) affords sensitized visible emission from the Tb III metal center 5 D 4 → 7 F J (J = 6, 5, 4, 3) transitions. Additional wavelengths selected to directly probe the 5 D 3 (λ ex = 366 nm) and 5 D 4 (λ ex = 488 nm) energy levels of Tb III also result in emission, however at a lower intensity due to poor extinction coefficients at these wavelengths. Only the primary 7 F 5 transition is visible at λ ex = 488 nm. Solid-state variable temperature spectra at 20, 77, 150, and 298 K highlights an increased intensity and resolution of m j state crystal field splitting at lower temperatures, most notably in the central 545 nm 5 D 4 → 7 F 5 transition ( Simultaneous NIR Yb III emission from the 2 F 5/2 → 2 F 7/2 transition (~980 nm) and the associated 4-fold splitting of the Kramer ground state are present under the same set of excitation regimes across a range of temperatures, with an expected loss in resolution, intensity and broadening of signal at higher temperatures ( Figure 2B, Supplementary Figure S2.2). The relationship between emission intensity and excitation wavelength is not concordant between the Yb III and Tb III centers. The signal intensity of the  Figure S2.5) show a broad ligandcentered band with a hypsochromic shift across the series from [Yb] (λ max = 331 nm) to [Yb 2 Tb] (λ max = 311 nm) and further again for the Tb III center (λ em = 545 nm, λ max = 296 nm). Solution-state excitation spectra for the series (Supplementary Figure S2.6) identify this primarily as a solid state effect and present a narrowing of each profile (λ max = 285 nm) with a similar~15 nm shift for Tb III excitation (λ max = 270 nm). UV-VIS absorption data show minimal change in signal when comparing complexes and resemble solution-state profiles (Supplementary Figure S2.7).
Solid-state lifetime measurements of the primary Tb III emission at 545 nm fit a bi-exponential decay profile across all temperatures and λ ex , determined via analysis of R 2 values and fit residuals (Supplementary Figures S2.8-S2.10). For simplicity, we report the global fluorescence lifetime, τ n , of each decay component which is an average across variable λ ex (280, 366, 488 nm) and temperature (20, 77, 150, and 298 K). This value for the two exponents of Tb III decay are 0.14 ± 0.01 ms and 0.70 ± 0.03 ms for τ 1 and τ 2 respectively (fitting after the IRF). The relative contribution of each component varies marginally across the variable excitation and temperature series, but the long-lived τ 2 component remains the most significant (Supplementary Table S2). Calculation of the average lifetime, τ avg , considers the variation in percentage contribution from each τ n component toward an overall average value (Supplemetary equations S1, S2). Factoring in all data across the excitation and Frontiers in Chemistry frontiersin.org 04 temperature range results in τ avg = 0.55 ± 0.03 m for the Tb III center ( Table 1).

Emission behavior of the dual-visible emitting complex [Eu 2 Tb]
Insertion of Eu III into the DO3A binding pocket results in strong visible emission (480-700 nm) from the same 7 F J Tb III transitions present in [Yb 2 Tb], in addition to the 5 D 0 → 7 F J (J = 0-4) transitions from Eu III . As above, excitation of the phenyl linker at λ ex = 280 nm results in simultaneous sensitized emission from the two metal centers ( Figure 3A). Direct excitation of the Tb III 5 D 3 band (λ ex = 366 nm) again yields emission from both metals proportional to respective excitation profiles. The second direct Tb III excitation ( 5 D 4 , λ ex = 488 nm) employed in [Yb 2 Tb] is relatively weak in this complex, meaning no significant emission was observed. Alternatively, direct excitation of the Eu III 5 L 6 level (λ ex = 395 nm) can be used to selectively produce Eu III -centered emission and minimize Tb III spectral features, namely, the J = 4, 3 transitions ( Figure 3A). Variable temperature emission spectra highlight the consistency of this wavelength-selective visible emission in addition to crystal field splitting at lower temperatures, in particular the Eu III 7 J 1 and 7 J 2 signals ( Figure 3B, Supplementary Figure S2.15). The lack of splitting in the 5 D 0 → 7 F 0 transition is indicative of a singular Eu III environment due to the non-degeneracy of both states, as expected from equivalent [Eu III (DO3A)] sites (Binnemans, 2015). Additionally, the hypersensitive 7 F 2 transition known to be highly dependent on coordination environment remains consistent throughout. Analysis of λ ex against temperature shows an appreciable change in intensity at 701 nm (Eu III 7 F 4 state) when changing from sensitized (λ ex = 280 nm) to direct excitation (λ ex = 395 nm) (Supplementary Figure S2.16). However, the relative intensity of the 7 F 4 signal is strong overall which is in agreement with the square antiprismatic geometry expected of macrocyclic species such as DO3A (Binnemans, 2015).  Frontiers in Chemistry frontiersin.org 05 Fitting of the 545 nm Tb III decay results in a bi-exponential fit analogous to the [Yb 2 Tb] species (τ 1 = 0.11 ± 0.01 ms, τ 2 = 0.56 ± 0.07 ms). The percentage contribution of each lifetime component is more evenly distributed than in the corresponding [Yb 2 Tb] complex, resulting in a shorter τ avg value of 0.33 ± 0.02 ms (Table 2). However, comparison of the longer-lived τ 2 component in both Tb III -containing complexes highlights similarity between the two (Supplementary Table S3).
[Eu 2 Tb] follows the trend with increased ligand-centered absorption and further absence of the other 5 D 4 feature at 362 nm; Eu III series spectra exhibit general solid-state broadening analogous to [Yb 2 Tb]. Solution-state excitation scans in D 2 O are ligand-centered across the series, with a marginal shift between samples (Δλ max = 20 nm) and a minor 5 L 6 feature at 395 nm (Supplementary Figure S2.21) and are in agreement with UV-VIS spectra (Supplementary Figure S2.22).  Figures S2.23, S2.24). Excitation at 366 nm was investigated despite the absence of Tb III to generate results comparable with other data sets. The Eu III 7 F 2 transition displays a minor spectral shift, and the 7 F 4 presents a change in splitting pattern compared to [Eu 2 Tb], both of which are sensitive to coordination environment and suggest Eu III is selectively bound in the DTPA site ( Figure 4A). The Eu III center exhibits a greater response at λ ex = 395 nm compared to [Eu 2 Tb], which is consistent across a range of temperatures ( Figure 4B). Solid-state excitation spectra (λ em = 615 nm) show a large ligandcentered signal in addition to distinct Eu III bands analogous to [Eu] and rationalize this behavior (Supplementary Figure S2.25). Solution-state excitation spectra of [Yb 2 Eu] present 5 D 4 ← 7 F 0 , 7 F 1 Eu III excitation signals which are absent in solution measurements of the other Eu III complexes (Supplementary Figure S2.26).

Photophysical properties of the near infra-red-visible emitting complex [Yb 2 Eu]
The Yb III emission intensity from the 2 F 5/2 → 2 F 7/2 transition shows an increased temperature-wavelength relationship compared to other Yb III complexes but maintains maximum emission at λ = 366 nm (Supplementary Figure S2.27). At T = 298 K both ligand and Eu III excitation (λ ex = 280 nm and 395 nm, respectively) result in similar emission intensity, however the latter begins to dominate at lower temperatures (T ≤ 150 K). Solid and solution-state excitation spectra at 980 nm exhibit the same relationship as [Yb 2 Tb] with ã 40 nm blue shift between the two metal centers and UV-VIS spectra is comparable to other Eu III species (Supplementary Figure S2.28).
Lifetime data for Eu III emission in [Yb 2 Eu] is similar to previous samples with τ 1 = 0.17 ± 0.02 ms, τ 2 = 0.74 ± 0.03 ms across the temperature and excitation range. The individual component contribution reflects the previous DTPA-bound metal (Tb III [Yb 2 Tb]), with a clear dominance of the long-lived τ 2 decay (Supplementary Table S5) which gives rise to a longer τ avg value of 0.63 ± 0.09 ms. Yb III lifetimes remain on the same order of magnitude as previous measurements; τ = 9.8 ± 0.6 µs across variable λ ex and temperature. A summary of temperature-dependent lifetime data for each Ln III across the trinuclear bimetallic series is presented in Table 3 and highlights the general observation of longer lifetimes at lower temperatures.

Solution-state measurements
Solution-state measurements in water and D 2 O result in lifetimes with mono-exponential decay profiles for all metals; the shorter-lived τ 1 component in the bi-exponential Tb III and Eu III solids are not present. Solution measurements are likely probing an average lifetime arising from minor changes in hydration state and exchange between the square and twisted square antiprismatic (SAP ↔ TSAP) isomers of the DO3A macrocycles (Miller et al., 2010;Tircso et al., 2011;Nielsen and Sørensen, 2019). Consequently, solution experiments exhibit significantly longer lifetimes on average, with significant gains in deuterated solvent due to reduced energetic overlap of X-D oscillators with emissive Ln III states (Supplementary Figures S2.29-S2.31) (Doffek et al., 2012). The number of bound solvent molecules can be calculated via the inner sphere hydration parameter, q, using a modified Horrocks equation (Supplementary Equation S3, Supplementary Table S1) (Beeby et al., 1999). This calculation takes into account the degree of vibrational quenching by proximate X-H (X = C, N, O) groups for each metal; O-H oscillators contribute significantly for all three, however the significance of N-H groups is relevant for Eu III only and is negligible in magnitude for Tb III and Yb III systems. Whilst C-H quenching arising from the acetate methylene and DO3A ring groups has a considerable influence on the luminescent lifetimes of near infra-red emitting lanthanides, in the case of Yb III this contribution is small when compared to closely diffusing O-H oscillators and mostly unobserved, due to the long X-H . . . Ln III distances of these species and a 1/r 6 dependence for quenching via energy transfer. This enables the estimation of q by a modified Horrock's equation (Beeby et al., 1999).
Comparison of the luminescence decay in water and D 2 O (Table 4) shows q values change depending on both metal choice and binding environment, with the lowest arising from DO3Abound Yb III in [Yb 2 Tb] (q = 0.21) and [Yb 2 Eu] (q = 0.29). The small size of Yb III due to the lanthanide contraction and availability of hard Lewis basic N-and O-donors in the Frontiers in Chemistry frontiersin.org 07 octadentate DO3A precludes access to nearby solvent molecules (Faulkner and Pope, 2003). In complex [Yb 2 Tb], the neighboring DTPA-bound Tb III is more readily accessible due to a larger ionic radius and reduction in kinetic stability of the DTPA binding pocket, resulting in~1 bound water molecule (Idée et al., 2009;Sørensen and Faulkner, 2018). Similarly, in the complex [Eu 2 Tb], q = 1.0 for the DO3A Eu III center as a result of the slightly larger 9 coordinate ionic radius (Sastri et al., 2003) which competes with the steric bulk of the DO3A macrocycle to allow inclusion of 1 solvent donor. In the case of the Tb III (DTPA) centre in [Eu 2 Tb], an apparent q value of 2.4 is determined. However, in the case of substantial energy transfer from the assumedly 5 D 4 excited state of Tb III to the 5 D 0 state of Eu III occuring, phonon assisted energy transfer processes through the O-H vibrational manifold are in direct competition with those that act to quench the 5 D 4 → 7 F J transitions. Given that both of these quenching pathways will possess different rate constants, it follows that Horrocks equation is no longer appropriate (Beeby et al., 1999). Analysis of lifetime data for the analogous [Yb 2 Eu] complex indicates that any competitive intermolecular energy transfer from the 5 D 0 Eu III excited state to the 7 F 5/2 excited state of Yb III is inconsequential, and q of Eu III bound in the DTPA coordination pocket is calculated as 1.1 as expected.

Discussion
The potential of this molecular scaffold to facilitate energy transfer between two bound Ln III ions was a key factor in both the ligand and experimental design. Measurement of the Gd-(DO3A)-aminophenyl acetamide complex [Gd] (analogous to [Yb] and [Eu]) facilitates characterization of the triplet state T 1 of the phenyl linker. The primary Gd 6 P 7/2 state is too high in energy to be sensitized, therefore emission of [Gd] is entirely ligandcentered. Solid-state measurements at 298 K display broad fluorescence (λ em = 431 nm), while phosphorescence from the triplet state is observed at 20 K (λ em = 476 nm, Supplementary Figure S2.32). Crucially, this is high enough in energy to sensitize Tb III , Eu III and Yb III emissive states (T 1 = 21,008 cm -1 , Tb III 5 D 4 = 20,453 cm -1 , Eu III 5 D 0 = 17,227 cm -1 , Yb III 2 F 7/2 = 2924 cm -1 ) (Carnall et al., 1968b;1968a;Sastri et al., 2003) (Figure 5). Additionally, the relative energies and stoichiometry of potential donor and acceptor states (2:1 acceptor:donor ratio) of the metal centers are conducive toward intermetallic ET pathways in [Yb 2 Tb] (Tb III 5 D 4 → Yb III 2 F 5/2 ) and [Eu 2 Tb] (Tb III 5 D 4 → Eu III 5 D 0 ). The potential [Yb 2 Eu] pathway (Eu III 5 D 0 → Yb III 2 F 5/2 ) requires too great an energy mismatch to be a competing process, even when considering phonon assistance via the vibrational manifold of O-H oscillators, as is possibly the case in the Tb III → Yb III systems.
Existence of energy transfer between bound metals is often strongly correlated with changes in the luminescence lifetime of excited states, where an energetic pathway from the donor excited state to an acceptor results in an overall reduction in lifetime in the former. An observed reduction in lifetime is apparent as temperature increases for each metal site due to increased X-H (X = C, N, O) vibrations and therefore quenching of excited states. Recorded lifetimes for the potential Tb III 5 D 4 donor state are largely consistent across a range of excitation bands. There is however a marked decrease in  Indeed, the 5 D J energy levels of Tb III (J = 4) and Eu III (J = 2, 1, 0) exhibit appreciable energetic overlap. The 395 nm excitation feature in [Eu 2 Tb] populates a Eu III 5 L 6 state which is energetically higher than the Tb III 5 D 4 level (ΔE = 4947 cm -1 ) and can populate the latter via non-radiative decay to 5 D 2 and subsequent ET (Figure 6) (Bispo et al., 2018). Additionally, there is the possibility of phonon-assisted ET facilitated by proximate X-H oscillators. The manifold provided by vibrational overtones of O-H and O-D oscillators can provide an alternate energetic pathway to populate 7 F J states of Tb III from direct Eu III 5 L 6 population. Irrespective of these pathways, the emissive states of both Tb III and Eu III are higher in energy than the 7 F J states of either metal (Supplementary Figure S2.33) and have been reported to communicate in similar systems (Zaïm et al., 2014). The simultaneous reduction in lifetime of both Eu III and Tb III is exclusive to [Eu 2 Tb] and is not present in the parent complexes. The ability of each to act as both donor and acceptor when paired together appears to facilitate a series of ET and possibly back energy transfer (BET) processes between the metal centers.
Despite evidence of solvent interactions and consequent vibrational quenching, solution lifetimes are comparable or often longer than those in the solid state. The additional components in solid-state data likely arise from a combination of SAP ↔ TSAP isomerism and intermolecular quenching between emissive centers in two or more molecules. Solid-state quenching effects appear to play a significant role in the ability of the complex to facilitate energy transfer, as preliminary evidence of Tb IIIsensitized Yb III emission has previously been reported in solution (Faulkner and Pope, 2003). However, intermetallic distance is another contributing factor and is especially pertinent in solution-state measurements (Sørensen et al., 2017). Single crystal X-ray diffraction structures reported for aminophenyl trifluoromethyl Yb III (DO3A) (CCDC ID: EGOWUV) and bis-aminocarboxyphenyl Gd III (DTPA) (CCDC ID: QEZGIM) complexes with analogous phenyl moieties allow approximate calculation of these distances via superimposition and preliminary energy minimization. Crystal structures were used without modification and manipulated into a feasible geometry measurements. Eu III , Tb III , and Yb III energies are taken from literature for comparison (Carnall et al., 1968a;Carnall et al., 1968b;Sastri et al., 2003).
Frontiers in Chemistry frontiersin.org 09 before undergoing optimization in Avogadro to yield an average Yb III -Gd III distance of 11.9 Å (Supplementary Figure S2.34) (Dutta et al., 2006;Pujales-Paradela et al., 2019). Examples of multimetallic architectures that exhibit intermetallic energy transfer often report shorter distances (≤10 Å) (Natrajan et al., 2009;Maniaki et al., 2023). The significant r −6 distance dependence of dipolar ET processes and the innate flexibility of molecular lanthanide systems both act to quench any potential intermetallic communication.

Conclusion
A series of three trinuclear bimetallic lanthanide complexes have been synthesized with selective introduction of metals into specific DO3A and DTPA binding sites. All three species exhibit strong sensitized emission when excited via a phenyl linker and represent a broad spectral range, from visible to NIR depending on the metal combination selected. Extensive photophysical measurements investigating the effects of temperature and excitation wavelength revealed communication between Eu III and Tb III centers when closely bound within the same complex, highlighting the potential for energy transfer between metals. The absence of evidence of intermetallic communication in the Yb III heterometallics emphasizes the impact of solid-state quenching and solution-state molecular motion on multimetallic lanthanide scaffolds that could facilitate energy transfer. Further work on optimizing the distance between metal centers and varying Ln III selection will be investigated in the future, with the aim of observing energy transfer and therefore accessing more intricate photophysics and applications.

Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.