Li
Sun
a,
Masoumeh
Keshavarz
a,
Giacomo
Romolini
a,
Bjorn
Dieu
a,
Johan
Hofkens
ac,
Flip
de Jong
a,
Eduard
Fron
*ac,
Maarten B. J.
Roeffaers
*b and
Mark
Van der Auweraer
*ac
aChem&Tech – Molecular Imaging and Photonics, KU Leuven, Celestijnenlaan 200F, 3001 Leuven, Belgium. E-mail: eduard.fron@kuleuven.be
bChem&Tech – Centre for Membrane Separations, Adsorption, Catalysis and Spectroscopy for Sustainable Solutions, KU Leuven, Celestijnenlaan 200F, 3001 Leuven, Belgium. E-mail: maarten.roeffaers@kuleuven.be
cKU Leuven Core Facility for Advanced Spectroscopy, Celestijnenlaan 200F, 3001 Leuven, Belgium. E-mail: mark.vanderauweraer@kuleuven.be
First published on 24th August 2022
Zeolite confined silver clusters (AgCLs) have attracted extensive attention due to their remarkable luminescent properties, but the elucidation of the underlying photophysical processes and especially the excited-state dynamics remains a challenge. Herein, we investigate the bright photoluminescence of AgCLs confined in Linde Type A zeolites (LTA) by systematically varying the temperature (298–77 K) and co-cation composition (Li/Na) and examining their respective influence on the steady-state and time-resolved photoluminescence. The observed polychromatic emission of the tetrahedral Ag4(H2O)n2+ clusters ranges from orange to violet and three distinct emitting species are identified, corresponding to three long-lived triplet states populated consecutively and separated by a small energy barrier. These long-lived species are at the origin of the polychromatic luminescence with high photoluminescence quantum yields. Furthermore, the Li-content dependence of decay times points to the importance of guest–host–guest interactions in tuning the luminescent properties with a 43% decrease of the dominating decay time by increasing Li content. Based on our findings, a simplified model for the photophysical kinetics is proposed that identifies the excited-state processes. The results outlined here pave the way for a rational design of confined metal clusters in various frames and inspire the specified applications of Ag-zeolites.
Recently, our group reported the structural and luminescence variations of the hydrated silver-Linde type A (Ag-LTA) system via tuning the exchange ratio of lithium and sodium ions (Li+ and Na+). Increased guest–host–guest interactions are proposed to determine the optical changes in the Li+ incorporated systems. Namely, the guest Li+ incorporation in the LTA host results in a reduced lattice parameter and a more rigid zeolite framework which in turn leads to a smaller interatomic distance between the silver atoms in the excited state guest AgCLs (Agc–Agc).33
However, due to the complex excited-state dynamics of the system, the fundamental photophysical processes and the origin of the bright luminescence of Ag-LTA composites could not yet be unequivocally determined. At this time, several theoretical models have been put forward to explain the complex photophysical kinetics.40,41 For instance, Grandjean et al. proposed a model in which the luminescent properties of Ag3K9-LTA originate from a confined two-electron superatom quantum system with hybridized Ag and water oxygen orbitals delocalized over the cluster.42 Upon photo-excitation, one electron of the s-type highest occupied molecular orbital (HOMO) is excited to a p-type lowest unoccupied molecular orbital (LUMO) and finally relaxes to the long-lived lowest-lying 3P triplet state to which the major emission is attributed. In this model, the properties of the luminescence depend on the interaction between Ag42+ and coordinated water molecules and/or lattice oxygens in the ground and excited states. Quantum mechanical calculations have also explored the impact of other factors such as different ligands and an asymmetric environment surrounding the core framework, altering the electronic properties.43 Investigating the sodium exchanged Ag3Na9-LTA, Fron et al. reported a fast decaying (∼3 ns) luminescence at room temperature which is incompatible with emission from a triplet state.39
In this work, we systematically study the photoluminescence (PL) of the partly Ag-exchanged AgLiNa-LTA zeolites by considering the impact of environmental temperature and co-cation composition. Samples are characterized by high PLQYs (up to 40% at room temperature) and show polychromatic emission upon adjusting the Li/Na ratio and the environmental temperature. The steady-state temperature-dependent PL reveals the existence of three distinct emitting species which are dominant in different temperature ranges. The kinetics underpinning the interplay of these three emissions are investigated by temperature-dependent time-resolved luminescence covering a time range from picoseconds (ps) to milliseconds (ms). The obtained results reveal the photophysical properties, excited-state dynamics and the origin of the bright luminescence observed in LTA confined AgCLs (for which the structure Ag4(H2O)n2+ has been suggested earlier based on structural analysis and density functional theory calculations33,42,44). In the next step, the long-lived decay times assigned to the main emitting species are tuned by varying the Li/Na composition, highlighting the importance of the co-cation. Armed with these results we feel confident introducing a clear, simplified model for the excited-state kinetics of this class of AgCL in LTA. Furthermore, we are confident that the described tunability of the PL at different temperatures and with different co-cations offers new tools for the rational design of such systems for various applications.
Fig. 1 (a) Two-dimensional excitation/emission profiles (at half of the maximum intensity) of hydrated Ag1-exchanged LixNa12−x-LTA (x = 0–12) zeolites at 298 K and (b) 77 K. (c) The diffuse reflectance spectra of hydrated Ag1-exchanged LixNa12−x-LTA (x = 1, 4, 6, 8, 11) zeolites at 298 K and 77 K; samples with different Li composition are simply denoted as Lix (x = 1, 4, 6, 8, 11), as depicted in the legend. (d) Temperature-dependent photoluminescence spectra of Ag1Li6Na5-LTA with temperatures ranging from 298 K to 83 K under 265 nm excitation; the peaks that appear sequentially as the temperature is decreased are called peak 1, peak 2, and peak 3. (e) The CIE (1931) color coordinates of Ag1Li6Na5-LTA emission varying from 298 K to 83 K under 265 nm excitation. (f) Contributions of the three peaks to the entire emission system at different temperatures of Ag1Li6Na5LTA. The procedure for the deconvolution is given in the ESI after Fig. S9.† (g) Photographs of Ag1Li6Na5-LTA powder at different temperatures under illumination by a UV lamp (255 nm). |
To explore the origin of this thermochromic behavior, temperature-dependent PL measurements were performed to monitor the evolution of the emission spectrum between 298 K and 83 K. Fig. 1d shows that the emission of Ag1Li6Na5-LTA changes in three distinctive steps upon decreasing the temperature: (1) the peak around 595 nm (peak 1: orange emission) gradually disappears while a peak around 495 nm (peak 2: cyan emission) gains intensity from 298 K to 233 K; (2) cyan emission dominates around 233–153 K and at lower temperatures a new peak around 395 nm (peak 3: violet emission) appears; (3) violet emission gradually enhances with decreasing temperature (153–83 K) while the cyan emission behaves oppositely and the orange emission disappears completely. Notably, this optical change is reversible by gradually heating the sample from 83 K to 298 K (Fig. S7†). These changes result in the color evolution seen in photographs under UV illumination (Fig. 1g) and the Commission International de I'Eclairage (CIE) color coordinates (Fig. 1e) which continuously vary from (0.50, 0.47) at 298 K, characteristic of orange, to (0.19, 0.39) at 213 K (cyan) and finally (0.18, 0.13) at 83 K (violet). These remarkable observations demonstrate the presence of three distinct emitting species with maxima at 595, 495 and 395 nm, and suggest the consecutive population of three distinct excited species. This also confirms that the blue shifted emission bands observed by lowering the temperature from 298 K to 77 K are not related to a hindered vibrational relaxation on the potential energy surface of the excited state corresponding to e.g. gradual changes of the Ag–Ag or Ag–O distances or associated bond angles. Such a process would take place if the excitation would populate higher vibrational levels in the same potential energy surface well from which the emission occurs at the minimum. Here, the cyan and orange emission bands (495 and 595 nm) are due to the presence of additional and different minima in the excited state potential energy surface possibly corresponding to different electronic states. The interconversion between the three emitting species is thermally activated, allowing the tuning of their contributions to the steady-state PL spectrum by changing temperature (Fig. 1f). As increasing the temperature only leads to a minor decrease of the total integrated intensity of the three peaks, the temperature dependence of the integrated intensity of each peak largely reflects the competition between the emission and conversion of the three different states, rather than the competition between radiative and non-radiative decays of each separate state.
Upon investigating the temperature-dependent PL of other Ag1-exchanged zeolites with different Li content (Lix, x = 1, 4, 8, 11), similar PL properties were observed, showing that this thermally activated luminescence is an inherent characteristic of the Ag-zeolite series under study. This means that such a phenomenon is mainly caused by intrinsic properties of the AgCLs inside zeolite cages rather than by a specific Li/Na ratio (Fig. S9†). We have previously observed that the presence of water molecules in the cage plays an important role in the spectroscopy of the AgCLs.39 As expected, for the samples investigated here, both excitation and emission spectra hinge on hydration level. For data and a detailed discussion please refer to ESI (Fig. S10†).
Although a detailed study of exchanging Li+ to other counterions than Na+ is beyond the scope of this work, the hydrated Ag1Li6K5-LTA sample was prepared in a similar way as Ag1Li6Na5-LTA and its emission and excitation spectra were compared to those of Ag1Li6Na5-LTA at RT and 77 K, respectively (Fig. S11†). Consistently, the data indicates the presence of several excited state species related to different minima of the exited state potential surface (see ESI† for more discussion).
To elucidate the mechanism of the interconversion between different emitting states, four temperatures i.e., 298 K, 253 K, 153 K and 77 K, were selected for ns–ms time-resolved luminescence for Ag1Li6Na5-LTA. The temperature-dependent time-resolved luminescence experiments were performed in a 200 μs and 2 ms time window while the excitation was set to 265 nm (4.68 eV). The decays obtained for solid materials such as this system often require a multi-exponential fit, which often is not unique and we have no underlying physical model. Hence, it is not always possible to unambiguously assign every recovered decay time to a specific species. In this context, we will focus our attention on the main emitting species and try to discuss other significant components based on their AWDs. From the decay traces detected in the range of 400–760 nm at 298 K (Fig. S13†), four decay components were resolved as 172 ns, 685 ns, 15.4 μs and 203.6 μs, and the correspondent AWDs are displayed in Fig. 3a. The component with the shortest decay time of 172 ns has a very low (<1%) contribution to the stationary emission and decay associated spectra (see DAS in Fig. S17a†), and can be attributed to an impurity or to clusters with a slightly different structure and low abundance. However, the amplitudes of the other three slower decaying components (685 ns, 15.4 μs and 203.6 μs) contribute mainly around 495 nm, in the range 495–560 nm, and around 595 nm respectively. Notably, the amplitude of the 685 ns component becomes negative at wavelengths above 600 nm, indicating that the species emits at 595 nm is populated by the excited state emitting at 495 nm (Fig. S18†).47 The inverse of this time constant (685 ns) corresponds to the sum of the decay rate constant of this species to the ground state and the rate constant for the interconversion. From the DAS in Fig. S17† it becomes very clear that the 203.6 μs component presents the dominant contribution (about 90%, 10% being attributed to the 15.4 μs one) to the steady-state PL which is corroborated by the close resemblance of its AWD and DAS to the RT stationary emission spectrum of Ag1Li6Na5-LTA. Such a long-lived species may suggest a triplet state decaying through spin-forbidden radiative and non-radiative processes. The component with a decay time of 15.4 μs, however, is more puzzling. Its AWD is overlapping with both the 495 nm and 595 nm species. Since the steady-state spectroscopy clearly showed three emitting states, of which the 595 nm and the 495 nm are of interest at this temperature, it is unlikely this is due to yet another species. It might however indicate some distribution of decay times for each emitting species owing to the slightly different local environments leading to different radiative and non-radiative rates. This distribution of decay times will be discussed further in more detail.
When the sample temperature is decreased to 253 K, the AWD shows similar features as at 298 K (Fig. 3b and S14 and S17b†), while the decay times are all longer than at 298 K. Using similar arguments as for 298 K, we can exclude the 398 ns component (<1% contribution) from the analysis and attribute the components with a 3.5 μs and 262.8 μs decay time to the 495 nm and the 595 nm species respectively. Interestingly, we see that the AWD shape of the 151.5 μs decay changed slightly compared to that of the 15.4 μs decay observed at 298 K. The decreased amplitude of the 151.5 μs component around 495 nm is accompanied by the appearance of a fifth component with a decay time of 21 μs whose AWD is overlapping with the 495 nm emission. This further strengthens our assumption that this additional component is due to an underlying distribution of decay times for the individual emission bands. This also accounts for the tenfold increase in decay time going from 15.4 μs to 151.5 μs, which is due to the splitting of the former component into a shorter 21 μs component and a longer 151.5 μs component.
Decreasing the temperature further to 153 K, the emission of the 595 nm species disappears which is in accordance with its disappearance in the steady-state emission spectra. We retrieved three decay components with maxima at 395 nm (595 ns, 4.4 μs and 19.8 μs) and one component with a maximum around 495 nm (108.8 μs) (Fig. 3c and S15†). Similar to the negative decay amplitudes observed at 298 K and 253 K, here the decay amplitudes of the 4.4 μs component also become negative around 495 nm. This is a clear indication that the species emitting at 395 nm is the precursor of the species emitting at 495 nm.
At 77 K finally, three decay components were retrieved with amplitudes in the blue part of the spectrum (Fig. 3d and S16†). The main component with the maximum at 415 nm has a decay time of 24.1 μs, and its AWD resembles the stationary emission spectrum recorded at the same temperature. This component contributes about 97% to the stationary emission intensity (Fig. S17†). The 2.2 μs component, also centered at 415 nm, has a much smaller amplitude and contributes about 2% to the stationary emission intensity. As discussed before this component can be attributed to AgCLs with a slightly different environment. The fastest decaying component (11 ns) is close to the time resolution of the setup (10 ns) and might correspond to the ns species observed by a dedicated nanosecond setup (TCSPC) as described earlier (Fig. 2b).
The analysis of the decays suggests that the bright main-emission of Ag1Li6Na5-LTA should be attributed to triplet species. Based on the results obtained from the temperature-dependent PL and time-resolved luminescence, one can conclude that the components with the long (tens or hundreds of μs) decay times of the three main emitting species are at the origin of the bright emission observed in Ag1Li6Na5-LTA with high PLQYs. Considering that even at RT the PLQY is still close to 0.4, the outspoken decrease in decay times recorded for the species responsible for the cyan and violet emission must be due to thermal activation of the rate of interconversion between these three species rather than to an increase of the non-radiative decay rate constant. This also means that at lower temperatures where we can observe peak 3 and peak 2 or peak 2 and peak 1 simultaneously, the decay of the corresponding excited species is mainly determined by the rate of interconversion between those three species.
At 77 K, a small fraction of the S3 state population directly decays to the S0 state via weak fluorescence in a few ns (1.1 ns and 3.32 ns), while a large fraction undergoes intersystem crossing (ISC3) to the T3 state emitting at 415 nm (2.99 eV). Presumably, the two nanosecond decay components are in fact a representation of a distribution of decay times induced by a dispersion in environmental factors affecting the cluster radiative properties. In this respect one should note that also for the processes occurring on μs time scale such as emission of from the triplet state or interconversion between the triplet states always a distribution of rate constants was observed. Based on the theoretical calculations43 and in agreement with the experimental results, the AgCLs under study possess asymmetric structures as a result of an asymmetric cluster core, an asymmetric cage, or an asymmetric placement of the core within the cage facilitating the ISC processes. This T3 state then decays to the S0 state with a decay time of 24.1 μs via violet phosphorescence. Notice that no emission from T2 and T1 is observed at 77 K, these states at 77 K are neither populated by relaxation from T3 or T2 nor by ISC from S2 and S1. Hence, this also means that S2 and S1 are not populated by direct excitation suggesting that the equilibrium geometry of those states differs strongly from that of the S0 state. Here, one should note that the excitation spectra at 77 K of the (weak) emission in the red tail of the emission spectrum at 495 and 595 nm do not show appreciable intensity above 375 nm (Fig. S8†). This indicates that the species responsible for orange and cyan emission cannot be populated by direct excitation, they are only accessible by the doorway species responsible for the violet emission.
Upon increasing the sample temperature to 153 K, it becomes possible for the major fraction of the excited AgCLs to overcome the barrier separating T3 from T2 (495 nm, 2.50 eV) leading to the formation of T2 with a time constant of 4.4 μs as indicated by the negative amplitude of this component at the wavelength where T2 emits. The T2 state formed at 153 K decays again by radiative and non-radiative processes to S0 yielding a bright cyan phosphorescence with a decay time of 108.8 μs. The barrier for the interconversion from T3 to T2, which must be overcome by thermal activation, can be related to the geometry change of AgCLs with T3 → T2 (ICT2) (Fig. 5). This geometric relaxation corresponds to a major change in the tetrahedral geometry of the Ag42+ cluster or a displacement of the Ag42+ cluster where e.g., the bond lengths to some surrounding oxygens are lengthened while those to other oxygens are shortened. Preliminary time-dependent density functional theory (TD-DFT) calculations suggest a flattening of the tetrahedral geometry upon intersystem crossing which in turn affects the bond lengths between the Ag and oxygen atoms. In this case, we would expect that this relaxation and the corresponding barrier are similar for S3 → S2 (ICS2) but with lower efficiency (detailed discussion is in the ESI†). This suggests that the geometric relaxation prefers to take the highly efficient ICT pathway. However, one cannot exclude that the barrier could also be overcome by a mechanism related to the reverse intersystem crossing (RISC) where efficient ISC3 to T3 is followed by the sequence T3 → S3 → S2 → T2 (RISC3 → ICS2 → ISC2) (Fig. S49†). Since the energy gap between the S3 and T3 states (ΔEST3 = 0.07 eV) is expected to be sufficiently small,48 this may make the RISC3 feasible when induced by thermal activation (the same case for RISC1 and RISC2 at high temperatures). Further discussion on this topic can be found in the ESI.†
For the other components of the T3 emission (595 ns and 19.8 μs), there is no direct evidence in the AWD plots that they also undergo a conversion to T2 besides deactivation to S0. However, at higher temperatures the species linked to those components must also be converted mainly to T2 (and in the next step to T1) as the total PLQY of T1 amounts to ∼0.4 at 298 K. This suggests that there is probably a distribution of the rate constants for decay of T3 to S0 and conversion of T3 to T2. As ICT2 has to compete with phosphorescence from T3 (kph3) and non-radiative decay of T3 to S0 (knrT3), the inverse decay time which amounts to 1/(4.4 μs) = 2.3 × 105 s−1 equals kICT2 + kph3 + knrT3. Hence 2.3 × 105 s−1 is an upper limit for kICT2 of the major component at 153 K. This also suggests a distribution in the environment of the AgCLs which could also explain the large width of the emission bands of the different emitting species. Such a distribution also means that the analysis of the luminescence decays as a triple or quadruple exponential decay and the recovered decay times are to some extent arbitrary with a continuous distribution of decay times. The distribution in rate constants kICT2 and the corresponding activation barriers is also indicated by the observation that it is not possible to fit the temperature dependence of the integrated intensity of peak 3 (Fig. 1f) to a simple combination of a temperature independent components, related to the phosphorescence and intersystem crossing to S0, and a component related to kICT2 showing an Arrhenius type temperature dependence. This does not impede that the recovered decay times indicate roughly the time range in which the different photophysical processes occur. Anyway, the conversion from T3 to T2 is much slower than what would normally be expected for IC occurring with a rate constant exceeding 1010 s−1. Otherwise, fluorescence from S3, ISC3 or phosphorescence from T3 would not be able to compete effectively with the interconversion.
At elevated temperatures of 253 K and 298 K, the T2 population undergoes a similar process being transferred into T1 state which relaxes to S0 state via bright orange phosphorescence with a decay time of 262.8 μs and 203.6 μs at 253 K and 298 K respectively. Similarly, the non-radiative interconversion process can either take the path of T2 → T1 (ICT1) facilitated via the geometric relaxation (Fig. S48†), or the path of low efficient ICS1 after the thermal barrier at RISC2 and followed by the ISC1 to arrive at T1 state (Fig. S49†).
If the AgCL reported here has a similar geometric structure corresponding to Ag4(H2O)n2+, considering the model proposed by Grandjean et al., it is tempting to assign the three excited states to a transition from the s-type orbital to the three different p-type orbitals. Although this process would in principle have a similar transition dipole for S0 → S1, S0 → S2 and S0 → S3, only the last transition is observed. This is most probably due to the different geometries of the relaxed excited singlet states of the three singlet species, only that of S3 having an equilibrium geometry close to that of S0 state. Any transitions from S0 to S1 or S2 will be strongly blue-shifted compared to the corresponding energy differences between the equilibrium geometries of these states. Furthermore, they will be strongly broadened and depressed in peak intensity. In this case, the discrepancy between the results obtained and those obtained by Fron et al. for Ag3Na9-LTA should be attributed to the formation of expanded AgCLs in Ag3Na9-LTA, where a larger average number of Ag+-ions per cage is present.39
Footnote |
† Electronic supplementary information (ESI) available: Experimental details, some auxiliary characterization results, and additional discussion. See https://doi.org/10.1039/d2sc03197c |
This journal is © The Royal Society of Chemistry 2022 |