Linlin Zenga,
Yitong Wang
b,
Junjun Tanc,
Quanbing Peia,
Jie Kong
a,
Wei Zhang
a,
Shuji Ye
a,
Rongchao Jin*b,
Yi Luo
*ac and
Meng Zhou
*a
aHefei National Research Center for Physical Sciences at the Microscale, University of Science and Technology of China, Hefei, Anhui 230026, P. R. China. E-mail: yiluo@ustc.edu.cn; mzhou88@ustc.edu.cn
bDepartment of Chemistry, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, USA. E-mail: rongchao@andrew.cmu.edu
cHefei National Laboratory, University of Science and Technology of China, Hefei, Anhui 230088, China
First published on 9th September 2025
Surface ligand modification is an effective strategy for enhancing the photoluminescence (PL) of atomically precise gold nanoclusters. Nevertheless, the underlying mechanism of PL enhancement, particularly the role of triplet states remains vague. In this work, we picture the formation and relaxation of triplet states in a series of four Au52(SR)32 nanoclusters with different –R groups by probing their excited state dynamics. Electronic pump–probe spectroscopy reveals that the nanoclusters with fewer methyl groups on the ligand exhibit a faster intersystem crossing (ISC) pathway, hence boosting the triplet-state-related emission. Additionally, the metal core of the four nanoclusters exhibits similar low-frequency vibrations, suggesting that the variation in non-radiative relaxation is primarily mediated by ligand vibrations, rather than the metal core vibrations. These findings establish that ligand-induced PL enhancement is driven by both accelerated ISC and reduced vibrational dissipation. This work sheds light on the mechanism by which ligand engineering enhances PL in gold nanoclusters, highlighting the critical role of triplet state dynamics in tailoring their emission properties, which hold promise in applications such as sensing, bioimaging, optoelectronics, and solar energy conversion.
With deeper insights into origins and mechanisms of PL in gold nanoclusters, triplet excited states may serve as key intermediates. Experiments involving triplet energy transfer (TET) and triplet–triplet annihilation upconversion have demonstrated that metal nanoclusters can efficiently populate triplet excited states and exhibit room-temperature phosphorescence.7,8,19–21 This underscores the need to consider triplet state dynamics when designing strategies to enhance PL. With this perspective, two fundamental questions arise: (1) Is ligand-induced PL enhancement solely a consequence of suppressed non-radiative relaxation through increased rigidity? (2) What role do triplet state formation and relaxation play in ligand-induced PL enhancement?
In this work, we choose a series of four Au52(SR)32 nanoclusters with an identical metal core but different thiolate ligands (i.e., different substituents on the aromatic –SPh ligand) as a platform to explore the mechanism by which ligands regulate PL enhancement. By employing ultrafast transient absorption spectroscopy, we probed the triplet state dynamics and coherent oscillation of the gold core. Our results reveal that the intersystem crossing (ISC) rate plays a crucial role in PL enhancement of Au52(SR)32: nanoclusters exhibiting the strongest NIR emission predominantly populate the T1 state through a faster ISC pathway. Furthermore, modifications to the carbon tails of the ligands have a negligible effect on the core vibrations. This work illustrates the important role of both efficient ISC and suppressed ligand vibration in enhancing the PL of metal nanoclusters, offering new insights into the fundamentals of ligand-regulated luminescence.
To explore the evolution of excited states of Au52(SR)32 nanoclusters, nanosecond transient absorption (ns-TA) experiments were first conducted. The ns-TA data maps show a broad excited-state absorption (ESA) signal spanning 450–700 nm, and a prominent ground-state bleaching (GSB) signal at ∼800 nm, which aligns well with the absorption peak in the steady-state UV-vis spectra (Fig. 2A and S1A–C). Only a long-lived excited-state is observed in ns-TA. By fitting the kinetic trace at 562 nm, the excited-state lifetime was determined (Fig. 2B, S1D–F and Table S1). Au52(p-MBT)32 exhibits the longest excited-state lifetime of 604 ns, while Au52(TBBT)32 has the shortest lifetime of 145 ns. For comparison, the PL lifetimes of Au52(SR)32 in toluene under ambient conditions were also measured by time-correlated single photon counting (TCSPC) (Fig. 2C and S2). Both the TA and PL lifetimes decrease progressively as the number of –CH3 on the carbon tail increases. In addition, the excited-state lifetimes obtained from ns-TA and PL measurements are comparable (Fig. 2D), indicating that the excited-state observed in ns-TA is the origin of the PL emission in Au52(SR)32.
Ligand-protected metal nanoclusters have emerged as novel triplet sensitizers for photon upconversion,24–28 primarily due to their ability to readily populate triplet excited-states facilitated by strong spin–orbit coupling (SOC). In the previous report,22 the singlet oxygen emission at 1274 nm was observed when Au52(SR)32 solutions were purged with O2, which occurs through energy transfer from Au52(SR)32 triplet states to O2 (triplet ground state), resulting in singlet O2 (excited state) and then phosphorescence emission at 1274 nm. Additionally, an abnormal and non-monotonic blue shift of the Au52(SR)32 emission peak was detected as the temperature decreased.22 These findings confirmed that both phosphorescence and thermally activated delayed fluorescence (TADF) coexist in the PL of Au52(SR)32. Here, we further investigate the triplet-state-related emission in Au52(p-MBT)32 through TET experiments. The UV-vis absorption spectra of mixed Au52(p-MBT)32 and rubrene in solutions are shown in Fig. 3A. No new absorption peaks were observed with the addition of rubrene, ruling out the formation of a ground-state complex between clusters and rubrene. As the rubrene concentration increased, the PL lifetime of Au52(SR)32 decreased (Fig. 3B). The data fit well with the linear Stern–Volmer equation, indicating that the PL quenching of the clusters occurs through collisional interactions with rubrene.29 This PL quenching is attributed to TET from Au52(SR)32 to rubrene during collisions, with a determined TET rate constant of 2.1 × 108 M−1 s−1. The successful observation of TET in Au52(p-MBT)32 provides direct evidence for the phosphorescence of Au52(SR)32.
Modifying ligands on Au52(SR)32 significantly influences their triplet excited state emission. Au52(p-MBT)32 exhibits a PLQY of 18.3%, whereas Au52(TBBT)32 shows a much lower PLQY of 3.8%.22 The intersystem crossing from singlet to triplet states plays a critical role in facilitating triplet excited state emission, and may account for the large differences in PLQY among the Au52(SR)32 nanoclusters. To investigate ISC in these clusters, femtosecond TA (fs-TA) spectroscopy was employed to further probe their excited-state dynamics. Upon excitation of Au52(SR)32 at 380 nm, an initial broad ESA signal spanning the entire probe wavelength range was observed, with two peaks at ∼550 and ∼700 nm (Fig. S3 and S4). At a delay time of approximately 1 ns, the ESA signal became narrower and the ESA signal intensity above 650 nm decreased significantly. The change of TA spectral profiles suggests that a state-to-state evolution occurred. Kinetic traces probed at ∼600 nm revealed a rapid decay within the first 1 ps, corresponding to this excited-state evolution (Fig. 4A). In contrast, when pumped at 800 nm, the relatively narrow ESA signal appeared instantaneously, and the broad ESA observed with 380 nm excitation was absent (Fig. S5). Furthermore, kinetic traces probed at ∼600 nm did not exhibit the fast decay (Fig. 4B). Since excitation at 800 nm corresponds to near-bandgap excitation and directly populates the lower excited-state, the fast ESA decay can be attributed to the internal conversion (IC) from a higher excited-state to a lower excited-state.
An evident slow rise in the ESA signal within 100 ps, following the rapid IC, was observed in the excited-state dynamics of Au52(SR)32 nanoclusters. After 100 ps, the ESA signal intensity remained constant throughout the measured time range (Fig. 4A). The long-lived excited-state, which does not decay within the fs-TA measurement window, corresponds to the emissive triplet excited-state identified in ns-TA. Thus, the slow rise in ESA is attributed to ISC. Interestingly, the amplitude of this rise (Fig. 4A) correlates well with the PLQY of the various Au52(SR)32 nanoclusters by an inverse relationship, that is, Au52(p-MBT)32, which exhibits the highest PLQY, shows the least observable rise in its kinetic trace (Fig. 4A), whereas the other three nanoclusters with lower PLQYs display more and more pronounced rise amplitudes. The time constant of the ISC is about 11 ps (Table S2). However, under 800 nm excitation, all four nanoclusters exhibit a slow rise in the ESA signal (Fig. 4B and Table S3). This suggests that the involvement of higher excited-states (Sn, Tn) is pivotal in facilitating the ISC process. A similar phenomenon has been reported in other metal clusters.25,27,30,31 In Au13 nanoclusters, negligible SOC exists between S1 and T1 according to the El-Sayed rule (i.e., a change in orbital type for achieving stronger ISC).32 However, large SOC matrix elements between S1 and higher triplet excited-states (e.g., T2, T3, T4) contribute to the ultrafast and near-unity ISC.25 In isomeric Au28(CHT)20 clusters (where CHT denotes cyclohexanethiolate), the ISC rate from S1 to T2 is several orders of magnitude faster than from S1 to T1, with T2 serving as a critical intermediate for excited-state relaxation.30 For the rise observed in Au52(SR)32 fs-TA dynamics, we ascribe it to a direct ISC process from S1 to T1. With near-bandgap excitation, the nanoclusters are directly excited to S1 and subsequently undergo ISC to T1, resulting in the observed rise in all four nanoclusters (Fig. 4D). In contrast, with 380 nm excitation, the nanoclusters are initially populated in higher singlet excited-states (Sn), which rapidly undergo IC to S1, but for Au52(p-MBT)32, ultrafast ISC from Sn to Tn may occur, which was not captured by our fs-TA (time resolution ∼ 100 fs). For the other three nanoclusters, the ISC process (∼11 ps) primarily occurs between S1 and T1, leading to the observed rise in their TA kinetic traces (Fig. 4C). As the proportion of the picosecond ISC pathway increases, the PLQY of Au52(SR)32 progressively decreases.
When these four Au52(SR)32 nanoclusters were embedded in PMMA films, both their excited-state lifetimes (Figure S7) and PLQYs were significantly increased compared to those in solutions.22 To understand the origin of PL enhancement, we further investigated the excited-state dynamics of the nanoclusters in films. The fs-TA data maps of Au52(SR)32 films exhibited spectral features similar to those observed in toluene solutions, including a broad ESA signal with peak around 550 nm (Fig. 5A and S6A–C). However, the dynamics of Au52(SR)32 films were distinctly different (Fig. 5B) compared to solutions. For instance, Au52(TBBT)32 showed the most pronounced ESA rise in toluene (1–100 ps), but no such distinct slow rise was observed in the film. This absence of rise indicates that the ISC from S1 to T1 occurs much more rapidly in the film than in the solution. Theoretically, the ISC rate can be expressed using first-order perturbation theory:33,34
The emission mechanism of Au52(SR)32 is strongly associated with triplet states. For metal nanoclusters involving TADF, the strong SOC induces ultrafast ISC (10−15 to 10−12 s), leading to a rapid equilibrium between the S1 and T1 states before emission occurs.38 Notably, the ISC dynamics observed in our transient absorption measurements already indicate the establishment of this equilibrium. Consequently, faster ISC not only increases the triplet state population but also accelerates the S1–T1 equilibration, thereby facilitating both phosphorescence and TADF.
The fewer methyl groups present on the ligand, the stronger the restriction on its motion.22 The correlation between ligand flexibility, ISC dynamics and PLQY observed in Au52(SR)32 nanoclusters reveals an important structure–property relationship. However, whether the general trend—that more flexible ligands lead to slower ISC and consequently reduced PLQY—can be extended to other metal nanoclusters requires further studies. Differences in core structure and ligand–metal interactions may result in distinct excited-state behaviors. Therefore, while our findings provide valuable insights, further investigations on a broader range of ligand-protected nanoclusters will be necessary to establish the universality of this mechanism.
A broadband laser pulse can not only populate electronic excited-states but also induce superpositions of vibrational states, resulting in coherent vibrational wavepackets.39–44 These wavepackets manifest as periodic oscillatory modulation in the amplitude of pump–probe signals.45 As shown in Fig. 6A and S8, the early-time TA maps of Au52(SR)32 films display clear oscillatory modulations across the entire ESA region. To determine the frequency of the coherent vibrational motion, we performed fast Fourier transformation (FFT) on the pure oscillation signals, which were obtained by subtracting the excited-state population dynamics (Fig. S9). The FFT analysis reveals a dominant frequency of ∼2.4 THz (80 cm−1) for all four Au52(SR)32 nanoclusters (Fig. 6B and S10). Since these oscillations appear in the ESA signals, the ∼2.4 THz wavepacket motion is attributed to vibrations on the potential energy surface of the excited electronic state.
![]() | ||
Fig. 6 (A) Fs-TA data map of Au52(IPBT)32 film within the first 3 ps upon 800 nm excitation. (B) FFT analysis of the kinetic trace at 562 nm. (C) Oscillation dynamics and the corresponding fits. |
Low-frequency coherent vibrations in metal nanoclusters are typically associated with the acoustic vibrations of the metal core.46–52 In spherical Au25 and Ag44 clusters, a similar oscillation frequency (∼2.4 THz) had been observed and assigned to the breathing mode, characterized by symmetric, in-phase radical oscillations of the metal atoms.46,51,53 Accordingly, the 2.4 THz vibration mode observed in Au52(SR)32 nanoclusters herein should be attributed to the breathing mode of the gold core. The similarity in coherent vibration frequency in these four Au52(SR)32 suggests that their metal cores share a similar structure, consistent with the X-ray crystallographic results.22 The damping dynamics of the coherent vibrations were extracted by fitting the residual oscillations with exponentially decaying cosine functions (Fig. 6C). The oscillation period was ∼0.4 ps for all Au52(SR)32 nanoclusters, consistent with the vibrational frequencies (∼2.4 THz) obtained from FFT analysis (Table S4). The damping time constants ranged from 0.19 to 0.69 ps. A previous report on Au10 nanoclusters highlighted that the rigidity of surface ligands can influence the metal core vibrations, hence, reducing the amplitude of acoustic vibrations and enhancing the core structural relaxation time in electronic dynamics.54 This suppression leads to enhanced PL due to reduced non-radiative relaxation caused by core structural changes. In contrast, such a suppression of coherent vibration amplitude or damping time was not observed in the Au52(SR)32 nanoclusters, although the reduction of methyl groups restricted surface ligand motions.22 This finding indicates that the rigidity of surface ligands has a negligible effect on the metal core vibrations of Au52(SR)32. Instead, non-radiative relaxation may be predominantly mediated by the high-frequency vibrational and rotational motions of the surface thiolate ligands.
The data supporting this article have been included as part of the SI. Supplementary information: Materials and methods, SI Text, Fig. S1–S13, Tables S1–S4. See DOI: https://doi.org/10.1039/d5sc03898g.
This journal is © The Royal Society of Chemistry 2025 |