Manipulating the Intersystem Crossing Process in Atomically Precise Gold Nanoclusters Authors: Guiying He, Sihan Chen, Rongchao Jin Published: 2025 ABSTRACT Atomically precise metal nanoclusters (NCs) have attracted wide research interest. In terms of electronic properties, a striking feature of such NCs is triplet excited state generation with remarkably high efficiency. Experimental and theoretical findings indicate that NCs are promising luminescent materials with room-temperature phosphorescence and thermally activated delayed fluorescence. However, the manipulation of triplet formation remains difficult due to the complexity of the electron dynamics in NCs. In this Perspective, we summarize recent advances in fundamental research on this topic. We first illustrate the typical spectral features of the triplet state and analytical methods such as time-resolved photoluminescence (TR-PL), transient absorption (TA), and temperature-dependent PL spectroscopies. We then focus on the recent understating of triplet states in NCs and how to manipulate the triplet states. Finally, we present the remaining challenges and future outlooks. This Perspective aims to contribute to the further design of NCs for efficient ISC processes and applications of the triplet states. With a fundamental understanding of the triplet states in NCs, one may develop star materials for triplet utilization in optoelectronics, photocatalysis, and near-infrared solar energy upconversion. Atomically precise metal nanoclusters (NCs) have attracted significant attention because of their potential applications in optoelectronics, − photocatalysis, − photodynamic therapy via singlet oxygen generation, − solar cells, − and photon upconversion. − Metal NCs (e.g., 1–3 nm in diameter) are typically composed of a metal core with a specific number of metal atoms and a protecting shell of organic ligands or metal–ligand staple-like motifs. , Such NCs bridge the nanoparticles in nanoscience research and small metal complexes in coordination chemistry. , The precise control of size, shape, and structure of NCs at the atomic level renders it possible to manipulate the excitons and photophysical properties with unprecedented atomic precision. A fundamental understanding of the optical properties and excited state dynamics of NCs is of paramount importance for both basic science and practical applications. The discrete electronic energy levels of NCs have been well-known since early crystal structure determinations. − Different from the metallic-state nanoparticles in which the surface plasmon resonance (SPR) is dominant, smaller nanoclusters (e.g., <2.2 nm) ,, are no longer plasmonic and instead exhibit single-electron transitions (i.e., excitons) upon photoexcitation, with manifestations of multiple peaks in the optical absorption spectrum because of the quantum confinement effect. − In 2008, with the determination of the X-ray structure of [Au25(SR)18]− (SR = SC2H4Ph), its electronic and optical properties were successfully explained by structure-based density functional theory (DFT) simulations. While the visible range photoluminescence quantum yield (QY) of Au25 was low in early research (∼0.2% for [Au25(SG)18]−), the lifetimes were recorded to be long (246 ns and 1.2 μs) by the time-resolved PL technique. It should be noted that early studies of the excited states of NCs were focused on the spin-0 states but ignored the spin properties. − It was not until recently that the triplet state was introduced to interpret the excited state dynamics by adopting the molecular photophysical model, as opposed to the quantum dot model, with the triplet-related phenomena such as the phosphorescence, ,− thermally activated delayed fluorescence (TADF), − triplet energy transfer (TET) and triplet–triplet annihilation upconversion (TTA-UC). − ,− Sakai et al. demonstrated the triplet behavior of Au25(PET)18 together with the determination of triplet yield (Φ T = 24%) and molar absorption coefficients (ε T ). The photoluminescence of Au52(TBBT)32 was previously attributed to fluorescence from S1 with a lifetime of 135 ns measured by transient absorption (TA), while recent research observed triplet state lifetime of 145 ns by TA and 100 ns by time-resolved photoluminescence (TR-PL) via intersystem crossing from the S1 state (∼11 ps). , Besides, the spin properties have also been recently included in theoretical simulations of NCs. ,, Here we illustrate an explicit system for spin properties. The Au42(PET)32 NC (see its structure in Figure a) was recently reported to have dual emission (fluorescence and phosphorescence), , which provides a perfect platform to investigate the interplay between singlet and triplet states. As shown in Figure b, the frontier orbitals were calculated based on the optimized geometries of S0, S1 and T1 states, respectively, indicating that the energy gap between HOMO and LUMO is altered by the electronic and spin properties. The calculated absorption and emission spectra are shown in Figure c,d by the time-dependent DFT (TDDFT) method, which are comparable to the experimental results. , To understand the triplet-related energy flow after photoexcitation, recent efforts have been devoted to probing the triplet formation, emission mechanism, and triplet–triplet annihilation upconversion for photon energy conversion. ,,,− ,,, (a) Au42 structure with simplified ligands (−SCH3) based on the X-ray structure of Au42(SCH2Ph)32, (b) its frontier molecular orbitals, (c) comparison of experimental and calculated absorption, and (d) photoluminescence. Panels (a,b), Copyright © 2023 American Chemical Society; panels (c,d), Copyright © 2025 The Authors, published by American Chemical Society. This Perspective focuses on the triplet excited states of NCs, including their formation and manipulation. We first briefly introduce spectroscopic techniques for probing the excited-state dynamics, including time-resolved spectroscopy and temperature-dependent spectroscopic methods. Then, the ISC manipulation is discussed, including the structure and size control, effects of alloying, and the ligand structure as well as the surrounding environment. Finally, we discuss some future perspectives. Spectroscopic Techniques for Probing ISC Upon photoexcitation with energy exceeding that of the HOMO–LUMO gap (E g ), the photoinduced excited state will spontaneously dissipate the absorbed energy and return to the ground state through a series of radiative and nonradiative processes, which occur in a time range of femtoseconds to microseconds. Transient absorption (TA) and time-resolved photoluminescence (TR-PL) spectroscopies are the most effective tools to investigate the excited state dynamics for a variety of materials. A detailed introduction to time-resolved spectroscopy techniques was reviewed recently. Below we briefly discuss their application to NCs. The TA technique, also known as pump–probe, can record the difference of optical absorption between the excited state and ground state as a function of the time delay. Thus, TA is a very useful tool to track the energy flow in NCs because of its broad spectral range (UV to IR) for detection and its time range from femtoseconds to milliseconds. Generally, for the TA spectra, the positive signals represent excited-state absorption (ESA) or photoinduced absorption (PIA), and the negative signals represent ground-state bleach (GSB) and stimulated emission (SE). As shown in Figure a, the TA spectra of Au18(SR)14 in the visible- to near-IR range are dominated by a broad ESA along with sharp GSB and SE peaks. GSB and SE are observed as dips in the positive signals due to overlapping with positive ESA signals. The positions of the GSB/SE peaks agree well with those in the steady-state absorption/emission spectra (top of Figure a) of Au18. (a) Data map of femtosecond and nanosecond TA of Au18(DMBT)14 in toluene probed in the visible to near-IR region. The dips are GSB peaks and SE, which overlap with the broad ESA. Temperature-dependent PL spectra (b) and lifetimes (c) of Au18(DMBT)14 in Me-THF solution. Panels (a–c) are licensed by The Royal Society of Chemistry 2025. Generally, TR-PL is restricted to the recording of PL-related exciton dynamics, either fluorescence (FL) or phosphorescence (PH). Time-correlated single photon counting (TCSPC) is the most widely applied tool to measure PL lifetimes with picosecond time resolution. TCSPC builds up a PL decay curve by analyzing the distribution of single photon arrival time in an excitation period but repeating the excitation many times. The photoluminescence lifetime of NCs ranging from ∼100 ps to μs can be recorded by TCSPC, while FL upconversion based on the pump–probe technique can measure ultrafast PL processes with femtosecond resolution. Considering the complicated origin of PL in NCs, temperature-dependent PL is also indispensible and has been used to reveal the exciton–phonon coupling and PL mechanism; in such measurements, PL spectra (Figure b) and lifetimes (Figure c) can be measured from room temperature down to 80 K (liquid nitrogen as the cryogen) , or 4 K (liquid helium as the cryogen). − In addition to TA and TR-PL, other types of ultrafast time-resolved spectroscopy can also be applied to recording the electronic relaxation of NCs. For example, two-dimensional electronic spectroscopy (2DES) has been used to reveal the symmetry-dependent dynamics of Au38. The strict “nondegeneracies” of both the ground and excited states of Au25 and Au144, corresponding to superatomic P and D orbitals, can be characterized by magnetic circular dichroism (MCD) spectroscopy. , The electron paramagnetic resonance (EPR) spectroscopy can reveal the paramagnetic properties of NCs. ,, Triplet Formation Mechanism The electronic structure of metal NCs differs from that of semiconductor quantum dots (QDs). In QDs composed of heavy atoms, strong spin–orbit coupling (SOC) causes the formation of exciton states that cannot be described in terms of pure singlet (spin 0) and triplet (spin 1) states. Therefore, the concept of spin multiplicity is less relevant in QDs, and their exciton states are usually characterized by the total angular momentum. Generally, the photoinduced band-edge excitons are generated through relaxation of hot excitons in bulk semiconductors, while in QDs a high-energy hot exciton may create two lower-energy excitons if the photon energy is more than twice the bandgap of QD, and similarly for multiexciton generation. − The hot excitons and multiexcitons decay via subpicosecond cooling or a nonradiative Auger recombination process (10 to 100 ps). , We note that no biexciton or multiexciton was observed in atomically precise metal NCs, and the exciton relaxation in metal NCs is laser pump power independent as opposed to the acceleration of relaxation in QDs with increasing pump power (i.e., fluence). − For gold nanoparticles with diameter larger than 3 nm (with distinct SPR), the energy exchange between hot electrons and the lattice occurs through electron–phonon coupling which can be described by the two-temperature model. In contrast, smaller-sized metal NCs that are composed of tens of metal atoms (e.g., Cu, Ag, or Au) exhibit efficient triplet excited state generation. ,,, Consequently, the excited-state dynamics of NCs can be explained by relaxation processes involving S1 and T1 states resulting from single-electron transitions, as in closed-shell dye molecules. This molecular model for NCs is supported by observations of (prompt) fluorescence, TADF, and phosphorescence from photoexcited metal NCs. ,,, Furthermore, photon upconversion based on triplet–triplet annihilation (TTA) can convert two low-energy photons into a high-energy one under low-intensity incoherent light irradiation, which is regarded as a promising technique for solar energy conversion. − With the advances in understanding the triplet states, NCs may become the star candidates for triplet utilization for many applications. The rate constant of ISC (k ISC ) based on the Marcus semiclassical theory , can be expressed as kISC=2πℏ|V|2ρFC where ℏ is the reduced Planck’s constant. ρ FC is the Franck–Condon-weighted density of states, which is expressed as ρFC=14πλkBTexp(−(λ−ΔEST)24λkBT) where λ is the total reorganization energy, k B is the Boltzmann constant, and ΔE ST = E S – E T is the adiabatic energy difference between the singlet and triplet states. In the k ISC , the term V corresponds to the coupling matrix element between the singlet and triplet states, including the direct SOC between the S and T states (V SO ), vibrational SOC (V vSO ), and spin-vibronic coupling (V SV ). Mechanistically, ISC can be enhanced by effective SOC coupling, small reorganization energy and small S-T energy gap. ,,, Considering the change in spin angular momentum, the ISC process must be accompanied by a compensating change in orbital angular momentum to conserve the total angular momentum. The ISC process is efficient if the electrons of metal atoms have large SOC constants, resulting in a significantly large SOC matrix element (SOCME). Conversely, the SOCME is relatively small without a change in orbital angular momentum, despite the heavy atom effects. This is known as the El-Sayed rule, which is effective when considering ISC mechanisms. , TDDFT calculations suggest that NCs usually possess the same electronic configuration in the S1 and T1 states, based on the identical molecular orbitals, which indicates no (or small) SOCME between the S1 and T1 states. ,, Nevertheless, these small SOCME values may still permit rapid ISC when the S1–T1 energy gap is sufficiently small. In contrast, a large ΔE ST makes the direct ISC from S1 to T1 nearly impossible. However, fast ISC may occur from S1 to Tn, which may become highly effective when the Tn state is directly coupled to S1 with a close energy. Thus, ISC in NCs may be driven by a combination of (i) direct spin–orbit coupling between S1 and T1, (ii) spin-vibronic coupling (i.e., vibrationally assisted SOC), and (iii) ISC mediated by higher triplet states (S1 → Tn) when S1–T1 coupling is weak but S1–Tn couplings are large (see example cases of Au13, Au25-rod and Au28 isomers in next section). Table lists the triplet properties of some NCs, including their ISC time constants, triplet state lifetimes determined by TA or TR-PL methods, and PLQY. 1Summary of the Triplet Properties of the Selected NCsNCsCompositionSolvent/FilmISC time τ T (TA) τ T (PL)PLQYref Au 11 [Au11(BINAP)4Cl2]+ EtOH3.5 ps , Au 13 [Au13(dppe)5Cl2]3+ DMF0.4 ps1.72 μs2.61 μs16% , , , Au 18 Au18(DMBT)14 toluene4 ns180 ns133 ns0.1% , Au18(CHT)14 tolueneNot obs. 0.1% Au 22 Au22( t BuPhCC)18 DCM148 ps600 ns485 ns11% Au 16 Cu 6 Au16Cu6( t BuPhCC)18 DCM<1 ps1.3 μs1.64 μs>99% Au 25 [Au25(PET)18]− toluene4 ps160 ns 0.048% , Au 25 -rod [Au25(PPh3)10(PET)5Cl2]2+ THF0.8 ps2.37 μs3.29 μs8% , , Au 25 (Se)-rod [Au25(PPh3)10(SePh)5Cl2]+ DCM0.8 ps>1 μs3 μs [Au25(PPh3)10(SePh)5Cl2]2+ DCM0.9 ps>1 μs2.3 μs Au 28 –C 2 Au28( t BuC6H4CC)20 toluene214 ps294 ns249 ns2.8% Au 28 –S Au28( t BuC6H4S)20 tolueneslow 150 ns0.3% Au 28 Au28i(CHT)20 DCM 1379 ns0.3% Au28ii(CHT)20 DCM 2281 ns3.7% Au 38 Au38(PET)24 toluene0.87 ps5.0 ns Au 38 Au38(PET)26 DCMultrafast 2.3 μs1.8% Au 38 Au38S2(S-Adm)20 toluene<2 ps4.7 μs3.0 μs15% , Au 42 Au42(PET)32 toluene/CHCl3/DCM500 ps (S1), 12.5 ps (Sn)2087 ns2384 ns12% NIR-TA, , , PS film1764 ns21.4% Au 44a Au44[CH3O(CH3)2CC]16(C6H5)6(tht)2·CH2Cl2·H2O2-Me-THF<1 ps1.3 μs/9.4 μs1.2 μs/9.5 μs0.8% Au 52 Au52(p-MBT)32 toluene7.97 ps604 ns564 ns18.3% , Au52(4-EBT)32 toluene16.58 ps539 ns547 ns14.7% Au52(IPBT)32 toluene14.10 ps383 ns334 ns11.1% Au52(TBBT)32 toluene23.37 ps145 ns100 ns3.8% Au52(p-MBT)32 PMMA film<200 fs988 ns808 ns33.9% Au52(4-EBT)32 PMMA film<200 fs857 ns768 ns32.2% Au52(IPBT)32 PMMA film<200 fs741 ns423 ns26.6% Au52(TBBT)32 PMMA film<200 fs720 ns409 ns22.1% AuCu-rod [Au25−x Cu x (PPh3)10(S-C2H4Ph)5Cl2]2+ THFultrafast2.6 μs3.1 μs Au 4 Cu 4 [Au4Cu4(S-Adm)5(DPPM)2]+ THF0.67 ns2.6 μs7.9 μs33% Au 2 Cu 6 Au2Cu6(S-Adm)6(TPP)2 THF19 ns3.0 μs4.9 μs44% , Au 2 Cu 6 DPA Au2Cu6(S-Adm)6[P(DPA)3]2 THF<0.1 ns150 μs Au@Cu 14 [Au@Cu14(SPh t Bu)12(PPh(C2H4CN)2)6]+ DCM<1 ps 1.23 μs71.3% Cl@Cu 14 [Cl@Cu14(SPh t Bu)12(PPh(C2H4CN)2)6]DCM 0.78 μs41.8% aBINAP = 2,2′-bis­(diphenylphosphino)-1,1′-binaphthyl.bdppe = 1,2-bis­(diphenylphosphino)­ethane.cDMBT = dimethylbenzenethiolate.dCHT = cyclohexanethiolate.ePET = 2-phenylethanethiolate.fS-Adm = 1-adamantanethiolate.g p-MBT = p-methylbenzenethiolate.h4-EBT = 4-ethylbenzenethiolate.iIPBT = isopropylbenzenethiolate.jTBBT = 4-tert-butylbenzenethiolate.kDPPM = bis­(diphenylphosphino)­methane.lTPP = triphenylphosphine.mDPA = 9,10-diphenylanthracene.nPLQY = 0.048% in benzene.oPLQY = 12% in DCM, excitation at 380 nm, including FL 3.2% and PH 8.7%. Summary of the Triplet Properties of the Selected NCs BINAP = 2,2′-bis­(diphenylphosphino)-1,1′-binaphthyl. dppe = 1,2-bis­(diphenylphosphino)­ethane. DMBT = dimethylbenzenethiolate. CHT = cyclohexanethiolate. PET = 2-phenylethanethiolate. S-Adm = 1-adamantanethiolate. p-MBT = p-methylbenzenethiolate. 4-EBT = 4-ethylbenzenethiolate. IPBT = isopropylbenzenethiolate. TBBT = 4-tert-butylbenzenethiolate. DPPM = bis­(diphenylphosphino)­methane. TPP = triphenylphosphine. DPA = 9,10-diphenylanthracene. PLQY = 0.048% in benzene. PLQY = 12% in DCM, excitation at 380 nm, including FL 3.2% and PH 8.7%. Manipulating the Intersystem Crossing Process Structure and Size Control Generally, the molecular orbitals and optical properties of NCs are largely dependent on their structure and size. , The steady-state optical properties and excited state dynamics have been observed to exhibit structure/size dependences. A typical example is the face-centered cubic (fcc) series (Au28, Au36, Au44, Au52), in which the E g decreases and carrier lifetime increases. Different from this fcc series of NCs, Kong et al. reviewed Au13-assembled oligomers (including the monomeric Au13, dimeric Au25, trimeric Au37, and cyclic pentameric Au60), which exhibit near-infrared (NIR) photoluminescence from the metal core. , A long-lived emissive state was observed in Au13 (1.72 μs) and Au25 rod (3 μs), ,, while the lifetime of Au37 rod was only 28 ns, and 100 ns for the cyclic Au60. In contrast to the rod-Au25, the spherical Au25 exhibits a few-picosecond CT process from the Au13 core to the staple surface, which may explain the dual emission in the visible and NIR region. ,,,, However, the origin of photoluminescence cannot be clearly explained without considering the spin properties. The Au25-rod (coprotected by phosphine and thiolate) exhibits only phosphorescence, without fluorescence, at room temperature with an ISC yield of almost 100%, suggesting the occurrence of ultrafast ISC process (rate ∼ 1012 s–1). , Recently, the dominant role of triplet state in the luminescent Au13 was demonstrated experimentally (Figure a). , As discussed above, various important triplet-state parameters for the ISC and TET processes of Au13 were determined. The El-Sayed rule (S 1 → T n≥2) was applied to reveal the underlying mechanism of highly efficient ISC in Au13, considering the slow ISC process between S1 and T1 with the S1-T1 energy gap of 0.73 eV. The highly efficient ISC process in Au13 was also demonstrated by TDDFT calculations. This mechanism is highly effective when the Tn state is directly coupled to S1 and is energetically close to S1. Similarly, the phosphorescence of Au25-rod (Figure b) from the triplet state with almost 100% ISC quantum yield was demonstrated by PL quenching and TTA-UC measurements. Moreover, the direct spin-forbidden transition from S0 to T1 in the NIR region (730–900 nm) was observed. , (a) Au13 excited-state relaxation pathway, in which the T2, T3 and T4 states are theoretically predicted to exist within the energy gap between S1 and T1 states and are shown as red dashed lines. Copyright © 2023 American Chemical Society. (b) Au25-rod excited state deactivation pathway and related photophysical parameters (solvent: deaerated THF). Copyright 2020 The Royal Society of Chemistry 2022. (c) Simulated mechanism of photoluminescence for the isomeric pair of Au28i and Au28ii. Copyright © 2024 The Authors, published by American Chemical Society. Besides the periodic series of NCs, Mazumder et al. reported the isomeric effects of Au28(CHT)20 on the photoluminescence and ISC, which is related to the electron–vibration coupling and also higher triplet states. Theoretical calculations by Pei et al. revealed the PL mechanism (Figure c), including the ISC process from S1 to the high triplet state T2. The larger ISC rate constant can be attributed to the larger SOCME of S1 and T1 (for Au28i) or a very small ΔE ST (for Au28ii) via theoretical calculations. The ISC process may be suggested as ∼11 ps determined by TA measurements, considering the triplet-related emission (TADF or phosphorescence). Recently, the structure/size dependency of dual emission mechanisms for the atomically precise one-dimensional (1D) ultrathin rod-shaped Au NCs (Au24, Au42, and Au60) were revealed by detailed DFT and TDDFT calculations. In earlier experimental work, Qi et al. reported an excited-state distortion for the generation of dual fluorescence (i.e., from S1 and distortion-caused S1′ states), which is now confirmed by Pei’s theoretical work. With increasing aspect ratio (AR, from 3.1 to 6.2 to 9.4 in the series of Au24, Au42, and Au60), the core structure transits from flexible Au24 to rigid rods, with enhanced x-axis transition dipole moments (μ x , from 0.58 or 1.00 (Au24 S1/S1′) to 6.31 (Au42) and then to 10.73 D­(Au60)), narrowing the adiabatic energy gap (ΔE ST ) between S1 and T1 states (from 0.520 or 0.851 eV (Au24 S1/S1′) to 0.37–0.57 eV (Au42 and Au60)). The manipulated energy levels lead to elevated fluorescence radiative rate constants (k r F ) and intersystem crossing rate constants (k ISC ) to the 108 s–1 regime (for Au42 and Au60), much higher than the typical 105–6 s–1 in Au24. ,, Consequently, as shown in Figure a, the shorter Au24 rod exhibits dual fluorescence emission (F1+F2), while the elongated systems (Au42 and Au60) show fluorescence-phosphorescence dual emission (F+P). Electronic structure analysis reveals that the increased length of the rods weakens charge transfer excitation while enhancing the excited state localization. Experimentally, Au24 exhibits dual fluorescence (670 and 1050 nm) by structural distortion of ∼1.6 ns time constant (Figure b). In contrast, Au42 exhibits fluorescence and phosphorescence (Figure c), , in which the phosphorescence (lifetime ∼ 3 μs) was confirmed by oxygen quenching measurements and TTA-UC process. (a) TDDFT simulated mechanism of photoluminescence for Au24, Au42 and Au60 NCs, in which F denotes fluorescence and P donates phosphorescence. Copyright © 2025 The Authors, published by American Chemical Society. (b) UV–vis absorption (black) and PL (gray) spectra of Au24 in DCM (the asterisk denotes the spectral artifact caused by the solvent overtone absorption of the NC emission). Copyright © 2020 The Authors, published by Springer Nature. (c) UV–vis absorption (green) and PL (gray) spectra of Au42 in DCM. Copyright © 2022 American Chemical Society. Doping or Alloying Effects Heteroatom doping has proven to be an effective strategy to enhance the ISC yield or accelerate ISC, which improves the PL and TTA-UC performance. ,,− In recent work, Liu et al. revisited the homometal Au25 and doped NCs, as well as the Ag25 and doped ones to disentangle the influencing factors and elucidate the PL mechanism. This series of M25 NCs with superatomic closed shell (1S2|1P6 electron configuration) exhibited emission from 800 to 1400 nm (edge to edge of PL), which was determined as predominant phosphorescence. Weak PL in the NIR region (∼1000 nm, QY = 1%) was observed for Au25 due to strong electron-vibrational coupling. The QYs of the three MAu24 NCs (M = Hg, Au, Cd) followed a linear relation with PL lifetimes by a mechanism of suppressed nonradiative decay for PL enhancement (Figure a). Heteroatom doping by the Cd or Hg atom occurs in an icosahedral shell site of Au25, which is found to suppress the nonradiative process (electron-acoustic phonon interaction), without affecting the radiative relaxation. An unusually strong electron–phonon interaction is mainly contributed by the Au2(SR)3 vibration. Trimetallic [PtCdAu23(PET)18]− with a Pt@CdAu11 kernel was synthesized by Suyama et al. to investigate the codoping effects of group X and XII atoms. The synergistic effects of codoping Pt and Cd atoms were observed in PLQY: the enhancement factor by codoping (46×) was expressed by the product of those by monodoping of Pt and Cd (9 and 5×, respectively). In contrast, Ag25 shows higher PL (QY = 3.5%) because of weaker electron-vibration coupling than in Au25. Single Au atom doping into Ag25 leads to a 5× enhancement of the radiative rate and a suppression of nonradiative decay rate in AuAg24 (QY rise to 35%). However, doping more Au atoms results in strong electron-vibration coupling via gold distribution to the staple motifs for Au x Ag25‑x (x = 3–10, QY drops to 5%). Furthermore, the center doping of the gold atom into Ag25 significantly improves the radiative recombination. Due to the increased ISC yield, PtAg24 exhibited an enhanced TTA-UC with perylene or TIPS-Ac as the annihilator. Alloying the bimetallic PtAg24 with Au-SR complex generated the trimetallic Pt1Au6.4Ag17.6 nanocluster, which quenched the weak fluorescence of PtAg24 (0.1%). However, the rod-like Pt2Au10Ag13(PPh3)10Br7 displayed a high PLQY (14.7%) via alloying with a large amount of AuI(PPh3)Br complex. In addition, the triplet energy level can be adjusted by Cu-substitution to promote triplet energy transfer and UC quantum yield. (a) Heterometal doping effects in the Au25 and Ag25 based series of NCs. Copyright © 2023 American Chemical Society. (b) Superatomic cores of MAu12. (c) Plot of the PLQYs of MAu12 and MAu12-m vs the energy of their PL peaks. Panels (b,c) Copyright © 2022 Wiley-VCH GmbH. Another interesting example is the doping-mediated energy level engineering of M@Au12 (MAu12(dppe)5Cl2) (where, dppe = 1,2-bis­(diphenylphosphino)­ethane) compared with undoped Au13 (Figure b). , The phosphorescence quantum yield was enhanced with an increase in the HOMO–LUMO gap and reached 46%–67% for MAu12 (M = Pt, Rh, or Ir) under deaerated conditions (Figure c). MAu12 can also act as a more efficient and stable photocatalyst than Au13 for intramolecular [2 + 2] cycloaddition of bisenone via the oxidative quenching cycle. Based on the results, doping with an element which is positioned at the lower left of the periodic table for M@Au12 is an efficient design for enhancing PLQY. , An enhanced ISC process was also observed by doping Au nanoclusters with Cu. Shi et al. reported that the ISC rate was increased by nearly 300-fold and nonradiative decay was suppressed by about 60-fold in Au16Cu6( t BuPhCC)18 [denoted Au16Cu6] relative to Au22( t BuPhCC)18 [Au22]. This alloy Au16Cu6( t BuPhCC)18 was synthesized by partially substituting Au with Cu in the Au22( t BuPhCC)18 template. As a result, Au16Cu6 exhibited near-unity NIR phosphorescence (PLQY > 99%, peak at 720 nm with tailing to 950 nm) in deaerated solution at room temperature, in constrast with the much lower efficiency of the parent Au22. Replacing six Au atoms in Au22 with Cu had a negligible effect on the SOC term but significantly decreased the singlet–triplet energy gap (ΔE ST) and the reorganization energy (λ), thereby increasing the Franck–Condon weighted density of states (ρ) and accelerating the ISC process. Xu et al. recently reported triple-functional Cu x Au61‑x NCs with NIR-II (>1000 nm wavelength) photoluminescence, photothermal and photodynamic properties. Crystallographic analysis of Cu x Au61‑x NCs revealed a penta-icosahedral, oblate structure, with the Cu atoms located in the core region, which led to a more compact structure. Thereby, an accelerated ISC process (3 ps) was observed in Cu x Au61‑x , while a slower process (12 ps) was observed in the undoped Au60. A structurally flexible Au–Cu alloy cluster (Au4Cu4) exhibited favorable sensitization properties and superior upconversion performance because of its near-unity ISC quantum yield and long triplet lifetime. In another case, a fast ISC process (<1 ps) and very high phosphorescence (QY = 71.3%) in nondegassed solution at room temperature were reported for Au@Cu14. The critical role of the central Au atom is illustrated by a QY drop to ∼ 40% when Au is replaced by Cl, indicating that Au enhances PL via the spin–orbit coupling effect. The energy gap between S1 and T1 for Au@Cu14 is smaller than that in Cl@Cu14, which contributes to a more efficient ISC process. Theoretical calculations by Pei and co-workers revealed that the enhanced SOC effect, larger transition dipole moments, and greater orbital overlap in Au@Cu14 collectively contribute to its significant radiative transition than the Cl@Cu14 counterpart. Overall, the heterometal doping strategy is quite versatile, as it not only alters the HOMO–LUMO gap but also modulates the ΔE ST, and the reorganization energy by manipulating the spin–orbit coupling, wave function overlap, and electron-vibration coupling, therefore promoting the ISC process. Ligand Effects Without modifying the energy gap by structure/size control or doping, triplet formation and the decay process can be affected by substitution of ligands. The past decade has witnessed significant progress in thiolate- and alkynyl-protected NCs, which provide excellent platforms to investigate ligand effects. The ligand electronic structure plays an essential role in modulating spin–orbit coupling and the ISC process in gold nanoclusters by reshaping metal–ligand orbital interactions and altering excited-state energetics. Electron-donating or electron-withdrawing substituents change the degree of orbital hybridization between ligand states and Au orbitals, thereby perturbing the symmetry and composition of superatomic frontier orbitals. Two isostructural Au38 exhibit totally different catalytic performance in the semihydrogenation of alkynes, unveiling the ligand effects in catalysis by gold NCs. The ligands with electron-rich atoms (e.g., N, O) or groups (e.g., −COOH, −NH2) can largely promote photoluminescence based on the studies of the ligand roles in Au25. Ligand substitution on Au25 altered both the quantum yields of high-spin emissive states and the extent of nonradiative decay arising from electronic state mixing. The origin of photoluminescence of Au25 was investigated by introducing four different types of thiolate ligands with different carbon tails. Visible PL from the surface state and NIR PL from the core state was observed with UV excitation indicating a rapid relaxation from higher core states to a lower core state and surface state. The energy barrier prevents the core-surface charge transfer when the NC is NIR excited, which displays only NIR emission. Recently, Shi et al. reported a detailed study of ISC process for alkynyl-protected (denoted Au28–C2) and its counterpart with thiolate ligand (Au28–S), which revealed 10-fold stronger luminescence in Au28–C2 than in Au28–S. Au28–C2 emits TADF and phosphorescence at room temperature, while Au28–S emits TADF and fluorescence. Both Au28–C2 and Au28–S (Figure a) have the same number of gold atoms with similar kernel structures, indicating an identical SOC value; however, a more compact core–shell structure in Au28–C2 facilitates its ISC process (214 ps) with a large triplet state population and hence higher PL (Figure b). (a) X-ray structures of Au28–C2 and Au28–S NCs. (b) Comparison of the emission spectra and PL decay curves of Au28–C2 and Au28–S. Panels (a.b) Copyright © 2025 American Chemical Society. (c) Comparison of PL spectra and decay profiles of Au52(SR)32 with different −R groups under ambient condition in DCM. Copyright © 2023 The Authors, published by American Chemical Society. Another interesting case of ligand-induced modulation of ISC pertains to Au52(SR)32. Since the PL of Au52 comprises TADF and phosphorescence, which is related to the triplet population, the PLQY can serve as a criterion for the ISC yield. Wang et al. studied the effects of carbon tails (−R) of the thiolate ligands on Au52(SR)32, in which the PLQY is largely enhanced with a decrease in the ligand’s para-bulkiness. As shown in Figure c, Au52(SR)32 capped with p-methylbenzenethiolate (p-MBT) exhibits the highest PLQY (18.3%) in dichloromethane (DCM), which results from the restriction of ligand vibrations and rotations via a stronger interligand π···π stacking on the Au52 core. Consistent with the PLQY results, Zeng et al. investigated the ISC process by TA spectroscopy. The results showed a faster ISC process (<200 fs) for Au52(p-MBT)32, but slower ISC (∼11 ps) for the other three Au52(SR)32. Of note, these ISC time constants are for the case of 380 nm excitation, while the values of Au52(SR)32 listed in Table are for band-edge excitation (800 nm). In this series of NCs, the variation in nonradiative relaxation is primarily mediated by ligand vibrations because the metal cores are similar and have similar low-frequency phonons. Thus, the ligand-induced PL enhancement is driven by both an accelerated ISC and reduced vibrational dissipation by ligands. In recent work, Smith and Knappenberger reported the −R effect in the ISC process of [Au25(SR)18]− and found that the aliphatic −C6H13 and −C12H25 tails accelerated ISC (2.4–2.6 ps) compared to −CH2CH2Ph (4 ps), which was suggested to arise from the coupling between the core electron and staple motif’s vibrations in which aliphatic groups resulted in less rigid staples. The excited-state analysis of Au18(SR)14 revealed state-resolved relaxation due to the presence of multiple excited states. Au18(captopril)14 exhibited a PLQY of ∼8.6%, but only 0.089% for Au18(CHT)14 (where, CHT = cyclohexanethiolate), which is consistent with the recently reported weak emission of Au18(DMBT)14 (where, DMBT = 2,4-dimethylbenzenethiolate) and Au18(CHT)14. , The TA measurements in the visible–NIR region combined with temperature-dependent PL (see Figure ) suggest a very slow ISC process (∼4 ns) for Au18(DMBT)14. In contrast, no obvious triplet population for Au18(CHT)14 was observed. This work presents a clear example that the excited states of gold NCs can be modulated by tailoring the −R groups of thiolate ligands, with the aromatic ligands exerting a large influence on the excited state dynamics and pathways owing to the networking interactions among the surface ligands’ tails. A triplet-mediator ligand (TL)-protected metal NC was introduced by Mitsui and co-workers to act as an efficient triplet sensitizer. , In Au2Cu6DPA (where DPA = 9,10-diphenylanthracene), the excitation of the Au2Cu6 core rapidly generates a metal-to-ligand charge transfer state, followed by the formation of the long-lived triplet state (∼150 μs) at a DPA site in the TL; note: the triplet state lifetime of the Au2Cu6 core is only 4.9 μs. In another example, efficient intra- and intermolecular triplet energy transfer from photoexcited Ag29 is determined to achieve a green-to-blue UC internal quantum yield of 24% and an exceptionally low threshold intensity by adding an excess amount of phosphine DPA derivative, P­(DPA)3, to Ag29. Overall, the ligand effects comprise three categories, (i) different types of ligands as exemplified with Au28L20 above (L = thiolate vs alkynyl), (ii) the same type but with different −R groups as exemplified with Au52(SR)32, [Au25(SR)18]−, and Au18(SR)14, which are quite intriguing and came as a surprise, and (iii) NCs with triplet-mediator ligands as exemplified by Mitsui et al. Effects of Surrounding Environment The optical properties of NCs can be largely affected by charge transfer between the ligand and the metal core. ,− Charge transfer from Au core to the peripheral ligands was identified by TA measurements using solvents with different polarizability. ,, Since the S1 energy level is more sensitive to solvent polarity, the energy gap between singlet and triplet (ΔE ST ) can be manipulated by solvent polarity. In highly polar solvents, the S1 energy drops, reducing the ΔE ST and thereby accelerating the ISC process. For instance, the cyclic penta-icosahedral Cu x Au61‑x NC showed that the ISC process occurred in 3 ps in DMSO but 30 ps in DCM. Similar results were observed for the cyclic penta-icosahedral Au60 (12 ps in DMSO vs 45 ps in DCM). As discussed above, Au42 exhibits dual emission of FL and PH, which provides a platform to investigate the effects of the microenvironment around the NCs. Dual emission (PLQY = 11.9%) in the NIR region of Au42 (Figure a) comprises fluorescence (QY = 3.2%) and phosphorescence (QY = 8.7%). After embedding Au42 in polystyrene films, the phosphorescence QY significantly increases to 20.3% while the fluorescence is suppressed to 1.1% (Figure b). The suppression of fluorescence and enhancement of phosphorescence from the solution to the film state was attributed to the dipole–dipole interaction. Dipolar interaction causes a splitting in the excited states of Au42 assemblies and thus narrows the gap between the S1 and T1 states. Such a model can be extended to assemblies where the split excited states become denser and yield a band-like electronic structure. Because of the decrease in ΔE ST , the ISC process is enhanced, leading to a higher QY of phosphorescence. Similarly enhanced PL was also observed in the Au52 series, i.e., the PLQY of Au52(TBBT)32 embedded in PMMA films is up to 22.1% while it is only 3.8% in DCM solution. The film state also suppresses the nonradiative decay, hence, higher PLQY. (a) PL spectra of Au42 in deaerated DCM under N2 (blue profile) and O2 (red profile), respectively. (b) PL spectra of Au42/PS film under N2 (blue profile) and O2 (red profile), respectively. Panels (a,b) Copyright © 2022 American Chemical Society. (c) PL spectra and (d) PLQY of Au42 in deaerated 2-Me-THF containing DMBA with different concentrations. Panels (c,d) Copyright © 2024 The Authors, published by American Chemical Society. Furthermore, an effective enhancement of QY from 18% to 50% in deaerated solution of Au42 was demonstrated by noncoordinative interactions with amide molecules (as solvents). The NIR absorption peak blue-shifts from 806 to 781 nm with increasing amide (N,N-dimethylbenzamide, DMBA) concentration from 0 to 2143.9 mM, and the integrated PL intensity of Au42 increases significantly (Figure c,d). Specifically, the PLQY initially remained unchanged with a concentration up to 53.6 mM (Stage I). It then exhibited a gradual rise to 50.1% at the DMBA concentration of 1286.4 mM (stage II) and maintained this intensity as the concentration was further increased (Stage III). Cryogenic photoluminescence measurements indicated that amide molecules effectively suppressed the vibrations associated with the Au–S staple motifs on Au42 and also enhanced the radiative relaxation, both of which led to stronger emission. In addition, 1H NMR-derived evidence of Au11 cluster-solvent hydrogen bonding was observed with a downfield shift and an exchange broadening of ethanol’s −OH proton resonance arising from hydrogen bonding with the cluster’s axial chlorine ligand. , The hydrogen-bonding interaction between Au11 and protic solvents depopulates the vibrationally mediated relaxation pathway in favor of radiative decay. , The strength of the cluster-solvent hydrogen bonding interaction can be modulated since there is stronger hydrogen bonding with ethanol than with butanol. The enhanced strength of Au11-ethanol hydrogen bonding revealed an additional radiative channel (phosphorescence), which was not observed for the Au11-butanol system. Thus, direct interaction between clusters and solvents (hydrogen bonding or physical adsorption) should be considered for structural change via rigidification. The water insolubility of a majority of NCs limits their application in the biomedical and aqueous photocatalysis fields. To overcome this issue, Liu et al. reported a polymer wrapping method for phase transfer of various organic soluble NCs into aqueous phase without degrading the electronic and optical properties. Such materials are further demonstrated for robust photocatalysis in water. With an amphiphilic copolymer (Pluronic F127), Liu et al. made aqueous soluble Au18(DMBT)14@F127 nanoparticles (<2.5 nm, measured by transmission electron microscopy, wrapping up to several Au18 NCs in a nanoparticle), which showed a 10-fold increase in PL intensity than that of Au18(DMBT)14 in the organic phase, unlike quantum dots that lead to severe quenching of PL upon transfer to aqueous solution. Summary and Future Outlook In this Perspective, we summarize recent advances in manipulating the ISC process in atomically precise metal NCs, which are analyzed by time-resolved and temperature-dependent spectroscopy along with theoretical calculations. From the fundamental standpoint, the ISC process and triplet yield can be controlled by the structure and size of NCs, ligands, heterometal doping, and the surrounding environments (such as solvent polarization or solid state). Currently, insight into the triplet excited-state dynamics of NCs is not yet as well understood as the case of dye molecules, and there are still some fundamentals to be addressed in future work. The relationship between the ISC process and a variety of structures of NCs should be investigated; the latter include the fcc, ,− hexagonal close-packed (hcp), ,, body-centered cubic (bcc), , and polyhedron-based cores such as M4 tetrahedron, M7 decahedron, and M13 icosahedron. ,, Besides, the solvent should be another important factor for NCs that can be modulated when targeting specific optical properties such as PL. ,, Recent efforts have introduced optically active organic-semiconductor ligands (including singlet fission and upconversion chromophores) into NCs to build a new family of hybrid materials. , These organic ligands may directly contribute to the frontier orbitals, reconfiguring the band-edge states and excited state dynamics. Furthermore, combining NCs with functional molecules into a single entity or a nanocapsule should exhibit an excellent optical response with a relative low concentration, especially for triplet related multiexciton processes. To tackle the spin properties of NCs, it is necessary to develop advanced theoretical calculations and spectroscopic methods. Given the complicated excited states (e.g., dense states) of NCs, the three-state model (S0, S1 and T1) is not sufficient to explain the complicated ISC process. Time-resolved EPR is a powerful technique to provide spin information on the triplet states, such as those formed after a laser pulse, by measuring their magnetic resonance signals over the time delay. For commercial CW-EPR spectrometers, a magnetic field modulation of 100 kHz is routinely used to improve the sensitivity of the spectrometers. The response time is limited to tens of μs, which is usually inadequate for the photoinduced process. With a focus on the short-lived species, transient EPR directly detects the changes in EPR absorption following a laser pulse, and microwave irradiation is continuous and at low power, with no filed modulation. This provides an absorption-type line shape as in most other spectroscopic techniques. An integration window at various time ranges after the laser pulse is used to improve the S/N ratio. Since the electron spin states are not in thermal equilibrium (deviated from the Boltzmann population) after the laser pulse, the spectra are spin polarized, showing absorption (A) and emission (E). While one can assign triplet excited states with criteria of long lifetimes (e.g., μs), singlet oxygen generation and TTA results, NCs of small E g and short-lived excited-states (e.g., lifetimes of one hundred ns or so) and insensitivity of PL to O2 become difficult to assign their triplet states; in such cases, transient EPR should be pursued. Moreover, transient EPR can be used to track the spin dynamics of the triplet state. However, the instrument response time for the transient EPR is hundreds of nanoseconds, , which is contingent and may not be of sufficient resolution for those nanoclusters with strong SOC, because their triplet lifetimes may be shortened to tens of nanoseconds. The characterization of such very short-lived triplet states requires high time-resolution by transient methods. Ultrafast magneto-optical methods may provide ultrafast time-resolution (∼ps), in which a magnetic field is introduced to the transient optical spectroscopic measurements. − Such ultrafast magneto-optical techniques even allow for the detection of spin coherence, which is important for the application of NCs in quantum information science. Overall, research on photoexcited exciton dynamics in atomically precise metal NCs has acquired rapid progress after the recognition of triplet states. However, the much more complex electronic structure of metal NCs than conventional molecules poses major challenges in understanding the complex exciton dynamics; for example, why do the ISC times of various NCs span so broadly (ranging from femtoseconds to nanoseconds)? Similarly, why do the triplet lifetimes span so broadly (ranging from the typical microseconds down to 10s of ns at room temperature)? In addition, for traditional dyemolecules, ISC typically occurs between S1 and T1, but NCs have shown many cases of S1 to higher triplet states, and there is also a possibility of intersystem crossing from higher singlet to higher triplet states (i.e., Sn to Tm, with n and m being 2 or higher). Understanding such complex dynamics requires joint efforts from experimentalists and theoreticians. We believe that future work will make further advances on the topic of intersystem crossing and triplet state manipulation, and such advances will promote a variety of applications, such as optoelectronics, photocatalysis, singlet oxygen generation, sensitization, NIR photon upconversion, NIR luminescence, bioimaging, and therapy.