Minimized ΔE_ST: drive thermally activated delayed fluorescence materials in photodynamic therapy

Abstract

The energy gap (ΔE_ST) between the singlet state (S₁) and the triplet state (T₁) and the T₁ lifetime are important factors in determining the efficiency of photodynamic therapy (PDT). Owing to intersystem crossing (ISC) between S₁ and T₁, a smaller ΔE_ST becomes a key factor in the design of photosensitizers (PSs). Attractively, thermally activated delayed fluorescence (TADF) materials exhibit a small ΔE_ST, fully utilizing photon energy and achieving simultaneous improvement in both the energy gap and lifetime. However, the applicability of TADF in complex intracellular environments remains to be explored. In this review, the ingenious design of TADF molecules is presented according to group classification, and elaborate strategies of TADF are discussed in promoting triplet energy conversion, improving the penetration depth of phototherapy, and regulating the dynamic balance of triplet excitons. This article comprehensively summarizes the development of TADF materials in PDT, which has important reference value for investigating TADF PSs.

---

1. Introduction

Since von Tappeiner et al. proposed the concept of photodynamic therapy (PDT) in 1907, it has attracted widespread attention in the treatment of solid tumors.¹ PDT has provided excellent clinical achievements in the biomedical domain owing to its high selectivity, low toxicity and non-invasiveness.^2,3 As the core of the strategy, the reactive oxygen species (ROS) yield of photosensitizers (PSs) is the key to determining PDT efficiency. According to the Jablonski diagram, PS transitioned from the ground state (S₀) to the excited state (S_n) after absorbing photons and underwent vibrational relaxation and internal transition to S₁. Then, the S₁ exciton either transitioned back to S₀ or passed to the excited triplet state (T_n) through intersystem crossing (ISC). Because of its better stability, the triplet exciton can react with the surrounding active substrates through electron or energy transfer to generate ROS. It is classified into two categories: type I ROS, which include superoxide anion (O₂˙⁻), hydroxyl radical (˙OH), or hydrogen peroxide (H₂O₂) generated by the interaction of proton or electron transfer processes involving nearby biological substrate molecules (such as cell membrane components, proteins, and lipids); and type II ROS, which include singlet oxygen (¹O₂) generated by energy transfer to the surrounding molecular oxygen (³O₂).^4–6 Thus, the number of triplet excitons plays an essential role in producing ROS, which is associated with the singlet–triplet energy gap (ΔE_ST) and the ISC rate constant (k_ISC). According to Fermi's Golden rule, the relationship between k_ISC and ΔE_ST can be described by the following equation:
where H_SO represents the Hamiltonian of the spin–orbit coupling effect (SOC) and is directly proportional to k_ISC.⁷ Moreover, k_ISC depends on two important parameters: SOC and ΔE_ST. Generally, heavy atoms (such as transition metal atoms and halogen atoms) with high nuclear charges enhance the SOC of molecules, thereby increasing the absorption transition from S₀ to T₁ and the ISC, ultimately leading to an increase in the T₁ population.^8–10 Furthermore, to improve k_ISC, the heavy atom strategy was refined to reduce ΔE_ST by introducing a donor (D)–acceptor (A) structure with a twisting conformation.^11–13 However, the inherent cytotoxicity of heavy atoms should be seriously considered before treatment.

Pure organic thermally activated delayed fluorescence (TADF) materials provide new opportunities for exploring safe and efficient PSs. At the molecular level, TADF systems employed a rationally engineered D–A architecture with a spatially orthogonal orbital configuration.^14,15 To achieve a highly distorted molecular structure, D and A were connected directly or indirectly by π bridges, changing their substitution positions and numbers or expanding π conjugation.¹⁶ The interaction of the D–A group modulated the aggregation property of molecules and prolonged the lifetime of excited states.^17,18 Moreover, efficient separation of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) is essential to achieve low ΔE_ST and accelerate reverse intersystem crossing (RISC).^19–22 Additionally, the strategic decoupling of HOMO and LUMO minimized the wavefunction overlap (ρ/H–L < 0.1). Such spatial segregation induced a through-space charge transfer (TSCT) character, which was different from conventional through-bond charge transfer (TBCT), and it simultaneously achieved two critical design objectives.²³ First, ΔE_ST was drastically reduced through enhanced charge transfer (CT) state hybridization. Meanwhile, SOC was amplified in favour of ISC elevation, extending the excited state lifetime, which increased the possibility of electron/energy transfer among the surrounding medium and heightened the ROS yield.²⁴ The mechanism of the TADF materials employed in PDT is illustrated in Fig. 1. TADF PSs were first activated from S₀ to S_n after irradiation. Afterwards, the excited TADF molecules underwent ISC from the S₁ to T_n, in which the energy gap was less than or equal to 0.1 eV.²⁵ The unique small ΔE_ST endowed TADF materials with extremely high internal quantum efficiency and maximized the use of photon energy, which mediated a faster ISC process and expanded the luminescence pathway, ultimately achieving more efficient ROS generation and imaging capability. Different from the common fluorescence lifetime (∼ns), the unique RISC process facilitated the up-conversion of thermally activated non-emitting triplet excitons to spin-permissible singlet excitons, thereby emitting time-delayed fluorescence (μ ∼ ms).^26–28

Fig. 1 Schematic of the mechanism of TADF materials in the PDT application.

Thus, the introduction of the TADF group into the PSs exhibited great potential for manipulating the dynamic balance between ISC and RISC, seeking the optimal solution for triplet excitons and maximizing ROS generation. Considering the advantages of TADF materials, such as structural flexibility, oxygen sensitivity and biocompatibility, it was necessary to summarize and explore the impact of TADF-based PSs on enhanced PDT. Hence, this review presents the PDT strategies of TADF-based PSs in tumor treatment and discusses the molecular structure design, precise regulation of different functions, and the highlights of each strategy.

2. Molecular design of TADF materials for PDT

Since 2016, multifarious TADF molecule emitters with extremely small ΔE_ST (0.01–0.05 eV) have broken the theoretical efficiency limit of RISC in heavy metal-free organic emitters.^29–31 Zhang et al. first discovered and prepared green fluorescent metal-free TADF PSs to achieve the ¹O₂-increasing type PDT.³² Among them, the oxygen-induced triplet quenching effect of TADF materials promoted enhanced PDT by preventing the RISC process from obtaining a long T₁ lifetime. However, there is currently no classification for TADF molecules in PDT. In this study, metal-free TADF molecules are reviewed from the aspects of molecular design concept and electronic configuration, including three common categories: fluorescein derivatives, carbazole derivatives and quinone derivatives.

2.1. Fluorescein derivatives

Fluorescein derivatives with appropriate hydrophilicity, minuscule dark toxicity, and lysosomal localization are normally employed as donors and connected with a strong electron-withdrawing group (such as cyano) to construct the D–A configuration.^33,34 Nevertheless, most reported TADF fluorescein derivatives are type II PSs. Owing to the presence of the phenol group, ¹O₂ was mainly produced in lysosomes during treatment, disrupting cellular structures and implementing type II PDT.

Owing to the lack of drug sites, TADF molecules were not specifically released in diseased tissues, causing irreversible damage to normal tissues. Therefore, the construction of a smart activation PDT strategy has become an inevitable trend. Recently, Tian et al. synthesized a fluorescein derivative, named DCFP-Cl, with two covalent reaction sites.³⁵ DCFP-Cl was produced using 2,7-dichlorofluorescein as the donor and pyridinium salt as the acceptor. Benzyl chloride subunits acted as reaction sites to covalently react with the sulfhydryl groups of peptides and proteins to achieve the mitochondria-stapling strategy (Fig. 2A). With biotin receptor overexpression on the cell surface, Biotin-PSA-coated DCFP-Cl endowed nano PS with tumor recognition and intelligent activation. Under visible light excitation, DCFP-Cl formed a triplet lifetime of about 1–2 μs through the ISC process (Fig. 2B). By monitoring the emission decay at 700 nm, DCFP-Cl showed delayed fluorescence with a microsecond luminescence lifetime (τ = 2.57 μs) (Fig. 2C). Liu et al. employed hypoxia over-expressed nitro reductase (NTR) as an activator to develop a series of activable fluorescein PSs that possessed the ability to target the tumor edge.³⁶ The smart activation strategy was based on the photo-induced electron transfer (PET) mechanism (Fig. 2D). The introduction of nitrobenzene led to intramolecular electron transfer and fluorescence quenching. The fluorescence signal was recovered when the nitro group was specifically reduced to an amino group by the NTR. Compound 3 formed a spironolactone structure with the hydroxyl group of fluorescein to modulate the fluorescence properties (Fig. 2E). The PET process controlled the fluorescence decay of the S₁ state, thereby regulating the PDT efficiency.

Fig. 2 (A) Strategy and mechanism for Biotin-PSA@DCFP-Cl to enhance antitumor efficacy in PDT. (B) Nanosecond transient absorption spectra of DCFP-Cl (10 μM) at different decay times. (C) Luminescence decay trace of DCFP-Cl in deaerated MeCN.³⁵ Reproduced with permission. Copyright 2022, American Chemical Society. (D) Design strategy for the smart theranostic molecule and the molecular structure (E) of compounds 1–3.³⁶ Reproduced with permission. Copyright 2019, American Chemical Society.

Considering the hypoxia microenvironment caused by the rapid proliferation process, Chen et al. designed a novel nanosystem PS@BSA with compound 3 as the “electron pump” and BSA as the “electron reservoir” to facilitate the conversion of type II ROS to type I ROS (Fig. 3A).³⁷ Electron-rich BSA was involved in the redox cycle, where BSA provided electrons and combined with PS to form an electrochemically nonactive complex. Subsequently, the electrons were transferred to O₂ to form O₂˙⁻, enabling type I photosensitization. As depicted in Fig. 3B and C, the long-life decay (11.36 ms) of PS/BSA upon ground state depletion (570 nm) indicated that the strong binding of BSA to PS achieved efficient electron transfer and promoted O₂˙⁻ generation.

Fig. 3 (A) Simple schematic of PS acting as an “electron pump” and BSA as an “electron reservoir” in the type I PDT process. (B) Transient absorbance of PS in aqueous solution under anaerobic conditions by argon bubbling. (C) Decay curve at an absorption wavelength of 650 nm by BSA/PS.³⁷ Reproduced with permission. Copyright 2023, American Chemical Society. (D) Synthesis route for biotinylation of PS1-Biotin. (E) Cyclic voltammograms of PS1-Biotin and PS1.³⁸ Reproduced with permission. Copyright 2022, Springer Nature. (F) Molecular structure of the organic molecules used in this work. (a) 2CzPN, (b) 4CzIPN, and (c) 4CzTPN-Ph. (G) Variation in fluorescence intensity of SOSG at 525 nm as a function of irradiation duration.⁴¹ Reproduced with permission. Copyright 2016, Royal Society of Chemistry. (H) Molecular formula for XCy and car-XCy. (I) Flash photolysis result of car-XCy. (J) Detection of ROS generation using DHR123 with car-XCy.⁴² Reproduced with permission. Copyright 2024, Wiley-VCH. (K) Chemical structures of BTMCz. (L) PL decay profile with the respective average lifetimes applied under air or vacuum for BTMCz. (M) ROS detected by ESR spectra for BTMCz in the presence of TEMP and white light.⁴³ Adapted with permission. Copyright 2025, Wiley-VCH.

An et al. synthesized the PS PS1-Biotin by combining fluorescein with biotin to achieve type I and type II PDT mechanisms (Fig. 3D).³⁸ Without energy transfer, biotinylating conferred tumor targeting capability and facilitated the electron transfer process of type I PDT, producing large amounts of O₂˙⁻. Using ferrocene (Fc) as the external standard, cyclic voltammetry studies found that PS1-Biotin exhibited a lower reduction potential compared with the original PS1, which was consistent with the strong electron acceptance characteristics of type I PS (Fig. 3E). The anodic shift favoured electron acceptance by PS1-Biotin, and O₂˙⁻ was produced by the type I process. Thus, the incorporation of electron-enriched substrates was beneficial for achieving an efficient PDT mechanism.

2.2. Carbazole derivatives

Carbazoles, which are TADF groups, contain dense electron donors at the 3,6- or 2,7-positions. The ring acted as an ingenious scaffold to immobilize different groups through intramolecular conjugation.^39,40 Therefore, based on the advantages of easy modification, high stability and good hole transport ability, the carbazole ring is often connected to functional molecules and constructed TADF molecules with multiple functions. In 2016, Zhang et al. first combined a series of carbazole-based TADF molecules into the PDT field, including 4,5-bis(carbazol-9-yl)-1,2-dicyanobenzene (2CzPN), 2,4,5,6-tetrakis (carbazol-9-yl)-1,3-dicyanobenzene (4CzTPN), and 2,3,5,6-tetrakis (3,6-diphenyl- carbazol-9-yl)-1,4-dicyanobenzene (4CzTPN-Ph) (Fig. 3F).⁴¹ The strategy expanded π-conjugation by changing the receptor cardinality. Under photoexcitation, all three TADF molecules demonstrated a significant growth trend of ¹O₂ (Fig. 3G). This result formally verifies the feasibility of TADF materials for PDT applications.

Song et al. modified cyanine PS with a carbazole group to obtain the D–A–D′ configuration, named car-XCy (Fig. 3H), where quaternary ammonium and xanthene acted as the electron acceptor and donor, respectively.⁴² The configuration changes affected the intramolecular charge transfer (ICT) and modulated the excited state properties. The large rigid conjugate structure adjusted the molecule aggregation property, extended the T₁ lifetime from 0.31 μs (XCy) to 29.77 μs (car-XCy) and achieved PDT (Fig. 3I and J). The introduction of carbazole groups changed the molecular structure and electronic distribution, effectively optimized the excited characteristics and prolonged the T₁ lifetime, thereby strengthening the ROS generation.

As a carbazole derivative, 9-phenylcarbazole (PCZ) with high triple exciton promotion efficiency is widely used in building organic materials with unique electronic or photonic properties. Barman et al. synthesized PS with D–π–A configuration, named BTMCz, to achieve multi-application integration (Fig. 3K).⁴³ The introduction of carbazole conferred TADF properties to BTMCz, resulting in a fluorescence delay lifetime of 74.6 μs (Fig. 3L). Moreover, carbazole provided BTMCz with strong electronic strength, accelerated the transition of triplet excitons to single excitons, and promoted ROS generation, making it a potential luminescent agent in PDT (Fig. 3M).

2.3. Quinone derivatives

Studies have shown that 9,10-anthraquinone (AQ), a strong electron acceptor with two carbonyl groups, was fully utilized in TADF material development.^44–46 The AQ group is equipped with good power absorption capacity and a significant SOC effect, promoting the generation of triplet excitons by triggering the multichannel ISC. Moreover, AQ, with its unique redox cycling activity, undergoes an n–π* transition in the excited state, promoting the ˙OH generation.^47–49 As new receptors for PS design, AQ derivatives have attracted attention for their excellent ability to generate type I ROS. However, AQ was prone to π–π stacking in physiological environments owing to its rigid and planar structure. Recently, research has begun to explore whether PSs based on AQ receptors can improve PDT performance.

To explore the mechanism of type I ROS, Zhang et al. designed an AQ-cantered PS, namely tBuT2AQ, for tumor-targeted two-photon PDT (Fig. 4A).⁵⁰ The strong D–A interaction between the triphenylamine donor and AQ endowed tBuT2AQ with efficient ICT and small ΔE_ST, which contributed to the generation of the T₁ state. The significant SOC effect induced by the rapid n–π* transition originating from AQ further ensured a competent ISC process. The high SOC value (ξ) of tBuT2AQ in the theoretical calculation further supports the above theory (Fig. 4B). The redox cycling behavior mediated by the carbonyl group allowed the triplet exciton to further undergo the electron transfer process. From the perspective of the receptor, AQ stably operates redox reactions with semiquinone anion radicals (AQ⁻) and 9,10-dihydroanthracene-9,10-diol (AQH2) while trapping and releasing electrons. The C=O bond was deemed the active site for the redox reaction. In detail, the triplet PS acquired an electron from the substrate to generate a semiquinone anion radical (tBuT2AQ˙⁻). Then, the electron transfer between tBuT2AQ⁻ and triplet O₂ generated O₂˙⁻, which further formed H₂O₂ through dismutase catalysis or the one-electron reduction of tBuT2AQ˙⁻. The two separated cathode waves depicted in Fig. 4C indicated that AQ and tBuT2AQ underwent two consecutive one-electron reductions, verifying the formation of tBuT2AQ˙⁻. Additionally, the combination of the AQ properties and the planarity of the molecular structure enabled tBuT2AQ to be excited by two photons at 940 nm, making it potentially suitable for deep tissue therapy. This work encouraged PS design through follow-up responses to expand the application of PDT in deep diseases.

Fig. 4 (A) Schematic of the molecular design philosophy and application of tBuT2AQ. (B) Calculated energy gap and related spin–orbit coupling constants (ξ) of tBuT2AQ. (C) Cyclic voltammograms of AQ and tBuT2AQ in DMF at a scan rate of 100 mV s⁻¹.⁵⁰ Reproduced with permission. Copyright 2024, American Chemical Society. (D) Schematic of the molecular structures. (E) Redox cycling of PQ. Relative PL intensities of (F) SOSG and (G) HPF. (H) ESR signals of BMPO in PQ-TPA, PQ-TPAOC1, PQ-TPAOC4, and PQ-TPAOC8.⁵¹ Reproduced with permission. Copyright 2021, American Chemical Society.

9,10-Phenanthrenevvquinone (PQ) performs similarly to the AQ receptor. Guo et al. designed a series of PSs based on the PQ receptor for synergistic PDT/photothermal therapy (PTT) treatment (Fig. 4D).⁵¹ To achieve efficient generation of type I ROS and excellent photothermal conversion ability, methoxy/alkoxy chains were introduced into the donor to further modulate ICT effects and intermolecular distances. As depicted in Fig. 4E, in the presence of NAD(P)H quinone oxidoreductase (NQO1) or aldehyde-ketone reductase (AKR), PQ was reduced by two electrons and formed 9,10-phenanthrene hydroquinone (PQH2). Then, PQH2 was oxidised to PQ˙⁻ by H₂O₂, forming a redox cycle that generated type I ROS. 9,10-Anthracenediyl bis(methylene) dimalonic acid (ABDA), singlet oxygen sensor green (SOSG), and hydroxyphenyl fluorescein (HPF) were utilized to verify the nature of the ROS generated from the PSs (Fig. 4F–H). These results confirmed that the series of PSs produced only type I ROS. Thus, this work has broadened the horizon for the development of high-performance type I PDT nanoagents. Recently, Wang et al. used PQ as the acceptor and triphenylamine (TPA)/PCZ as the donor to design type I PSs in a D–A-D configuration, named TPQ and PPQ, respectively (Fig. 5A).⁵² As illustrated in Fig. 5B, multiple energy transition paths between the S₁ and T_n were presented, which aimed to enhance the population of triplet excitons. Furthermore, the SOC effect between S₁ and T_n accelerated the spin flip of the excited states. As shown in Fig. 5C, the TPQ and PPQ acted as type I PSs when activated, and the PPQ showed a higher ˙OH generation. Analysing the SOC path depicted in Fig. 5D, the stronger π–π interaction between PPQ molecules and energy conversion through multiple paths (S₁–T₁, S₁–T₂ and S₁–T₃), achieved synergistic SOC, enhanced the triplet exciton number, and led to a superior ˙OH generation rate compared to TPQ.

Fig. 5 (A) Molecular structures of TPQ and PPQ. (B) Photophysical and photochemical mechanisms of type-I processes and antitumor application of type-I PS PPQ. (C) Calculated frontier orbital amplitude plot of PPQ. (D) PL intensity change (I–I₀) of HPF (10 μM) in the presence of 10 μM TPQ NPs and PPQ NPs in PBS.⁵² Reproduced with permission. Copyright 2023, Elsevier B.V. (E) Transient phosphorescence decay curve of NIR-TADF NPs. (F) Theoretical geometric calculation of APDC-DTPA. (G) Fluorescence intensity increase in SOSG at 525 nm upon light irradiation with or without the NIR-TADF NPs.⁵³ Reproduced with permission. Copyright 2022, Wiley-VCH. (H) Molecular structure of FO and time-resolved photoluminescence decay curves of FO measured under nitrogen atmospheres. (I) Decay kinetic curves of FO at 822 nm. (J) EPR spectra of FO (0.1 mM in toluene) for ¹O₂ characterization with 2,2,6,6-tetramethyl-4-piperidone (TEMP, 50 mM).⁵⁴ Reproduced with permission. Copyright 2025, American Chemical Society.

2.4. Other molecular design

In 2022, Fang et al. first developed metal-free PSs with two-photon excitation (TPE) characteristics, named APDC-DTPA.⁵³ Attributed to the introduction of acenaphtho[1,2-b] pyrazine-8,9-dicarbonitrile (APDC) acceptor, APDC-DTPA possessed a TADF property that enabled it to exhibit a delayed fluorescence lifetime of 5.8 μs under hypoxic conditions (Fig. 5E). As shown in Fig. 5F and G, HOMO and LUMO achieved effective separation and made an effort for a narrow ΔE_ST value (0.10 eV), which favoured the ISC process to obtain outstanding ¹O₂ generation. This molecular strategy expanded the design direction of pure organic TADF-based PSs and drove progress towards high performance PDT.

Gong et al. employed the extended π-conjugated strategy and developed an arc-shaped oligomer BODIPY with a fully fused thiophene ring, named FO, consisting of eight BODIPY units.⁵⁴ Owing to the heavy atom effect of the sulphur element in thiophene, the SOC of FO was significantly enhanced, and the matrix element (T₁H_SOCS₁) was amplified, combined with its relatively small ΔE_ST, achieving a high ISC efficiency of 24%. The delay in the fluorescence lifetime of FO was 0.54 μs, providing conclusive experimental evidence for the TADF property (Fig. 5H). The kinetic decay curve of FO in toluene showed that the lifetime of its long triplet state reached 0.61 μs, which was attributed to the RISC (Fig. 5I). The octamer FO achieved a high fluorescence quantum yield of 32% at 829 nm. When FO was irradiated in the presence of TEMP (a specific spin agent), a strong EPR signal confirmed the generation of ¹O₂ during the irradiation process (Fig. 5J). The extended π-conjugation structure provides an effective method for regulating the inherent properties of TADF materials with rigid conjugation.

3. Function and regulation of TADF materials for PDT

For conventional organic PSs, excessively large ΔE_ST values (> 0.3 eV) imposed severe kinetic barriers to ISC and RISC, effectively trapping ∼75% of triplet excitons in non-productive T₁. This thermodynamic bottleneck squandered excited state energy and limited the effective ROS quantum yield, necessitating prolonged irradiation times and thus exacerbating photo-bleaching.⁵⁵

The resultant “triplet-to-singlet conversion” mechanism enabled near-unity RISC rates (k_RISC > 10⁻⁶), allowing dynamic repopulation of singlet states for iterative ISC-RISC cycling. This triplet exciton recycling paradigm achieved unprecedented triplet utilization efficiencies and surpassed the theoretical limit for conventional fluorescence. Crucially, the enhanced SOC has two advantages: orders-of-magnitude amplification of ROS quantum yields, and intrinsic resistance to photobleaching due to suppressed non-radiative decay pathways. The convergence of ultra-low ΔE_ST and metal-free SOC enhancement properties catalysed a paradigm shift in PS design.^56,57 These advances underscored TADF's potential to transcend its origins in photonics, offering a versatile molecular toolkit for precision theranostics that harmonized therapeutic potency, biosafety, and diagnostic functionality.

The unparalleled versatility of TADF materials lies in their inherently tunable photophysical properties, which can be rationally engineered through molecular-level structural modifications to meet diverse therapeutic and diagnostic demands. This adaptive design flexibility has positioned the TADF as a cornerstone for next-generation PDT. Until now, researchers have systematically synthesized multiple TADF-based PSs with tailored properties. Such innovations underscored the critical role of structure-performance correlations in advancing PDT efficacy.⁵⁸ In the subsequent sections, we categorized state-of-the-art TADF-based PSs based on different principles, systematically analyzed the structure–property relationships behind these enhanced functions and elucidated how molecular engineering breakthroughs had redefined the therapeutic ceiling of PDT.

3.1. Facilitation of energy conversion for triplet excitation

The CT state formed between electron donors and acceptors inherently exhibited a diminished ΔE_ST owing to its spatially separated frontier molecular orbitals. Crucially, the spatial arrangement between donor and acceptor units played a pivotal role in modulating CT state formation. Introducing a significant steric hindrance between these moieties facilitated the preferential stabilization of the CT state while suppressing competing localized excited states (LE), thereby minimizing ΔE_ST.⁵⁹ To optimize this effect, molecular design strategies should focus on pairing strong electron donors with strong acceptors to establish robust ICT. For structural modifications, such as adopting non-planar molecular geometries or incorporating sterically bulky substituents, the torsion angle between the donor and acceptor units is further enhanced, effectively suppressing LE-state interference and promoting the dominance of the CT state.

By integrating a strong electron-donating TPE unit with a robust electron-accepting AQ moiety, Hu et al. developed a novel TADF emitter, 2-(4-(triphenyl ethenyl)phenyl)anthraquinone (TPE-AQ), which exhibited an exceptionally small ΔE_ST of 0.19 eV (Fig. 6A).⁶⁰ Upon molecular self-assembly into well-defined nanostructures, the resulting TPE-AQ-based nanoparticles (NPs) demonstrated enhanced photosensitizing performance. Remarkably, under 450 nm laser irradiation (100 mW cm⁻², 5 min), these NPs achieved efficient ¹O₂ generation with a quantum yield comparable to conventional PSs at a low dosage (40 mg mL⁻¹) while exhibiting negligible collateral tissue damage (Fig. 6B). The superior cancer cell ablation efficacy was attributed to the synergistic effects of TADF-mediated ISC enhancement and precise energy transfer. This study established a paradigm for designing TADF-based nanoplatforms with optimized PDT performance.

Fig. 6 (A) Synthetic routes to TPE-AQ molecular geometries. (B) Decomposition rate constant of ADPA by Rose bengal, TPE-AQ NPs and water.⁶⁰ Reproduced with permission. Copyright 2021, the Royal Society of Chemistry. (C) Torsion angle results of oxa-XCy, thia-XCy, and car-XCy. (D) Calculation of ¹O₂ quantum yield and (E) flash photolysis results of XCys. (F) Illustration of the immunostaining analysis of relevant markers in cells treated with car-XCy NPs after NIR irradiation (660 nm, 20 mW cm⁻², and 10 min).⁴² Reproduced with permission. Copyright 2024, Wiley-VCH.

Song et al. designed a series of TADF molecules by strategically integrating quaternary ammonium structures as electron acceptors and xanthene derivatives as electron donors, with carbazole, phenothiazine, and phenoxazine serving as modifying groups for the xanthene-based cyanine PSs.⁴² This molecular engineering approach systematically modulated both spatial geometry and electronic structure, yielding three D–A–D* type TADF molecules (car-XCy, thia-XCy, and oxa-XCy) (Fig. 6C). Notably, structural modifications significantly reduced the ΔE_ST across all derivatives, where car-XCy exhibited the smallest ΔE_ST value. This reduction correlated with its minimized torsional angle (53.8°) between the modifying group and parent structure, facilitating enhanced S₁–T₁ state transition efficiency and subsequent ROS generation (Fig. 6D). The flash photolysis was further conducted to shed light upon the affected excited state lifetime after structure modification, where car-XCy demonstrated a remarkable 29.77 μs lifetime approximately 100-fold longer than unmodified XCy (0.31 μs) (Fig. 6E). This dramatic enhancement was attributed to the introduction of a rigid-conjugated structure. The synergistic effects of optimized spatial configuration and electronic modulation endowed car-XCy with both minimized ΔE_ST and a prolonged excited-state lifetime. Intriguingly, the combined advantages of superior ROS generation and endoplasmic reticulum (ER)-targeting capability in car-XCy induced robust ER stress, subsequently activating immunogenic cell death (Fig. 6F).

Complementary to steric engineering, optimizing the electron push–pull interplay between donor and acceptor could make an effort to reduce ΔE_ST. Rational alignment of the donor's HOMO with the acceptor's LUMO enhanced orbital coupling and strengthened CT efficiency. Additionally, ΔE_ST arose predominantly from exchange energy, which correlated with the spatial overlap of HOMO and LUMO orbitals.⁶¹ By strategically designing spatially segregated D–A systems with minimal orbital overlap, exchange energy contributions could be significantly reduced, leading to a further reduction in ΔE_ST.

Based on this design principle, Wang et al. developed a novel type I PS PPQ through rational molecular engineering by employing a spatially separated D–A strategy with PCZ as the conventional TADF donor and PQ (Fig. 7A).⁵² Compared with TPQ, the HOMO and LUMO orbitals of PPQ were completely decoupled, and the spatial separation significantly reduced ΔE_ST (Fig. 7B). Moreover, in the Natural Transition Orbitals (NTO) analysis, there were multiple energy conversion pathways between the S₁ state and different T_n states of PPQ, ensuring the stable transfer of photons to triplet excitons (Fig. 7C). This multi-channel ISC mechanism by PPQ ensured robust population transfer to triplet states, thereby optimizing photodynamic efficiency. Furthermore, the ΔE_S1-T2 value of PPQ was extremely as small as 0.12 eV, which achieved more efficient ISC conversion and enhanced the generation of triplet excitons. Thus, PPQ exhibited higher efficiency in promoting triplet excitons compared with TPQ. The synergistic combination of orbital decoupling and enhanced ISC pathways established PPQ as a promising candidate for precision PDT, demonstrating how strategic molecular design could manipulate both electronic structures and excited-state dynamics to achieve superior therapeutic performance. Zhang et al. strategically designed and synthesized two multifunctional TADF PSs, An-TPA and An-Cz-Ph, employing a molecular engineering approach focused on minimizing orbital overlap between the HOMO and LUMO (Fig. 7D).⁶² As illustrated in Fig. 7E, both molecules adopt non-planar conformations with spatially separated HOMO–LUMO distributions. Comparative analysis revealed that An-TPA exhibited a significantly smaller ΔE_ST than An-Cz-Ph, attributed to two synergistic factors: (1) the TPA unit served as a more electron-rich donor compared to the carbazole derivative, and (2) its enhanced orbital delocalization further promoted the spatial separation of frontier molecular orbitals. This optimized electronic configuration effectively facilitated efficient ISC and subsequent triplet-state energy conversion. The distinct structure–property relationships demonstrated in this study provided valuable insights for the rational design of TADF-based PSs, highlighting the importance of donor engineering and orbital decoupling strategies in advancing PDT efficacy. PSs based on TADF demonstrated exceptionally efficient ISC processes and superior photosensitization capabilities, making them highly attractive for PDT applications. Notably, ΔE_ST could be further minimized by enhancing the CT state and reducing the exchange energy through meticulous molecular design. This strategic optimization provided novel opportunities for an efficient triplet-state population and advanced the development of adequate ROS-mediated PDT, offering promising prospects for enhanced therapeutic efficacy.

Fig. 7 (A) Schematic of the molecular structures of TPQ and PPQ. (B) Calculated frontier orbital amplitude plots of TPQ and PPQ. (C) Calculated energy gaps and ξ of PPQ (left) and TPQ (right).⁵² Reproduced with permission. Copyright 2019, American Chemical Society. (D) Synthesis routes for An-TPA and An-Cz-Ph. (E) Molecular geometries and HOMO–LUMO distributions of An-TPA and An-Cz-Ph.⁶² Reproduced with permission. Copyright 2025, Wiley-VCH.

3.2. Dynamic equilibrium between ISC and RISC

Molecular engineering strategies enabled photosensitizing capability, target recognition, and real-time imaging functions to be integrated within single molecules, thereby overcoming limitations of diagnosis-therapy separation inherent to conventional treatments.⁶³ However, in traditional organic PSs, imaging functionality relied on radiative transitions while ROS generation depended on ISC rates. Owing to competing energy decay pathways, radiative and non-radiative transitions exhibited an inverse relationship. This necessitated sacrificing partial imaging capacity for efficient ROS production, making it difficult to achieve the integration of diagnosis and therapy at the single-molecule level. Additionally, conventional PSs face challenges, including poor reproducibility, suboptimal photosensitization efficiency, and inadequate imaging contrast, which hinder clinical translation.

In contrast, the emergence of the TADF provided a viable solution to this limitation. In TADF materials, minimized ΔE_ST coupled with long-lived triplet excitons ensured potent ROS generation, while modulable delayed fluorescence emission delivered robust optical signals for real-time tracking.⁶⁴ These synergistic photophysical properties positioned TADF-based materials as ideal candidates for single-molecule theranostics. By establishing ISC/RISC equilibrium through molecular engineering, ROS yields were enhanced while maintaining satisfactory imaging performance, thereby maximizing molecular theranostic efficacy. This advancement has created unprecedented opportunities for the clinical translation of PDT. In contrast, the emergence of the TADF provided a viable solution to this limitation. In TADF materials, minimized ΔE_ST coupled with long-lived triplet excitons ensured potent ROS generation, while modulable-delayed fluorescence emission delivered robust optical signals for real-time tracking.⁶⁵ These synergistic photophysical properties positioned TADF-based materials as ideal candidates for single-molecule theranostics. By establishing ISC/RISC equilibrium through molecular engineering, ROS yields were enhanced while maintaining satisfactory imaging performance, thereby maximizing molecular theranostic efficacy. This advancement has created unprecedented opportunities for the clinical translation of PDT. Therefore, in this section, we summarize the means of achieving single-molecule diagnostic and therapeutic functions through molecular engineering.

Incorporating heavy-atom substitution into TADF systems offered two advantages: (1) reduced ΔE_ST and (2) enhanced SOC, thereby promoting ISC and ROS generation. Building on this principle, Xiao et al. designed two TADF PSs, AQCz and its brominated derivative AQCzBr₂, both exhibiting ultralow ΔE_ST values (0.11 eV).⁶⁶ Crucially, bromine substitution induced a 2.8-fold increase in SOC and an 8-fold enhancement in k_ISC compared to the parent molecule AQCz, resulting in significantly improved ISC efficiency and ROS quantum yield (Fig. 8A). The heavy-atom engineering strategy achieved a synergistic balance between the ISC and RISC processes, enabling a 3.4-fold boost in ROS production while preserving intrinsic TADF characteristics (Fig. 8B). This work demonstrated that integrating heavy atoms into TADF scaffolds with inherently small ΔE_ST values could circumvent the traditional trade-off between ROS generation efficiency and TADF emission performance. Such molecular design principles provide a robust framework for developing multifunctional phototheranostic agents capable of simultaneous high-contrast imaging and superior therapeutic efficacy.

Fig. 8 (A) Exciton dynamics in TADF molecules. (B) DCFH activated rates of AQCz NPs, AQCzBr₂ NPs and MB with the same area of integral absorption.⁶⁶ Reproduced with permission. Copyright 2021, the Royal Society of Chemistry. (C) Structure of compound FL-RGD. (D) Normalized emission spectra of FL and FL-RGD in deaerated ethanol. (E) Frontier molecular orbital plots of FL and FL-RGD. (F) Depth imaging of FL-RGD in tumor tissue slices with two-photon or one-photon excitation. Scale bar = 30 μm. (G) Phototoxicity of FL-RGD. Error bars indicate mean ± SD (n = 6).⁶⁷ Reproduced with permission. Copyright 2019, Elsevier B.V.

In addition to the introduction of heavy atoms, the separation of HOMO and LUMO was also an effective means of achieving a narrow ΔE_ST and optimizing the ISC/RISC balance. More importantly, despite significant advancements in TADF-based cancer theranostics, achieving tumor-specific diagnosis and treatment remains a persistent challenge owing to the limited targeting specificity of conventional TADF PSs. Addressing this critical limitation, Liu et al. engineered a dual-targeting theranostic agent (FL-RGD) by covalently conjugating cyclic arginine-glycine-aspartate (cRGD) tripeptide to a fluorescein-derived TADF emitter (Fig. 8C).⁶⁷ The TADF nature of FL-RGD was confirmed by its time-resolved fluorescence spectrum, which aligned with its steady-state spectrum (Fig. 8D). The DFT calculations revealed spatially separated HOMO and LUMO orbitals, indicating minimized ΔE_ST (Fig. 8E). Crucially, the integration of cRGD did not perturb the TADF properties of FL, as evidenced by the nearly identical normalized steady-state/time-resolved spectra and DFT profiles between FL and FL-RGD. The intravenous administration of FL-RGD in tumor-bearing mice enabled two-photon confocal imaging of tumor sections with exceptional depth penetration (40–160 μm) and high signal-to-noise ratio (SNR > 15), demonstrating its potential for deep-tumor visualization (Fig. 8F). Concurrently, FL-RGD exhibited potent photodynamic activity, achieving a remarkably low half-maximal inhibitory concentration (IC₅₀ = 5.55 μM) under irradiation (Fig. 8G). This therapeutic potency correlated with its long T₁ lifetime and efficient ¹O₂ generation, which were facilitated by the narrow ΔE_ST that optimized the ISC/RISC equilibrium. The separated HOMO and LUMO achieved a balance between the ISC and RISC, allowing FL-RGD to achieve precise tumor ablation by controlling ¹O₂ generation while maintaining high-contrast deep tissue imaging. This work established a groundbreaking advancement in molecular theranostics, demonstrating how the rational integration of tumor-targeting ligands with TADF photophysics could overcome the specificity limitations of conventional PSs. The FL-RGD system represents a paradigm-shifting approach to image-guided PDT, offering dual functionality without compromising either imaging resolution or therapeutic efficacy.

Barman et al. demonstrated a groundbreaking strategy for fine-tuning k_RISC through donor engineering and second-order spin-vibronic coupling mechanisms.⁴³ Unlike the small ΔE_ST value, if RISC achieves an effective SOC, it is not conducive to efficient TADF owing to the purely CT characteristics and the large S₁ and T₁ states. In contrast, if second-order spin-vibronic coupling is given priority consideration, the S₁ and higher T₂ states will be closely adjacent, facilitating the occurrence of RISC. By systematically incorporating carbazole, phenothiazine (PTZ), and phenoxazine (PXZ) donors, they synthesized three TADF PSs (BTMCz, BTMPTZ, and BTMPXZ) with distinct photophysical behaviors (Fig. 9A). The carbazole-containing BTMCz exhibited superior TADF characteristics, including high PLQY (68%), and enhanced PDT efficiency compared to PTZ/PXZ-based analogs. This enhancement stemmed from the rigid carbazole donor's ability to suppress non-radiative decay while optimizing k_RISCvia second-order spin-vibronic coupling (Fig. 9B–D). Meanwhile, aggregation-induced self-assembly yielded monodisperse NPs with controlled size (80–120 nm) (Fig. 9E–G). These nanostructures demonstrated improved cellular uptake efficiency (1.8-fold increase vs. molecular counterparts) and high-contrast imaging capability, addressing critical challenges in bioimaging-guided PDT (Fig. 9H). Experiments have shown that by regulating the donor intensity, an ISC/RISC balance can be achieved while maximizing the imaging and therapeutic effects, allowing BTMCz to achieve a satisfactory imaging level and maintain stable ¹O₂ generation. The synergistic integration of donor engineering, spin–orbit manipulation, and nanoscale self-assembly has opened new frontiers for developing next-generation theranostic platforms for image-guided tumor ablation.

Fig. 9 (A) Chemical structures of BTMCZ, BTMPTZ, and BTMPXZ. (B)–(D) Schematic triplet harvesting routes and obtained excited state kinetics for BTMCZ, BTMPTZ, and BTMPXZ. (E)–(G) Nanoaggregate visualization by TEM. Scale bar = 1 μm. (H) Confocal images of MCF-7 cancer cells after incubation with BTMCz (10 μg mL⁻¹).⁴³ Reproduced with permission. Copyright 2025, Wiley-VCH.

3.3. Extended application in deep tumors

The photophysical dynamics of conventional PSs are intrinsically limited by penetration depth, severely restricting their utility in deep tumors. A primary challenge stemmed from the spectral mismatch between conventional TADF-based PSs and the optical transparency window of biological tissues.⁶⁸ This discordance resulted in suboptimal light penetration depths (< 2 mm), severely compromising therapeutic outcomes in deep-tissue malignancies.⁶⁸ NIR-responsive systems (700–1000 nm) offered a compelling solution through enhanced tissue penetrability (> 5 cm) and minimized phototoxicity.

Nevertheless, the development of NIR-activatable TADF PSs remains substantially underexplored, hindered by fundamental challenges in molecular design. These included stringent energy gap engineering constrained to balance narrow ΔE_ST with redshifted absorption, and the intricate interplay between SOC and RISC kinetics. To date, only a few NIR-excitable TADF PSs have been systematically characterized for PDT applications, underscoring the pressing need for innovative design paradigms. In addition, the emission wavelengths of existing TADF PSs were typically below 700 nm, making their imaging capabilities susceptible to autofluorescence and tissue scattering.

To overcome this limitation, Zhang et al. pioneered a breakthrough by developing two-photon activatable TADF PSs (PT and AT) through rational molecular engineering of ΔE_ST and oscillator strength (f) (Fig. 10A).⁶⁹ Compared with the donor 9,9-dimethyl-9,10-dihydroacridine of AT, the benzoxazine of PT exhibited stronger electron donor characteristics, which enabled it to achieve more efficient HOMO and LUMO separation from A. Therefore, the conjugation between D and A of PT was very weak, resulting in a small ΔE_ST value of 60 meV and f of 0.03, in favor of the ISC process of PDT. These results indicate that the enhancement of electronic supply capacity effectively reduces the ΔE_ST value while keeping A constant, thereby accelerating PDT. The TADF molecules were physically encapsulated by nanoprecipitation using amphiphilic DSPE-PEG2000 to form nanoparticles, namely PT NPs and AT NPs, to improve biocompatibility (Fig. 10B). Upon NIR (760–800 nm) two-photon excitation, both AT and PT NPs demonstrated bright emission, showcasing exceptional two-photon absorption and photostability (Fig. 10C and D). These attributes positioned them as promising candidates for deep-tissue NIR bioimaging and two-photon excitation-PDT. The systematic evaluation of their PDT efficacy revealed pronounced ROS accumulation and marked cytotoxicity in HeLa cells under an 800 nm femtosecond laser (Fig. 10E and F).

Fig. 10 (A) Molecular structures, configuration and the calculated spatial distribution of the HOMO and LUMO and the value of ΔE_ST and f of AT and PT. (B) Schematic of the nanoprecipitation for nanoparticle preparation. TPE emission spectra of (C) AT NPs (0.1 mg mL⁻¹) and (D) PT NPs (0.1 mg mL⁻¹) under 760, 770, 780, 790, and 800 nm femtosecond laser excitation. (E) Fluorescence images of DCFH-DA in HeLa cells. Scale bar = 20 μm. (F) Photo-cytotoxicity of PT NPs and AT NPs.⁶⁹ Reproduced with permission. Copyright 2020, the Royal Society of Chemistry. (G) Schematic of the fabrication of the NIR-TADF NPs via self-assembly for efficient TPE-PDT. (H) Normalized absorbance and FL spectra of free APDC-DTPA molecules in THF and the NIR-TADF NPs in aqueous medium, respectively. (I) Cellular uptake and SPE fluorescence imaging of the NIR-TADF NPs in A549 cells.⁵³ Reproduced with permission. Copyright 2025, the Royal Society of Chemistry.

Similarly, the above-mentioned NIR-TADF NPs developed by Fang et al. also possessed two-photon excitation capability (Fig. 10G).⁵³ Benefiting from the π–π* interaction of the conjugated backbone and the charge transfer between D and A, NIR-TADF NPs were endowed with the ability of two-photon absorption. Under two-photon excitation with femtosecond NIR pulses (880–1060 nm), the NPs exhibited strong NIR emission extending to 1000 nm in aqueous environments, accompanied by an absolute photoluminescence quantum yield (PLQY) of 3.4% (Fig. 10H). What is more, cellular studies revealed distinct red fluorescence signals in NIR-TADF NP-treated cells under two-photon irradiation (Fig. 10I), confirming their exceptional two-photon fluorescence imaging potential. The observed emission (650–1000 nm) effectively bypasses biological autofluorescence, achieving subcellular resolution in deep-tissue simulations. This work established a paradigm for designing NIR-responsive TADF systems, simultaneously overcoming the spectral limitations of conventional TADF agents and preserving their inherent advantages in a triplet-state population. The integration of two-photon excitation with NIR-TADF emission has opened new avenues for precision theranostics, particularly in deep-tissue PDT and high-contrast bioimaging applications.

4. Conclusions

PDT, as a minimally invasive therapeutic modality, has garnered substantial attention in precision oncology. However, conventional organic PSs face critical limitations that hinder their clinical translation, including suboptimal ROS quantum yields and compromised imaging fidelity, owing to tissue autofluorescence interference. These intrinsic drawbacks stem from inefficient triplet exciton utilization and rapid non-radiative decay pathways inherent to traditional fluorophores. In this context, TADF emitters have emerged as a revolutionary class of photonic materials, offering unparalleled advantages through their unique RISC mechanisms. Characterized by exceptionally small ΔE_ST and rationally tunable ISC/RISC rates, TADF materials enable amplified ROS generation capacity via enhanced triplet population dynamics. Moreover, through clever structural design, the absorption/emission of TADF materials is red-shifted to the NIR region, which provides a reasonable solution for the ablation of deep-seated tumors and establishes TADF as an ideal platform for theranostic integration.

Despite these theoretical advantages, the exploration of TADF materials in PDT remained largely unexplored until seminal breakthroughs in molecular design strategies around 2010 catalyzed systematic investigations. This review presents a comprehensive taxonomy of TADF-based PSs developed for PDT applications, categorizing them according to their molecular engineering paradigms. Nevertheless, the development of TADF-PSs for biomedical applications remains in its infancy, facing multifaceted challenges that demand urgent resolution.

The clinical translation of TADF-PSs confronts three fundamental bottlenecks. First, the inherent hydrophobicity of conjugated organic systems leads to poor aqueous dispersibility, resulting in nonspecific biodistribution and limited tumor accumulation efficiency. Second, most current TADF-PSs operate in the visible spectral range (400–650 nm), suffering from shallow tissue penetration depths that restrict efficacy against deep-seated malignancies. Third, the complex photophysical interplay among the molecular structure, ISC/RISC kinetics, and microenvironmental factors (e.g., oxygen tension, pH, and biomolecular interactions) remains poorly understood, hampering the rational design of tumor-selective agents. Although emerging strategies, such as PEGylation, NP encapsulation, and π-extension for red-shifted absorption, have shown promise, current solutions remain inadequate for addressing the stringent requirements of clinical PDT.

To bridge these gaps, future studies must prioritize three strategic directions: (1) employing quantum chemical calculations to establish quantitative structure–property relationships governing ΔE_ST modulation, ISC/RISC dynamics, and ROS generation pathways; (2) developing more NIR-II responsive (1000–1700 nm) TADF systems; and (3) implementing rigorous preclinical assessments of pharmacokinetics, biodegradation pathways (e.g., hepatic vs. renal clearance), and long-term biosafety. The convergence of molecular photonics and nanomedicine holds transformative potential for next-generation TADF-based PSs. By establishing standardized evaluation protocols and fostering interdisciplinary collaboration, we anticipate that TADF-based theranostics will overcome current limitations and usher in a new era of image-guided precision PDT. This paradigm shift could ultimately redefine clinical standards for non-invasive cancer treatment, offering hope for patients with deep tissue or metastatic malignancy refractory to conventional therapies.

Conflicts of interest

There are no conflicts to declare.

Data availability

No primary research results, software or code have been included, and no new data were generated or analyzed as part of this review.

Acknowledgements

The authors are thankful for financial support from the National Natural Science Foundation of China (22205159). This work was partially supported by the Hebei Natural Science Foundation (B2024110008, B202411033) and the research funding provided by the Cangzhou Institute of Tiangong University (TGCYY-F-0305).

References

1. J. H. Correia, J. A. Rodrigues, S. Pimenta, T. Dong and Z. Yang, Pharmaceutics, 2021, 13, 1332.
2. Y. Yu, H. Wang, Z. Zhuang, C. Ji, L. Zhang, Y. Li, Z. Zhao and D. Ding, ACS Nano, 2024, 18, 13019.
3. X. Xiong, J. Liu, L. Wu, S. Xiong, W. Jiang and P. Wang, Coord. Chem. Rev., 2024, 510, 215863.
4. A. Soleimany, D. K. Aghmiouni, M. Amirikhah, M. A. Shokrgozar, S. Khoee and B. Sarmento, Adv. Funct. Mater., 2024, 34, 2408594.
5. Y. Li, M. Tian, T. Yang, J. Cao, H. Chen, J. Guo, P. Liu and Y. Liu, Nano Res., 2024, 17, 5501.
6. A. Grebinyk, O. Chepurna, M. Frohme, J. Qu, R. Patil, L. O. Vretik and T. Y. Ohulchanskyy, J. Photochem. Photobiol., C, 2024, 58, 100652.
7. J. F. Berbigier, L. C. Luz and F. S. Rodembusch, Chem. Rec., 2024, e202400146.
8. Y. Li, G. Li, Q. zhang, Y. Li, Q. Jia, W. Zhang, X. Feng, W. Xu and J. Liu, Chem. Phys. Lett., 2021, 784, 139091.
9. S. Scoditti, F. Chiodo, G. Mazzone, S. Richeter and E. Sicilia, Molecules, 2022, 27, 4046.
10. C. Kim, H. Kim, J. Jo, S. Kim, A. M. Bongo, H. J. Kim and J. Yang, ACS Appl. Bio Mater., 2024, 7, 8294.
11. J. Chen, K. Wen, H. Chen, S. Jiang, X. Wu, L. Lv, A. Peng, S. Zhang and H. Huang, Small, 2020, 16, 2000909.
12. M. Yang, S. Wang, X. Ou, J. Ni, S. Segawa, J. Sun, F. Xu, R. T. K. Kwok, J. Zhao, J. W. Y. Lam, G. Jin and B. Z. Tang, ACS Nano, 2024, 18, 30069.
13. Z. Xu, Y. Ding, M. Jia, Y. Gao, Y. Zhu, C. You, Z. Wu, J. Liu, Y. Pan, D. Yan, J. Zhu, W. Hu, D. Wang and B. Z. Tang, Small, 2025, 2502354.
14. R. Tang, S. Xu, L. Du, F. F. Hung, T. L. Lam, G. Cheng, K. H. Low, Q. Wan, S. Wu, Y. Chen and C. M. Che, Adv. Opt. Mater., 2023, 11, 2300950.
15. F. B. Dias, J. Santos, D. R. Graves, P. Data, R. S. Nobuyasu, M. A. Fox, A. S. Batsanov, T. Palmeira, M. N. B. Santos, M. R. Bryce and A. P. Monkman, Adv. Sci., 2016, 3, 1600080.
16. F. Xie, J. Zhou, Y. Li and J. Tang, J. Mater. Chem. C, 2020, 8, 9476.
17. T. Wang, A. K. Gupta, S. Wu, A. M. Slawin and E. Z. Colman, J. Am. Chem. Soc., 2023, 145, 1945.
18. M. Sheokand, A. Z. Alsaleh, F. D. Souza and R. Misra, J. Phys. Chem. B, 2023, 127, 2761.
19. C. Zhou, W. Chen, H. Liu, X. Cao, N. Li, Y. Zhang, C. Lee and C. Yang, J. Mater. Chem. C, 2020, 8, 9639.
20. X. Yang, G. I. N. Waterhouse, S. Lu and J. Yu, Chem. Soc. Rev., 2023, 52, 8005.
21. K. Naveen, P. Palanisamy, M. Chae and J. H. Kwon, Chem. Commun., 2023, 59, 3685.
22. X. Luo, X. Xiao and Y. Zheng, Chem. Commun., 2024, 60, 1089.
23. M. Lu, S. Liao, J. Li, Z. Yu, N. Zhao, Z. Xie, S. Chen, L. Dang and M. Li, Chin. Chem. Lett., 2025, 36, 110066.
24. Y. Zhang, T. S. Lee, J. M. Favale, D. C. Leary, J. L. Petersen, G. D. Scholes, F. N. Castellano and C. Milsmann, Nat. Chem., 2020, 12, 345.
25. S. Li, X. Tang, C. H. Ng, J. Y. Lim, W. K. Tang, W. C. Chen, Y. Huo, M. Ng, S. S. Chen, D. Zhang, L. Duan and M. C. Tang, Adv. Opt. Mater., 2025, 2402479.
26. Y. Tao, K. Yuan, T. Chen, P. Xu, H. Li, R. Chen, C. Zheng, L. Zhang and W. Huang, Adv. Mater., 2014, 26, 7931.
27. Y. Im, M. Kim, Y. J. Cho, J. A. Seo, K. S. Yook and J. Y. Lee, Chem. Mater., 2017, 29, 1946.
28. F. B. Dias, Philos. Trans. R. Soc., 2015, 373, 20140447.
29. X. Luo, F. Li, J. Zou, Q. Zou, J. Su, M. Mao and Y. Zheng, Adv. Opt. Mater., 2021, 9, 2100784.
30. H. Zhou, D. Zhang, M. Li and C. Chen, Angew. Chem. Int. Ed., 2022, 61, e202117872.
31. H. Xu, Q. Wang, J. Mao, B. Tong and Q. Zhang, Prog. Chem., 2024, 36, 393.
32. J. Zhang, W. Chen, R. Chen, X. Liu, Y. Xiong, S. V. Kershaw, A. L. Rogach, C. Adachi, X. Zhang and C. Lee, Chem. Commun., 2016, 52, 11744–11747.
33. X. Xiong, F. Song, J. Wang, Y. Zhang, Y. Xue, L. Sun, N. Jiang, P. Gao, L. Tian and X. Peng, J. Am. Chem. Soc., 2014, 136, 9590.
34. Y. Wu, F. Song, W. Luo, Z. Liu, B. Song and X. Peng, Chem. Photo. Chem., 2017, 1, 79.
35. M. Tian, W. Chen, Y. Wu, J. An, G. Hong, M. Chen, F. Song, W. Zheng and X. Peng, ACS Appl. Mater. Interfaces, 2022, 14, 12050.
36. Z. Liu, F. Song, W. Shi, G. Gurzadyan, H. Yin, B. Song, R. Liang and X. Peng, ACS Appl. Mater. Interfaces, 2019, 11, 15426.
37. W. Chen, Z. Wang, M. Tian, G. Hong, Y. Wu, M. Sui, M. Chen, J. An, F. Song and X. Peng, J. Am. Chem. Soc., 2023, 145, 8130.
38. J. An, S. Tang, G. Hong, W. Chen, M. Chen, J. Song, Z. Li, X. Peng, F. Song and W. Zheng, Nat. Commun., 2022, 13, 2225.
39. C. Feng, K. Zhang, B. Zhang, L. Feng, L. He, C. Chen and M. Li, Angew. Chem., Int. Ed., 2025, 64, e202425094.
40. T. Li, S. Zheng, P. I. Djurovich and M. E. Thompson, Chem. Rev., 2024, 124, 4332.
41. J. Zhang, W. Chen, R. Chen, X. Liu, Y. Xiong, S. V. Kershaw, A. L. Rogach, C. Adachi, X. Zhang and C. Lee, Chem. Commun., 2016, 52, 11744.
42. J. Song, H. Fang, X. Wang and W. Zhong, Small, 2024, 20, 2402439.
43. D. Barman, P. Rajamalli, A. P. Bidkar, T. Sarmah, S. S. Ghosh, E. Z. Colman and P. K. Iyer, Small, 2025, 21, 2409533.
44. L. Lv, K. Yuan, T. Zhao, H. Lia and D. Wang, New J. Chem., 2023, 47, 18972.
45. B. Huang, W. Chen, Z. Li, J. Zhang, W. Zhao, Y. Feng, B. Z. Tang and C. Lee, Angew. Chem., Int. Ed., 2018, 57, 12473.
46. Z. Deng, C. Huang, Y. Luo, J. He, L. Li, X. Pang, G. Huang and D. L. Phillips, Nat. Commun., 2025, 16, 258.
47. B. Huang, Z. J. Li, H. Yang, D. Hu, W. J. Wu, Y. Feng, Y. M. Sun, B. P. Lin and W. Jiang, J. Mater. Chem. C, 2017, 5, 12031.
48. B. Huang, W. C. Chen, Z. Li, J. Zhang, W. Zhao, Y. Feng, B. Z. Tang and C. S. Lee, Angew. Chem., Int. Ed., 2018, 57, 12473.
49. W. C. Chen, B. Huang, S. F. Ni, Y. Xiong, A. R. Rogach, Y. P. Wan, D. Shen, Y. Yuan, J. X. Chen, M. A. Lo, C. Cao, Z. L. Zhu, Y. Wang, P. F. Wang, L. S. Liao and C. S. Lee, Adv. Funct. Mater., 2019, 29, 1903112.
50. W. Zhang, X. Li, M. Kang, Z. Zhang, Y. Pei, M. Fan, D. Yan, Y. Zhang, C. Yang, G. Xu, D. Wang, Z. Xu and B. Z. Tang, ACS Mater. Lett., 2024, 6, 2174.
51. J. Guo, J. Dai, X. Peng, Q. Wang, S. Wang, X. Lou, F. Xia, Z. Zhao and B. Z. Tang, ACS Nano, 2021, 15, 20042.
52. W. Wang, Y. Liu, H. Wang, F. Lin, H. Huang, Z. Chen, Z. Yang, Z. Chi and X. Zhou, Dyes Pigments, 2023, 218, 111444.
53. F. Fang, Y. Yuan, Y. Wan, J. Li, Y. Song, W. Chen, D. Zhao, Y. Chi, M. Li, C. Lee and J. Zhang, Small, 2022, 18, 2106215.
54. Q. Gong, J. Shao, W. Li, X. Guo, S. Ling, Y. Wu, Y. Wei, X. Xu, X. Jiang, L. Jiao and E. Hao, J. Am. Chem. Soc., 2025, 147, 21041–21052.
55. A. Polgar and Z. Hudson, Chem. Commun., 2021, 57, 10675.
56. X. Xiao, K. Ye, M. Imran1 and J. Zhao, Appl. Sci., 2022, 12, 9933.
57. V. Nguyen, A. Kumarc, M. Lee and J. Yoon, Coord. Chem. Rev., 2020, 425, 213545.
58. S. Gao, S. Yu, Y. Zhang, A. Wu, S. Zhang, G. Wei, H. Wang, Z. Xiao and W. Lu, Adv. Fun. Mater., 2021, 31, 2008356.
59. Z. Li, J. Lu and X. Li, Chem. – Eur. J., 2024, 30, e202401001.
60. S. Hu, B. Huang, Y. Pu, C. Xia, Q. Zhang, S. Guo, Y. Wang and X. Huang, J. Mater. Chem. B, 2021, 9, 5645.
61. J. Berbigier, L. Luz and F. Rodembusch, Chem. Rec., 2024, 24, e202400146.
62. J. Zhang, F. Fang, B. Liu, J. Tan, W. Chen, Z. Zhu, Y. Yuan, Y. Wan, X. Cui, S. Li, Q. Tong, X. Meng, C. Lee and J. Zhao, ACS Appl. Mater. Interfaces, 2019, 11, 41051.
63. F. Fang, L. Zhu, M. Li, Y. Song, M. Sun, D. Zhao and J. Zhang, Adv. Sci., 2021, 8, 2102970.
64. H. Wang, Y. Gao, J. Chen, X. Fan, Y. Shi, J. Yu, K. Wang, S. Li, C. Lee and X. Zhang, ACS Nano, 2025, 19, 2549.
65. X. Li, L. Huang, G. Baryshnikov, A. Ali, P. Dai, Z. Yang, Y. Sun, C. Dai, Z. Guo, Q. Zhao, F. Zhang and L. Zhu, Adv. Mater., 2025, 37, 2500236.
66. Y. Xiao, J. Chen, W. Chen, X. Zheng, J. Tan, X. Cui, Z. Yuan, S. Ji, C. Cao, G. Lu, W. Liu, P. Wang, S. Li and C. Lee, Chem. Commun., 2021, 57, 4902.
67. Z. Liu, W. Shi, G. Hong, W. Chen, B. Song, X. Peng, X. Xiong and F. Song, J. Control. Release, 2019, 310, 1.
68. L. Zhu, Z. Zhang, Y. Liu, Z. Deng, Y. Luo, Y. Liu, Z. Zhao and B. Z. Tang, J. Phys. Chem. C, 2024, 128, 15763.
69. Y. Xiao, J. Chen, K. Wang, S. Li, W. Tao, S. Tian, X. Cui, Z. Huang, X. Zhang and C. Lee, Chem. Sci., 2020, 11, 888.

---

Footnote

† Yaning Li and Mengyan Tian contributed equally to this manuscript.

---