Triplet-Mediated Up-Conversion Model

Updated 10 July 2026

Triplet-mediated up-conversion is a family of models that convert long-lived triplet states into higher-energy emissive singlet states using various energy-transfer pathways.
The model spans distinct mechanisms including sensitized triplet–triplet annihilation in molecules, energy-transfer upconversion in rare-earth materials, and triplet–polaron interaction in OLEDs.
Design rules, kinetic formulations, and interface engineering strategies are central to optimizing triplet transport, energetic alignment, and overall upconversion efficiency.

The term triplet-mediated up-conversion model is used across several distinct but related settings in which long-lived triplet excitations, triplet-pair manifolds, or triplet-labeled atomic manifolds act as an energy reservoir that is converted into higher-energy emissive states. In molecular photon upconversion, the dominant realization is sensitized triplet–triplet annihilation, where two annihilator triplets fuse to form a singlet; in lanthanide systems it denotes energy-transfer upconversion between ions in triplet-multiplicity manifolds; in electrically driven organic devices it can denote conversion of triplet excitons into singlets by triplet–polaron interaction without a spin flip (Zimmermann et al., 2012, Simpson et al., 2010, Obolda et al., 2015). Taken together, these works define not a single microscopic mechanism but a family of kinetic and spin-selective models in which triplet population, triplet transport, and triplet-selective interconversion determine the upconverted output.

1. Conceptual scope and terminology

In molecular upconversion, a triplet is usually the lowest triplet exciton $T_1$ of an annihilator or emitter. In that setting, the standard sequence is sensitizer absorption, intersystem crossing or another triplet-generation step, triplet energy transfer to an annihilator, and then triplet–triplet annihilation (TTA) to form a higher-energy singlet $S_1$ , followed by fluorescence. This is the framework used in non-coherent TTA models, hybrid nanocrystal–organic systems, perovskite–rubrene bilayers, and bulk-heterojunction thin films (Zimmermann et al., 2012, Ronchi et al., 2020, Nienhaus et al., 2019, Narayanan et al., 16 Oct 2025).

In rare-earth systems, the same phrase refers to a different object. For $\mathrm{Tm}^{3+}$ -doped silica fibers, “triplet” is the multiplicity label $2S+1=3$ of atomic manifolds such as ${}^3H_4$ and ${}^3F_4$ , not a molecular exciton. The operative mechanism is energy-transfer upconversion (ETU), specifically $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ , identified through its pump-off decay kinetics (Simpson et al., 2010).

Electrically driven systems broaden the meaning further. In deep-blue fluorescent OLEDs based on TPA-TAZ and TCP, triplet-mediated upconversion refers to triplet–polaron interaction (TPI): a triplet exciton interacts with a neighboring positive or negative polaron, a one-electron transfer forms a transient charge-transfer configuration, and a singlet exciton is produced without invoking reverse intersystem crossing (Obolda et al., 2015). At the single-molecule level in an STM nanocavity, PdOEP exhibits anti-Stokes singlet emission through a triplet shelving state $T_1$ , where a first tunneling event populates $T_1$ and a second one converts it into $S_1$ (Zheng et al., 10 Sep 2025).

Mechanistic class	Triplet reservoir	Up-conversion event
Molecular sTTA-UC	Annihilator $S_1$ 0 excitons	$S_1$ 1
Rare-earth ETU	$S_1$ 2 ion population	$S_1$ 3
OLED TPI	Triplet exciton plus polaron	One-electron transfer to a singlet exciton
STM UCEL	Molecular $S_1$ 4 shelving state	Second tunneling event produces $S_1$ 5

A common misconception is that all triplet-mediated upconversion is TTA. The literature does not support that reduction. TTA is only one member of a broader class that also includes ETU, charge-transfer-assisted sensitization, doublet–triplet energy transfer, and triplet–polaron interaction (Simpson et al., 2010, Han et al., 2017, Obolda et al., 2015).

2. Elementary processes and spin physics

The canonical molecular model is sensitized TTA. A sensitizer generates triplets, usually through absorption followed by ISC, and transfers them to an annihilator. Two annihilator triplets then fuse to form a higher-energy singlet if the energetic condition $S_1$ 6 is satisfied. In the bimolecular formulation, the delayed fluorescence intensity is quadratic in excitation at low flux and linear at high flux because the singlet-generation rate is proportional to the square of the triplet density until TTA becomes the dominant triplet sink (Mattiello et al., 2022).

Several papers show that this sequence can be modified without abandoning the triplet-mediated character. In Spiro-4-DPA, two triplets can be created on the same annihilator molecule, enabling diffusion-free intramolecular TTA. The relevant energetic statement is explicit: $S_1$ 7, so a doubly occupied molecule can upconvert without intermolecular encounter (Mattiello et al., 2022). In TIPS-Ph-Naph, excited-state engineering changes the branching of the triplet-pair manifold itself: $S_1$ 8 is resonant with $S_1$ 9, $\mathrm{Tm}^{3+}$ 0, and $\mathrm{Tm}^{3+}$ 1, and the measured singlet-formation probability reaches $\mathrm{Tm}^{3+}$ 2 (Ronchi et al., 3 May 2026).

Triplet generation can also proceed without ISC on the sensitizer. In the radical-sensitized model, the neutral $\mathrm{Tm}^{3+}$ 3-radical TTM-1Cz transfers energy from an excited doublet to an acceptor triplet through Dexter-like doublet–triplet energy transfer,

$\mathrm{Tm}^{3+}$ 4

after which the acceptor follows the ordinary TTA route (Han et al., 2017). This mechanism avoids the ISC energy loss characteristic of phosphorescent sensitizers.

The OLED TPI mechanism is even more distinct. A positive polaron $\mathrm{Tm}^{3+}$ 5 is a hole on a donor-rich HOMO and a negative polaron $\mathrm{Tm}^{3+}$ 6 is an electron on an acceptor-rich LUMO. A triplet exciton on one molecule interacts with a neighboring polaron, and a one-electron hop produces a charge-transfer intermediate. If the transferred carrier is antiparallel to the resident carrier on the receiving site, the final local exciton is a singlet. The multiplicity change therefore arises from carrier rearrangement rather than a spin flip (Obolda et al., 2015). This is why TPI can operate with $\mathrm{Tm}^{3+}$ 7 in TPA-TAZ and $\mathrm{Tm}^{3+}$ 8 in TCP, whereas TADF would require a much smaller singlet–triplet gap (Obolda et al., 2015).

In $\mathrm{Tm}^{3+}$ 9-doped silica, the spin physics is atomic rather than excitonic. The experimentally identified ETU channel,

$2S+1=3$0

is a pairwise energy-exchange process aided by spectral overlap and phonons. The “triplet-mediated” label there refers to transfer between manifolds of triplet multiplicity and is explicitly distinct from molecular TTA (Simpson et al., 2010).

3. Kinetic formulations and scaling laws

A minimal steady-state TTA model for annihilator triplets is

$2S+1=3$1

where $2S+1=3$2 is the triplet-generation rate, $2S+1=3$3 the first-order triplet decay rate, $2S+1=3$4 the bimolecular annihilation constant, and $2S+1=3$5 the singlet-formation factor (Mattiello et al., 2022). The threshold triplet density is

$2S+1=3$6

and the corresponding threshold generation rate is

$2S+1=3$7

When $2S+1=3$8, $2S+1=3$9; when ${}^3H_4$ 0, ${}^3H_4$ 1 (Mattiello et al., 2022).

A more structural TTA model appears in the 1D non-coherent upconversion framework, where emitters are absorbing boundaries separated by ${}^3H_4$ 2 sensitizers. There the characteristic escape, decay, and coagulation times are

${}^3H_4$ 3

and the optimal sensitizer spacing obeys

${}^3H_4$ 4

in the decay-limited regime and

${}^3H_4$ 5

in the coagulation-limited regime (Zimmermann et al., 2012). These scaling laws formalize a central design trade-off: increasing sensitizer fraction raises absorption but also raises transport losses.

Liquid-phase photochemical models add exchange equilibrium and dynamic quenching. When triplet energy transfer between sensitizer and emitter is fast, the total triplet population partitions as

${}^3H_4$ 6

with overall first-order loss

${}^3H_4$ 7

In the strongly exothermic low-intensity limit, the optimal sensitizer concentration is

${}^3H_4$ 8

which evaluates to ${}^3H_4$ 9 for the ZnOEP/DPA values used in that model (Jefferies et al., 2019).

Triplet-mediated OLED upconversion uses a different balance equation. Under electrical excitation,

${}^3F_4$ 0

${}^3F_4$ 1

with ${}^3F_4$ 2 and ${}^3F_4$ 3. The resulting singlet-formation fraction is

${}^3F_4$ 4

so the singlet yield grows with polaron density and saturates at ${}^3F_4$ 5 (Obolda et al., 2015).

The rare-earth ETU case is diagnosed through pump-off kinetics. With ${}^3F_4$ 6 the ${}^3F_4$ 7 population and ${}^3F_4$ 8 the ${}^3F_4$ 9 population,

$({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 0

Under the empirical approximation $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 1, the upconverted level obeys

$({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 2

which is the source of the characteristic $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 3 and $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 4 decay components (Simpson et al., 2010).

4. Experimental discrimination and spectroscopic signatures

Because several triplet-mediated pathways can produce superficially similar anti-Stokes emission, mechanistic discrimination relies on kinetics, magnetic response, and exclusion tests. In TTA systems, the decisive low- to high-power crossover is the transition from quadratic delayed fluorescence to linear delayed fluorescence, accompanied by the threshold relation $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 5 (Mattiello et al., 2022, Narayanan et al., 16 Oct 2025). In the PbS–TES-ADT BHJ, for example, measured thresholds increase strongly with decreasing QD bandgap, from $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 6 for 850 nm QDs under 808 nm excitation to $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 7 for 1150 nm QDs, indicating that energetic alignment and nonradiative loss in the mediator manifold strongly affect $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 8 (Narayanan et al., 16 Oct 2025).

The rare-earth ETU signature is different. The relation $({}^3F_4,{}^3F_4 \rightarrow {}^3H_4,{}^3H_6)$ 9 establishes a two-photon population pathway for $T_1$ 0, but the decisive ETU test is post-pump decay. Under indirect 1586 nm pumping, ESA vanishes immediately after pump-off, so continued feeding of $T_1$ 1 can only arise from ETU. The observed 800 nm decay is therefore fit by a double exponential containing both $T_1$ 2 and $T_1$ 3, and the fitted $T_1$ 4 agrees with the independently measured $T_1$ 5 lifetime (Simpson et al., 2010).

Magnetic-field measurements provide an orthogonal probe of spin-dependent channels. In TPI OLEDs, the magneto-current is fit by

$T_1$ 6

with $T_1$ 7 and $T_1$ 8. For TPA-TAZ at 8 V, $T_1$ 9, $T_1$ 0, and the MC amplitude reaches $T_1$ 1; for TCP at 8 V, $T_1$ 2, $T_1$ 3, and MC is $T_1$ 4 (Obolda et al., 2015). In the DIB-SQ:PCBM/rubrene:DBP NIR upconverter, the UC magnetic-field-dependent photoluminescence is negative, about $T_1$ 5 at $T_1$ 6, consistent with a TTA pathway described by Merrifield’s picture, whereas direct PL under 450 nm excitation is positive and exceeds $T_1$ 7 at $T_1$ 8, reflecting singlet-fission physics instead (Klein et al., 30 Sep 2025).

Electrical upconversion in the STM nanocavity is diagnosed by current scaling rather than optical power scaling. For PdOEP/NaCl/Ag(111), the minimal three-state model predicts $T_1$ 9, because the second current-driven event that converts $S_1$ 0 to $S_1$ 1 scales with electron arrival rate. Experimentally, $S_1$ 2 intensity is superlinear, with exponent $S_1$ 3 at $S_1$ 4, while $S_1$ 5 emission is approximately linear, with $S_1$ 6 at $S_1$ 7, matching a two-electron UCEL mechanism rather than a direct one-electron threshold process (Zheng et al., 10 Sep 2025).

Mechanism exclusion is equally important. The deep-blue OLED study explicitly rules out TADF and TTA by combining large $S_1$ 8, the absence of delayed fluorescence up to $S_1$ 9, a five-order intensity drop with no late component, a strictly linear PL versus excitation slope of $S_1$ 00, and 77 K transient PL with no delayed tail (Obolda et al., 2015). Likewise, the perovskite–rubrene bilayer assigns triplet sensitization primarily to interfacial charge-transfer-assisted triplet formation rather than direct Dexter transfer from a stable exciton, because bulk perovskite excitons dissociate within picoseconds and carrier diffusion lengths exceed the 15 nm sensitizer thickness (Nienhaus et al., 2019).

5. Representative material platforms and performance

The literature spans molecular solutions, neat organic films, hybrid nanocrystal systems, donor–acceptor heterojunctions, rare-earth fibers, OLEDs, and single-molecule junctions. The table lists representative implementations and reported figures of merit.

Platform	Representative system	Reported output
Deep-blue OLED TPI	TPA-TAZ / TCP	EQE $S_1$ 01 $S_1$ 02 for TPA-TAZ; singlet fraction $S_1$ 03 (Obolda et al., 2015)
Rare-earth ETU fiber	$S_1$ 04-doped silica	800 nm UC from $S_1$ 05; $S_1$ 06, $S_1$ 07 (Simpson et al., 2010)
Diffusion-free intramolecular TTA	PtOEP / Spiro-4-DPA	$S_1$ 08; UC lifetime $S_1$ 09 (Mattiello et al., 2022)
Excited-state-engineered sTTA	Ir(ppy) $S_1$ 10(acac) / TIPS-Ph-Naph	$S_1$ 11; absolute UCQY $S_1$ 12; $S_1$ 13 broadband (Ronchi et al., 3 May 2026)
Hybrid nanocrystal sensitizer	Au:CdSe–9-ACA–DPA	$S_1$ 14; $S_1$ 15 absolute; $S_1$ 16 (Ronchi et al., 2020)
CT-mediated solid-state UC	DIB-SQ:PCBM / rubrene:DBP	$S_1$ 17 at 690 nm; $S_1$ 18 (Klein et al., 30 Sep 2025)
Perovskite-sensitized bilayer	MA $S_1$ 19FA $S_1$ 20PbI $S_1$ 21 / rubrene:DBP	Internal upconversion efficiency in excess of $S_1$ 22 under 785 nm (Nienhaus et al., 2019)
NIR-I / NIR-II BHJ thin film	PbS–TCA / TES-ADT / DBP	IQE $S_1$ 23 at 808 nm, $S_1$ 24 at 1130 nm, $S_1$ 25 at 1208 nm (Narayanan et al., 16 Oct 2025)

These platforms illustrate several distinct optimization strategies. In TIPS-Ph-Naph, the decisive gain comes from manipulating the annihilator excited-state manifold so that singlet formation after triplet fusion is intrinsically favored (Ronchi et al., 3 May 2026). In Au-doped CdSe hybrids, the gain comes from suppressing parasitic hole transfer: the hole is routed to the Au state in $S_1$ 26, much faster than the $S_1$ 27 ligand hole-transfer channel, which raises the NC-to-ligand triplet transfer yield from about $S_1$ 28 in undoped NCs to about $S_1$ 29 (Ronchi et al., 2020). In DIB-SQ:PCBM/rubrene:DBP, optimization instead targets the sensitizer bulk heterojunction and its charge-transfer manifold, where the 1:3 DIB-SQ:PCBM blend maximizes the ratio of UC photoluminescence to direct rubrene PL and delivers the lowest threshold (Klein et al., 30 Sep 2025).

The thin-film PbS–TES-ADT system extends TTA-UC into the NIR-II. There, 5-TCA ligands simultaneously improve QD solubility, film morphology, and triplet sensitization. The UCPL increases by about 15-fold, the TES-ADT triplet lifetime is $S_1$ 30, and visible imaging of incoherent 1200 nm light is demonstrated at mask intensities of about $S_1$ 31 (Narayanan et al., 16 Oct 2025). By contrast, the perovskite–rubrene bilayer achieves its efficiency through the combination of strong spin–orbit coupling, long carrier diffusion lengths, and an interfacial CT route to rubrene triplets, rather than through molecular sensitizer ISC (Nienhaus et al., 2019).

6. Design rules, limitations, and open problems

Across these models, energy alignment is the primary design variable. For TTA, the annihilator must satisfy $S_1$ 32, while the sensitizer triplet or sensitized intermediate must lie above the annihilator $S_1$ 33 by a modest margin (Mattiello et al., 2022, Ronchi et al., 3 May 2026). For liquid photochemical systems, the partition of triplets between sensitizer and emitter shows that even small $S_1$ 34 can be detrimental if dynamic quenching is strong; this is why the model predicts an optimal sensitizer concentration below the solubility limit in ZnOEP/DPA (Jefferies et al., 2019). For CT-mediated solid-state systems, the relevant energies are not only $S_1$ 35 and $S_1$ 36, but also the CT and defect-state manifolds. In DIB-SQ:PCBM/rubrene, $S_1$ 37 and $S_1$ 38 both exceed $S_1$ 39, enabling downhill triplet funneling (Klein et al., 30 Sep 2025).

Interfacial coupling and morphology are the second major control parameter. The TPI OLED work states that close donor–acceptor contacts of $S_1$ 40 and transfer integrals in the 10–30 meV range are favorable for one-electron triplet harvesting (Obolda et al., 2015). The PbS–TES-ADT BHJ identifies an optimal TCA ligand density of $S_1$ 41 at 4 vol/vol equivalents, linking surface chemistry directly to triplet injection and morphology (Narayanan et al., 16 Oct 2025). The perovskite–rubrene bilayer benefits from the absence of long insulating ligands, a limitation that constrains colloidal sensitizers (Nienhaus et al., 2019). Hybrid nanocrystal upconversion further shows that interfacial parasitic charge transfer can dominate unless the sensitizer is redesigned at the electronic-structure level, as in Au:CdSe (Ronchi et al., 2020).

Triplet lifetime and quenching control remain universal constraints. Oxygen exclusion is essential in essentially all molecular TTA systems, and in solution the upconversion threshold scales inversely with the square of the sensitization yield. The Au:CdSe model makes this dependence explicit through the statement that $S_1$ 42 grows linearly with $S_1$ 43 while $S_1$ 44 scales inversely with $S_1$ 45 (Ronchi et al., 2020). The liquid equilibrium model emphasizes that sensitizer concentration simultaneously raises generation and enhances sensitizer-induced dynamic quenching, which is why $S_1$ 46 can be far below the maximum soluble concentration (Jefferies et al., 2019). The thin-film PbS platform adds a different trade-off: higher QD loading increases absorption but lowers IQE because QDs act as sinks for singlets and UC photons via parasitic FRET or back transfer (Narayanan et al., 16 Oct 2025).

Spin engineering is increasingly treated as a design objective rather than a passive constraint. The TIPS-Ph-Naph work shows that $S_1$ 47 can be driven from the conventional $S_1$ 48 range into the $S_1$ 49 range by deliberately aligning $S_1$ 50 with high-lying $S_1$ 51 and $S_1$ 52 states (Ronchi et al., 3 May 2026). Magnetic-field studies in both OLEDs and solid-state TTA systems show that field-dependent spin mixing can be used as a mechanistic discriminator and, plausibly, as a future optimization axis (Obolda et al., 2015, Klein et al., 30 Sep 2025).

Several open problems recur. In TPI OLEDs, direct spectroscopic observation of the CT intermediate during operation is still lacking, and the competition between productive TPI upconversion and destructive exciton–polaron annihilation remains difficult to disentangle (Obolda et al., 2015). In CT-mediated solid-state TTA systems, absolute values of $S_1$ 53, $S_1$ 54, $S_1$ 55, and $S_1$ 56 are often not directly extracted, which limits parameter-level comparison across platforms (Klein et al., 30 Sep 2025). At long wavelengths, mediator or CT triplets that enable sensitization can also increase nonradiative decay and raise thresholds, as seen in NIR-II BHJs (Narayanan et al., 16 Oct 2025). More broadly, the collected literature suggests that triplet-mediated upconversion has matured from a single TTA paradigm into a broader discipline of excited-state, spin-manifold, and interfacial-charge engineering, in which the optimal model depends on whether the limiting step is triplet generation, triplet transport, triplet fusion, or triplet-to-singlet conversion.