Anomalously bright single-molecule upconversion electroluminescence

Upconversion electroluminescence (UCEL) emits photons with energies higher than the excitation electron energy and has been widely observed in ensemble systems, including OLEDs. Prior UCEL mechanisms in ensembles include triplet–triplet annihilation, thermally assisted processes, and Auger processes. At the single-molecule level, UCEL has been demonstrated at low temperatures using scanning tunneling microscope-induced luminescence (STML). Because thermally activated and intermolecular energy transfer mechanisms are unlikely in single-molecule junctions, prior single-molecule UCEL was attributed to a spin-triplet-mediated two-electron mechanism: an inelastic electron–molecule scattering (IES) step to populate an intermediate triplet T1, followed by carrier injection (CI) to generate a singlet exciton. Plasmon- and vibration-assisted mechanisms are ruled out due to insufficient lifetimes of relay states. However, single-molecule UCEL reported previously is orders of magnitude weaker than normal-bias electroluminescence (EL) at voltages exceeding the exciton energy, limited by the inefficiency of the IES step. This motivates the central question: can one engineer the molecule–substrate energy-level alignment to activate a more efficient UCEL pathway that avoids the IES bottleneck? The authors propose that appropriate level alignment enables a spin-triplet-mediated UCEL mechanism using only carrier injection steps, and test this by tuning substrate work function and spacer thickness, supported by differential conductance and bias-dependent EL measurements, and a quantum master equation model that maps mechanism regimes via EL diagrams.

The study situates UCEL within known ensemble mechanisms: triplet–triplet annihilation, thermally assisted delayed fluorescence, and Auger-type processes. For single molecules under STML, prior works reported UCEL via a two-electron sequence requiring an inelastic electron–molecule scattering (IES) to create T1, then CI to reach S1; plasmon- and vibration-assisted upconversion were deemed implausible due to short-lived intermediates. Previous single-molecule systems such as H2Pc/NaCl/Ag(100) showed weak UCEL relative to normal EL, reinforcing the inefficiency of IES. Reports of bias-polarity dependence and charging-enhanced emission in phthalocyanines highlight sensitivity to level alignment and charging. The authors build upon this by proposing and validating a pure CI+CI mechanism contingent on energy-level alignment, extending prior theoretical many-body descriptions and experimental STML studies.

Experimental: A custom optical STM (Unisoku) operated at ~7 K and ultra-high vacuum (<1e-10 Torr) was used. Clean Au(111) and Ag(100) substrates were prepared via sputter/anneal. Ultrathin NaCl(100) films (2–4 monolayers) were grown as insulating spacers to decouple molecules from metal quenching. Isolated phthalocyanines (H2Pc, PtPc) were deposited onto NaCl at ~7 K via Knudsen cell evaporation. Ag tips were made by electrochemical etching and cleaned by electron bombardment and Ar+ sputtering. STML was collected with a built-in lens near the junction and analyzed using a liquid-nitrogen-cooled CCD spectrometer; STM imaging and STML spectra were in constant-current mode. Spectra were not corrected for wavelength-dependent detector sensitivity. Differential conductance (dI/dV) spectra used lock-in detection (20 mV, 329 Hz). Measurements probed bias-dependent EL at both polarities and various NaCl thicknesses to tune level alignment. Key systems included H2Pc on 3ML-NaCl/Au(111) (also Ag(100) reference), and PtPc on 3–4ML-NaCl/Au(111). Analysis leveraged onset voltages of EL and of molecular orbitals (LUMO) in dI/dV to infer injection barriers and alignment, and current–photon intensity scaling to infer multi-electron processes. Theoretical: A microscopic quantum master equation model described excitation and decay dynamics among molecular many-body states (S0, T1, S1, transient anion D0− and cation D0+), with levels pinned to the substrate. Parameters included ET1≈1.2 eV, ES≈1.8 eV, and electron injection barrier Φe≈1.1 eV (representative for H2Pc/3ML-NaCl/Au). The model computed the exciton excitation efficiency Nex as a function of bias V and Φe, constructing electroluminescence (EL) diagrams that demarcate dominant mechanisms (IES, weak-IES, CI, and combinations forming UCEL pathways: IES+CI, w-IES+w-IES, w-IES+CI, CI+CI). These diagrams were used to interpret experiments and predict optimal UCEL regimes and driving voltages under different level alignments (e.g., spacer thickness variations).

• Observation of anomalously bright single-molecule UCEL from H2Pc on 3ML-NaCl/Au(111), with emission efficiency improved by more than one order of magnitude over previous single-molecule UCEL studies and even exceeding normal-bias EL intensities at comparable conditions.
• At positive bias V=1.7 V (eV < ES=1.81 eV), UCEL intensity is over two orders of magnitude stronger than at negative bias with similar magnitude and surpasses normal EL at V=2.0 V; at V=1.5 V UCEL is much weaker, indicating a narrow efficient upconversion window.
• UCEL onset voltage aligns with T1 energy (~1.2 eV), implicating triplet mediation; differential conductance shows LUMO onset near 1.6 V, and the sharp EL increase coincides with this onset, implicating carrier injection into LUMO.
• Photon intensity scales nonlinearly with current with exponent ~1.8 at both 1.7 V (UCEL) and 2.0 V (normal EL), evidencing multi-electron excitation dominance in both regimes.
• Proposed and supported a spin-triplet-mediated CI+CI mechanism: two sequential carrier-injection steps generate T1 via a transient anion (D0−) and hole injection; a second pair of CI steps promotes T1 → S1 via a transient cation (D0+). This avoids the inefficient IES step and explains the anomalously high UCEL efficiency.
• EL diagrams from a quantum master equation model map mechanism domains versus bias and injection barrier, identifying a red triangular region (ES > Φ ≥ ET and eVb ≥ Φe) where CI+CI operates, yielding markedly higher Nex than IES- or w-IES-involving pathways.
• The mechanism change from IES+CI to CI+CI around V≈+1.6 V explains the experimental intensity jump; even above ES (normal EL region), one-electron CI is energetically inhibited, so CI+CI still dominates.
• Tuning NaCl spacer thickness adjusts Φe and correspondingly shifts the optimal UCEL driving voltage: for H2Pc/NaCl/Au(111), reducing spacer from 3ML to 2ML shifts LUMO onset from ~1.6 to ~1.5 V and the maximal UCEL bias from ~1.7 to ~1.6 V; for PtPc/NaCl/Au(111), 4ML→3ML shifts LUMO onset ~1.8→~1.65 V and optimal UCEL bias ~1.9→~1.8 V (PtPc ES≈1.95 eV), corroborating the model’s predictions.
• The absence of bright UCEL on Ag(100) is explained by different work function/level alignment that violates the CI+CI criterion (Φ < ET).

The findings demonstrate that engineering molecule–substrate energy-level alignments can activate a highly efficient UCEL pathway at the single-molecule level. By ensuring ES > Φ ≥ ET and enabling electron injection into the LUMO at sub-gap biases, the molecule undergoes two sequential carrier-injection cycles that first populate a long-lived T1 relay state and then convert it to S1, circumventing the inefficient IES step. The coincidence of the EL intensity surge with the LUMO onset in dI/dV, the multi-electron current scaling, and the EL diagram predictions collectively substantiate the CI+CI mechanism as the dominant process in both the UCEL and even normal EL regions for the studied alignment. The model rationalizes contrasting behaviors between Au(111) and Ag(100) substrates and explains sensitivity to NaCl spacer thickness, providing a practical route to tune optimal driving voltages and enhance UCEL. These insights generalize to organic EL by clarifying how specific alignment conditions enable multi-electron, triplet-mediated excitation with superior efficiency, offering guiding principles for designing energy-efficient nanoscale optoelectronic devices.

The work experimentally realizes anomalously bright upconversion electroluminescence from single phthalocyanine molecules by tailoring energy-level alignment in STML junctions and introduces a spin-triplet-mediated UCEL mechanism comprising only carrier injection steps (CI+CI). A quantum master equation framework produces electroluminescence diagrams that link mechanism regimes to level alignment, identifying the conditions ES > Φ ≥ ET and eVb ≥ Φe for highly efficient UCEL. The approach explains prior weak UCEL results, accounts for substrate-dependent behavior, and enables tuning of optimal driving voltages via spacer thickness. These results provide microscopic understanding of single-molecule electro–optic conversion pathways and offer guidance for optimizing single-molecule and organic optoelectronic devices. Future research could explore broader molecular families, substrate and spacer engineering strategies, room-temperature operation, and plasmon–exciton coupling optimization to further enhance efficiency and generality.

• The demonstrated phenomenon relies on precise energy-level alignment (ES > Φ ≥ ET) and the presence of a long-lived T1 relay state; deviations (e.g., different substrate work functions or adsorption configurations) can suppress the CI+CI pathway, as indicated by variable behaviors on Ag(100) and reports of configuration-dependent STML.
• Experiments were performed at cryogenic temperature (~7 K) and UHV in an STM nanocavity; generalizability to ambient conditions or device geometries is not established.
• Spectra were not corrected for wavelength-dependent detector sensitivity, potentially affecting quantitative comparisons of absolute intensities across wavelengths.
• The theoretical model assumes level pinning to the substrate and uses specific parameter choices; while predictive, accuracy depends on correct parameterization for each system, and some excited cationic-state contributions at negative bias require additional considerations.
• The efficiency gains were shown for specific phthalocyanines (H2Pc, PtPc) on NaCl/Au(111); extension to other materials may require careful interface engineering.