BNOO

Structure

A B/N-based framework with O/O heteroatoms. This series changes spin–orbit coupling while retaining a related molecular framework.

Properties and function

Excited-state pathways

Singlet and triplet states exchange population through competing spin-conversion and decay channels.

Performance under load

Compare measured efficiency at each reported luminance.

Device and processing

1. Molecular assembly

   Published aromatic C–N coupling and boron ring formation assemble the emitter framework.

2. Host–guest film

   The emitter is diluted in a host rather than measured as an isolated gas-phase molecule.

3. OLED operation

   Injected charges produce excited states in the emitting layer; device measurements report efficiency and stability separately.

Measured host-film photophysics

1 wt% emitter in DMIC-TRZ host film, inert atmosphere

Experimentally inferred rate from measured prompt/delayed PL lifetimes and yields; not a direct elementary-state measurement

2PXZBN is named BNOO in the 2024 computational comparison.

Observed OLED performance

Device A: 1 wt% emitter in DMIC-TRZ, 50 nm emitting layer.

Within this four-device series, faster inferred RISC accompanies better efficiency at high luminance, while the best-EQE device has a shorter reported operational lifetime.

This series is not an isolated causal test of RISC versus lifetime. Heteroatom substitution changes other molecular and device properties.

Representative device values are kept separate from ensemble means. SI tables report averages for more than fifteen devices but the downloaded curves do not include replicate-level data or pointwise error bars.

Reported LT50 values are condition-specific laboratory device measurements, not service-life predictions.

Lines connect published observations only; no model is fitted to these device curves.

Compare efficiency at the operating brightness

All four Hu devices report EQE at 1000 and 10000 cd m−2. These tabulated checkpoints support a common-luminance comparison; each device’s peak EQE is its own reference, not a shared brightness.

Absolute loss = peak EQE − checkpoint EQE, in percentage points (pp). Relative roll-off = 100 × (peak EQE − checkpoint EQE) / peak EQE, in percent. Retained peak efficiency = 100 × checkpoint EQE / peak EQE.

2PXZBN at 1000 cd m−2: EQE 24% versus peak 30.7%; the loss is 6.7 pp, relative roll-off 21.82%, and retained peak efficiency 78.18%. These calculations use representative-device values, not ensemble means.

• Hu et al. (2022), SI Table 10: representative-device checkpoints

Measured electrical operating points

Hu et al. 2022 devices A–D; vacuum-deposited common stack; 1 wt% emitter in DMIC-TRZ, 50 nm emitting layer. The dataset contains 83 observed operating points joined by exact luminance across the Figure 3d–f source sheets; no voltage or current is interpolated.

With J in mA cm−2 and V in volts, electrical input per area is J × V in mW cm−2; multiply by 10 for W m−2. Equal voltage is not equal brightness. Switching emitters in the interactive view selects the nearest observed brightness and displays its actual value.

Source lm W−1 values are photometric luminous efficacy, not dimensionless optical power efficiency. The separate πL/(10JV) check assumes Lambertian angular emission; agreement checks arithmetic without establishing the angular pattern. The selected display domain requires L ≥ 1 cd m−2, positive V/J and EQE within 0–100%; all raw rows remain downloadable.

No pointwise experimental error bars are supplied. These bipolar-device observations do not identify mobility, injection barriers or the cause of efficiency loss.

• Hu Figure 3 source workbook
• Electrical operating-point data
• Electrical analysis and source discrepancies
• Electrical-analysis reproducibility bundle
• Electrical-analysis source manifest

Device-ensemble evidence

Published device-ensemble summary, distinct from representative-device values and the single source curve

Averages are reported as based on over fifteen independent devices. Exact n per condition is not supplied.

The retrieved SI table and adjoining text do not define the ± statistic; do not relabel it as SD, SEM or a confidence interval.

Percentage points of EQE; not a relative percentage of the mean

These ensemble means must not silently replace the paper’s representative-device maxima or be used as pointwise error bars on the downloaded Figure 3e curves.

• Supplementary Table 10, printed page 32 (PDF page 33)
• Supplementary Table 8 footnote, printed pages 29–30 (PDF pages 30–31)

Computed excited-state model

Exact integrated three-excited-state linear kinetics from rounded published computational rates; no new experiment or electronic-structure calculation.

The model uses published elementary rates for S1, T1 and T2. dp/dt = −A p; the integrated state residence is u = A⁻¹p0. Outcome probabilities are decay rates multiplied by residence times.

For optical initial excitation p0 = (1, 0, 0), the reproduced total radiative probability is 93.09%. The measured host-film PLQY is 71%. These are calculated and measured results from different evidence contexts; agreement with a published calculation is an implementation check.

The 1:3 singlet/triplet generation ratio is an assumed spin-statistical limit, not a measurement for these devices.

The default 25% extraction and 100% charge balance are illustrative scenario assumptions. Their product with model internal emission is not measured EQE.

Added first-order triplet loss is a hypothetical sensitivity parameter; it is not a fitted TTA rate, current density, oxygen concentration, or brightness.

Removing T2→S1 is a mathematical pathway test, not the prediction of a synthesizable molecule.

Omitted processes: Bimolecular triplet–triplet and triplet–polaron annihilation; Charge transport and spatial recombination; Host/morphology changes; Optical stack and dipole orientation; Chemical degradation and operational lifetime.

No statistical confidence interval is assigned to published computed elementary rates; two-significant-figure input rounding limits output precision. Slider results are sensitivities, not confidence bounds.

Chalcogen substitution and excited-state kinetics

Element records list O, S and Se as group-16 elements.

The O/S/Se comparison tests how embedded atoms change spin–orbit coupling and transitions involving higher triplet states. Molecular geometry and competing rates remain part of the explanation.

The 2024 kinetic calculations cover BNOO, BNSS and BNSeSe; BNSSe is listed in the experimental comparison.

• Oxygen
• Sulfur
• Selenium
• Quantitative prediction of rate constants and its application to organic emitters
• Efficient selenium-integrated TADF OLEDs with reduced roll-off

Conjugation and emission energy

Benzene and biphenyl provide concrete aromatic structures; the existing oligophenyl investigation examines one specific absorption trend.

The 2025 pair challenges a simple longer-conjugation/redder-emission rule: precisely placed additional boron changes the electronic pattern and gives a blue shift.

The oligophenyl result concerns absorption in a different series. Emission energy, the frontier-orbital energy difference and the singlet–triplet gap are different quantities.

• Benzene
• Biphenyl
• Boron
• Late-stage direct double borylation of B/N-based multi-resonance framework enables high-performance ultra-narrowband deep-blue organic light-emitting diodes
• Quantitative prediction of rate constants and its application to organic emitters

Sources and reproducible analysis

Published measurements, source-derived rates and Mol2Mat calculations are identified separately. The downloads reproduce the analysis without claiming new synthesis, device fabrication or electronic-structure calculations.

• Reproducibility bundle
• Analysis JSON
• Source manifest
• Python calculation
• Measured device observations CSV
• Condition-annotated records
• Published computed elementary rates
• Efficient selenium-integrated TADF OLEDs with reduced roll-off
• Supporting information, Tables 8–10
• Figure 3 numerical source workbook
• Quantitative prediction of rate constants and its application to organic emitters
• Physical data and inferred kinetic parameters in 1 wt% DMIC-TRZ films
• Final published Figure 3: A–D device architecture and performance
• Author-shared final article, device architecture text

Density-dependent excited-state losses

conditional-normalized-steady-state-model

All defaults are illustrative dimensionless assumptions; no rate is assigned to a selected BNOO, BNSS, BNSSe or BNSeSe emitter.

Spatially uniform, steady generation in a two-excited-state kinetic system. S1/T1 are coarse-grained states, not a reduction fitted to the separate S1/T1/T2 calculation.

Fixed first-order coefficients, no state filling, heating, diffusion gradients, charge transport, polaron annihilation, phosphorescence, singlet-singlet annihilation or chemical degradation.

fS=.25 is an assumed electrical 1:3 singlet:triplet generation fraction. fS=1 describes pure singlet generation, not an entire optical experiment.

TTA f=.25 is the stated effective branch convention. It is not a measured universal spin yield.

STA destroys one singlet and leaves one triplet after rapid higher-triplet relaxation.

Both a and b scale proportionally with G for a fixed material; changing only one load is a channel-isolation counterfactual.

The interactive comparison varies dimensionless generation load kTT G / KS² and scales kST G / KS² with it. The two-state sensitivity model is separate from the material-specific three-state calculation. It does not convert a measured OLED curve into collision-rate constants.

kTT*T^2 is effective annihilating encounters per volume per time. At f=1/4 the singlet source is +(1/4)kTT*T^2 and triplet sink is -(5/4)kTT*T^2. The NET excitation-number loss is kTT*T^2, not 2*kTT*T^2.

• Reproducible density-loss calculation
• Density scenarios and method limits

An operando test of competing mechanisms, 2026

King et al., Uncovering the Origin of Efficiency Roll-Off in TADF OLEDs. 15 wt% 4CzIPN in mCBP; 30 nm emissive layer.

TTA was negligible in this device; STA and polaron quenching explained the measured roll-off.

SPA dominated below about 100 mA cm^-2.

STA dominated above about 100 mA cm^-2.

The adjacent normalised TTA/STA exercise is a sensitivity calculation. It neither uses these fitted coefficients as presets nor reproduces the measured 4CzIPN device, which also requires SPA, TPA, charge densities and device geometry.

The ± values below are one standard deviation from fitting. An additional estimated 20% uncertainty in generation rate is excluded. The TTA upper bound is based on the fit’s 95% confidence interval. No independent reprocessing of the source raw dataset is claimed.

• Uncovering the Origin of Efficiency Roll-Off in TADF OLEDs
• Published experimental dataset record

Evidence and limits

Inspect the matched device curves from Hu et al. (2022) and the state-resolved calculations of Shizu and Kaji (2024) below.

来源说明

References

• Hu et al. · Nature Photonics 16, 803–810 (2022)
• Shizu and Kaji · Nature Communications 15, 4723 (2024)