Light-intensity-dependent photoresponse time of organic photodetectors and its molecular origin

Organic photodetectors (OPDs) are attractive for imaging due to narrow absorption bands, mechanical flexibility, low weight, transparency, and processability. Sublimed small-molecule bulk heterojunctions (BHJs), typically using C60 acceptors, enable efficient exciton separation but suffer from slower photoresponse than inorganic devices, largely due to lower mobility, polaron formation, and energetic disorder that induces charge trapping. While donor design with push–pull motifs has improved EQE, dark current, detectivity, and linearity, the molecular origins of trap states and their impact on photoresponse—particularly under low-light conditions—remain unclear even for co-evaporated small-molecule systems. This work investigates how donor molecular structure (planar MPTA vs twisted NP-SA) governs shallow and deep trap formation in C60 blends and how these traps control the light-intensity-dependent photoresponse times of BHJ OPDs.

Recent OPD advances with sublimable donor molecules featuring donor–acceptor (push–pull) structures have enabled tuning of energy levels and optical bandgaps, achieving sharp/narrow absorption and fast response with merocyanine-like dyes. Strategies including donor fusion and acceptor functionalization/elongation have delivered devices with EQE in the green of 60–75% at 5 V reverse bias, dark current ~1 nA/cm2, detectivity ~1e13 Jones, and linear detection ranges over seven orders of magnitude. Despite these, OPDs lag inorganic detectors in response time due to low mobilities and energetic disorder leading to charge trapping. Prior work indicates traps can originate from chemical impurities, conformational isomers, and morphology (packing, orientation, domains). Even in sublimed BHJs with C60, the detailed molecular origins of trapping and their influence on light-intensity-dependent response, especially at low light, are unresolved. This study directly compares a twisted donor (NP-SA) and a fully planar donor (MPTA) to address this gap.

Materials and device fabrication: All organic semiconductors were purified via high-vacuum sublimation (<1e−7 Torr). Thin films were thermally evaporated under high vacuum at ~0.35 Å/s on cleaned glass or quartz. OPD stacks on ITO-coated glass: hole-extraction layer (5 nm), Donor:C60 BHJ (100 nm, 1:1 w/w), electron extraction layer, Yb (2 nm), and sputtered ITO top electrode (10 nm). Devices were encapsulated (glass, 98.5% transmittance); pixel area 0.04 cm2. Post-fabrication thermal annealing for thin films: 160 °C for 3 h. Transient optoelectronic measurements: Fully encapsulated devices were probed using an oscilloscope, with illumination from a ring of 12 white LEDs (spectrum provided). Bias from a Keithley 2400. Low-light achieved by LED power control and neutral density filters; low currents amplified (FEMTO DLPCA-200). Techniques included frequency response (bandwidth at set intensities/bias), transient photocurrent (light-on/off), transient photovoltage, and charge extraction (CE) to assess carrier densities and effective mobilities versus light intensity. Surface photovoltage (SPV) measured with a Kelvin probe under varying light intensities. Optical and structural characterization: UV–Vis transmittance/absorbance, steady-state and time-resolved PL, Raman spectroscopy (backscattering, 488 nm for Raman, 514 nm for PL), AFM (tapping mode), and TEM. PL under reverse bias was also measured to resolve charge-transfer (CT) emissions. Energetic characterization: Ambient photoemission spectroscopy (APS) on films (gold tip, grounded ITO) to extract HOMO onsets and density of states (DOS), including sub-gap tail states. Computational methods: DFT (Gaussian09, B3LYP/6-311G(d,p)) gas-phase optimizations, dihedral potential scans, simulated Raman modes (0.97 scaling). DFT was also used to evaluate possible MPTA–C60 adduct structures and energy levels. Systems studied: Two donors—NP-SA (twisted, tertiary amine with benzene and naphthalene substituents on a selenophene core) and MPTA (highly planar, fused heterocycle donor unit incorporating amine linkage and thiophene), each co-evaporated with C60 to form BHJs.

- Spectral selectivity: NP-SA:C60 detects 460–620 nm; MPTA:C60 detects 450–600 nm. Both show field-dependent responsivity due to bias-dependent CT dissociation. - J–V and dark current: Both devices exhibit similar dark and light J–V behavior with very low dark currents (~1e−8 A cm−2) from 0 to −5 V. MPTA:C60 shows increased dark shunt at strong reverse bias, consistent with trap-related shunt. - High-light response times: At 0.7 mW/cm2 and 0 V, −3 dB cut-off frequencies are ~3 kHz (NP-SA:C60) and ~4 kHz (MPTA:C60). Bandwidth increases at −3 V due to faster extraction (reverse-bias drift). - Transient extraction at high light (0.6–2.2 mW/cm2): MPTA:C60 exhibits faster photocurrent extraction than NP-SA:C60, indicating higher effective carrier mobility and fewer shallow traps in MPTA blends at high intensity. - Low-light dynamics and trapping: MPTA:C60 exhibits biphasic photocurrent transients at low intensities, with biexponential decay constants ~5.9 μs (fast) and ~210 μs (slow). The slow component indicates a population extracted from relatively deep traps. Activation analysis suggests deep traps are ~0.1 eV below shallow states (indicative, based on simplified assumptions). - Trap densities: Integration of the slow CE component at short circuit under 0.3 mW/cm2 yields deep trapped charge density ~1×10^15 cm−3; at open circuit, up to ~5×10^15 cm−3 accumulate, reflecting total deep trap density in MPTA:C60. - SPV: MPTA:C60 shows stronger light-intensity dependence and reduced SPV at low-light (e.g., 0.5 mW/cm2) versus NP-SA:C60, indicating stronger trap-assisted recombination and slower return to equilibrium in the dark, consistent with deep traps. - Optical/structural insights: MPTA neat films show strongly red-shifted, intense PL with ~5 ns lifetime, consistent with excimeric emission and strong intermolecular coupling; NP-SA decays fast (<400 ps). Upon blending with C60, MPTA’s PL is quenched and CT emission (>800 nm) appears; residual high-energy shoulders suggest some neat-like MPTA domains. Raman shows that blending induces extensive changes in MPTA vibrational modes (broadened, intensity changes), indicating substantial disruption of intermolecular coupling; NP-SA blends show limited changes mostly in end-group vibrations. - Adduct formation: DFT indicates feasible MPTA–C60 adduct formation via lone pair–π interaction (enabled by MPTA planarity and lack of bulky substituents). Simulated Raman of the adduct matches experimental trends. The adduct has a HOMO ~90 meV shallower than MPTA, adding localized energy levels that can act as deep traps. - Energetics (APS): Neat MPTA has a shallower HOMO (5.16 eV) than NP-SA (5.37 eV), reflecting strong intermolecular coupling in MPTA. Blending with C60 deepens HOMO for both, but far more for MPTA (shift ~370 meV) than NP-SA (~100 meV), and narrows the DOS in MPTA blends (fewer shallow tail states). Sub-gap analysis shows intrinsically low tail-state density in MPTA (narrow band-tails) preserved after blending. - DFT dihedral scans: NP-SA exhibits larger conformational energy differences (~200 meV) between 0° and 180° isomers and broader HOMO dispersion (~70 meV considering ±20° around minima), indicative of higher conformational energetic disorder. MPTA shows similar energies for 0°/180° and narrow HOMO dispersion (~30 meV), implying low intrinsic shallow-trap density. Overall: Planar MPTA enables fast, high-light response via strong coupling and low shallow-trap DOS but, when blended with C60, suffers from disrupted packing and possible MPTA–C60 adducts that create relatively deep trap states dominating low-light dynamics. Twisted NP-SA yields more consistent, trap-resilient low-light response albeit slower at high light.

The study disentangles how donor molecular structure governs trap formation and, consequently, OPD photoresponse across light intensities. Planar MPTA’s low intrinsic energetic disorder minimizes shallow traps and enhances carrier mobility, supporting faster extraction and higher bandwidth under high-light operation. However, blending with C60 disrupts MPTA’s strong intermolecular coupling, lowering the HOMO by ~370 meV and creating energetic inhomogeneity. Residual low-bandgap neat-like states and formation of MPTA–C60 adducts introduce relatively deep trap states. Under low-light, when photogenerated carrier densities are low, these deep traps dominate dynamics, yielding biphasic transients and reduced bandwidth. In contrast, the twisted NP-SA’s more amorphous packing experiences minor energetic/morphological perturbation upon blending; steric hindrance also disfavors adduct formation. As a result, NP-SA:C60 maintains more uniform energetics with fewer deep traps, preserving response time at low light despite reduced high-light mobility and bandwidth. These findings clarify why OPD design rules differ for high- versus low-light applications: maximizing planarity and coupling can benefit high-light speed, whereas mitigating deep-trap formation (via steric design and morphology control) is critical for low-light operation.

This work correlates donor molecular structure with trap-state formation and light-intensity-dependent response in OPDs. A highly planar donor (MPTA) paired with C60 enables fast, high-light response due to low shallow-trap DOS and strong intermolecular coupling, but blending-induced disruption and potential MPTA–C60 adducts create relatively deep traps that slow low-light response. A twisted donor (NP-SA) yields slower high-light response but maintains faster, more stable low-light dynamics due to fewer deep traps. Design implications: for high-speed OPDs under bright conditions, planar donors with strong coupling are advantageous; for low-light applications, donors should minimize deep-trap pathways by resisting morphology disruption and sterically hindering adduct formation with acceptors. Future research should explore donor designs combining planar backbones with steric protection to prevent adducts, optimize donor–acceptor interactions to preserve uniform energetics upon blending, and generalize these insights across different acceptors beyond C60.

Quantification of deep-trap energies (~0.1 eV below shallow states) is indicative and based on simplified biexponential detrapping assumptions rather than full distributions. Deep-trap densities are inferred from transient CE and open-circuit accumulation and may depend on device architecture and measurement conditions. The study focuses on two donors with C60; generalizability to other donor families or non-fullerene acceptors requires further validation. Gas-phase DFT and APS/ambient measurements provide insights but may not fully capture solid-state morphological complexity. Microscopic morphology changes were below AFM/TEM detection limits, so conclusions rely on spectroscopic proxies (PL/Raman/APS).