Biomimetic nanocluster photoreceptors for adaptive circular polarization vision

The study addresses the challenge of building compact, multifunctional artificial vision systems (AVSs) that can perform color recognition, adaptive photoperception, and circularly polarized light (CPL) detection within a single, simple device. Current AVSs often require complex multilayer stacks or separate detector-processor integrations, increasing cost and fabrication complexity. Drawing inspiration from the mantis shrimp’s visual system—which integrates UV-to-visible sensitivity, adaptive filtering, and CPL perception—the authors propose a biomimetic approach using atomically precise chiral silver nanoclusters interfaced with an organic semiconductor (pentacene). The research aims to elucidate the optoelectronic behavior of nanoclusters in such heterostructures, engineer an in-sensor charge reservoir and light-valve mechanism for spectral- and intensity-dependent adaptation, and demonstrate wafer-scale, all-in-one photoreceptors capable of color vision, photoadaptation, and CPL recognition with memory.

The paper situates its contribution against prior efforts in bioinspired vision hardware: photoadaptation devices and neuromorphic phototransistors often employ sophisticated multilayers or external integration of sensing and processing, raising manufacturing complexity and costs (refs 1–3). Individual functions similar to mantis shrimp capabilities have been shown separately using panchromatic absorbers for color sensing (refs 7,8), circularly polarized molecular assemblies (refs 9,10) or chiral compounds (refs 11,12) for CPL detection, and other adaptive devices; however, parallel, all-in-one integration of color recognition, tunable adaptation, CPL perception, and multi-state readout in a single cell remains scarce. Nanoclusters with discrete energy levels and wide bandgaps have been recognized for strong photon-to-electron conversion and tunability (refs 13–15), yet their electronic engineering and device-level optoelectronic behavior have been underexplored. The work also compares its adaptive performance against state-of-the-art adaptable devices (refs 1,26–34), highlighting advances in perception range and adaptation speed.

Device architecture and fabrication:

• Wafer-scale ACP arrays were fabricated on p-type Si/SiO2 substrates (≈300 nm SiO2) cleaned via piranha solution, sequential ultrasonication (DI water, acetone, ethanol), N2 blow-dry, and O2 plasma (10 min).
• Chiral Ag nanoclusters were synthesized following literature protocols, using silver nitrate and (R/S)-thiazolidine-2-thione ligands; crystals were dissolved in tetrahydrofuran (5 mg mL⁻1) and spin-coated at 6000 rpm for 30 s to form ultrasmooth nanocluster films. Volatile solvent ensured minimal aggregation; resulting film roughness RMS ≈0.45 nm.
• Pentacene (≈30 nm) was thermally evaporated atop the nanocluster layer at ≈0.1 Å s⁻1 under ≈6×10⁻4 Pa. Au source/drain electrodes (≈30 nm) were deposited through a shadow mask, defining channels W/L = 4500 µm / 20 µm. Arrays on a 4-inch wafer included 200 tested photoreceptors.

Controls and comparative devices:

• Control devices without Ag nanoclusters or with a 7 nm tetratetracontane interlayer exhibited faint photoresponse and negligible hysteresis even under 10 mW cm⁻2 white light, underscoring the necessity of the nanocluster–molecule interface (NMI). Additional controls with quantum dots or semiconductor interlayers corroborated the indispensability of the NMI.

Characterization techniques:

• Structural and morphological: Optical microscopy, scanning near-field optical microscopy (SNOM), cryo-TEM; XRD to assess crystallinity (nanoclusters showed no significant peaks; pentacene crystallinity suppressed on nanoclusters vs SiO2).
• Electrical/optical: Semiconductor parameter analyzer (Keithley 4200A SCS) and sourcemeters (Keysight B2912A) for transfer characteristics, hysteresis windows, and real-time photoresponse under varied Vg, VDS, light wavelength/intensity; CPL measurements using circular dichroism (CD) spectroscopy and CPL illumination (e.g., 270 nm, up to 1 mW cm⁻2), with both S- and R-enantiomer Ag nanoclusters.
• Spectroscopy and dynamics: Time-resolved photoluminescence (TRPL) for interfacial charge transfer; femtosecond transient absorption spectroscopy (pump 580 nm and 365 nm) with global target analysis to resolve localized singlet (LE), charge-transfer (CT), and triplet (T) states, extracting lifetimes (τ) and pathways (including singlet fission competition).
• Surface potential and charge modulation: Kelvin probe force microscopy (KPFM) under varied wavelengths (473, 532, 633 nm) and intensities to monitor photoadaptation-induced hole depletion in pentacene.

Operation protocols and metrics:

• Hysteresis window vs light intensity/wavelength and Vg scan; synergistic Vg–light pulsing for light-valve operation; adaptation dynamics fitted to Iadaptation = Aadaptation exp(−t/τdec) + Iinitial.
• Definition and extraction of perception range (PR = 20 log10(Imax/Imin)), adaptation time (time from Imax to Iadapt), and accuracy = (Imax − Iadapt)/(Imax/Imin).
• Imaging demo: 5×5 ACP pixel array exposed to colored letters (R/G/B/UV) with preconditioning pulse (Vg = 60 V, 1 s; light 2 mW cm⁻2, 1 s) and readout after defined intervals to demonstrate spectral-dependent adaptation and pattern recognition.

• Wafer-scale integration: Functional ACP arrays fabricated on 4-inch wafers with a smooth pentacene/Ag nanocluster heterointerface; nanocluster film RMS roughness ≈0.45 nm.
• Charge reservoir and light-valve behavior: Under illumination, counterclockwise hysteresis with VT shift ≈45 V indicates light-amplified trapping/recombination. Photocurrent increases at positive Vg but maximum on-state current decreases with higher light intensity, evidencing dual modulation. Synergistic Vg+light pulses increased IDS from ~1×10⁻10 A to ~1×10⁻5 A.
• Spectral-dependent photoadaptation: Adaptation dynamics Iadaptation = Aexp(−t/τdec) + Iinitial with wavelength-dependent τdec: τdec(365 nm) ≈ 0.25 s, τdec(460 nm) ≈ 46.4 s, τdec(525 nm) ≈ 286 s, τdec(620 nm) ≈ 290 s. High-energy photons induce faster adaptation via interfacial recombination and hole depletion.
• Dynamic range and retention: ~10⁵ dynamic current range maintained over 10,000 s; repeatability confirmed over 165 photoresponse/photoadaptation cycles. Multi-level photoresponses achieved by varying Vg impulse amplitude without changing light intensity. Shelf life: retention preserved after one year; humidity >40% elevates off-state current, reducing dynamic range.
• Performance metrics: Imax/Imin up to 10⁶ gives PR ≈ 120. Under Vg = −30 V and UV (365 nm, 0.8 mW cm⁻2), adaptation time ≈ 0.45 s with accuracy ≈ 99.75%. Pentacene mobility on nanoclusters ~0.027 cm² V⁻1 s⁻1 vs 0.055 cm² V⁻1 s⁻1 on SiO2 (limited impact on biomimetic functions).
• Interfacial charge-transfer evidence: TRPL lifetimes shortened when interfaced (from 1.54 µs and 5.74 µs to 0.78 µs and 3.99 µs), indicating efficient hole injection into pentacene. Transient absorption/global analysis: CT lifetimes reduced from 74 ps to 12 ps (pump 580 nm) and from 21 ps to 4.9 ps (pump 365 nm) in pentacene/Ag nanoclusters vs pentacene alone, accelerating recombination to triplet states and governing adaptation. KPFM shows surface potential decreases with intensity and at shorter wavelengths, consistent with hole depletion.
• Ligand-assisted transfer pathway: Phenyl ligands enable π–π interactions with pentacene, forming efficient electron-capture/charge-transfer pathways; alkyl ligands hinder transfer. Spin density localized on Ag cores supports electron reservoir role.
• All-in-one functions: ACPs demonstrate color vision (UV–RGB), tunable photoadaptation (photopic/scotopic modes via Vg), and CPL perception with memory.
• CPL recognition and memory: Chiral Ag nanoclusters exhibit gCD ≈ ±1.3×10⁻3 at 277 nm. Devices with S-Ag show higher photocurrent under RCPL than LCPL; R-Ag shows opposite handedness selectivity. Upon CPL exposure (e.g., 270 nm, 100 µW cm⁻2), currents evolve to a steady state and decay exponentially after light off, enabling CPL memory; retention and discrimination sustained and reproducible, with maintenance and distinguishability after 100,000 s inferred. Adaptive response to low-density CPL yields ΔI ~0.19 with Tadapt ~7 s.
• Imaging demonstration: A 5×5 ACP array distinguishes colored letter patterns after defined adaptation intervals, showcasing spectral-dependent adaptation and spatially resolved recognition.

The findings validate a biomimetic strategy that leverages a nanocluster–molecule interface to emulate key aspects of mantis shrimp vision in a compact, single-cell device. The chiral Ag nanoclusters act as an in-sensor electron reservoir with light-tunable Fermi levels, enabling a light-valve mechanism that co-modulates channel conductance with gate voltage and illumination. This mechanism underpins the observed spectral- and intensity-dependent photoadaptation, allowing the device to self-regulate across a wide perception range and rapidly adapt to changing light conditions. The interface’s ligand-mediated charge transfer accelerates carrier dynamics (shortened CT lifetimes), directly dictating photoadaptation kinetics and providing a physical basis for the multi-state outputs. Incorporating chirality confers CPL sensitivity and, notably, memory—capabilities not commonly present together in CPL detectors—allowing discrimination and retention of handedness information. Collectively, the ACP integrates color vision, adaptive regulation, and CPL detection with memory on a wafer scale, addressing the need for structurally simple, multifunctional AVSs and advancing nanocluster-based neuromorphic optoelectronics.

This work introduces wafer-scale artificial nanocluster photoreceptors that unify color vision, spectral- and intensity-dependent photoadaptation, and circular polarization vision with memory in a simple pentacene/chiral Ag nanocluster heterostructure. The core-shell nanoclusters serve as an electron reservoir and, assisted by aromatic ligands, establish efficient interfacial charge-transfer pathways that realize a light-valve mechanism for adaptive conductance control. Quantitatively, the devices achieve PR ≈ 120 with Imax/Imin ≈ 10⁶, adaptation times down to ~0.45 s (accuracy ~99.75%), dynamic range spanning ~10⁵ over 10,000 s, and robust repeatability. Transient spectroscopy and KPFM substantiate the interfacial dynamics governing adaptation. The demonstrated CPL sensing with memory (gCD ≈ ±1.3×10⁻3; handedness-dependent photocurrents; long retention) further expands functionality. These advances provide a concise platform to interrogate nanocluster optoelectronics and offer design guidelines for compact, multi-task AVSs and neuromorphic devices. Future directions include optimizing ligand chemistry and nanocluster energetics for enhanced selectivity and speed, improving environmental stability (e.g., humidity tolerance), scaling to high-density pixel arrays, and integrating on-chip processing for advanced, encrypted, event-driven vision.

• Environmental sensitivity: Device off-state currents increase significantly at humidity levels above ~40%, which can reduce dynamic range.
• Mobility/crystallinity trade-offs: Pentacene mobility is reduced on nanocluster layers (0.027 vs 0.055 cm² V⁻1 s⁻1 on SiO2), reflecting suppressed crystallinity; while not severely impacting functions here, it may limit ultimate speed or gain in other regimes.
• Interface formation challenges: Ag nanoclusters can aggregate if not rapidly processed with volatile solvents; achieving uniform, ultrasmooth films requires careful control of nucleation and spin-coating conditions.
• CPL metrics: The dissymmetry factor (|gCD| ~1.3×10⁻3) is modest, which may limit CPL contrast without optimization of nanocluster structure or optical coupling.
• Retention claims for CPL memory at very long timescales (e.g., 100,000 s) are inferred from decay behavior; extended long-term stress testing under varied environments would strengthen reliability assessments.