Circularly polarized light-sensitive, hot electron transistor with chiral plasmonic nanoparticles

Abstract

The quantitative detection of circularly polarized light (CPL) is necessary in next-generation optical communication carrying high-density information and in phase-controlled displays exhibiting volumetric imaging. In the current technology, multiple pixels of different wavelengths and polarizers are required, inevitably resulting in high loss and low detection efficiency. Here, we demonstrate a highly efficient CPL-detecting transistor composed of chiral plasmonic nanoparticles with a high Khun's dissymmetry (g-factor) of 0.2 and a high mobility conducting oxide of InGaZnO. The device successfully distinguished the circular polarization state and displayed an unprecedented photoresponsivity of over 1 A/W under visible CPL excitation. This observation is mainly attributed to the hot electron generation in chiral plasmonic nanoparticles and to the effective collection of hot electrons in the oxide semiconducting transistor. Such characteristics further contribute to opto-neuromorphic operation and the artificial nervous system based on the device successfully performs image classification work. We anticipate that our strategy will aid in the rational design and fabrication of a high-performance CPL detector and opto-neuromorphic operation with a chiral plasmonic structure depending on the wavelength and circular polarization state.

Fig. 1. Schematic of the chiral gold nanoparticles and CPL-detecting transistor.

a Schematic of the chiral gold nanoparticle array used as a medium to distinguish LCP light and RCP light. b Circular dichroism and extinction data of the chiral gold nanoparticle array. c Schematic of the CPL detector consisting of a chiral gold nanoparticle array integrated with an InGaZnO transistor that acts as a hot electron acceptor. d Corresponding energy band diagram of the CPL-detecting transistor.

Fig. 2. Photocurrent and activation energy of the IGZO-chiral gold nanoparticle array.

a, e Output curve of the CPL-detecting transistor under illumination with 635 nm and 780 nm CPL, respectively. The power density of LCP and RCP light is precisely calibrated at 3.7 mW/cm², and the active area is 100 μm $\times$ 50 μm. b, f Magnified output curve when the gate voltage is zero. c, g Photocurrent with respect to gate voltage under 635 nm and 780 nm CPL illumination, respectively. d, h Activation energy depending on the gate voltage under LCP and RCP light illumination at 635 nm and 780 nm wavelengths, respectively.

Fig. 3. Simulation analysis of a chiral gold nanoparticle and the IGZO/chiral gold nanoparticle interface.

a Multipole expansion of the total extinction of a single chiral gold nanoparticle of 180 nm. b, c Surface charge distribution under linear excitation (x-polarization) with 635 nm and 780 nm, respectively. d, e E-field enhancement and its corresponding current distribution in the yz plane (yellow arrow) at x = −80 nm under linear excitation (x-polarization) with 635 nm and 780 nm, respectively. f, g Difference between the E-field enhancement under LCP and RCP excitation in the xy plane, which is the interface between the chiral gold nanoparticle and IGZO under 635 nm and 780 nm light excitation, respectively. The integrated value is −1.20 $\times$ 10⁻¹⁵ m² (635 nm) and 7.10 $\times$ 10⁻¹⁵ m² (780 nm).

Fig. 4. Polarization-dependent broadband transient absorption (TA) dynamics in chiral gold nanoparticle arrays.

a Time-resolved TA (ΔT/T) surface maps of the chiral gold nanoparticle array measured with RCP (top) and LCP (bottom) pump beams. b TA spectral cut taken at a pump-probe time delay of 0.5 ps under RCP (blue) and LCP (red) pump illumination. Note that the TA signal values in the range of 630–680 nm (between the dashed lines), where strong pump scattering noise (λ_pump = 650 nm) interferes with the weak TA signal, are intentionally set to zero for a better view of the TA signals in other spectral regions of interest. c, d Photobleaching (PB) traces as a function of pump-probe time delay with RCP (blue) and LCP (red) pump beams at 580 nm and 770 nm, respectively.

Fig. 5. Photocurrent of the CPL-detecting transistor with a tunneling oxide, and suggested hot electron transport mechanism depending on the gate voltage.

a Photocurrent of the CPL-detecting transistor under 635 nm CPL illumination when the device has a tunneling HfO₂ film of various thickness at the interface between the chiral gold nanoparticles and InGaZnO layer. b Corresponding photocurrent with respect to tunneling length, and fit to an exponential decay. c–e Hot electron migration schematics when the gate voltage is zero (c) and positive (e). Magnified energy band bending of the Schottky barrier depending on the gate voltage-dependent InGaZnO energy band bending of the CPL-detecting transistor (d).

Fig. 6. Synaptic Characteristics of CPL-Detecting Transistor.

a Delta post synaptic current under CPL pulse illumination at 635 nm depending on frequency. The light intensity was set to 4.9 mW (b) PPF(paired pulse facilitation) as a function of pulse interval time. The graph is fitted into double exponential function: y = A₁ exp(−t/τ₁), + A₂ exp(−t/ τ₂), where t is the interval time. c Delta post synaptic current and (d) synaptic weight trend after light is turned off according to CPL pulse number.

Fig. 7. Polarization Dependent Image Classification.

a Artificial neural network structure based on circularly polarized light sensory neuron to perform handwritten digit classification work using the MNIST database, and schematic of machine vision featuring differences in learning efficiency depending on the circular polarization state. b Potentiation curve under 1 Hz CPL pulse illumination at a 635 nm wavelength, and decaying curve without light illumination. The intensity of CPL was set to 4.9 mW (c), Magnified post synaptic current between 5.0 and 10.0 s, in which the current well responds to the 1 Hz CPL pulse input. d Handwritten digit classification simulation results based on potentiation curve at 635 nm LCP and RCP light pulses. e Potentiation curve under 1 Hz CPL pulse illumination at a 780 nm wavelength, and decaying curve without light illumination. f Magnified post synaptic current between 5.0 and 10.0 s, in which the current well responds to the 1 Hz CPL pulse input. g Handwritten digit classification simulation results based on the potentiation curve at 780 nm LCP and RCP light pulse.