Nanoscale light field imaging with graphene

Nanoscale light manipulation and characterization underpin modern nano-optics. Considerable progress has been made using plasmonic structures, superlenses, plasmonic waveguides, graphene plasmonics, and optical phononics. However, most light-field imaging still relies on microscopy of scattered propagating fields and is subject to the diffraction limit, constraining spatial resolution to a fraction of the wavelength. Near-field optical microscopy can improve resolution but suffers from the fundamental sixth-power scaling of scattered power with probe size, practically limiting direct light mapping to around 50 nm at optical wavelengths, and it is inherently invasive. Furthermore, common s-SNOM techniques require harmonic demodulation to suppress background, which can distort images of fields with different spatial frequencies. Electron microscopies such as EELS and CL provide optical-response information with unmatched spatial resolutions via tightly focused electron beams, but they give indirect access to optical fields and typically probe resonant excitations, requiring elaborate data processing. Other indirect methods based on scattered fields can achieve extremely high spatial resolution down to 1–10 nm (e.g., fluorescent hotspot mapping, SERS, s-SNOM), but are often invasive and rely on assumptions to translate signals into field maps. This work introduces a conceptually different, direct, and virtually non-invasive nanoscale light-field imaging method with ~20 nm lateral resolution. It uses the photoelectric effect in a graphene p-n junction whose position can be accurately controlled by gate voltages. The graphene surface defines the imaging plane and is placed near the nanostructure producing the nanoscale fields. As a demonstration, the electric-field profile of a tightly confined plasmonic slot-waveguide mode at telecom wavelengths is mapped and found to agree excellently with simulations. The method circumvents fundamental limitations in near-field microscopy by converting light directly to electrical signals without mechanical scanning, promising precise, high-resolution light-field mapping.

The paper reviews limitations of conventional and advanced light-field imaging approaches. Far-field microscopy of scattered light is diffraction-limited. Near-field techniques improve resolution by accessing evanescent fields but are constrained by sixth-power scaling of scattered power with probe size, practically limiting direct mapping to ~50 nm and introducing invasiveness and potential image distortions due to high-harmonic demodulation in s-SNOM. Electron-based techniques (EELS, CL) achieve sub-nanometer spatial resolution and map resonant polaritonic excitations indirectly through electron energy loss or cathodoluminescence, requiring complex data interpretation. Additional indirect optical methods achieve nanometer-scale mapping (e.g., single-molecule fluorescence localization of hotspots with ~1 nm resolution, SERS-based field evaluation ~5 nm, s-SNOM backscatter mapping <10 nm), but they are invasive and rely on translating measured signals into field distributions under specific assumptions. The literature context motivates a direct, minimally invasive approach that electrically reads out local optical fields without relying on scattering or non-linear processes.

Concept and operating principle: A monolayer graphene sheet is placed adjacent to the optical nanostructure of interest and contacted by source and drain electrodes made of the same material and kept at the same temperature to eliminate thermoelectric parasitics. A thin high-k gate dielectric (30 nm HfO2) covers graphene, and two parallel metallic gates are fabricated on top to induce and laterally position a graphene p-n junction via applied voltages. Illumination generates electron–hole pairs; at the p-n junction, built-in fields separate carriers, producing a photoelectric current/voltage proportional to the local optical field intensity where the junction resides. By sweeping a gate voltage, the junction is displaced laterally across the region, enabling electrical scanning of the optical field without moving parts. Electrostatic modeling (established methods) determines Fermi energy variation, junction position versus gate voltage, and built-in electric field profile, yielding a junction (“photo-active”) width ~20 nm that defines spatial resolution. Device design and fabrication: Two CVD graphene flakes were wet-transferred onto Si substrate with 1.5 μm SiO2. Graphene was patterned into two 4 μm-wide strips (detector and modulator/detector). A 30 nm HfO2 layer was e-beam evaporated above the graphene to serve as the gate dielectric. A 100 nm Ag plasmonic slot waveguide with a 300 nm-wide slot (including a 90° bend) and Ag contacts to graphene were fabricated by e-beam lithography and lift-off; a 1 nm Cr adhesion layer was used under Ag. Dipole nano-antennas with back reflectors were added at waveguide ends for efficient free-space coupling. The device was encapsulated in a 1.5 μm PMMA layer to prevent Ag oxidation. Fabrication steps included PMMA-assisted transfer, copper etch in ammonium persulfate, cleaning, lithography, O2:Ar plasma etch, HfO2 evaporation (0.6 Å/s), high-resolution e-beam patterning (100 kV), low-temperature development, metal deposition (Ag 0.9 nm/s), and SEM inspection. Optical excitation and electrical detection: A 1550 nm laser (focused with a 40×, NA 0.65 objective) polarized along the input nano-antenna excited the gap surface plasmon (GSP) mode. Output was monitored with cross-polarized detection and a SWIR InGaAs camera to verify in/out coupling. Photovoltage between source and drain was measured as a function of gate voltage. Two modalities were used: lock-in detection with an optical chopper (800 Hz) to isolate photocurrent due to GSP absorption in graphene, and direct measurements using a sourcemeter to record dark and illuminated I–V and V–V characteristics. Graphene gating characteristics were obtained by applying the same gate voltage to both G1 and G2 while grounding graphene. Light modulation via Pauli blocking was demonstrated by applying a 1 Hz, 6 V amplitude, 1 V offset square-wave to both gates, achieving 12% output modulation (0.12 dB·μm−1), establishing that EF can reach half the photon energy at ~5.5 V. Field mapping protocol: To map the transverse field across the slot, gate G2 was swept while G1 and source were grounded, forming and moving a p-n junction under the illuminated waveguide. The photovoltage vs gate (with lock-in) was recorded at 1550 nm with 0.5 mW input. The gate–position calibration for the junction was obtained from electrostatic modeling, enabling reconstruction of the GSP electric-field intensity profile in the graphene plane. Numerical simulations: 3D FDTD (Lumerical) modeled laser coupling, GSP propagation, and out-coupling, yielding Pout/Pin ≈ 0.12% (experiment ~0.1%). For accurate cross-sectional mode profiles, 2D FEM mode analysis (COMSOL RF) was used, with rounded Ag edges (20 nm radius) to avoid singularities, material parameters matching FDTD, graphene neglected for mode profile, triangular mesh (≤10 nm in Ag, ≤190 nm/n elsewhere), refined mesh along graphene position (down to ~1 nm), PEC boundaries, and convergence checks. Alternative FDTD software cross-validated waveguide profiles. Thermoelectric contributions were analyzed in linear approximation (Supplementary) and found insufficient to explain data.

- Demonstrated direct, electrical nanoscale light-field imaging using a movable graphene p-n junction, achieving an experimental spatial resolution of approximately 22 nm FWHM (Gaussian apparatus width ~9.5 nm), consistent with calculated photo-active junction width below 20 nm across gate voltages. - Successfully mapped the transverse electric-field intensity profile of a gap surface plasmon mode in a 300 nm-slot Ag waveguide at 1550 nm. Reconstructed profiles from photovoltage vs gate agreed excellently with COMSOL/FDTD simulations; discrepancies at high gate voltages were attributed to dielectric leakage affecting capacitance. - Photovoltage was negligible below ~0.2 V on G2 (no junction in illuminated region), increased sharply with gate, and peaked around 5 μV at ~0.4 V under 0.5 mW input. Photocurrent/photovoltage exhibited linear dependence on incident power and linear I–V characteristics near zero bias; measured photocurrents were on the order of nA, close to a simulated maximum ~3 nA. - Device optics: Measured Pout/Pin ≈ 0.1% matched FDTD prediction ~0.12%. Electro-optic modulation via Pauli blocking reached 12% (0.12 dB·μm−1) with a 6 V gate amplitude; Pauli blocking threshold inferred near 5.5 V. - Graphene absorption (free-space ~2.3%, guided GSP 0.01–0.1 dB·μm−1) minimally perturbed fields, supporting the method’s non-invasiveness. - Thermoelectric effects were analyzed and, under the symmetric contact and isothermal conditions used, were not able to account for the observed photoresponse; the photoelectric effect at the p-n junction dominated.

The study addresses the challenge of direct, non-invasive nanoscale optical field mapping beyond diffraction and practical near-field limits by converting local optical intensity into an electrical signal at a controllably positioned graphene p-n junction. By sweeping a gate, the junction traverses the optical mode cross-section without mechanical motion, enabling precise spatial registration. The electric field profile reconstructed from photovoltage vs gate position matches simulated GSP mode intensity, validating the method as a faithful field mapper. Linear power dependence and agreement in photocurrent magnitudes support a photoelectric origin of the response; under equal-material, isothermal contacts, thermoelectric contributions should cancel in metals, and calculated thermoelectric signals could not explain the measurements. Graphene’s weak absorption ensures minimal perturbation of the field, preserving non-invasiveness. Remaining deviations at higher gate biases likely arise from dielectric leakage and breakdown limits that affect position calibration. The approach is naturally compatible with planar nanophotonic circuitry and can extend to other gating symmetries and potentially to 2D mapping with more complex gate architectures, suggesting broad utility for characterizing strongly confined on-chip optical modes.

The work proposes and experimentally demonstrates a direct, electrical method for nanoscale light-field imaging using a graphene p-n junction that is induced and moved by gate voltages. The technique achieves ~20 nm spatial resolution (∼22 nm FWHM) determined by the junction width and reconstructs the transverse field profile of a plasmonic slot-waveguide mode at telecom wavelengths in excellent agreement with simulations. The configuration also exhibits efficient electro-optic modulation (0.12 dB·μm−1 at 6 V). By eliminating mechanical scanning and relying on minimal field perturbation, the method establishes a new paradigm for precise, non-invasive nanoscale optical characterization. Future directions include improving resolution by reducing dielectric thickness and operating temperature, optimizing gate geometries for 2D mapping, mitigating dielectric leakage to extend gate range, and integrating fully graphene-based gating to further minimize invasiveness.

- Current implementation demonstrates one-dimensional mapping; extending to full 2D imaging requires more complex gate architectures. - Spatial resolution is limited by the p-n junction width, which depends on dielectric thickness and operating conditions; thinner dielectrics or lower temperatures are needed for further improvement. - Maximum junction displacement is limited by dielectric breakdown and leakage currents at higher gate voltages, which can distort the capacitance-based position calibration and introduce discrepancies. - Fabrication is relatively complex, involving high-quality graphene transfer, precise dielectric deposition, and nanoscale metal patterning; device geometry constraints may limit the range of structures that can be probed. - Thermoelectric effects are minimized by design but could contribute in semiconducting or asymmetric contact scenarios, requiring careful control of thermal and material symmetry conditions.