Infrared neural stimulation and inhibition using an implantable silicon photonic microdevice

Temperature critically influences physiological processes in the brain, yet the spatial and temporal patterns of brain temperature and their roles in homeostasis remain insufficiently characterized. Both endogenous fluctuations and exogenous sources (e.g., optogenetics, multiphoton imaging, electrical and magnetic stimulations) can produce significant heating that may affect neural function. Infrared neural stimulation (INS; 1400–2100 nm) enables contactless thermal transients without genetic modification and has been shown to modulate neural systems, including GABAergic neurotransmission, cardiac pacing, visual cortex activation, and auditory neurons. Infrared neural inhibition (INI) can suppress neural activity and may require lower radiant exposures than INS, potentially benefiting portable applications; its mechanism likely involves baseline temperature increases affecting temperature-dependent ion channel kinetics. Despite progress, in vivo demonstrations targeting intracortical and hippocampal structures with integrated tools have been lacking. Multifunctional implantable tools that can both induce controlled thermal changes and record neural signals with high spatial precision are needed. The authors present an implantable silicon-based multimodal photonic neural probe that delivers 1550 nm IR light and simultaneously records electrophysiological signals and local temperature, enabling reversible excitation or inhibition of neuronal firing in deep brain regions. They establish safe operation via systematic optical, thermal, and electrochemical characterization and provide in vivo validation in rat cortex and hippocampus.

Device design and fabrication: A multifunctional implantable microsystem was fabricated from single-crystalline silicon, integrating photonic delivery, temperature sensing, and electrophysiological recording on a single probe. The probe shaft is 5 mm long, 170 µm wide, and 190 µm thick, acting as a multimode optical waveguide for IR delivery (λ = 1550 nm) and carrying thin-film platinum structures for recording electrodes and a temperature sensor. Low-loss IR transmission is achieved by wet chemical nanomachining to reduce surface roughness. IR light from a pigtailed laser diode (max 70 mW) is coupled via an optical fiber and micro-optical structures; light exits at the probe tip to be absorbed in tissue. The assembled chip mounts on a custom PCB providing optical and electrical connections. Electrochemical optimization includes depositing porous platinum on sputtered platinum to lower recording site impedance.

Optical and electrical characterization: The overall optical efficiency delivering IR at chip scale averaged 32.04 ± 4.10% (max packaged 41.5 ± 3.29% from prior work). For the device set used here, optical efficiency was 24.2 ± 6.9% (n = 5). Beam spot size at the tip averaged 0.024 ± 0.006 mm². The integrated platinum temperature sensor had a temperature coefficient of resistance (TCR) of 2636 ± 75 ppm/K (n = 5). Electrode impedance at 1 kHz was reduced from 678 ± 198 kΩ to 46 ± 9 kΩ (n = 4) after porous platinum deposition.

In vitro thermal mapping and calibration: The probe was immersed in water while a calibrated external platinum temperature sensor was positioned with micrometric control to map spatial temperature distribution around the tip. Temperature rise profiles were measured along axes perpendicular to the shaft to determine the decay of heat from IR absorption. The full width at half maximum (FWHM) of the temperature distribution along the y-axis (x = 200 µm) was 1020 ± 184 µm. Two-dimensional distributions were reconstructed from measurements. Calibration curves relating laser diode operating current to optical power, and optical power to temperature rise, were established prior to in vivo experiments. The integrated temperature sensor’s readings were compared to external and fiber-optic temperature sensors; external measurements exceeded integrated readings by 24 ± 6%, and in vivo temperature data were rescaled accordingly. The effect of fiber core diameter on temperature profile shape along y was assessed, showing slight distribution changes attributable to beam shape.

In vivo experiments in rats: Acute experiments were conducted in rat neocortex and hippocampus to test neuromodulation and recording. A multimodal optrode and a laminar silicon probe (control) were implanted. Stimulation protocols used continuous-wave 1550 nm IR light in randomized trials with 2-minute ON and 4-minute OFF periods, at multiple optical power levels. At least 10 trials per power were recorded. Multiunit and single-unit activity near the irradiated region were monitored; the integrated temperature sensor concurrently measured local temperature dynamics for model validation. The electrophysiological waveforms were examined for artifacts and stability over hours. Spatial positioning included cortical depths around 1300 µm and 1600 µm and hippocampal CA1 pyramidal layer placements. A coupled optical-thermal multiphysics model estimated local temperature distributions and dynamics at recording sites for each arrangement.

Safety monitoring: Rectal body temperature was tracked during stimulation sequences to assess systemic thermal effects; no correlation with local brain temperature elevations was observed.

• Device performance: Optical efficiency for the devices used was 24.2 ± 6.9% (n = 5); integrated temperature sensor TCR was 2636 ± 75 ppm/K (n = 5). Recording site impedance at 1 kHz decreased from 678 ± 198 kΩ to 46 ± 9 kΩ after porous platinum deposition (n = 4).
• In vitro thermal spread: The FWHM of the temperature rise distribution perpendicular to the shaft (y-axis at x = 200 µm) was 1020 ± 184 µm, indicating sub-millimeter to millimeter-scale thermal spread. External temperature measurements were 24 ± 6% higher than integrated sensor readings; in vivo data were rescaled accordingly.
• Artifact-free recordings: No visible stimulus-onset artifacts were detected in raw electrophysiological data; action potential waveforms remained stable over several hours, suggesting no acute damage.
• Cortical excitation: At cortical depth ~1300 µm, multiunit activity significantly increased during IR illumination at powers 6.9, 8.5, and 10.5 mW (p < 0.01).
• Cortical inhibition: At cortical depth ~1600 µm, multiunit activity was significantly suppressed during IR illumination at higher powers 6.9, 8.5, and 10.5 mW (p < 0.01).
• Hippocampal excitation: In the CA1 pyramidal layer, multiunit activity significantly increased at powers 2.8, 7.1, 10.7, and 13.4 mW (p < 0.01).
• Thermal modeling: Concurrent temperature sensing and a validated optical-thermal model estimated local temperature dynamics at recording sites and supported interpretation of excitation versus inhibition regimes.
• Systemic safety: Rectal body temperature showed no significant changes during stimulation sequences, consistent with localized brain heating.

The study demonstrates that precisely delivered infrared light via an integrated silicon photonic microprobe can modulate neural activity bidirectionally in deep brain tissue while enabling simultaneous, artifact-free electrophysiological and local temperature measurements. These results align with known temperature influences on neuronal physiology and metabolism, where modest temperature changes can affect ion channel kinetics, synaptic transmission, and network excitability. The observed excitation at certain depths and powers and inhibition at deeper cortical sites at higher powers are consistent with temperature-dependent mechanisms, including differential sensitivity of neuronal elements and gating kinetics that may lead to reversible suppression when baseline temperature rises sufficiently. The combination of in vitro thermal mapping, in vivo sensing, and a validated multiphysics model provides a coherent framework to estimate temperature distributions at recording locations and to design stimulation paradigms that target excitation or inhibition while maintaining safety. The platform advances experimental capability to interrogate thermally evoked responses with spatial and temporal precision in vivo, offering a versatile tool for probing the role of temperature in neural function and pathology.

This work introduces and validates a multimodal implantable silicon photonic microprobe that delivers 1550 nm infrared light and concurrently records local temperature and extracellular neural activity. The device enables reversible excitation and inhibition of neuronal firing in rat cortex and hippocampus without recording artifacts, with characterized optical efficiency, sensitive on-chip thermometry, and low-impedance electrodes. In vitro and modeled thermal profiles guide safe, targeted neuromodulation. To the authors’ knowledge, this is the first integrated microdevice to achieve deep-tissue IR neuromodulation with simultaneous electrophysiological readout. The approach is a promising experimental toolset to reveal thermally evoked neural responses in vivo and to inform the safe operation of thermally impactful diagnostic and therapeutic modalities. Future work can leverage this platform to map parameter spaces for excitation versus inhibition across brain regions and to explore chronic and behavioral applications.

• Temperature sensing required calibration: external measurements exceeded on-probe sensor readings by 24 ± 6%, necessitating rescaling; residual discrepancies may affect absolute temperature estimates in vivo.
• In vitro thermal characterizations were performed in water; while supported by a validated model, translation of exact spatial temperature profiles to brain tissue may introduce uncertainties.
• Neuromodulation used continuous-wave 1550 nm illumination in acute rat preparations; results may not directly generalize to other wavelengths, pulsed regimes, chronic implants, or other species.
• Reported optical efficiency and electrode metrics (e.g., impedance) were characterized on small sample sizes (e.g., n = 4–5), which may limit generalizability of device performance statistics.