Battery-free energy harvesting for implanted neural devices — inductive coupling, ultrasound, and RF mid-field powering through living tissue
Every implantable neural device — deep brain stimulators, cochlear implants, peripheral nerve cuffs, closed-loop seizure monitors — has historically been limited in size by one component: the battery. A lithium battery capable of powering a stimulator for 5–10 years occupies a volume many times larger than the electronics it feeds. Removing the battery entirely and harvesting power wirelessly from outside the body unlocks devices small enough to thread into a peripheral nerve, sit on the cortical surface, or float free as "neural dust."
A conventional implantable pulse generator (IPG) — the kind used in deep brain stimulation or spinal cord stimulation — packages a primary or rechargeable lithium cell alongside its stimulation ASIC, capacitors, and header. The battery alone typically accounts for 60–80% of total device volume. It also dictates surgical strategy: IPGs are placed in a subcutaneous pocket (chest or abdomen) with a lead tunneled to the target, because the battery itself is too bulky to sit at the stimulation site.
Removing the battery collapses the whole design problem. Without a multi-year energy reservoir to carry, the implant only needs electronics sized to a stimulation ASIC, a small antenna/coil/transducer, and a modest energy-buffering capacitor. This is what allows implants to shrink from centimeter-scale cans to devices measured in cubic millimeters — small enough to be injected through a needle rather than surgically implanted.
The "neural dust" concept proposed by the Maharbiz/Carmena/Alon groups at UC Berkeley (2013) envisioned free-floating, ultrasonically powered sensor motes as small as 0.8 mm³ — roughly the size of a large grain of sand — scattered across the surface of peripheral nerves or the cortex, each one battery-free and individually addressable.
Smaller, battery-free implants change what is surgically and clinically feasible:
• Peripheral nerve interfaces: cuffs or intrafascicular electrodes small enough to wrap individual nerve fascicles without the mass of a lead-and-can system pulling on delicate nerve tissue • Cortical surface arrays: micro-scale ECoG or penetrating arrays that sit directly at the recording/stimulation site instead of tethering to a distant battery pack via percutaneous or subcutaneous leads • Injectable / minimally invasive placement: devices delivered via needle or catheter rather than open surgery, reducing infection risk and recovery time • Elimination of battery-replacement surgery: primary-cell IPGs require replacement procedures every 3–9 years; a battery-free device that harvests power on demand never needs to be explanted for power reasons • Distributed sensor/stimulator networks: dozens to hundreds of independent battery-free motes can be seeded across a neural structure, something impossible if each required its own battery and percutaneous lead
Removing the battery does not remove the need for energy — it relocates the energy source outside the body. A wearable or nearby external transmitter must continuously (or periodically) beam power through skin, fat, muscle, and bone to reach the implant, and the implant must harvest, rectify, and store just enough of that energy to perform its job: sense a signal, process it, and deliver a stimulation pulse or transmit data back out.
This reframes implant design as a power-budget problem. Every microjoule delivered wirelessly is precious, tissue heating limits how much power can safely be transmitted, and transfer efficiency falls steeply with depth and misalignment. The remaining four stages of this simulation follow that energy on its journey — from an external source, through tissue, into on-chip storage, and finally into a delivered neural stimulation pulse.
Three physical mechanisms dominate wireless power transfer into tissue, each with a different depth/size/efficiency tradeoff. The external transmitter must be selected — and positioned — to match the implant's depth, size, and orientation tolerance. There is no single best modality; the choice depends entirely on where in the body the implant sits.
Inductive coupling uses two coils — an external transmit coil (Tx) and a small implant receive coil (Rx) — forming a loosely coupled transformer across the skin. An oscillating current in Tx (typically 1–13.56 MHz) creates a time-varying magnetic field that induces a voltage in Rx via Faraday's law.
Strengths: mature technology (used in cochlear implants and pacemaker chargers for decades), very high efficiency (>50%) when Tx and Rx are close (a few mm) and well aligned, simple circuitry.
Weaknesses: near-field coupling strength falls off roughly as the cube of coil separation, so efficiency collapses rapidly with depth — workable at a few millimeters, marginal beyond 1–2 cm for mm-scale receive coils. Also highly sensitive to lateral misalignment and rotation between Tx and Rx coils.
Ultrasonic powering uses a piezoelectric transducer array on the skin to launch an acoustic pressure wave (typically 1–10 MHz) that travels through tissue and is converted back to electrical energy by a piezoelectric receiver on the implant.
Strengths: acoustic waves attenuate far more slowly in soft tissue than electromagnetic fields at comparable frequencies, enabling useful power delivery at 5–10 cm depth. Piezoelectric receivers can be extremely small (sub-mm), matching neural dust scale. Wavelength at MHz frequencies is short enough to focus energy tightly, improving spatial selectivity when many implants are present.
Weaknesses: acoustic impedance mismatch at the skin-air and bone interfaces requires gel coupling or careful transducer design; bone strongly scatters/absorbs ultrasound, complicating intracranial use; transducer alignment still matters, though less critically than inductive coupling.
Conventional far-field RF (radiative) power transfer is inefficient at delivering power to sub-wavelength implants deep in tissue, because most of the radiated energy never reaches the small receive antenna. Mid-field powering (Poon, O'Driscoll, Meng, 2010) exploits the fact that tissue itself behaves as a high-permittivity medium: at frequencies around 1–2 GHz, a specially shaped external source can create a field pattern that behaves like near-field coupling at the body surface but propagates like a wave once inside tissue — concentrating energy at depth onto a millimeter-scale antenna.
Strengths: can power mm-scale or smaller implants several centimeters deep — the deepest reach of the three modalities for very small receivers — with an external source that itself is compact and skin-mounted.
Weaknesses: RF absorption in tissue directly produces heat, so specific absorption rate (SAR) safety limits cap deliverable power; efficiency is lower than inductive coupling at short range; antenna design is more complex than a simple coil.
Rule of thumb used by implant engineers: inductive coupling for shallow, larger-coil implants (cochlear, shallow SCS leads); ultrasound for deep, very small, alignment-tolerant implants (neural dust, deep peripheral nerves); RF mid-field for deep, mm-scale implants where ultrasound coupling through bone is impractical (some intracranial applications).
| Product | Indication | Trial Design | Key Result |
|---|---|---|---|
| Inductive Coupling | |||
| Ultrasound | |||
| RF Mid-Field | |||
| Far-Field RF (radiative) |
Once launched from the external transmitter, energy must cross layers of skin, fat, and muscle — each with different absorption and scattering properties — before reaching the implant. Two constraints govern how much power can actually arrive: physical attenuation with depth, and regulatory limits on how much energy can be deposited in living tissue without causing thermal damage.
For inductive coupling, the coupling coefficient between transmit and receive coils falls off roughly with the cube of separation distance in the near field; captured power scales even more steeply because both the field strength at the receiver and the achievable coil quality factor degrade with distance and tissue loading. This is why inductive links that transfer >50% of energy at 2–3 mm can drop below 1% by 2–3 cm.
Ultrasound and RF mid-field degrade more gently — closer to exponential attenuation with depth, governed by the tissue absorption coefficient at the chosen frequency — which is precisely why they are favored when the implant sits several centimeters below the skin. In all three modalities, higher operating frequency generally improves coupling to a small receiver but increases tissue absorption, so implant engineers must balance receiver size against penetration depth when choosing frequency.
This simulation models received power falling off with depth using an exponential-decay approximation of near-field attenuation. Drag the Implant Depth slider and watch Transfer Efficiency collapse from tens of percent near the surface to a fraction of one percent at several centimeters — this is the central engineering challenge of the entire field.
Depth is not the only variable that matters. Inductive links are highly sensitive to lateral offset and angular misalignment between transmit and receive coils — a few millimeters of lateral shift, or a 30° tilt, can cut coupling efficiency by more than half. This is a real clinical problem: implants shift slightly after healing, patients move, and external transmitters worn on a belt or patch can drift out of position.
Engineering responses include: resonant multi-coil relay links that widen the effective coupling volume, coil arrays that electronically steer the effective transmit position to track the implant, closed-loop power control that raises transmit power when a telemetry channel reports low received power, and — for ultrasound — beamforming transducer arrays that electronically refocus the acoustic beam onto a moving implant without physically repositioning the transducer.
Wireless power transfer through tissue cannot be scaled up indefinitely — absorbed energy becomes heat, and living tissue tolerates only a small, sustained temperature rise (generally held under about 2 °C for chronic implants) before risking protein denaturation and cell damage.
For RF and inductive powering, regulatory bodies (IEEE C95.1, FCC) cap Specific Absorption Rate (SAR) at 1.6 W/kg averaged over any 1 gram of tissue (US limit; ICNIRP uses 2 W/kg over 10 g). This directly limits how much transmit power can be pushed into the body regardless of how much the implant "wants" to receive.
For ultrasound, the FDA's diagnostic-use guideline caps spatial-peak temporal-average intensity (ISPTA) at 720 mW/cm² for most applications, and mechanical index limits guard against cavitation. Implant power-transfer systems designed for chronic, continuous operation are typically engineered with substantial margin below these ceilings, since diagnostic limits assume brief, intermittent exposure rather than hours of continuous power beaming.
Arriving energy is useless to the implant until it is captured, converted from an oscillating field into stable DC, and buffered so that a stimulation pulse or radio transmission — which draws far more instantaneous current than the trickle of harvested power — can be delivered on demand. This conversion chain happens entirely on, or immediately adjacent to, the implant's application-specific integrated circuit (ASIC).
The implant's energy-receiving element mirrors the external transmitter's modality:
• Inductive coupling: a micro-fabricated receive coil (often a spiral trace on the implant substrate or wound around a ferrite core) tuned to resonance with an on-chip matching capacitor to maximize power transfer at the transmit frequency • Ultrasound: a small piezoelectric element (e.g., lead zirconate titanate, PZT, or a biocompatible piezo-polymer) that flexes under the incoming pressure wave, generating an AC voltage via the piezoelectric effect • RF mid-field: a miniature dipole or loop antenna sized to the mid-field wavelength inside tissue, feeding a matching network tuned for the ~1–2 GHz operating band
In every case, the raw signal captured is an oscillating AC waveform at the transmit frequency — not yet usable to power digital logic or drive a stimulation electrode.
A rectifier converts the AC signal from the coil/transducer/antenna into DC. Implant power-management ICs typically use a CMOS diode-bridge or active synchronous rectifier built from low-threshold transistors, since conventional silicon diodes waste too much of the already-scarce harvested voltage as forward-voltage drop.
After rectification, a voltage regulator (often a low-dropout linear regulator, LDO, for simplicity, or a switched-capacitor converter for efficiency) stabilizes the DC output against fluctuations caused by patient movement, coupling drift, or transmitter duty cycling — the digital and analog circuits downstream need a clean, stable supply rail even though the input power is intermittent and noisy.
Because harvested power can be as low as a few microwatts, every stage of the power-management chain is designed for extreme energy efficiency: sub-threshold CMOS logic, ultra-low-quiescent-current regulators, and aggressive power-gating that shuts down unused circuit blocks between operations.
Neural stimulation and RF data transmission both draw short, high-current bursts — far more instantaneous power than a continuously harvested microwatt-scale supply can provide directly. A storage element buffers energy between these bursts:
• On-chip capacitors (10 nF to a few μF, sometimes larger off-chip ceramic capacitors) charge continuously from the rectified supply and discharge rapidly during a stimulation pulse or transmission burst • Some designs add a thin-film rechargeable microbattery or supercapacitor for longer energy reserves, trading some of the "no battery at all" simplicity for extended operation between transmitter sessions
The capacitor's state of charge directly determines what the implant can do next: below a threshold voltage, stimulation is withheld; once sufficient charge accumulates, the power-management circuit releases a regulated pulse to the stimulation driver. This charge-and-release rhythm is exactly what produces the duty-cycled operation seen in the final stage.
A battery-free implant does not operate continuously by default — it operates whenever the energy budget allows. Stimulation pulses, sensor sampling, and wireless data transmission are all scheduled around the rhythm of harvested and stored energy, producing a duty-cycled pattern that is the defining behavioral signature of wirelessly powered neural devices.
When received power is high (shallow implant, strong transmitter, good alignment), the storage capacitor recharges quickly between operations and the implant can stimulate or sense at close to its target rate with minimal delay. When received power is low (deep implant, weak coupling, misalignment), the capacitor recharges slowly, and the implant's firmware/power-management logic must stretch out the interval between operations, skip cycles, or reduce pulse amplitude to stay within the available energy budget.
This is not a flaw to be engineered away — it is the fundamental physical constraint of battery-free operation. Well-designed systems make this degradation graceful: closed-loop stimulators prioritize the most clinically important pulses, and sensing implants can adaptively lower sampling rate rather than fail outright when power is scarce.
Peripheral nerve stimulators: battery-free cuff or intrafascicular electrodes deliver charge-balanced biphasic current pulses to excite or block nerve fiber activity — for applications ranging from bladder control to chronic pain management to bioelectronic treatment of inflammatory disease. Each pulse consumes a discrete, predictable packet of stored energy, making the power budget easy to reason about.
Closed-loop sensing implants: rather than only stimulating on a fixed schedule, these implants sample local neural activity (e.g., local field potentials or nerve action potentials), process it on-chip, and trigger stimulation only when a target biomarker crosses a threshold (e.g., detecting a seizure onset or the biomarker of upcoming bladder distension). This closed-loop model must budget energy for sensing and computation as well as stimulation — and because a stimulation event may need to fire the moment a trigger is detected, systems often reserve a minimum capacitor charge as a "readiness" margin rather than draining it fully between operations.
Most wirelessly powered implants also need to send data outward — confirming a stimulation pulse fired, or streaming recorded neural signals for diagnostic use. Actively transmitting an RF signal is energy-expensive, so many battery-free implants instead use backscatter or load-modulation telemetry: rather than generating their own RF carrier, the implant modulates the electrical load it presents to the incoming power field (varying its coil impedance, for instance), and the external receiver detects this modulation as a data signal riding on the power link itself.
This approach lets the implant "talk" using only a small fraction of the energy a full active radio transmitter would require — an essential trick for a device with a power budget measured in microwatts and a storage capacitor measured in nanofarads.
The combined picture across all five stages: a battery-free implant is fundamentally a real-time energy-management system first, and a stimulation/sensing device second. Every clinical function it performs is gated by whether enough harvested energy has accumulated to pay for it.