RES EARCH

◥ and magnetoelectric materials are being eval-

REVIEW SUMMARY uated for use in nontraditional energy transfer
techniques because they have unique advan-
BIOELECTRONICS tages in the context of power transfer through

Miniature battery-free bioelectronics

the human body. Similarly, materials such as
triboelectrics, magnetoelastics, and thermo-
electrics are being explored to harvest energy
Vishnu Nair, Ashley N. Dalrymple, Zhanghao Yu, Gaurav Balakrishnan, Christopher J. Bettinger, from the human body. Furthermore, the em-
Douglas J. Weber, Kaiyuan Yang, Jacob T. Robinson* erging methods for power transfer through
the human body are also being explored as a
means for efficient data communication chan-
BACKGROUND: Bioelectronics, electronic devices clude ultrasound, electromagnetic, magnetic, nels that will support wireless networks or
that sense or manipulate biological signals, and optical methods to deliver power and data miniature implantable bioelectronics.
can substantially improve medical outcomes safely through the body, as well as energy-
by modulating the activity of excitable cells harvesting methods that extract chemical, ki- OUTLOOK: The extreme miniaturization and
and tissues in the nervous system, cardiovas- netic, and thermal energy from the body itself. long functional lifetime that are enabled by
cular system, and beyond. This specific stim- These emerging materials and technologies battery-free technologies point toward a fu-

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

ulation and monitoring can reduce side effects for data and power transfer are allowing bio- ture in which networks of tiny bioelectronics
and adapt in real time to the needs of each electronic devices to be miniaturized with tiny could be distributed throughout the body to
patient. However, despite the potential for bio- rechargeable batteries or, in the extreme case, accurately sense physiological states and apply
electronics to treat numerous conditions, only made entirely battery free. therapies when and where they are needed.
few implanted bioelectronic devices are cur- Reaching this vision will require additional
rently the standard of care. These include car- scientific advances in manufacturing and pack-
diac pacemakers that use electrical impulses Electronics aging of the materials that support this wire-
to synchronize the contraction of heart cells less bioelectronic network. For example, to
and continuous glucose monitors that generate reach submillimeter scale, alternatives to tra-
an electronic alert when a user needs insulin. ditional hermetic seals must be explored to
Large battery packs, cumbersome tethers, protect these electronics from the harsh envi-
and intricate packaging architectures compli- 1 cm Housing ronment of the human body. In addition to
cate device design and present numerous pos- typical material biocompatibility, implant-
sible failure modes, which limits the widespread able bioelectronics will have a longer ope-
deployment of therapeutic implantable bio- rating lifespan if there is little to no foreign
electronic devices. Wires can break and pro- body tissue response and scar formation to
vide a pathway for infection. Furthermore, the degrade the recording or stimulation efficacy.
need to route wires and leads limits the num- Energy receiving/ Advances in nanomaterials, soft electron-
ber and location of bioelectronic sensing and harvesting material ics, and coatings with growth factors, anti-
stimulating sites. Battery packs also create inflammatory drugs, and peptides are underway,
challenges for wearable bioelectronics such but ultimately these technologies need more
as heart rate monitors or pulse oximeters be- in vivo testing throughout different regions
cause the added size and weight of batteries of the body. This is especially critical if we
+
limit the overall comfort, appearance, and ease are to propel bioelectronic devices into true
of use. Magnetic Thermal Electrical Acoustic distributed networks inside and nearby dif-
Bioelectronic devices that operate without ferent organs and tissues throughout the body,
batteries could be miniaturized to millimeter- each with their own movements, vascula-
scale dimensions and easily implanted or ture, immune cells, and repair mechanisms.
comfortably worn. This approach would allow Furthermore, distributed implanted bioelec-
distributed networks of sensors and actuators Radio Optical Mechanical Chemical tronics will require specialized data and
to measure and manipulate physiological ac- power transfer modalities that will need
tivity throughout the body to enable precise Battery-free bioelectronics. Miniature bio- systems and standards for these emerging
and adaptive bioelectronic therapies with mini- electronics could support myriad therapeutics. communication and energy transfer meth-
mal risks or interference with daily activities. ods. For continued long-term success, power
To reach this level of miniaturization, bioelec- and data transfer modalities must be effi-
tronics must receive power from an external ADVANCES: We review progress in the mate- cient, safe, secure, reliable, and able to be up-
transmitter or harvest energy from the body
itself. Unfortunately, traditional methods of
rials and methods that are enabling wireless
power transfer and energy harvesting safely
dated as new implants and methods arise.
▪
transmitting wireless power over long distances through and within the human body. These
rely on electromagnetic waves that are absorbed methods must contend with the fact that many The list of author affiliations is available in the full article online.
*Corresponding author. Email: jtrobinson@rice.edu
or reflected by the body. To overcome this chal- forms of energy are absorbed or reflected by
Cite this article as V. Nair et al., Science 382, eabn4732
lenge, recent innovations in wireless power bone and soft tissues, placing limits on the (2023). DOI: 10.1126/science.abn4732
transfer and energy-harvesting technologies safety and efficiency of traditional power and
have been leveraged to support the power re- data transmission technologies. To overcome READ THE FULL ARTICLE AT
quirements of bioelectronic devices. These in- these challenges, piezoelectric, photovoltaic, https://doi.org/10.1126/science.abn4732

Nair et al., Science 382, 662 (2023) 10 November 2023 1 of 1

RES EARCH

◥ tude, 50 to 130 Hz in frequency, and 100 to

REVIEW 130 ms in pulse width, necessitating an esti-
mated power of 80 mW to 6.5 mW (assuming
BIOELECTRONICS stimulation electrodes with an impedance of

Miniature battery-free bioelectronics

1 kW) (14). Thus, for stimulation applications that
require microwatt-level powers, batteries must
be significantly larger than those used for min-
Vishnu Nair1, Ashley N. Dalrymple2,3,4, Zhanghao Yu5, Gaurav Balakrishnan6, iature cardiac pace makers such as the Micra.
Christopher J. Bettinger6, Douglas J. Weber2,7,8, Kaiyuan Yang5, Jacob T. Robinson1,5* Compared with stimulation, bioelectronic sens-
ing can be supported with a power of 100 mW
Miniature wireless bioelectronic implants that can operate for extended periods of time can transform or less, and thus can be supported with smaller
how we treat disorders by acting rapidly on precise nerves and organs in a way that drugs cannot. batteries, but extreme miniaturization will still
To reach this goal, materials and methods are needed to wirelessly transfer energy through the body or require battery-free approaches. A modern pulse
harvest energy from the body itself. We review some of the capabilities of emerging energy transfer oximeter made using a flexible organic semi-
methods to identify the performance envelope for existing technology and discover where opportunities conductor consumes only 24 mW against an
lie to improve how much—and how efficiently—we can deliver energy to the tiny bioelectronic implants over-the-counter device consuming ~60 mW
that can support emerging medical technologies. (15). Another sensing application that involves

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

reading the neural biopotentials for seizure
detection consumes only 140 mW of power (16).

E
lectronic sensing and stimulation at multi- most every commercial bioelectronic implant: The total power consumption of the system
ple sites throughout the body enable bio- a battery, a pulse generator circuit typically in must also include the supporting electronics,
electronic therapies that go far beyond the battery housing, and lead wires that con- which often consume power in the range of 10
what is possible with pharmaceuticals. nect the battery housing to the stimulation to 100 mW (17). Therefore, 200 mW is often a
For example, by recording from sensory and/or recording site. Among these basic com- sufficient power budget for bioelectronics that
neurons in the dorsal root ganglia to control ponents, the battery remains the largest vol- only support sensing functions.
stimulation in peripheral nerves (1) or spinal ume element in the system, so it prevents
cord, it is possible to restore walking in patients bioelectronic sensors and stimulators to be Battery-free technologies support
with severe spinal cord injuries (2). Electrical miniaturized to the size of injectables and dis- extreme miniaturization
measurements of impedance from the colon tributed throughout the body, especially in To achieve higher power levels without in-
can detect inflammation and be used to trig- sites where space is a premium. Recent im- creasing the size of the implant, next-generation
ger the start of vagus nerve stimulation to re- provements in battery technology and more miniature bioelectronic devices must receive
duce intestinal inflammation (3, 4). Bidirectional power-efficient electronics have enabled more power from an external transmitter or from
bioelectronics can also be used to interact with miniaturization, which provides hints of what the body itself. We can quantify when battery-
external objects and the person’s environment. is possible when the bioelectronic device can free solutions should be considered by calcu-
For instance, implanted electromyography (EMG) be small enough that it can be placed directly lating the volume of an implant for which
electrodes can be used to control a myoelectric at the stimulation or recording site. For exam- batteries would be too small to support >24 hours
prosthesis (5, 6). As the person interacts with ple, the Medtronic Micra is one of the first of operation (V24h) (Fig. 1C). In general, the
and manipulates an object, sensors on the ro- leadless pacemakers that integrates the bat- amount of charge in an ideal battery is equal
botic arm can trigger stimulation through elec- tery, pulse generator, and electrodes into a to the volume of the battery (V) multiplied by
trode arrays on or in the peripheral nerves to single capsule that is small enough to be de- the volumetric charge storage capacity (k),
provide tactile feedback to the user (5, 6). Al- livered to the heart through a catheter (10). thus we can write the battery lifetime (T) for
ternatively, neural recordings from electrodes This implant, roughly the size of a pen cap, is a specific application drawing a current (I)
implanted into the motor cortex can be de- ~10 times smaller than modern pacemakers, as follows (18):
coded to control a robotic limb, and sensory which are roughly the size of an Oreo cookie
k
feedback can be provided by electrodes im- (Fig. 1B). By combining the lead wires, bat- T¼ V ð1Þ
I
planted in the somatosensory cortex (7). tery, and electronics into this miniature pack-
For these types of medical advances to be age, the Micra avoids some of the primary Where k depends on the battery technology
deployed outside of controlled research set- failure modes related to lead wire migration, used. On the basis of Eq. 1, we can plot the bat-
tings, commercial bioelectronics must advance failure, or infection (10). tery lifetime versus battery volume for a given
beyond the historical device architectures that As promising as the Micra is, many applica- application (average current draw) and a given
are based on cardiac pacemakers developed in tions require more power than cardiac pacing, battery technology (Fig. 1C). We can then find
the 1950s (8, 9). Indeed, the core components making it difficult to create similarly sized the battery volume that would support a V24h
of the pacemaker architecture still exist in al- miniature devices that are based on battery (Fig. 1C)
power. The power requirements of bioelec-
I
V24h ¼ 24  ð2Þ
1
Rice Neuroengineering Initiative, Rice University, Houston, tronic devices vary depending on the applica-
TX, USA. 2Department of Mechanical Engineering, Carnegie k
Mellon University, Pittsburgh, PA, USA. 3Department of
tion. For example, cardiac pacing typically
Biomedical Engineering, University of Utah, Salt Lake City, draws 100 to 700 mW of power (11, 12), whereas To reduce V24h and enable smaller battery-
UT, USA. 4Department of Physical Medicine and tonic spinal cord stimulation of 2 to 8 V in powered devices, one must either increase the
Rehabilitation, University of Utah, Salt Lake City, UT, USA. volumetric charge capacity of the battery or
5 amplitude, 60 to 100 Hz in frequency, and
Department of Electrical and Computer Engineering, Rice
University, Houston, TX, USA. 6Department of Materials 100 to 500 ms in pulse width requires an esti- reduce the power draw on the battery. Figure
Science & Engineering, Carnegie Mellon University, mated power of 24 mW to 3.2 mW (assuming 1C shows approximate V24h for a hypothetical
Pittsburgh, PA, USA. 7Neuroscience Institute, Carnegie stimulation electrodes with an impedance of application drawing 8.3 mA current from a
Mellon University, Pittsburgh, PA, USA. 8Center for Neural
Basis of Cognition, Pittsburgh, PA, USA. 1 kΩ) (13). Deep-brain stimulation (DBS) stim- battery having a k = 0.05 (estimated from the
*Corresponding author. Email: jtrobinson@rice.edu ulation typically requires 4 to 10 V in ampli- data sheet of battery LITH 75 from Dantona

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 1 of 10

RES EARCH | R E V I E W

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

Fig. 1. Rationale for developing miniature battery-free bioelectronics. (lithium manganese oxide LITH-75 battery from Dantona Industries, k = 0.05
(A) Schematic of miniature battery-free bioelectronics that harvest energy calculated from data sheet) drawing an 8.3-mA current. The pink horizontal line for
from transmitters or the human body but also communicate wirelessly. an operation time of 24 hours meets the line defined by the power density and
[Illustration by Maayan Harel] (B) Operational lifetime as a function of power current draw at V24h. Thus, for this example device, it is not possible to create an
requirements for devices operating under 1 kΩ load. Example shown is a Boston implant smaller than V24h with that can last more than 24 hours between charges
Scientific Accolade pacemaker [photo courtesy of Vishnu Nair] with a 1600-mAh (pink shaded region). This zone is expected to be accessible only by devices
battery (green curve) and a Medtronic Micra pacemaker [photo reproduced powered wirelessly or through energy harvesting. Images of a rice grain (10 to
with permission from (116)] with a 120-mAh battery (blue curve). Pacemaker 15 mm3), pea (1 to 2 cm3), almond (3 to 4 cm3), and Oreo cookie (9 to 10 cm3)
photographs are shown with a 1 € coin (diameter = 23.25 mm) for size are shown at the approximate locations along the x axis to indicate the sizes of
comparison. [Photos copyright Oxford Univ. Press, used with permission] the corresponding batteries ([Images are from iStock: almond, mrtekmekci; cookie,
(C) Operation time versus battery volume plot for a cylindrical commercial battery TheCrimsonMonkey; rice, dizainera; pea, nilsz]).

Industries). This analysis does not consider energy-harvesting technologies, and WPT in support battery-free bioelectronics (Fig. 2), but
battery capacity fade over time, which would terms of the figures of merit that best describe there are several challenges that must be ad-
have increased V24h further. Thus, to supply each technology. We can then compare these dressed before widespread adoption. These
the energy that cannot be stored in a battery, approaches by examining how technologies include a need for the components of the en-
we consider both energy harvesting from the would perform if they were applied to the ergy harvesting system to be biocompatible to
human body and wireless power transfer (WPT) same bioelectronic application. prevent toxicity and inflammation. Further,
to devices on or in the human body. One chal- Our goal with this analysis is to understand the biological environment is aqueous and
lenge in making these comparisons, however, the performance envelopes for existing tech- ionic, which creates the need to encapsulate
is that the figures of merit for batteries, energy nologies. This will enable us to explore current devices to prevent material degradation or en-
harvesters, and WPT are all different because state-of-the-art technologies and future oppor- ergy losses (19). Finally, with all technologies,
they scale differently as a function of size. Al- tunities to develop materials and methods for there are manufacturing challenges associated
though batteries store more charge with in- energy transfer or harvesting that will expand with miniaturization.
creasing volume, energy harvesters create more the capabilities of battery-free bioelectronics.
power with increasing volume. We cannot com- Biomechanical energy harvesting
pare these technologies without specifying the Materials and methods for harvesting energy Triboelectric generators (TEGs) and piezoelec-
load they are driving. WPT technologies gener- from the human body tric generators (PEGs) offer two strategies to
ally do not scale linearly with volume, so we Harvesting energy from thermal, mechanical, convert biomechanical work into electrical en-
must characterize the performance of batteries, and chemical sources in the human body can ergy. TEGs harness the electrical field produced

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 2 of 10

RES EARCH | R E V I E W

Thermoelectric Magnetoelastic Piezoelectric Triboelectric Electrochemical with those that exhibit triboelectricity. Prom-
Heart Lungs Cochlear duct inent inorganic piezoelectric materials include
exotic compounds such as lead zirconate tita-
nate (PZT) and barium titanate (BT), which
ΔF

+
ΔT - V
ΔA
are challenging to implement in biomedical
[K+] >> [Na+]
M
ΔC applications because of their intrinsic toxicity
+
ΔF [K+] << [Na+] V and brittle mechanical properties. Zinc oxide
+V- -
(ZnO) thin-film and aligned nanowire struc-
-
+
V tures have been used as a biocompatible in-
organic material with a strong piezoelectric
in vitro implants Shallow in vivo implants ( < 1 cm) Deep in vivo implants ( > 1 cm) response. For instance, a flexible ZnO nano-
wire array generated power densities as high
as ~10 mW/cm3 (26). The b phase of flexible,
nontoxic PVDF fluoropolymers also exhibits a
well-documented piezoelectric response (27, 28).
b-PVDF has been used to harvest mechanical
energy and to power implantable bioelectronics

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

since as early as 1984 (29). Centimeter-scale
PVDF energy-harvesting patches (200 nm in
thickness) generate a peak power of 40 nW
in vivo (30). Thin-film PVDF devices are wrap-
Fig. 2. Comparing the power densities of energy-harvesting technologies. Shown are schematic ped around blood vessels to measure arterial
illustrations of thermoelectric (43, 46, 117–120), magnetoelastic (47), piezoelectric (41, 121), triboelectric pressure, but the materials and device architec-
(24, 122–124), and electrochemical (35, 39) energy-harvesting technologies. The plots are divided into ture could also be scaled as an energy-harvesting
three groups: in vitro implants (24, 31, 35, 39, 47, 123, 124, 117, 119), which consist of devices with subsystem that can supply power to other bio-
energy-harvesting capabilities that are measured by instrument-driven changes in the environment versus electronic devices. Additionally, hybrid piezo-
in vivo implants implanted in tissue or organisms. The in vivo implants are further divided into shallow electric and electret nanogenerators have been
implants in which the energy-harvesting element is implanted at a depth ≤1 cm and deep implants fabricated using various structures composed
that are implanted at a depth of >1 cm. An envelope (dashed line) is constructed in each graph by joining of biocompatible poly(L-lactide) (31). That study
the highest power-generating device for a given specific area. The in vitro envelope (black dashed reported instantaneous maximum power den-
line) overlayed with the two in vivo envelopes (gray dashed line) show that the performance of deeper sities of ~65 mW/m2 for in vitro operation for
implants is lower than those for their shallow and in vitro counterparts, suggesting room for application in wearable devices (31). More re-
improvement for these deeply implanted technologies. The y axis shows a plot of the range of powers cently, a piezoionic response, or pressure-driven
needed for common bioelectronic applications such as pulse oximetry, DBS, cardiac pacing, and on- ionic flux, was engineered by patterning hy-
board integrated electronics. We expect this envelope to expand as new technologies are invented or drogels with fixed charge densities, such as
existing technologies are improved to enable higher power densities. [Figure created with BioRender] anionic polyacrylic acid within neutral poly-
acrylamide hydrogels to create voltage gra-
by contact electrification and charge separa- alcohol) (PVA) (21). Many of these materials dients on deformation. Power densities of up
tion. PEGs use materials that convert mech- have been used in implantable devices, which to 0.85 mW/cm3 were generated and harnessed
anical strain into an internal electric field. These increases the likelihood for regulatory approval for fully hydrogel-based transducers capable
generators can be used to harness voluntary and eventual clinical adoption. of force sensing and direct peripheral nerve
mechanical work such as physical activity and TEG prototypes typically adopt thin-film stimulation (32).
involuntary mechanical work such as cardiac form factors with nominal footprints of 1 to TEGs and PEGs of 1 cm3 can theoretically
and circulatory activity. The typical power as- 100 cm2 that generate large peak voltages meet the energy requirements of many im-
sociated with natural limb motion of an 80-kg (>100 V) and low peak currents (10 to 100 mA) plantable medical devices, including cardiac
human walking at 1 Hz ranges from ~0.80 W (21, 22). There have been many advances in pacemakers (>100 mW), drug pumps (>0.1 mW),
in the elbow to ~34 W in the knee to ~1 W in TEGs to increase areal and volumetric power and neuromodulation devices such as cochlear
the blood circulation (20). For powering bio- densities and energy conversion efficiencies. implants (~1 mW) (33). Biomechanical gen-
electronics, TEGs and PEGs must maximize These have included the synthesis and de- erators that convert voluntary movements to
conversion efficiency and power densities to ployment of nanoscale generators in various power are limited by their dependence on
achieve miniaturized form factors. architectures and modes of operation and op- physical activity and their sensitivity to hu-
TEGs are created by manufacturing electron timization of materials selection (22). The re- midity. As a result, the peak power genera-
donors and acceptors into films, fibers, or other sulting specific areal peak power generation tion varies substantially from average power
form factors that maximize charge separation can approach 10 mW/cm2, but is typically production. Coupling these energy-harvesting
and subsequent field generation. For deployment ~0.01 mW/cm2 (23, 24). Areal peak power den- approaches with energy storage systems or
in bioelectronics, they typically use nontoxic sities vary because of substantial variations in interfacing them with mechanically dynamic
materials that are amenable for implantable materials selection, fabrication approaches, tissues such as the heart or blood vessels re-
bioelectronics. Examples of electron acceptors and device structural design. These areal power mains a promising power-harvesting strategy.
include poly(dimethylsiloxane) (PDMS), poly densities during in vitro operation project to To deploy biomechanical generators in chron-
(tetrafluoroethylene) (PTFE), and poly(vinyl- volumetric peak power densities of 1 to 10 mW/ ic applications, there is also a critical challenge
idene fluoride) (PVDF), whereas electron donors cm3 (23, 24). Further, TEGs can be optimized in designing functionally stable structures re-
may include metal films (e.g., Ag, Au, Al, Cu) or to operate at efficiencies as high as 55% (25). sistant to cyclic mechanical loading. The amount
polymers such as cellulose, silk, poly(ethyl- The choice of materials with piezoelectric of cyclic fatigue, and therefore lifetime, depends
eneterephthalate) (PET), nylon, and poly(vinyl properties for PEGs is much smaller compared on many factors such as the implant, site, the

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 3 of 10

RES EARCH | R E V I E W

device geometry, the materials chemistry, mi- activity, suboptimal loading, and other losses electric power generation for bioelectronics is
crostructure, strain rate, and maximum strain. reduce this projected in vitro current output to that thermal gradients are always present in
A typical implant with a target chronic lifetime values between 1 mA/cm3 and 1 mA/cm3. Such epidermal and subdermal applications. Although
of 10 years is calculated to accommodate ~108 a wide range of reported values for current reasonable specific power densities are already
cycles at a deformation frequency of 1 Hz. density is attributed to the variability in initial achievable, materials with increased power fac-
enzyme activity, enzyme stability, and enzyme tor of S2s (44) could further increase the ef-
Electrochemical cells and other immobilization techniques combined with typ- ficiency of thermoelectric energy-harvesting
energy-harvesting strategies ical variances in electrode materials and de- devices. It is essential to design systems that
Electrochemical cells that integrate biological vice fabrication. The power density is also maximize the thermal gradient by selecting
components can be used for onsite power gen- limited by the typically low open-circuit volt- high ZT materials, using heat sinks, and opti-
eration in bioelectronic devices. For instance, age (Voc) of redox reactions occurring in aque- mizing device geometries and structures to
there have been attempts to harness an exist- ous environments and the thermodynamic prevent lateral heat loss. Further, the effective-
ing biologically maintained electrochemical potential window of water. These circum- ness of ThEGs varies by climate in which it is
gradient in the cochlea, known as the endoco- stances yield areal power densities of 0.1 to deployed. If efficient thermoelectric materials
chlear potential, to power bioelectronic im- 1 mW/cm2 for in vitro operation. Fuel cells can be made that are mechanically flexible,
plants. An in vivo demonstration showed that that optimize enzyme packaging, maximize nontoxic, and easily processable into minia-
the endocochlear potential of a guinea pig was charge transfer at interfaces, and facilitate turized or thin-film form factors, they would

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

able to provide ~6 nW of power with operating charge transport can achieve in vitro areal be compelling candidates for generators for
voltages near 55 mV (34). Although the power power densities of ~4 mW/cm2 (41). However, bioelectronic applications.
and voltages extracted are too low for most practical limitations need to be addressed be- Magnetoelastic materials harness elastic de-
conventional bioelectronic applications, cus- fore the deployment of biofuel cells as energy- formation to produce local variations in mag-
tom low-power integrated circuit design could harvesting subsystems, including the efficient netic field. A recent demonstration showed that
increase the feasibility of using such power removal of reaction by-products, enzyme re- a soft composite of magnetic iron-neodymium-
sources in specific cases. Another approach generation, and form factors to minimize fi- boron microparticles in a silicone matrix pro-
for biological batteries has been pursued, spe- brous capsule formation (42). duce a magneto-mechanical coupling factor of
cifically to power transient ingestible bioelec- Thermoelectric generators (ThEGs) use spe- ~7 × 10−8 T /Pa, almost four times greater than
tronic devices with the gastrointestinal fluids cialized materials in which a thermal gradient rigid metallic alloys (47). Coupling composite
serving as the electrolyte. Using magnesium promotes diffusion of charge carriers to create materials with magnetic inductors allows soft
or zinc anodes and copper cathodes, gastro- an electric potential. This need for a temper- magnetoelastic generators to supply ~20 W/m2
intestinal batteries produced average power ature gradient limits the use of ThEGs mainly from biomechanical energy for wearable appli-
densities of 0.23 mW/mm2 and were able to to peripheral devices such as wearables or cations. MEGs can achieve peak powers of
power ingestible central temperature sensors subdermal implants, locations where the tem- 30 mW with ultrasonic stimulation.
with wireless data transmission capabilities perature gradients between the body and the
(35). A similar approach was used to design ambient environment are greatest (43). Ther- Materials and methods for WPT technologies
fiber-based injectable batteries with biofluid moelectric materials are defined by their Seebeck For cases in which power harvested from the
electrolytes in a variety of tissues such as the coefficients S (specific magnitude of induced body cannot adequately power a miniature bio-
dermis, heart, and brain while producing power thermoelectric voltage), typically ~100 mV/K. A electronic device, wireless power transmission
densities as high as ~80 mW/cm3 (36). Prac- further figure of merit for thermoelectric mate- can often achieve greater power densities. How-
rials is a dimensionless figure ZT (ZT = S sk T ),
2
tical limitations such as electrode degradation, ever, care must be taken to design the mate-
large fluctuations in produced power, battery where s is the electrical conductivity, T is the rials and methods so that this energy transfer
size and footprint, and device packaging still average temperature, and k is the thermal is safe for the human body. The safety limita-
need to be addressed before the adoption of conductivity that should be maximized (44). tion results primarily from the fact that energy
bioelectrochemical energy harvesting systems Conventional thermoelectric materials are typ- intended for a bioelectronic device can also be
for bioelectronics (37). ically rigid, inorganic semiconductors such as absorbed by the body, and this can lead to un-
Biofuel cells use biological motifs often im- bismuth-telluride, silicon and silicon-germanium safe levels of tissue heating (48). Tissue heating–
mobilized on an anode that oxidize biomole- alloys, and lead tellurides. Some of these mate- associated safety limits are set by specific ab-
cules and generate a flow of electrons to a rials, particularly bismuth-telluride and sili- sorption rates that depend on the conductivity
corresponding cathode. These cells, including con, have been used as thick films, thin films, of the tissue, density of the tissue, frequency of
anodes with immobilized microbial and enzy- or nanomaterial building blocks for ThEGs in the electromagnetic wave, and root mean square
matic species, have been explored as energy- bioelectronic applications. For instance, a bis- of the electric field. Unfortunately, the high-
harvesting devices for bioelectronic implants muth telluride ThEG harvested power at frequency electromagnetic waves preferred
for >50 years (38). Enzyme-based biofuel cells ~40 mW/cm2 density to sustainably power a for wirelessly powered electronics are relative-
such as glucose oxidase fuel cells are attrac- cardiac pacemaker (45). However, these mate- ly easily absorbed by the body (49). This has
tive as energy-harvesting systems for implant- rials have limited applications in vivo because led to a renaissance in research to identify
able devices because glucose is an abundant of their toxicity and rigid mechanical properties. materials and methods to receive energy from
energy source. Advances in electrode materials Conjugated polymers and ionic liquid-based nontraditional energy sources such as low-
and enzyme immobilization and stabilization ThEGs have shown promise for mechanically frequency magnetic fields, light, and ultrasound.
have improved implantable glucose oxidase compliant ThEGs. More recent in vivo dem- These technologies have shown substantial
fuel cells that can operate stably for up to onstrations of thermoelectric generators pro- improvements in the amount of power that
1 year in vitro (39). The theoretical upper bound duce specific areal power densities that approach can be received by miniature bioelectronic im-
on volumetric current density of an enzyme- 1 mW/cm2 (46). Thus, centimeter-scale energy- plants (Fig. 3) (50). Unlike batteries and ener-
based fuel cell, assuming a specific activity of harvesting devices could meet the demands of gy harvesters, however, WPT depends on the
1 U/mg and a mass density of 1 g/cm3, is 1.6 A/ many types of bioelectronic implants. Perhaps size and orientation of the transmitter, the
cm3 (40). However, a lower specific enzyme the most intriguing advantage of thermo- depth of the device in the tissue, and the size

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 4 of 10

RES EARCH | R E V I E W

Ultrasound Radiofrequency Inductive Magnetoelectric Light Midfield Capacitive

coupling
Skin
- + - +
- +
- +
p n
- +

V I V I I I
I

Shallow implants ( < 1 cm) Deep implants ( > 1 cm)

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

Fig. 3. Comparing the power densities of WPT technologies. Schematics of harvesting” represents the in vitro envelope obtained from the corresponding
WPT technologies correspond to the plots below. Scatter plots show power analysis in Fig. 2. Here, we only consider demonstrations of wireless power within
versus receiver area for various WPT devices (54, 55, 58, 61–71, 73–81, 95) human safety limits. The y axis shows a plot of the range of powers needed
classified into deep versus shallow implants, where a deep implant has for common bioelectronic applications such as pulse oximetry, DBS, cardiac
stimulation electrodes or the receiver >1 cm below the surface of the skin. pacing, and on-board integrated electronics. Technologies to the top left of these
Shallow implants have the receiver element ≤1 cm below the surface of the skin. plots have the highest power densities. We expect this envelope to expand as
The light gray dashed line labeled “wireless powering” connects the highest new technologies are invented or existing technologies are improved to enable
power devices for each size. The dark gray dashed line labeled “energy higher power densities.

of the receiver (51). Therefore, power density rial in the center of the receiver coil (52–57). extensively elsewhere (51). We compiled data
is only one of the many properties to consider However, using ferromagnetic material in re- from published work to determine how much
when evaluating the best WPT technology for ceivers can interfere with diagnostic methods power can be delivered as a function of the
a particular application. such as computed tomography (CT) and mag- size of the receiver and compared that with
netic resonance imaging (MRI) scans. Typical the power levels needed for common bioelec-
Near-field inductive coupling: The gold standard frequencies for inductive coupling are in the tronics applications. In Fig. 3, we plot the re-
Near-field inductive coupling (NIC) is by far the range of 1 to 13 MHz, which approaches the ceived power (in microwatts) versus receiver
most common method for power delivery to frequencies at which magnetic fields begin to area (in square millimeters) for ultrasound
biomedical implants and is the standard against be absorbed or reflected by the ionic fluids (61–63), radiofrequency (64, 65), magnetoelec-
which any new modality will be evaluated. This within the body (60). Although the safety of tric (66–70), light (71–76), midfield (59, 77, 78),
technique uses alternating magnetic fields WPT is determined by the specific absorption and capacitive (79–81) power transfer with
generated by a coil of wire to induce an electric rate, regulatory authorities such as the FCC shallow implants (<1 cm in tissue) and deep
current in a receiver coil (52–57). The induced allocate a spectrum of frequencies deemed safe implants (>1 cm in tissue). One centimeter is
electric current can then be used to power the for humans whenever the US Food and Drug used as a reference to distinguish between
electronic elements. Efficient power transfer Administration approves a new biomedical deep versus shallow implant because it is ap-
for inductive coupling depends on the total technology for clinical use. Currently, various proximately the depth from the scalp to the
magnetic flux that is captured by the receiver bands within the range of 400 to 2400 MHz dura in an adult human or the thickness of
coil (52–57). Thus, the efficiency in NIC with a have been approved for medical radio com- skin across epidermis to the subcutaneous
millimeter-scaled receiver scales with the area munications, micropower and body area net- layer. Examining these plots revealed where
of the receiver (AR) and the separation be- works. Further the decay of the transmitter new methods will be needed to achieve com-
tween receiver and transmitter (STR) as follows: coil magnetic field along its axis limits the pact devices capable of high-power bioelec-
3=2 3
AR STR (58). This leads to power transfer ef- miniaturization of implanted devices to depths tronic applications. For example, we found no
ficiencies <1% when the receiver is deeper than of 5 to 10 mm, for power requirements greater deeply implanted battery-free bioelectronics
1 cm under tissue (58). Inductive coupling is than 10 mW (54). implants smaller than ~1 mm2 that could re-
also sensitive to angular misalignment and ceive ~1 mW of power or more. Thus, improve-
axial separation of coils (59). The angular mis- Emerging WPT techniques ments in the materials and methods are needed
alignment can be overcome to an extent by There are several modalities of WPT that seek to make ultraminiature devices with the power
designing a transmitter coil with larger flux to improve upon NIC in terms of power den- needed for applications such as DBS and pulse
volume compared with the receiver coil and sity, depth, misalignment tolerance, and effi- oximetry. We also found numerous instances
can be mitigated by using a ferromagnetic mate- ciency. These techniques have been reviewed in which magnetoelectrics, light, ultrasound,

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 5 of 10

RES EARCH | R E V I E W

and radiofrequency demonstrated higher powers Materials and methods for wireless the data demodulation easier, leading to lower
compared with similarly sized NIC-powered communication and telemetry power consumption and design overhead in
devices. These data support the trend that For many applications, users and physicians the implants.
these emerging technologies can begin to re- must communicate with the bioelectronic de- Uplink communication is more challenging
place NIC for miniature bioelectronic power vices both to reprogram their operation and to because of the broad data rate and form factor
delivery to enable more advanced function- receive sensor data that can inform treatment. targets for different sensing applications. The
ality. For this analysis, we restricted the data Wireless data transmission through or near data rates needed to measure the dynamics of
to applications that fell within the safety limits the human body suffers many of the chal- physiological signals such as neural activity and
for human exposure. It is also critical to ac- lenges related to wireless power transmission, temperature, and biochemical concentrations
knowledge that there are many other metrics namely absorption and reflection of radiating can vary from around one to tens of thousands
upon which one can compare WPT technolo- waves by the body. Fortunately, the same mate- of kilobits per second. For example, measuring
gies. Alignment tolerance, power transfer effi- rials and methods used to overcome the wire- the spiking waveforms from one channel of
ciency, and comfort of the wearable transmitter less power transmission problem can also be neural activity typically requires sampling at
could be more important factors when consid- used to send and receive data near or through 10 KHz, but this rate can be greatly com-
ering bioelectronic systems (51). the human body. This includes transmitting pressed to 1 KHz if one were only interested in
data to the devices (downlink) and receiving capturing principle components of the wave
Wireless powering versus energy harvesting at a data from the device (uplink). Sharing the same forms (87), down to 3.3 KHz to measure the

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

miniature scale method for power and data transmission is timing of action potentials, or down to 300 to
By plotting the received power versus the de- highly desirable for device miniaturization. 1000 Hz to simply measure the spectral power
vice size in Figs. 2 and 3, we could identify However, considering various power and data in the spiking band (88–91). We summarized
which bioelectronic applications can be en- rate targets in different applications, sharing uplink schemes for wireless battery-free bio-
abled using emerging power technologies of the same method is not always feasible nor the electronics into three categories based on their
different sizes. The dashed lines in Fig. 3 con- optimal solution. relation to the wireless power transmission
nect the highest-power devices at each size for Downlink communication is typically used method. The major figures of merit for com-
WPT. A similar line plotted for in vivo energy- for infrequent device programming at low data munication are the amount of energy (in
harvesting technologies shown in Fig. 2 re- rates. Therefore, almost all battery-free bio- picojoules) consumed per bit and the data rate
veals that state-of-the-art WPT achieves 100 to electronic systems transmit downlink data by (in megabytes per second). When multiplied
1000 times more received power in miniature directly modulating the wireless power trans- together these constraints determine the amount
bioelectronic devices for both deep and shal- mission signal. The simplest modulation scheme of power consumed for data transmission. In
low implants. This discrepancy is likely because is on-off keying (OOK), which turns the power Fig. 4, we plot these two figures of merit for
energy-harvesting efficiency typical scales with transmitter on and off to represent binary several different communication strategies de-
device volume rather than device area. There- data. OOK downlink has been exploited with scribed below.
fore, without significant advances in energy- inductive coupling (82–84), ultrasound (62),
harvesting materials and methods, WPT will and light (85) power transmissions. Amplitude- Passive backscattering
be the best solution for applications with high- shift keying (ASK) modulation is commonly Passive backscattering is a low-energy wire-
power requirements. Nevertheless, when we adopted in inductive coupling (86), radio (87), less technology that exploits the physical process
look to the power levels required for applica- and magnetoelectrics (67, 68). ASK encodes of reflecting waves coming from an external
tions such as cardiac pacing, DBS, and pulse data on the amplitude of the power signal to transmitter by modulating the input imped-
oximetry plotted along the y axis (Fig. 2), we avoid complete on/off switching cycles, thus ance of receivers at negligible energy cost. It is
can see that millimeter-sized devices are be- increasing the data transmission rate and re- well suited for many bioelectronic applica-
ginning to meet the power requirements nec- ducing energy costs. ASK, along with OOK, the tions because it exploits the highly asymmet-
essary to enable these types of applications simplest form of ASK, are both popular in rical energy budget between the battery-free
with tiny implants. downlink communications because they make device and the external transmitter. Passive

Fig. 4. Uplink communi- Inductive, Radiofrequency (PB)

Incid Passive Backscatter (PB) Ultrasound (PB)
cation mechanisms e
Wave nt Power Magnetoelectric (PB)
10
for bioimplants. Shown Radiofrequency (AB)
are recent methods Coil/ Ultrasound (AB)
Antenna/ Transmitter Body Coupling (AB)
demonstrated in vivo for Data
d
Transducer Light (AB)
ecte
Energy/Bit (pJ/bit)

uplink communication, Refl ave Radiofrequency (Hybrid)

W
along with their tradeoffs
Active Backscatter (AB)
between energy and Antenna/ Antenna/
data rate. The communi- Transducer Transducer
Transmitter
cation schemes are Power Data
categorized into passive
backscattering (PB), Same Modality

active backscattering 10
Hybrid
(AB), and hybrid mode Antenna/ Antenna
Transducer /LED
depending on their rela-
Transmitter
tion to the wireless Power Data 10
power technology. 10 10 10 10 10 10 10 10
Different Modality Data Rate (Mbps)

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 6 of 10

RES EARCH | R E V I E W

backscattering of inductive coupling (86, 87, 92) fied and improved for communication to and densities 10 to 100 times those of ideal in vitro
and ultrasound waves (61, 62, 93) have been from implants, such as biphasic quasistatic energy-harvesting devices. At these power
demonstrated in vivo for sensing applications. (100, 101) and ionic (102) communication. levels, many therapeutic stimulation applica-
Recent work also demonstrated the feasibility tions such as DBS and spinal cord stimulation
of leveraging magnetoelectric converse effects Hybrid communication and become accessible. Combining stimulation with
for backscattering (94). Load shift keying is the power transmission measurement will enable miniature and dis-
most common modulation scheme to transmit The last category of bioelectronic devices uses tributed networks for providing closed-loop
digital binary data using different strengths of different materials and methods for power and therapies envisioned by bioelectronics (110).
reflection. Specifically, changing the load im- communication. Although integrating different To facilitate real-time feedback control, im-
pedance to an antenna, or simply shorting it in mechanisms may increase the device size and plants should also be equipped with low-power
certain cases, will result in a change in the assembly complexity, it offers the freedom to uplink communications. Because transmitting
amplitude or frequency of the backscattered separately optimize power and communica- data through the body typically expends be-
signal. Such load modulation can be easily im- tion links based on specific applications and tween 0.1 and 52 pJ/bit at target data rates of
plemented using a switch controlled by the naturally supports simultaneous power and 2.5 to 20.48 Mbps, we would expect instan-
uplink data bit, making it a straightforward data transmission. Radiofrequency transmit- taneous power consumption to range from
and practical approach. The choice of load mod- ters are widely used in implantable bioelec- 1 to 655 mW (100, 102, 111). The total power
ulation method and receiver design must be tronics that demand high bandwidth and are consumed by data transmission will depend

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

optimized for a specific type of wireless signal. often combined with WPT through inductive on the quantity of data and the frequency of
Beyond digital data uplink, directly modulat- coupling (103–107) and tissue energy harvest- transmission required for a specific applica-
ing the reflected waves using amplified bio- ing (34). Commercial cardiac pacemakers and tion, but instantaneous power places design
signals has shown promise for reducing the DBS systems use bluetooth low energy (BLE) constraints on closed-loop bioelectronic sys-
device size and power, leading to sub-cubic- for communication. This technology usually tems. Thus, for WPT technologies to support
millimeter neural recording devices (62, 95). achieves a data rate of 1 Mbps, accompanied these types of close-loop systems, we would
The limitations of such analog backscattering by a power consumption exceeding 1 mW. LED- expect target power densities to be ~103 mW/
methods are lower signal-to-noise ratio and in- based optical uplink is another popular uplink mm2 and above. However, power densities
compatibility with additional recording chan- method in hybrid systems (52, 83, 108) because cannot be used as the only measure to deter-
nels. As shown in Fig. 4, passive backscattering of its small footprint, simple design, and easy mine the practicality of a given WPT technol-
devices present the lowest energy consumption integration. ogy. Other factors factors such as transmitter
among all uplink technologies, with inductive size and weight, transmitter-receiver coupling,
backscattering having superior data rates be- Outlook: Miniature bioelectronics embedded in electromagnetic radiation safety, and implant-
cause of its higher frequency compared with wireless networks as the future ability must also be considered when design-
ultrasound. The wealth of energy-harvesting, WPT, and ing future bioelectronic systems.
communication technologies provides a rich In the near term, the most likely clinical ap-
Active backscattering toolbox for creating next-generation battery- plications for miniature battery-free implants
Active backscattering refers to the process of free bioelectronics that can be miniaturized will be less-invasive alternatives for existing
broadcasting uplink data using the same method and long-lasting in ways that battery-powered neuromodulation and cardiac therapies. Min-
as for wireless power transmission. Active back- devices cannot. Energy-harvesting technolo- iature battery-free implants for cardiac pacing,
scattering is most common when wireless power gies provide major advantages because the pain, overactive bladder, sleep apnea, and mood
transmission is performed using radio waves human body can supply this power without disorders are all in commercial development
(85, 96, 97) or light (71, 85). Radiofrequency the need for external transmitters. Energy har- (112–115). Other applications that do not re-
electromagnetics provide a high bandwidth to vesters demonstrated to date typically capture quire long-term implantation, such as post-
transmit data at high speed, allowing multi- 10−3 to 103 mW with physical footprints of 101 operative monitoring, drug delivery, and cancer
channel biosignal recordings. Radiofrequency to 103 mm2 (109). As shown in Fig. 3, the in vitro therapy, may be able to move more quickly
active backscattering is largely limited by low- envelopes of energy-harvesting devices reveal through the regulatory approval process and
power transmission due to tissue absorption. power generation in the range of 102 to 104 mW reach the market in the next 5 years. Just as
Further, most bioelectronic devices with opti- over harvesting element areas of 10 to 103 mm2, the battery-powered implanted pulse gener-
cal power transfer adopt active backscattering corresponding to power densities of ~10 mW/ ators developed for cardiac pacing in the 1960s
because of the difficulty in reflecting light mm2 (109). However, deep implant in vivo de- created a platform that enabled decades of in-
and the small footprint required by a data- vices can only generate powers in the range of novation in bioelectronic medicine, miniature
transmitting LED. Additionally, ultrasound com- 1 to 10 mW for similar sized devices with power battery-free bioelectronics and networks of
munication can be performed in the active densities in the range of 0.01 to 0.1 mW/mm2 wireless implants may be the next great plat-
backscattering approach by applying an alter- (109). Thus, the focus of future work should be form technology for the decade’s innovation in
nating voltage to the energy-harvesting piezo- toward designing devices to bridge this gap in bioelectronic medicine.
electric transducer to generate OOK-modulated power-harvesting capabilities. Although there
ultrasound waves (98). In principle, active back- are many applications in sensing that are sup-
REFERENCES AND NOTES
scattering consumes more energy than its pas- ported at these power levels, stimulation at 1. B. J. Holinski, D. G. Everaert, V. K. Mushahwar, R. B. Stein,
sive counterpart but has greater potential for energies beyond cardiac pacing are typically Real-time control of walking using recordings from dorsal
higher signal-to-noise ratio and thus deeper orders of magnitude higher. In these cases, root ganglia. J. Neural Eng. 10, 056008 (2013). doi: 10.1088/
1741-2560/10/5/056008; pmid: 23928579
implants. Finally, there is growing interest in WPT may be suitable.
2. T. M. Bruns, J. B. Wagenaar, M. J. Bauman, R. A. Gaunt,
tissue-coupled communication schemes that Current WPT technology envelopes (Fig. 4) D. J. Weber, Real-time control of hind limb functional
use biological tissue as a medium, which require reach power levels of ~0.1 to 100 mW in 0.1- to electrical stimulation using feedback from dorsal root ganglia
only electrodes on the miniaturized implants. 1000-mm2–sized devices, achieving power den- recordings. J. Neural Eng. 10, 026020 (2013). doi: 10.1088/
1741-2560/10/2/026020; pmid: 23503062
Conventional intrabody capacitive and galvanic sities of 100 to 1000 mW/mm2 (109). Thus, 3. S. C. Payne et al., Anti-inflammatory effects of abdominal
communication schemes (99) have been modi- WPT power densities can generate real power vagus nerve stimulation on experimental intestinal

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 7 of 10

RES EARCH | R E V I E W

inflammation. Front. Neurosci. 13, 418 (2019). doi: 10.3389/ 25. L. Lin et al., Robust triboelectric nanogenerator based on 45. D. Bhatia, S. Bairagi, S. Goel, M. Jangra, Pacemakers
fnins.2019.00418; pmid: 31133776 rolling electrification and electrostatic induction at an charging using body energy. J. Pharm. Bioallied Sci. 2, 51–54
4. S. C. Payne et al., Transmural impedance detects graded instantaneous energy conversion efficiency of ~55%. (2010). doi: 10.4103/0975-7406.62713; pmid: 21814432
changes of inflammation in experimental colitis. R. Soc. Open ACS Nano 9, 922–930 (2015). doi: 10.1021/nn506673x; 46. D. Rozgic, D. Markovic, A miniaturized 0.78-mW/cm2
Sci. 7, 191819 (2020). doi: 10.1098/rsos.191819; pmid: 25555045 autonomous thermoelectric energy-harvesting platform for
pmid: 32257338 26. S. Lee et al., Super-flexible nanogenerator for energy biomedical sensors. IEEE Trans. Biomed. Circuits Syst. 11,
5. J. A. George et al., Biomimetic sensory feedback through harvesting from gentle wind and as an active deformation 773–783 (2017). doi: 10.1109/TBCAS.2017.2684818;
peripheral nerve stimulation improves dexterous use of a sensor. Adv. Funct. Mater. 23, 2445–2449 (2013). pmid: 28541912
bionic hand. Sci. Robot. 4, eaax2352 (2019). doi: 10.1126/ doi: 10.1002/adfm.201202867 47. Y. Zhou et al., Giant magnetoelastic effect in soft systems for
scirobotics.aax2352; pmid: 33137773 27. S. Panda et al., Biocompatible CaTiO3-PVDF composite- bioelectronics. Nat. Mater. 20, 1670–1676 (2021).
6. D. W. Tan et al., A neural interface provides long-term stable based piezoelectric nanogenerator for exercise evaluation doi: 10.1038/s41563-021-01093-1; pmid: 34594013
natural touch perception. Sci. Transl. Med. 6, 257ra138 and energy harvesting. Nano Energy 102, 107682 (2022). 48. IEEE, “IEEE standard for safety levels with respect to human
(2014). doi: 10.1126/scitranslmed.3008669; pmid: 25298320 doi: 10.1016/j.nanoen.2022.107682 exposure to electric, magnetic, and electromagnetic fields,
7. S. N. Flesher et al., A brain-computer interface that evokes 28. N. A. Shepelin et al., New developments in composites, 0 Hz to 300 GHz,” in IEEE Standard C95.1-2019 (IEEE,
tactile sensations improves robotic arm control. Science 372, copolymer technologies and processing techniques for 2019), 2019 Revision of IEEE Std C95.1-2005/Incorporates
831–836 (2021). doi: 10.1126/science.abd0380; flexible fluoropolymer piezoelectric generators for efficient IEEE Std C95.1-2019/Cor 1-2019); doi: 10.1109/
pmid: 34016775 energy harvesting. Energy Environ. Sci. 12, 1143–1176 (2019). IEEESTD.2019.8859679
8. S. K. Mulpuru, M. Madhavan, C. J. McLeod, Y.-M. Cha, doi: 10.1039/C8EE03006E 49. F. T. Alrashdan et al., Wearable wireless power systems for
P. A. Friedman, Cardiac pacemakers: Function, 29. E. Häsler, L. Stein, G. Harbauer, Implantable physiological ‘ME-BIT’ magnetoelectric-powered bio implants. J. Neural
troubleshooting, and management. J. Am. Coll. Cardiol. 69, power supply with PVDF film. Ferroelectrics 60, 277–282 Eng. 18, 045011 (2021). doi: 10.1088/1741-2552/ac1178;
189–210 (2017). doi: 10.1016/j.jacc.2016.10.061; (1984). doi: 10.1080/00150198408017528 pmid: 34229314
pmid: 28081829 30. X. Cheng et al., Implantable and self-powered blood pressure 50. R. Hinchet et al., Transcutaneous ultrasound energy

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

9. L. D. Abrams, W. A. Hudson, R. Lightwood, A surgical monitoring based on a piezoelectric thinfilm: Simulated, harvesting using capacitive triboelectric technology. Science
approach to the management of heart-block using an in vitro and in vivo studies. Nano Energy 22, 453–460 (2016). 365, 491–494 (2019). doi: 10.1126/science.aan3997;
inductive coupled artifical cardiac pacemaker. Lancet 1, doi: 10.1016/j.nanoen.2016.02.037 pmid: 31371614
1372–1374 (1960). doi: 10.1016/S0140-6736(60)91151-X; 31. S. Gong, B. Zhang, J. Zhang, Z. L. Wang, K. Ren, 51. A. Singer, J. T. Robinson, Wireless power delivery techniques for
pmid: 13791559 Biocompatible poly(lactic acid)-based hybrid piezoelectric miniature implantable bioelectronics. Adv. Healthc. Mater. 10,
10. N. Bhatia, M. El-Chami, Leadless pacemakers: A and electret nanogenerator for electronic skin applications. e2100664 (2021). doi: 10.1002/adhm.202100664;
contemporary review. J. Geriatr. Cardiol. 15, 249–253 (2018). Adv. Funct. Mater. 30, 1908724 (2020). doi: 10.1002/ pmid: 34114368
pmid: 29915613 adfm.201908724 52. A. Burton et al., Wireless, battery-free subdermally
11. P. Abiri et al., Inductively powered wireless pacing via a 32. Y. Dobashi et al., Piezoionic mechanoreceptors: Force- implantable photometry systems for chronic recording of
miniature pacemaker and remote stimulation control system. induced current generation in hydrogels. Science 376, neural dynamics. Proc. Natl. Acad. Sci. U.S.A. 117, 2835–2845
Sci. Rep. 7, 6180 (2017). doi: 10.1038/s41598-017-06493-5; 502–507 (2022). doi: 10.1126/science.aaw1974; (2020). doi: 10.1073/pnas.1920073117; pmid: 31974306
pmid: 28733677 pmid: 35482868 53. G. E. Loeb, R. A. Peck, W. H. Moore, K. Hood, BION system
12. W. E. Hill, A. Murray, J. P. Bourke, L. Howell, R. G. Gold, 33. H. A. Yiğit, H. Uluşan, M. B. Yuksel, S. Chamanian, B. Çiftci, for distributed neural prosthetic interfaces. Med. Eng. Phys.
Minimum energy for cardiac pacing. Clin. Phys. Physiol. Meas. A. Koyuncuoğlu, A. Muhtaroğlu, H. Külah, “A pulse-width 23, 9–18 (2001). doi: 10.1016/S1350-4533(01)00011-X;
9, 41–46 (1988). doi: 10.1088/0143-0815/9/1/003; modulated cochlear implant interface electronics with pmid: 11344003
pmid: 3359743 513 µW power consumption,” in 2019 IEEE/ACM International 54. G. Shin et al., Flexible near-field wireless optoelectronics as
13. D. M. Moore, C. McCrory, Spinal cord stimulation. BJA Educ. Symposium on Low Power Electronics and Design (ISLPED), subdermal implants for broad applications in optogenetics.
16, 258–263 (2016). doi: 10.1093/bjaed/mkv072; Lausanne, Switzerland, 29–31 July 2019 (IEEE, 2019); Neuron 93, 509–521.e3 (2017). doi: 10.1016/j.
pmid: 27458592 doi: 10.1109/ISLPED.2019.8824838 neuron.2016.12.031; pmid: 28132830
14. R. Ramasubbu, S. Lang, Z. H. T. Kiss, Dosing of electrical 34. P. P. Mercier, A. C. Lysaght, S. Bandyopadhyay, 55. H. Lyu et al., Synchronized biventricular heart pacing in a
parameters in deep brain stimulation (DBS) for intractable A. P. Chandrakasan, K. M. Stankovic, Energy extraction from closed-chest porcine model based on wirelessly powered
depression: A review of clinical studies. Front. Psychiatry 9, 302 the biologic battery in the inner ear. Nat. Biotechnol. 30, leadless pacemakers. Sci. Rep. 10, 2067 (2020).
(2018). doi: 10.3389/fpsyt.2018.00302; pmid: 30050474 1240–1243 (2012). doi: 10.1038/nbt.2394; pmid: 23138225 doi: 10.1038/s41598-020-59017-z; pmid: 32034237
15. H. Lee et al., Toward all-day wearable health monitoring: An 35. P. Nadeau et al., Prolonged energy harvesting for ingestible 56. G. L. Barbruni, P. M. Ros, D. Demarchi, S. Carrara, D. Ghezzi,
ultralow-power, reflective organic pulse oximetry sensing devices. Nat. Biomed. Eng. 1, 0022 (2017). doi: 10.1038/ Miniaturised wireless power transfer systems for
patch. Sci. Adv. 4, eaas9530 (2018). doi: 10.1126/sciadv. s41551-016-0022; pmid: 28458955 neurostimulation: A review. IEEE Trans. Biomed. Circuits Syst.
aas9530; pmid: 30430132 36. Y. Zhao et al., Injectable fiber batteries for all-region power 14, 1160–1178 (2020). doi: 10.1109/TBCAS.2020.3038599;
16. A. Bahr et al., Epileptic seizure detection on an ultra-low- supply in vivo. J. Mater. Chem. A Mater. Energy Sustain. 9, pmid: 33201828
power embedded RISC-V processor using a convolutional 1463–1470 (2021). doi: 10.1039/D0TA10357H 57. D. K. Freeman et al., A sub-millimeter, inductively powered
neural network. Biosensors (Basel) 11, 203 (2021). 37. T. J.-P. Ivase, B. B. Nyakuma, O. Oladokun, P. T. Abu, neural stimulator. Front. Neurosci. 11, 659 (2017).
doi: 10.3390/bios11070203; pmid: 34201480 M. N. Hassan, Review of the principal mechanisms, prospects, doi: 10.3389/fnins.2017.00659; pmid: 29230164
17. P. M. Lee, Z. Xiong, J. Ho, Methods for powering bioelectronic and challenges of bioelectrochemical systems. Environ. Prog. 58. M. Zargham, P. G. Gulak, Fully integrated on-chip coil in
microdevices. Bioelectron. Med. (Lond.) 1, 201–217 (2018). Sustain. Energy 39, 13298 (2020). doi: 10.1002/ep.13298 0.13 mm CMOS for wireless power transfer through biological
doi: 10.2217/bem-2018-0005 38. A. T. Yahiro, S. M. Lee, D. O. Kimble, Bioelectrochemistry: media. IEEE Trans. Biomed. Circuits Syst. 9, 259–271 (2015).
18. N. E. Galushkin, N. N. Yazvinskaya, D. N. Galushkin, A I. Enzyme utilizing bio-fuel cell studies. Biochim. Biophys. Acta doi: 10.1109/TBCAS.2014.2328318; pmid: 25099630
critical review of using the Peukert equation and its 88, 375–383 (1964). pmid: 14249845 59. K. L. Montgomery et al., Wirelessly powered, fully internal
generalizations for lithium-ion cells. J. Electrochem. Soc. 167, 39. B. Reuillard et al., One-year stability for a glucose/oxygen optogenetics for brain, spinal and peripheral circuits in mice.
120516 (2020). doi: 10.1149/1945-7111/abad69 biofuel cell combined with pH reactivation of the laccase/ Nat. Methods 12, 969–974 (2015). doi: 10.1038/nmeth.3536;
19. S.-W. Hwang et al., A physically transient form of silicon carbon nanotube biocathode. Bioelectrochemistry 106 (Pt A), pmid: 26280330
electronics. Science 337, 1640–1644 (2012). doi: 10.1126/ 73–76 (2015). doi: 10.1016/j.bioelechem.2015.04.009; 60. T. Nan et al., Acoustically actuated ultra-compact NEMS
science.1226325; pmid: 23019646 pmid: 25934467 magnetoelectric antennas. Nat. Commun. 8, 296 (2017).
20. R. Riemer, A. Shapiro, Biomechanical energy harvesting from 40. G. T. R. Palmore, G. M. Whitesides, in Enzymatic Conversion doi: 10.1038/s41467-017-00343-8; pmid: 28831042
human motion: Theory, state of the art, design guidelines, of Biomass for Fuels Production, M. E. Himmel, J. O. Baker, 61. S. Sonmezoglu, J. R. Fineman, E. Maltepe, M. M. Maharbiz,
and future directions. J. Neuroeng. Rehabil. 8, 22 (2011). R. P. Overend, Eds. (American Chemical Society, 1994), Monitoring deep-tissue oxygenation with a millimeter-scale
doi: 10.1186/1743-0003-8-22; pmid: 21521509 vol. 14, pp. 271–290. ultrasonic implant. Nat. Biotechnol. 39, 855–864 (2021).
21. Z. L. Wang, Triboelectric nanogenerators as new energy 41. C. H. Kwon et al., High-power hybrid biofuel cells using doi: 10.1038/s41587-021-00866-y; pmid: 33782610
technology for self-powered systems and as active layer-by-layer assembled glucose oxidase-coated metallic 62. D. K. Piech et al., A wireless millimetre-scale implantable
mechanical and chemical sensors. ACS Nano 7, 9533–9557 cotton fibers. Nat. Commun. 9, 4479 (2018). doi: 10.1038/ neural stimulator with ultrasonically powered bidirectional
(2013). doi: 10.1021/nn404614z; pmid: 24079963 s41467-018-06994-5; pmid: 30367069 communication. Nat. Biomed. Eng. 4, 207–222 (2020).
22. B. Yang, Y. Xiong, K. Ma, S. Liu, X. Tao, Recent advances in 42. S. Cosnier, A. J. Gross, A. Le Goff, M. Holzinger, Recent doi: 10.1038/s41551-020-0518-9; pmid: 32076132
wearable textile-based triboelectric generator systems for advances on enzymatic glucose/oxygen and hydrogen/ 63. K. H. Lee et al., Ultrasound-driven two-dimensional Ti3C2Tx
energy harvesting from human motion. EcoMat 2, e12054 oxygen biofuel cells: Achievements and limitations. MXene hydrogel generator. ACS Nano 14, 3199–3207 (2020).
(2020). doi: 10.1002/eom2.12054 J. Power Sources 325, 252–263 (2016). doi: 10.1016/ doi: 10.1021/acsnano.9b08462; pmid: 32078295
23. P. Bai et al., Integrated multilayered triboelectric j.jpowsour.2016.05.133 64. A. Khalifa et al., The Microbead: A 0.009 mm 3 implantable
nanogenerator for harvesting biomechanical energy from 43. C.-G. Han et al., Giant thermopower of ionic gelatin near wireless neural stimulator. IEEE Trans. Biomed. Circuits Syst.
human motions. ACS Nano 7, 3713–3719 (2013). room temperature. Science 368, 1091–1098 (2020). 13, 971–985 (2019). doi: 10.1109/TBCAS.2019.2939014;
doi: 10.1021/nn4007708; pmid: 23484470 doi: 10.1126/science.aaz5045; pmid: 32354840 pmid: 31484132
24. Y. Sun et al., Flexible hybrid piezo/triboelectric energy 44. E. W. Zaia, M. P. Gordon, P. Yuan, J. J. Urban, Progress and 65. W. S. Kim et al., Organ-specific, multimodal, wireless
harvester with high power density workable at elevated perspective: Soft thermoelectric materials for wearable and optoelectronics for high-throughput phenotyping of
temperatures. J. Mater. Chem. A Mater. Energy Sustain. 8, internet-of-things applications. Adv. Electron. Mater. 5, peripheral neural pathways. Nat. Commun. 12, 157 (2021).
12003–12012 (2020). doi: 10.1039/D0TA04612D 1800823 (2019). doi: 10.1002/aelm.201800823 doi: 10.1038/s41467-020-20421-8; pmid: 33420038

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 8 of 10

RES EARCH | R E V I E W

66. A. Singer et al., Magnetoelectric materials for miniature, telemetry. IEEE J. Solid-State Circuits 57, 1061–1074 (2022). 105. C. S. Mestais et al., WIMAGINE: Wireless 64-channel ECoG
wireless neural stimulation at therapeutic frequencies. doi: 10.1109/JSSC.2022.3141688; pmid: 36186085 recording implant for long term clinical applications.
Neuron 107, 631–643.e5 (2020). doi: 10.1016/ 86. J. Kim et al., A soft and transparent contact lens for the IEEE Trans. Neural Syst. Rehabil. Eng. 23, 10–21 (2015).
j.neuron.2020.05.019; pmid: 32516574 wireless quantitative monitoring of intraocular pressure. doi: 10.1109/TNSRE.2014.2333541; pmid: 25014960
67. J. C. Chen et al., A wireless millimetric magnetoelectric Nat. Biomed. Eng. 5, 772–782 (2021). doi: 10.1038/ 106. Y. Jia et al., A trimodal wireless implantable neural interface
implant for the endovascular stimulation of peripheral nerves. s41551-021-00719-8; pmid: 33941897 system-on-chip. IEEE Trans. Biomed. Circuits Syst. 14,
Nat. Biomed. Eng. 6, 706–716 (2022). doi: 10.1038/ 87. J. Lee et al., Neural recording and stimulation using wireless 1207–1217 (2020). doi: 10.1109/TBCAS.2020.3037452;
s41551-022-00873-7; pmid: 35361934 networks of microimplants. Nat. Electron. 4, 604–614 (2021). pmid: 33180731
68. Z. Yu et al., Magnetoelectric biol.-implants powered and doi: 10.1038/s41928-021-00631-8 107. B. Lee et al., An implantable peripheral nerve recording and
programmed by a single transmitter for coordinated 88. K. Sahasrabuddhe et al., The Argo: A high channel count stimulation system for experiments on freely moving
multisite stimulation. IEEE J. Solid-State Circuits 57, recording system for neural recording in vivo. J. Neural Eng. animal subjects. Sci. Rep. 8, 6115 (2018). doi: 10.1038/
818–830 (2022). doi: 10.1109/JSSC.2021.3129993; 18, 015002 (2020). doi: 10.1088/1741-2552/abd0ce; s41598-018-24465-1; pmid: 29666407
pmid: 36275505 pmid: 33624614 108. J. Ausra et al., Wireless battery free fully implantable
69. Z. Yu et al., MagNI: A magnetoelectrically powered and 89. J. T. Costello et al., A low-power communication scheme for multimodal recording and neuromodulation tools for
controlled wireless neurostimulating implant. IEEE Trans. wireless, 1000 channel brain-machine interfaces. J. Neural songbirds. Nat. Commun. 12, 1968 (2021). doi: 10.1038/
Biomed. Circuits Syst. 14, 1241–1252 (2020). doi: 10.1109/ Eng. 19, 036037 (2022). doi: 10.1088/1741-2552/ac7352; s41467-021-22138-8; pmid: 33785751
TBCAS.2020.3037862; pmid: 33180732 pmid: 35613546 109. Data for: V. Nair, A. N. Dalrymple, Z. Yu, G. Balakrishnan,
70. W. Kim et al., Magnetoelectrics enables large power 90. R. R. Harrison et al., A Low-Power Integrated Circuit for a C. J. Bettinger, D. J. Weber, K. Yang, J. T. Robinson, Miniature
delivery to mm-sized wireless bioelectronics. J. Appl. Phys. Wireless 100-Electrode Neural Recording System. IEEE J. battery-free bioelectronics, Zenodo (2023); doi: 10.5281/
134, 094103 (2023). doi: 10.1063/5.0156015; Solid-State Circuits 42, 123–133 (2007). doi: 10.1109/ zenodo.8310306
pmid: 37692260 JSSC.2006.886567 110. V. Nair, B. Tian, “Perspectives for seamless integration of
71. M. Silverå Ejneby et al., Chronic electrical stimulation of 91. S. R. Nason et al., A low-power band of neuronal spiking bioelectronic systems in neuromedicine,” in Neural Interface

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

peripheral nerves via deep-red light transduced by an activity dominated by local single units improves the Engineering, L. Guo, Ed. (Springer, 2020); pp. 365–381;
implanted organic photocapacitor. Nat. Biomed. Eng. 6, performance of brain-machine interfaces. Nat. Biomed. Eng. http://link.springer.com/10.1007/978-3-030-41854-0_16.
741–753 (2022). doi: 10.1038/s41551-021-00817-7; 4, 973–983 (2020). doi: 10.1038/s41551-020-0591-0; 111. C. Lee et al., “A miniaturized wireless neural implant with
pmid: 34916610 pmid: 32719512 body-coupled data transmission and power delivery for freely
72. J. Charthad et al., A mm-sized wireless implantable device for 92. R. Muller et al., A Minimally Invasive 64-Channel Wireless behaving animals,” in 2022 IEEE International Solid-State
electrical stimulation of peripheral nerves. IEEE Trans. mECoG Implant. IEEE J. Solid-State Circuits 50, 344–359 Circuits Conference (ISSCC), San Francisco, CA, 20–26
Biomed. Circuits Syst. 12, 257–270 (2018). doi: 10.1109/ (2015). doi: 10.1109/JSSC.2014.2364824 February 2022 (IEEE, 2022); doi: 10.1109/ISSCC42614.2022.9731733
TBCAS.2018.2799623; pmid: 29578414 93. C. Shi et al., Application of a sub-0.1-mm3 implantable 112. O. Padron et al., PD66-01 treatment of urinary urgency
73. L. Lu et al., Biodegradable monocrystalline silicon mote for in vivo real-time wireless temperature sensing. incontinence (UUI) with an ultra-miniaturized sacral nerve
photovoltaic microcells as power supplies for transient Sci. Adv. 7, eabf6312 (2021). doi: 10.1126/sciadv.abf6312; modulation (snm) system: Preliminary outcomes of the
biomedical implants. Adv. Energy Mater. 8, 1703035 (2018). pmid: 33962948 SANS-UUI study. J. Urol. 206, e1158 (2021). doi: 10.1097/
doi: 10.1002/aenm.201703035 94. Z. Yu et al., “Magnetoelectric backscatter communication for JU.0000000000002110.01
74. A. Saha, S. Iqbal, M. Karmaker, S. F. Zinnat, M. T. Ali, A millimeter-sized wireless biomedical implants,” in 113. D. Seo et al., Wireless recording in the peripheral nervous
wireless optical power system for medical implants using low Proceedings of the 28th Annual International Conference on system with ultrasonic neural dust. Neuron 91, 529–539
power near-IR laser. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. Mobile Computing and Networking, Sydney, NSW, Australia, (2016). doi: 10.1016/j.neuron.2016.06.034; pmid: 27497221
2017, 1978–1981 (2017). doi: 10.1109/EMBC.2017.8037238; 17–21 October 2022 (ACM, 2022); doi: 10.1145/ 114. K. Patch, Neural dust swept up in latest leap for bioelectronic
pmid: 29060282 3495243.3560541 medicine. Nat. Biotechnol. 39, 255–256 (2021). doi: 10.1038/
75. A. Ahnood et al., Diamond encapsulated photovoltaics for 95. M. M. Ghanbari et al., A sub-mm3 ultrasonic free-floating s41587-021-00856-0; pmid: 33692519
transdermal power delivery. Biosens. Bioelectron. 77, implant for multi-mote neural recording. IEEE J. Solid-State 115. T. Okabe et al., Leadless left ventricular stimulation with
589–597 (2016). doi: 10.1016/j.bios.2015.10.022; Circuits 54, 3017–3030 (2019). doi: 10.1109/ WiSE-CRT System - Initial experience and results from phase
pmid: 26476599 JSSC.2019.2936303 I of SOLVE-CRT Study (nonrandomized, roll-in phase).
76. K. Song, J. H. Han, H. C. Yang, K. I. Nam, J. Lee, Generation 96. D. A. Borton, M. Yin, J. Aceros, A. Nurmikko, An implantable Heart Rhythm 19, 22–29 (2022). doi: 10.1016/
of electrical power under human skin by subdermal solar wireless neural interface for recording cortical circuit j.hrthm.2021.06.1195; pmid: 34332966
cell arrays for implantable bioelectronic devices. Biosens. dynamics in moving primates. J. Neural Eng. 10, 026010 116. P. Ritter et al., Early performance of a miniaturized leadless
Bioelectron. 92, 364–371 (2017). doi: 10.1016/ (2013). doi: 10.1088/1741-2560/10/2/026010; cardiac pacemaker: The Micra Transcatheter Pacing Study.
j.bios.2016.10.095; pmid: 27836601 pmid: 23428937 Eur. Heart J. 36, 2510–2519 (2015). doi: 10.1093/eurheartj/
77. D. R. Agrawal et al., Conformal phased surfaces for wireless 97. M. Monge, A. Lee-Gosselin, M. G. Shapiro, A. Emami, ehv214; pmid: 26045305
powering of bioelectronic microdevices. Nat. Biomed. Eng. Localization of microscale devices in vivo using addressable 117. G. Yao et al., Effective weight control via an implanted
1, 0043 (2017). doi: 10.1038/s41551-017-0043; transmitters operated as magnetic spins. Nat. Biomed. Eng. self-powered vagus nerve stimulation device. Nat. Commun.
pmid: 29226018 1, 736–744 (2017). doi: 10.1038/s41551-017-0129-2; 9, 5349 (2018). doi: 10.1038/s41467-018-07764-z;
78. J. S. Ho et al., Wireless power transfer to deep-tissue pmid: 31015668 pmid: 30559435
microimplants. Proc. Natl. Acad. Sci. U.S.A. 111, 7974–7979 98. S. Baltsavias et al., In vivo wireless sensors for gut 118. H. Ouyang et al., Symbiotic cardiac pacemaker. Nat. Commun.
(2014). doi: 10.1073/pnas.1403002111; pmid: 24843161 microbiome redox monitoring. IEEE Trans. Biomed. Eng. 67, 10, 1821 (2019). doi: 10.1038/s41467-019-09851-1;
79. A. Aldaoud et al., Near-field wireless power transfer to 1821–1830 (2020). pmid: 31634824 pmid: 31015519
stent-based biomedical implants. IEEE J. Electromagn. 99. M. S. Wegmueller, M. Oberle, N. Felber, N. Kuster, W. Fichtner, 119. H. Ryu et al., Self-rechargeable cardiac pacemaker system with
RF Microw. Med. Biol. 2, 193–200 (2018). doi: 10.1109/ Signal transmission by galvanic coupling through the human triboelectric nanogenerators. Nat. Commun. 12, 4374 (2021).
JERM.2018.2833386 body. IEEE Trans. Instrum. Meas. 59, 963–969 (2010). doi: 10.1038/s41467-021-24417-w; pmid: 34272375
80. R. Jegadeesan, K. Agarwal, Y.-X. Guo, S.-C. Yen, N. V. Thakor, doi: 10.1109/TIM.2009.2031449 120. C. Dagdeviren et al., Conformal piezoelectric energy
Wireless power delivery to flexible subcutaneous implants 100. S. Sriram, S. Avlani, M. P. Ward, S. Sen, Electro-quasistatic harvesting and storage from motions of the heart, lung, and
using capacitive coupling. IEEE Trans. Microw. Theory Tech. animal body communication for untethered rodent diaphragm. Proc. Natl. Acad. Sci. U.S.A. 111, 1927–1932
65, 280–292 (2017). doi: 10.1109/TMTT.2016.2615623 biopotential recording. Sci. Rep. 11, 3307 (2021). (2014). doi: 10.1073/pnas.1317233111; pmid: 24449853
81. R. Erfani, F. Marefat, A. M. Sodagar, P. Mohseni, Modeling doi: 10.1038/s41598-021-81108-8; pmid: 33558552 121. W. Ren et al., High-performance wearable thermoelectric
and experimental validation of a capacitive link for wireless 101. B. Chatterjee et al., “A 1.15 mW 5.54 mm 3 implant with generator with self-healing, recycling, and Lego-like
power transfer to biomedical implants. IEEE Trans. Circuits a bidirectional neural sensor and stimulator SoC utilizing reconfiguring capabilities. Sci. Adv. 7, eabe0586 (2021).
Syst. II Express Briefs 65, 923–927 (2018). doi: 10.1109/ bi-phasic quasi-static brain communication achieving 6kbps- doi: 10.1126/sciadv.abe0586; pmid: 33568483
TCSII.2017.2737140 10Mbps uplink with compressive sensing and RO-PUF based 122. S. Hong et al., Wearable thermoelectrics for personalized
82. Y. Zhang et al., Battery-free, lightweight, injectable collision avoidance,” in 2021 Symposium on VLSI Circuits, thermoregulation. Sci. Adv. 5, eaaw0536 (2019).
microsystem for in vivo wireless pharmacology and Kyoto, Japan, 13–19 June 2021 (IEEE, 2021); doi: 10.23919/ doi: 10.1126/sciadv.aaw0536; pmid: 31114803
optogenetics. Proc. Natl. Acad. Sci. U.S.A. 116, 21427–21437 VLSICircuits52068.2021.9492445 123. K. Nan et al., Compliant and stretchable thermoelectric
(2019). doi: 10.1073/pnas.1909850116; pmid: 31601737 102. Z. Zhao, G. D. Spyropoulos, C. Cea, J. N. Gelinas, coils for energy harvesting in miniature flexible devices.
83. Y. Zhang et al., Battery-free, fully implantable optofluidic D. Khodagholy, Ionic communication for implantable Sci. Adv. 4, eaau5849 (2018). doi: 10.1126/sciadv.aau5849;
cuff system for wireless optogenetic and pharmacological bioelectronics. Sci. Adv. 8, eabm7851 (2022). doi: 10.1126/ pmid: 30406207
neuromodulation of peripheral nerves. Sci. Adv. 5, sciadv.abm7851; pmid: 35385298 124. C. S. Kim et al., Self-powered wearable electrocardiography
eaaw5296 (2019). doi: 10.1126/sciadv.aaw5296; 103. J. P. Wright et al., A fully implantable wireless bidirectional using a wearable thermoelectric power generator. ACS Energy
pmid: 31281895 neuromodulation system for mice. Biosens. Bioelectron. Lett. 3, 501–507 (2018). doi: 10.1021/acsenergylett.7b01237
84. P. Gutruf et al., Fully implantable optoelectronic systems 200, 113886 (2022). doi: 10.1016/j.bios.2021.113886;
for battery-free, multimodal operation in neuroscience pmid: 34995836 AC KNOWLED GME NTS
research. Nat. Electron. 1, 652–660 (2018). doi: 10.1038/ 104. A. D. Mickle et al., A wireless closed-loop system for V.N. thanks A. Singer, J. Woods, W. Kim, A. Tuppen, and G. Duret
s41928-018-0175-0 optogenetic peripheral neuromodulation. Nature 565, for useful scientific discussions. Funding: This work received
85. J. Lim et al., A light-tolerant wireless neural recording IC for 361–365 (2019). doi: 10.1038/s41586-018-0823-6; no funding. Author contributions: V.N. and J.T.R. designed the
motor prediction with near-infrared-based power and data pmid: 30602791 concept, wrote the introduction and the sections on battery-free

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 9 of 10

RES EARCH | R E V I E W

concepts and WPT, and prepared Figs. 1 to 3. G.B. and C.J.B. financial compensation from Motif Neurotech, Inc. K.Y. holds equity information: Copyright © 2023 the authors, some rights reserved;
wrote the section on energy harvesting. Z.Y. and K.Y. wrote the in Motif Neurotech, Inc. J.T.R. and K.Y. are inventors on patent exclusive licensee American Association for the Advancement of
section on communications and prepared Fig. 4. A.N.D. and application number PCT/US2020/026688 submitted by Rice Science. No claim to original US government works. https://www.
D.J.W. wrote the sections on wireless networks and current University that covers magnetoelectric power and data transfer to science.org/about/science-licenses-journal-article-reuse
technologies. All authors reviewed the manuscript and provided bioelectronic implants. The remaining authors declare no
input. Competing interests: J.T.R. is a co-founder and CEO of competing interests. Data and materials availability: Datasets Submitted 10 June 2022; accepted 28 September 2023
Motif Neurotech, Inc., and holds an equity stake in and receives used in this study are available on Zenodo (109). License 10.1126/science.abn4732

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

Nair et al., Science 382, eabn4732 (2023) 10 November 2023 10 of 10

 Miniature battery-free bioelectronics
Vishnu Nair, Ashley N. Dalrymple, Zhanghao Yu, Gaurav Balakrishnan, Christopher J. Bettinger, Douglas J. Weber,
Kaiyuan Yang, and Jacob T. Robinson

Science 382 (6671), eabn4732. DOI: 10.1126/science.abn4732

Downloaded from https://www.science.org at Carnegie Mellon University on November 10, 2023

Editor’s summary
Bioelectronic devices that can sense or manipulate biological signals, such as pacemakers that can detect and
regulate irregular cardiac cycles, can dramatically improve the health and lifestyle of the patients who use them.
However, these devices are often limited by the storage capacity of an onboard battery or are tethered to wires that
can cause infection. Nair et al. reviewed advances in developing alternate methods to generate, transmit, and store
electrical charge that enable wireless power transfer and energy harvesting safely through and within the human body.
In addition to removing power limitations, these advances often also enable communication from a device or possibly
even between devices. —Marc S. Lavine

View the article online

https://www.science.org/doi/10.1126/science.abn4732
Permissions
https://www.science.org/help/reprints-and-permissions

Use of this article is subject to the Terms of service

Science (ISSN 1095-9203) is published by the American Association for the Advancement of Science. 1200 New York Avenue NW,
Washington, DC 20005. The title Science is a registered trademark of AAAS.

Copyright © 2023 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim
to original U.S. Government Works