The human heart beats roughly 100,000 times a day, throughout a person's life. If that constant motion could be converted into electricity, it might ease the battery-life problem facing implantable medical devices such as pacemakers. Most implantable devices today rely on chemical batteries. Replacing a depleted battery requires surgery, which adds physical burden for patients, infection risk, and medical costs.
To address this, Associate Professor Jun Yao and doctoral student Siqi Wang of the University of Massachusetts Amherst (UMass Amherst), working with the University of Connecticut, developed a new generator that harvests electricity from the beating of cardiac muscle cells. Rather than attaching a generator to the surface of an organ, as in conventional approaches, the team cultured cardiomyocytes inside an ultrathin mesh embedded with tiny piezoelectric elements, integrating the device with the tissue. The results were published in the October 2026 issue of Science Advances (Vol. 12, Issue 41, DOI: 10.1126/sciadv.aei5963).
The key feature is that the team boosted power generation by redesigning the device structure rather than improving the piezoelectric material itself. However, this work is an in vitro proof of concept using cultured cardiomyocytes derived from human cells. It has not demonstrated operation inside a living body, nor that the device can replace a pacemaker battery.
A Generator Mesh Integrated Into Cardiac Tissue, Not Attached to Its Surface
"Energy harvesting," which converts motion from heartbeats and breathing into electricity, has long been studied as a power source for implantable medical devices. Conventional piezoelectric generators, however, suffer from a mismatch in stiffness with biological tissue. A typical generator patch forms piezoelectric material and electrodes on a supporting substrate such as a polymer film, then attaches it to the surface of the heart or a muscle. If an artificial material is too stiff relative to soft tissue, it not only hinders the tissue's natural movement but also fails to extract enough of the deformation needed for power generation.
Another problem is that the device volume is not used efficiently for generation. In surface-mounted devices, the contact area with tissue is limited, so deformation arising inside the tissue is not fully transmitted to the piezoelectric elements. In addition, the supporting substrate, which does not directly contribute to power generation, occupies most of the device's volume, so output per unit volume tends to be low. Making the device thinner increases flexibility, but then it becomes difficult to apply enough strain to the piezoelectric material.
The team devised a "biohybrid mesh harvester," in which tiny piezoelectric elements are distributed across a porous, net-like structure. The mesh serves as both the support for the electrical circuitry and a scaffold on which cells adhere and form tissue. Cardiomyocytes grow into the gaps of the mesh and wrap around the fine wiring, so the mesh ends up embedded inside the tissue. This transmits cell contraction directly into bending of the piezoelectric elements while reducing the volume of substrate that contributes nothing to power generation.
Whereas conventional devices received motion from the surface of tissue, this mesh receives motion from within it. The basic idea of the study is that this structure can reconcile flexibility with power-generation performance.
Just 3.7 μm Thick: A Flexible Structure Cardiomyocytes Can Incorporate
The mesh the team fabricated covers an area of 7 mm by 8 mm, with tiny ribbons of lead zirconate titanate (PZT), a piezoelectric material, arranged in 16 rows by 4 columns. Each PZT element is about 50 μm wide and about 470 μm long, and four neighboring elements along the long axis are connected in series to form a generation unit. PZT is a piezoelectric ceramic that converts mechanical deformation into electric charge and is widely used in sensors and actuators.
The wiring ribbons that make up the mesh are about 20 μm wide and only 3.7 μm thick. The basic structure combines a 1.5 μm-thick polyimide (PI) bottom layer, a platinum (Pt) thin-film wiring layer in the middle, and a 2 μm-thick SU-8 insulating protective layer on top. Thanks to this thinness, the bending stiffness of the wiring ribbon is about , giving it the flexibility to follow the deformation of biological tissue. The arrangement of the mesh is also designed so that it bends easily in a particular direction as cardiac muscle contracts.
The PZT thin film was formed by a sol-gel method and then poled, applying a DC electric field of 150 kV/cm at 150°C to bring it into a state where it exhibits piezoelectric properties. Before using cells, the team checked power-generation performance by mechanically deforming a 1-row by 9-column test array cut from the mesh. In a buckling test applying roughly 0.1% strain, each PZT element produced open-circuit voltage pulses averaging , with little variation between elements.
After fabrication, the mesh was released from the silicon substrate by dissolving a germanium sacrificial layer between them with a hydrogen peroxide solution. It was then washed and sterilized, and its surface was treated with poly-D-lysine and Matrigel to make it easier for cardiomyocytes to adhere. This prepared the flexible, few-micrometer-thick mesh for integration with biological tissue.
Power Confirmed From Cultured Heart Muscle Beating, but Initial Design Fell Short
The team cultured cardiomyocytes differentiated from human embryonic stem cells (hESC-CMs) on the mesh. They used cells 10 to 12 days after differentiation. After adhering to the mesh surface, the cells spread through the gaps to the top, side, and bottom surfaces. As culture progressed, the cells bonded to one another to form layered tissue. The contractile force of the tissue partially peeled it from the culture dish, turning it into a free-standing composite tissue with the mesh embedded inside.
Three-dimensional observation with confocal laser scanning microscopy also confirmed that the mesh was not merely attached to the surface of the cells but enveloped by the cardiac tissue. The cultured tissue beat spontaneously at 0.3 to 0.5 Hz, and the mesh bent periodically in time with that movement. Measurements of five representative generation units showed voltage pulses synchronized with the heartbeat, confirming that the mechanical motion of the cells could be converted into electricity via the piezoelectric effect.
However, the output of the initial mesh design was insufficient. With the uniform 3.7 μm-thick structure, the measured voltage amplitude averaged only about 1.4 mV, far smaller than the roughly 0.49 V obtained in the mechanical buckling test.
One cause is parasitic capacitance in the measurement system. Tiny piezoelectric elements generate little charge, and the capacitance of the cables and the instrument's input stage lowers the observed voltage. After the team corrected for the effect of parasitic capacitance, the intrinsic open-circuit voltage of the element was estimated at about 9 mV. Even so, this falls well short of the value from the mechanical test.
The team therefore measured changes in curvature of the cardiac tissue through image analysis and estimated the strain applied to the PZT layer using multilayer composite beam theory. The result showed that the average strain generated by the heartbeat was only about (0.00064%). That is more than two orders of magnitude smaller than the roughly $10^{-3}$ (0.1%) applied in the buckling test. In other words, the beating cells were bending the mesh, but not deforming the piezoelectric material enough.
Shifting the Neutral Plane Alone Raised Output Voltage Up to 80-Fold
To solve the output shortfall, the team looked not at increasing the force applied to the piezoelectric material but at extracting greater strain from the same bending. The key was the position of the "structural neutral plane" that exists in a stack of multiple materials.
When a thin plate or beam bends, a neutral plane exists inside it that neither stretches nor shrinks along its length. The farther a position is from the neutral plane, the greater the tensile or compressive strain from bending. In a simple bending model, the relationship among strain , curvature , and distance from the neutral plane $y$ is expressed as follows.
This relationship shows that increasing the distance between the piezoelectric layer and the neutral plane can impose greater strain on the piezoelectric material even at the same curvature. But thickening the entire mesh also raises its bending stiffness, so soft cardiac tissue can no longer deform the mesh sufficiently. Because the bending stiffness of a beam is generally proportional to the cube of its thickness, simply making the film thicker could backfire.
The team therefore adopted "local structural engineering," which thickens only the power-generating portions. While keeping the wiring at 3.7 μm thick, they locally thickened only the SU-8 protective layer above the PZT elements to about 8.5 μm. This preserves the flexibility of the mesh as a whole while moving the neutral plane away from the center of the PZT layer in the generating portions, amplifying the bending strain.
With this change, the theoretical average strain applied to the PZT layer during heartbeats reached about . That is more than three times the theoretical value derived from the curvature without local thickening (about ), and more than 10 times the roughly estimated for the initial mesh.
Power-generation performance also improved significantly. The voltage amplitude observed directly by the instrument rose from about 1.4 mV in the initial design to 30 to 50 mV. The open-circuit voltage estimated after correcting for parasitic capacitance was 450 to 750 mV, roughly 50 to 80 times the approximately 9 mV of the initial design. The energy obtained from a single heartbeat was calculated at about 7 pJ (picojoules).
The main measurement results are summarized below. Note that the measured or estimated strain values and the voltage measurement conditions differ in each case, so the benchtop test and the tests using cultured tissue cannot be compared as identical conditions.
| Condition | PZT layer strain | Voltage amplitude | Notes |
|---|---|---|---|
| Initial mesh | About | Measured about 1.4 mV; estimated about 9 mV after calibration | Flexible, but insufficient strain reaches the piezoelectric layer |
| Locally engineered mesh | About (theoretical) | Measured 30–50 mV; estimated 450–750 mV after calibration | Neutral plane position adjusted to amplify local strain |
| Benchtop buckling test | About $10^{-3}$ | Average $0.49\pm0.011$ V | Mechanical test applying relatively large deformation with external equipment |
The team further reports that the improved mesh achieved an effective power density exceeding 10 μW/cm³ by volume. Most conventional surface-mounted energy harvesters are said to produce less than 1 μW/cm³, so this is more than a 10-fold improvement. Since μW/cm³ expresses power per unit volume, it is strictly more appropriate to call it "volumetric power density" rather than "energy density."
This improvement did not come from developing a new piezoelectric material. The PZT thin film used has an energy conversion efficiency of about 3.5%, so the material itself saw no special performance gain. In conventional structures, a thick supporting substrate that contributes nothing to generation occupies most of the volume; the team replaced it with a porous mesh. By also locally controlling the position of the neutral plane, they made it possible to transmit the small deformations produced by cells efficiently to the piezoelectric elements. The key achievement of this study is that it raised power density through structural and mechanical design rather than material performance.
Lead Safety and Higher Absolute Output Are Essential for Practical Use
The results show the potential of generators integrated with biological tissue, but major challenges remain before they could power medical devices such as pacemakers. Two stand out: the safety of lead-containing piezoelectric material and the absolute amount of power that can actually be extracted.
First, the PZT used here contains toxic lead (Pb). If lead leaks into the body, it could harm the nervous system and organs, so strict safety verification is needed for devices implanted for long periods. The team protected the elements with polymer insulating layers of polyimide and SU-8 and confirmed that the mesh maintained stable electrical insulation in cell culture medium for more than two months. In staining tests that distinguish live from dead cells, cardiomyocytes cultured on the mesh showed high viability and metabolic activity comparable to normal culture conditions.
However, these results guarantee only short- to medium-term compatibility in an in vitro culture environment. Inside the body, the device would face harsher conditions, including long-term exposure to body fluids, repeated deformation from heartbeats, and immune responses. For use over years or decades, it must be verified that a polymer coating only a few micrometers thick develops no fine cracks or defects and keeps the lead reliably contained. Replacing it with lead-free piezoelectric materials such as potassium sodium niobate (KNN) will also be an important research topic.
The other major challenge is the shortage of power itself. Although the volumetric power density exceeded 10 μW/cm³, the energy extracted from a single generation unit is only about 7 pJ per heartbeat. This means that even though power density normalized to the tiny element volume is high, the power obtained by the device as a whole is still very small. Continuous operation of a typical pacemaker is said to require several microwatts to several tens of microwatts, leaving a large gap relative to the output of a single element here. The team has not demonstrated driving an external medical device with the generator alone.
In addition, the study measured multiple generation units in the mesh individually and did not implement a circuit to collect power from the entire array. In cardiac tissue, different regions contract with time lags as electrical excitation propagates, so the phase of voltage pulses may differ between generation units. To combine many elements in the future, the output of each unit will need to be rectified and integrated, together with power management circuitry that efficiently delivers it to an energy storage device.
The team also envisions three-dimensional stacking of the mesh as a way to scale up output. If multiple mesh layers can be placed within tissue, taking advantage of the ultrathin, flexible structure, generated power might be increased without greatly expanding the planar footprint. Moreover, the cultured cardiac tissue used here was only several tens of micrometers thick, whereas the actual heart wall is several millimeters to more than 1 cm thick. If the device can eventually be integrated safely with such thick tissue, larger mechanical deformations might be exploited. At present, however, neither output gains from stacking nor long-term operation in vivo has been demonstrated.
What this study shows is a new design approach: instead of attaching a generator to an organ's surface, embed it inside living tissue and draw power from its natural motion. Using conventional piezoelectric material, the team greatly raised power density by reducing the supporting structure and controlling the neutral plane, which is significant from an engineering standpoint.
Whether this technology leads to battery-free implantable medical devices depends on future verification. Can it remain integrated with tissue in animal experiments in vivo? Can it withstand repeated deformation while preventing leakage of toxic substances? Can it stably extract enough power to run an actual medical device? Solving these challenges will be the key to moving from laboratory proof of concept to medical application.