The paper battery dissolved. Not all the swallowable electronics did.

A swallowable electronic capsule must survive exactly what should eventually destroy it: stomach fluid. Keep the liquid out too briefly and the battery dies before its work is done. Keep it out indefinitely and the patient is left carrying ordinary electronic waste through the gut.

Researchers have now built a paper-based battery intended to negotiate that contradiction. Protected by natural wax, it powered tracking and electrical-stimulation prototypes inside pigs before its materials began breaking apart. That matters because ingestible electronics could someday monitor whether medicine was swallowed or provide temporary stimulation without requiring a retrieval procedure.

The crucial qualification is equally interesting: the battery was designed to disappear, but the demonstrated capsule did not disappear completely. Its conventional circuit board passed intact, and investigators sometimes recovered metallic remnants. This is evidence that a transient power source can operate in a large animal—not that a fully resorbable medical device is ready for people.

A battery wearing a raincoat with a timer

The cell described in Nature Chemical Engineering paired a magnesium-alloy anode with a cathode made from molybdenum trioxide, activated carbon and cellulose nanofibrils. The last material binds the cathode into a paper-like sheet. A gelatin gel containing choline chloride and lactic acid carried ions between the electrodes.

Beeswax supplied the temporary raincoat. For longer passage experiments, the researchers added slower-degrading candelilla wax. The coating had to exclude fluid while the battery operated, then gradually delaminate so gastric liquid could reach the metals.

That is what bioresorbable means here: the selected materials can break down into products the body can process or excrete. The paper explicitly says the term does not mean nutritional or food-grade. This battery was not edible in the everyday sense.

The larger cell reached a peak open-circuit voltage of 1.84 volts and a maximum measured capacity of 3.5 milliamp-hours. Under continuous discharge it maintained about 1.6 volts for a day. Wax-coated cells retained useful output after as long as three days in pigs’ stomachs, although voltage and capacity declined as moisture penetrated the package.

Those measurements answer different questions:

A small calculation shows why these quantities should not be casually blended. Multiplying the peak voltage by the maximum capacity gives 1.84 V × 3.5 mAh = 6.44 mWh. Yet multiplying the separately reported areal energy density by the cell’s 8-by-24-millimetre footprint gives 4.13 mWh/cm² × 1.92 cm² = 7.93 mWh.

Those cannot both be the exact delivered energy of one cell under one test. The likely explanation is that maximum voltage, capacity and areal energy came from different operating conditions or active-area conventions. The useful result is not a homemade 6.44- or 7.93-millwatt-hour label; it is the warning that battery figures are only comparable when the load and measurement method match.

One cell, three kinds of evidence

The paper combines several demonstrations that are easy to blur together.

On the bench, the researchers measured voltage, capacity and discharge behaviour. A low-power stimulation board ran for up to three days from one cell and seven days from two cells connected in parallel. Its output was a 0.5-milliamp pulse lasting 0.3 milliseconds, repeated 14 times each second.

That waveform was on for only 0.0003 × 14 = 0.0042, or 0.42% of the time. Multiplying that fraction by the pulse current gives an average stimulation component of roughly 0.0021 milliamps, before allowing for the control electronics and conversion losses. This severe duty-cycling helps explain how a modest cell could support a multiday demonstration.

For wireless tracking, a battery-assisted RFID tag communicated across four metres in air. In a controlled animal setup, with the reader fixed 1.5 metres away, the signal changed after the capsule was placed in the esophagus. That establishes a detectable event under fixed geometry. It does not establish reliable medication tracking while a person walks, turns or moves among rooms; the authors identify body orientation, distance and individual anatomy as unresolved complications.

For gastric stimulation, the capsule delivered 20 minutes of electrical pulses while investigators measured the hunger-related hormone ghrelin. The figure and source data identify four independent animals, while the accompanying prose inconsistently says three. The paper reports an average increase of 36.3% with a standard deviation of 12.85%, but that internal counting discrepancy should remain visible rather than being tidied away.

Tissue from two pigs was examined after stimulation, and the researchers reported no significant damage at the treated sites. Four hormone measurements and two tissue examinations are feasibility evidence. They cannot establish treatment effectiveness, rare harms or human safety.

Why runtime comparisons can play tricks

An earlier FLASH capsule used two conventional 1.5-volt, 28-milliamp-hour silver-oxide cells in series. In series, voltage adds while capacity does not: 3 V × 28 mAh = 84 mWh nominally. Its pig experiment used 20 minutes of stimulation, and the designers budgeted power for up to about an hour.

The newer paper battery appears to hold far less energy, yet its revised circuit produced pulses for days. That is not evidence that the paper chemistry somehow defeated arithmetic. The devices had different circuits, loads, encapsulation and operating schedules. Runtime describes the entire system; energy capacity describes the cell under specified conditions.

The same caution applies to battery-free ingestible designs. If a device receives energy from an external transmitter, onboard milliwatt-hours are no longer a meaningful basis for comparison. The pertinent questions become how reliably energy reaches the capsule, at what distance and orientation, and what equipment must remain nearby.

What actually disappeared

In simulated gastric fluid at 37°C, the stimulation capsule began coming apart slowly. After the first 14 days, the researchers raised the fluid temperature to 75°C to accelerate degradation. The capsule was mostly dissolved by 40 days and reduced to small particles by 90 days—but most of that timeline came from an accelerated test, not an ordinary stomach.

Passage experiments supplied a different observation. Over one to two weeks in pigs, capsules softened, fragmented and disassembled. Small metallic battery remnants were occasionally detected, and molybdenum current collectors were recovered. The conventional stimulation circuit board remained intact and was excreted in every passage test.

The tracking capsule had another exception: its roughly 18-square-millimetre RFID chip was not resorbable and was expected to pass naturally. “Battery dissolved” and “device vanished” are therefore different claims.

The authors also report variation among hand-built battery units and list manufacturing consistency, shelf life, fed-state behaviour, changing acidity and mucus as work still to be done. Their animals were fasted, and the cohorts were small. Real meals and ordinary movement could alter both the wax’s lifetime and the electronics’ contact with tissue.

The achievement is narrower—and more useful—than a disappearing electronic pill. Researchers showed that a wax-protected, bioresorbable battery can remain functional in a pig long enough to power controlled tracking and stimulation experiments. The next engineering contradiction is waiting inside the same capsule: making the circuitry disappear as convincingly as its power source.