A team working in MIT’s Department of Biological Engineering has kept a micro-LED continuously lit for 72 hours using nothing but electricity generated by living bacteria growing on an electrode. No battery, no external power line, just a colony of exoelectrogenic microbes metabolizing nutrients and shedding electrons into a circuit. The light output is faint, the setup is bench-scale, and the whole system needs to be fed like anything else that’s alive. But 72 hours is long enough to matter, because the previous ceiling for this kind of demonstration was measured in hours, not days.
Here’s why you should care about a dim LED in a Cambridge lab. Bioelectronic medicine, the field building vagus nerve stimulators, retinal implants, and closed-loop neuromodulation devices, has a power problem that nobody talks about as much as the electrodes or the algorithms. Every implant needs a source of current, and right now that almost always means a battery: lithium, rigid, finite, and foreign to the tissue around it. A power source that grows from biological material, tolerates the body’s chemistry, and refuels itself from ambient nutrients would solve a problem that battery engineers have been unable to touch. This demonstration doesn’t solve it. It gestures at the shape of a solution.
The Electricity Was Never the Hard Part, the Duration Was
Bacteria that generate usable current are not new or exotic. Species like Shewanella oneidensis and Geobacter sulfurreducens have been studied for over two decades because they do something most organisms don’t bother with: they move electrons outside their own cell walls, using external metal or mineral surfaces as an electron dump instead of oxygen. Put a colony of these microbes on an anode submerged in a nutrient-rich solution, put a cathode nearby, and you have a microbial fuel cell. Current flows because the bacteria are, in effect, breathing onto the electrode.
The physics of this has been demonstrated in a lab since the early 2000s. Derek Lovley’s group at UMass Amherst has spent years characterizing how Geobacter builds conductive protein nanowires to move those electrons, work that eventually led to a separate device called the Air-gen, which pulls current from the humidity in ordinary air using those same protein filaments. Seokheun Choi’s lab at Binghamton University has built paper-based microbial fuel cells cheap enough to be disposable, powering small LEDs and sensors for diagnostic strips. None of this is the hard part anymore. The hard part is keeping the bacteria alive, fed, and electrochemically active long enough for the output to be useful for something other than a demo.
Seventy two hours isn’t a power spec. It’s a survival spec.
That’s the actual news here. Most published microbial fuel cell demonstrations that power an LED or a small sensor run for a few hours before the biofilm’s output degrades, the nutrient solution depletes, or the electrode surface fouls. Getting to 72 hours means the MIT team solved, at least partially, the maintenance problem: a slow-release nutrient reservoir and an electrode geometry that keeps the biofilm from smothering itself in its own waste products. It’s an engineering result dressed up as a biology result.
What The Bacteria Are Actually Doing In That Circuit
Picture a thin biofilm, a living layer maybe a few cell-widths thick, coating a conductive surface. The bacteria inside it are consuming an organic substrate, breaking it down for energy the same way any microbe would, except that a fraction of the electrons produced in that metabolic process get routed out through the cell membrane instead of staying inside to reduce oxygen. Those electrons travel through conductive pili or membrane proteins to the electrode surface, and from there into an external circuit, where they can do work, in this case driving a micro-LED that draws current in the microwatt range.
Microwatts sound trivial next to the milliwatts a hearing aid battery delivers, and they are. A modern pacemaker battery runs for eight to twelve years on a single charge doing far more electrical work than lighting an LED. The comparison isn’t meant to flatter the bacteria. It’s meant to locate this result honestly: this is a proof that the mechanism can be sustained, not a proof that it can compete with lithium on power density. Those are different claims, and conflating them is how a modest lab result turns into an overhyped headline.
The bacteria aren’t the battery. They’re the living catalyst inside the fuel cell.
Why Anyone Bothers Trying To Replace A Battery At All
You might reasonably ask why this is worth the trouble when batteries already work fine. The answer is that they work fine until the device has to live inside a body for years, or until it has to be small enough to sit on a retina or thread through a blood vessel. Lithium batteries are rigid, they degrade with charge cycles, and their failure mode in an implant is a surgical problem, not just an inconvenience. Encapsulation to keep battery chemistry away from tissue adds bulk to devices that are already fighting for space. Neural interfaces in particular have spent the last decade shrinking electrode arrays down to the point where the battery, not the electronics, is now the largest single component in some designs.
A power source built from biological material sidesteps some of that. It could in principle be biodegradable, refuel itself from glucose or other metabolites already present in the body, and avoid the rigid-device-in-soft-tissue mismatch that causes scarring and signal loss over time. Glucose biofuel cells, a chemical cousin of the microbial approach that skips the living organism and uses enzymes directly, have already been tested as implantable power sources in animal models, generating enough current to trickle-charge a capacitor for low-draw sensors. The bacterial version adds a layer of complexity, a living organism that has its own needs and its own failure modes, but it also adds something the enzyme systems don’t have: self-renewal. Enzymes degrade and stop working. Bacteria, fed properly, keep dividing and keep working.
Where This Sits Next To Other Attempts At The Same Problem
The MIT result is part of a small but persistent line of research into biologically derived power, most of it aimed at low-draw sensors and disposable diagnostics rather than implants. Here’s roughly how the recent landmarks stack up against each other.
| Project | Institution | Power source | What it drove | Sustained duration |
|---|---|---|---|---|
| Micro-LED demonstration | MIT, Dept. of Biological Engineering | Exoelectrogenic biofilm microbial fuel cell | Single micro-LED | 72 hours |
| Paper-based MFC | Binghamton University | Dried bacterial spores rehydrated on paper substrate | LED and biosensor circuits | Hours, single-use |
| Air-gen protein nanowire generator | UMass Amherst | Geobacter-derived protein nanowires, no living cells | Small electronic loads from ambient humidity | Continuous, no feeding required |
| Pacemaker lithium battery | Standard clinical devices | Lithium-iodide chemistry | Full pacing and sensing circuitry | 8 to 12 years |
That last row is the point. Nothing biological is anywhere near competing with a pacemaker battery on raw endurance, and it’s worth saying plainly that it may never need to. Bioelectronic devices don’t all require decade-long power. Some need to run for a few days after implantation while tissue heals, or for the duration of a diagnostic window, or intermittently rather than continuously. A biological power source that lasts 72 hours and then gets reabsorbed or excreted might be exactly the right tool for a transient sensor, even while being the wrong tool for a permanent neural implant.
The Part Of This Story That Deserves Some Skepticism
The honest limitation is that a microbial fuel cell needs an ongoing supply of organic substrate, meaning food, and in a lab that’s a reservoir of nutrient solution refreshed on a schedule. Inside a body, that substrate would have to come from the surrounding tissue itself, which raises questions the MIT team hasn’t answered yet: what happens to local glucose or lactate concentrations near an implant that’s actively feeding a bacterial colony, and how does the immune system respond to a permanent colony of foreign bacteria sitting inside a device, even one that’s contained behind a membrane. Encapsulating living bacteria well enough to keep them separate from the immune system while still letting nutrients in and electrons out is its own unsolved engineering problem, arguably harder than the fuel cell chemistry itself.
There’s also the matter of scale. A single micro-LED draws an amount of current that a determined engineer could generate a dozen other ways, including a coin battery smaller than a fingernail. The interesting question is whether this approach scales to something that matters clinically, a sensor transmitting data wirelessly, a drug pump, a stimulation electrode, and whether it can do that while staying small enough to implant. Seventy two hours of a dim light proves the concept survives its own metabolism. It doesn’t prove the concept scales.
What Seventy Two Hours Actually Buys The Field
What this result buys the field is a benchmark. Every subsequent group working on living power sources for bioelectronics now has a number to beat, and a demonstrated engineering approach, nutrient reservoirs paired with fouling-resistant electrodes, that others can adapt or improve on. That’s how slow research areas move: not through a single breakthrough that solves implantable power overnight, but through a sequence of demonstrations that each push the durable-operation window a little further, from hours to days, and eventually, maybe, to weeks.
The question worth sitting with is not whether bacteria will replace lithium in your next pacemaker. They almost certainly won’t, not in your lifetime, not for a device that needs to run reliably for a decade in your chest. The real question is whether the next generation of short-lived, dissolvable, or minimally invasive sensors, the kind meant to do one job for a few days and then disappear, will end up powered by something closer to biology than chemistry. The MIT lab’s 72 hours is a small, specific answer to a much bigger question the field hasn’t fully asked yet: does an implant’s power source have to outlive the implant itself, or can it be allowed to live and die on the same clock.
Credit: masakazu sasaki on Unsplash