Introduction: The Subtraction Has Two Terms
Solid-state fusion (SSF) is a contested area of research in which some scientists claim to have measured more heat coming out of a metal-and-hydrogen electrochemical cell than the electricity flowing in can account for. The excess, if real, would suggest an unknown energy source — possibly nuclear — operating inside the metal. The case for it rests on a subtraction: measure the thermal power the cell gives off, subtract the electrical power you put into running it, and whatever remains is the claimed anomaly. Almost three decades of argument have fixed on the first term — the heat coming out — asking whether the calorimeter was designed correctly, whether heat losses were properly calibrated, whether evolved gases recombined inside the cell or escaped. The second term, the electricity going in, tends to get waved through as though reading it were obvious. It is not.
The input to a working electrochemical cell is electrical power. For a steady, unchanging current at a fixed voltage — what engineers call clean DC — power really is just voltage multiplied by current. The moment the drive becomes anything more complex, whether it ripples slightly, gets pulsed on and off, or carries an alternating component on top of the DC, that simple multiplication fails. Real power is the time average of the instantaneous product of voltage and current at each moment: (1/T)∫ v(t) i(t) dt. That integral and the product of separately averaged readings can disagree by several percent, and the gap grows with the complexity of the waveform.1 A few percent error in the input term, in a cell where the claimed excess is itself only a few percent of the total power throughput, is the whole difference between a genuine anomaly and a measurement artifact.
That is the concern, and it is worth stating at its true width, because that width is narrower than it first sounds. On clean, constant-current DC — the drive used in the most widely cited palladium experiments — multiplying voltage by current already gives the real power, and this entire objection evaporates. The problem bites only for experiments that deliberately leave clean DC behind: those using pulsed current, current ramps, or AC superimposed on DC, often on the premise that far-from-equilibrium conditions are where any anomalous effect will appear. For those experiments, a mismeasured input can manufacture an apparent excess, or bury a real one. How large that subset of experiments is remains, honestly, unquantified. The defensible claim is not that the whole controversy is a power-measurement artifact. It is narrower and sturdier: a real and overlooked class of input-power error can invalidate non-DC excess-heat results, a power engineer can eliminate it, and eliminating it raises the floor of the debate without settling the nuclear question.
Section I — What “Power” Means When the Current Isn’t Clean
Power systems engineers draw careful distinctions between types of power that the rest of the world treats as a single concept. Start with the simplest case. Real power, measured in watts, is the rate at which energy is actually consumed — the useful work done. Reactive power sloshes back and forth between the source and the load without being consumed, a consequence of inductors and capacitors storing and releasing energy on each half-cycle. Apparent power is the product of the RMS voltage and the RMS current — the number you get if you measure each separately and multiply. Under steady sinusoidal conditions these three quantities stand in a tidy, predictable relationship and a meter reports them without fuss. Under nonsinusoidal or distorted conditions — the ordinary state of anything switched, pulsed, or driven through power electronics — they come apart. Multiplying separate meter readings overestimates the real power delivered, sometimes by enough to matter. IEEE Standard 1459 pins down correct definitions for distorted waveforms precisely because the intuitive shortcuts fail in ways that quietly skew energy accounting.
An electrolysis cell might not seem like a distorted-waveform problem, and run on laboratory-grade constant current it mostly is not. But loading hydrogen into palladium or nickel is a moving target. The cell's electrical impedance — its resistance to the flow of alternating current, analogous to the familiar DC resistance but frequency-dependent — shifts as loading proceeds. That shift is actually useful: the ratio of the loaded cathode's resistance to its unloaded value tracks how much hydrogen has gone in, which is why it has become the field's standard loading diagnostic. Excess-heat reports cluster at high loading, above roughly a deuterium-to-palladium atom ratio of 0.85.2 Gas bubbles evolving at the electrode surface also modulate the impedance on short timescales. And several protocols that report the most striking results deliberately leave steady DC behind, using current ramps, pulsing, or a superimposed AC component — on the premise that transient, far-from-equilibrium conditions are where any effect lives. Whatever one makes of that premise, the consequence is mundane and unavoidable: the input power of such a cell cannot be read from one voltage and one current reading. It must be computed as real power, continuously, with instruments fast enough to capture the full waveform. This is routine in power engineering, and easy to get wrong anywhere else.
Section II — What Power Systems Brings to the Bench
The first contribution is the one already named. A wideband power analyzer — an instrument that samples voltage and current simultaneously thousands of times per second and integrates their product in real time, rather than multiplying their averages — removes an entire category of dispute from excess-heat measurement. It is unglamorous work, and on its own it settles nothing. What it does is put a defensible number on the input side, so that whatever the calorimetry reports on the output side can be argued on its merits instead of waved away on the electrical bookkeeping. That is a precondition for a verdict, not the verdict itself.
Power delivered to a cell is not the same as heat available inside it, and the second contribution is keeping that distinction straight. Electrolysis — using electricity to split a molecule — is thermodynamically uphill: the electrical energy drives a reaction that absorbs energy, and some of it leaves the cell as chemical energy carried away by escaping gas rather than becoming heat. Here, heavy water (D₂O) is split into deuterium gas and oxygen gas, which carry that energy away. The electrochemistry handles this by crediting only the power fed in above the thermoneutral potential — the voltage threshold, about 1.54 V for heavy water, below which the reaction cannot run — to available heat, and by tracking what fraction of the evolved gas recombines back inside the cell rather than escaping.3 The electrical accounting is a shared job: power metrology measures what enters, thermodynamics apportions where it goes. An engineer fluent on both sides of that handoff is exactly what a calorimetric program needs, and the field has identified completeness of energy-input accounting as a central worry.4
Third is grounding, shielding, and measurement integrity. High-current electrolysis read out at low signal levels is an electromagnetic interference problem before it is anything else. Ground loops — spurious currents that form when two parts of a measurement circuit connect to the same ground at different points and different potentials — inject false signals. Common-mode pickup is interference that appears identically on both signal leads and slips through as though it were real. Thermoelectric offsets are small voltages generated wherever two different metals meet at slightly different temperatures. All three mimic small, slow power imbalances — exactly the signature the experiment is trying to isolate. The same discipline that keeps a substation's protective relays from tripping on phantom signals is what keeps a calorimeter from reporting a phantom watt.
Fourth is stimulation without self-deception. Power electronics can synthesize almost any drive waveform a protocol calls for, deliver it cleanly, and record exactly what was delivered. That matters for a specific reason. If a protocol claims that pulsing or a particular waveform correlates with excess heat, the only way to test that claim rigorously is to control the drive precisely and measure its real power precisely — so the drive itself cannot be the hidden source of the apparent excess. A boundary is worth stating plainly here: a well-characterized waveform lets you test a correlation. It is not evidence that any waveform alters a nuclear reaction rate, a claim nothing in the mainstream literature supports and that this article does not make. Power electronics in this setting is a tool for rigor, not a proposed mechanism.
The scale is what makes all of this matter. One watt of steady power from deuterium fusing to helium-4 would require about 2.6×10¹¹ fusion events per second, each releasing 23.8 MeV of energy.5 The heat these experiments claim is real, and small: milliwatts to a few watts, riding on input power often ten or a hundred times larger. That ratio is the entire problem. It is why the input number cannot be approximate, and why a field whose whole signal lives in a residual has something concrete to gain from a discipline organized around not mismeasuring power.
Section III — What the Traffic Carries Back
The return cargo is lighter than the outbound load, and saying so is part of the case. Two items are solid, and a third is worth naming only in the conditional.
The first is materials. The active materials of SSF are metals — usually palladium or nickel — loaded with hydrogen (or its heavier isotope, deuterium) to extreme fractions, approaching one hydrogen atom per metal atom. The most durable finding from the modern era of SSF research is that this extreme-loading regime remains poorly understood.6 Metal hydrides — compounds formed when hydrogen atoms embed themselves into a metal lattice — are also the working substance of a serious grid-storage concept: hold energy as hydrogen absorbed into a solid, and release it on demand.7 The thermodynamic behavior of hydrogen in palladium and nickel near saturation, the phase changes at that boundary, the kinetics of loading and unloading under electrical drive — an engineer working on hydrogen storage and an SSF experimentalist are asking these questions of the same materials, and neither field has mapped that territory well. Measurements made for one purpose pay off for the other.
The second is measurement pushed somewhere uncomfortable. A program that must resolve a few milliwatts of real power against watts of throughput, continuously, for weeks, under a distorted drive, is asking for power metrology at the edge of what commercial instruments comfortably deliver. Problems posed at that edge tend to sharpen the instruments, and low-level wideband power measurement has uses well beyond this one argument.
The third item lives in the conditional, and should stay there. If SSF ever produced reliable, scalable excess heat — an if the present evidence does not support — what it produced would be a low-grade, distributed thermal source. Integrating low-grade distributed sources into the grid is itself a live, unfinished engineering problem. It would require thermal-to-electric conversion, power conditioning, and interconnection under standards such as IEEE 1547, the same framework being built out today for rooftop solar and behind-the-meter battery storage.8 That is a reason the outcome would matter to the field, not a reason to expect it. The engineering of the connection is worth pursuing only after the physics of the source is settled, and the physics is not settled.
Section IV — The Honest Ledger
Nothing here argues that the excess heat is real, and the strongest reasons for doubt are not about calorimetry at all. A nuclear energy source releasing heat at the rate these experiments claim should also be producing energetic byproducts — neutrons and gamma rays — in predictable proportion, because those are the known outputs when deuterium nuclei fuse at these energies. Every detector nearby should register them. The contested reports go the other way: helium-4 turns up in rough correlation with the heat, but without the energetic radiation those reactions demand. That missing radiation, not the helium, is the field's hardest problem, and its absence is an objection that stands independent of any calorimeter.9 When a team assembled by Google reexamined the claims with modern instruments and a real budget, it did not reproduce excess heat; what it flagged as genuinely open was the materials science of extreme hydrogen loading, not a confirmed effect.6 A power engineer brought onto one of these experiments cannot settle whether a nuclear process is occurring, and should not pretend otherwise.
What a power engineer can do is strip away the cheapest objections. Measure the input as real power with traceable, wideband instrumentation. Credit the electrolysis enthalpy and the recombination correctly. Engineer out the ground loops and the common-mode artifacts. Do that, and whatever the residual turns out to be becomes a number worth arguing about. For the non-DC experiments, that is the difference between a result and a rumor: today such a claim can be set aside before anyone reaches the physics, because its electrical bookkeeping leaves room for doubt. Closing that room is not an endorsement of the effect — it is the precondition for reaching a verdict on it, either way. That work is now funded and underway: in 2023 the U.S. Advanced Research Projects Agency for Energy committed roughly ten million dollars across eight teams, among them groups at MIT, Stanford, and Lawrence Berkeley, to settle the low-energy-nuclear-reaction question with modern instruments and either find the effect or close the book.10
Section V — Why Take This On
The pitch here is not that solid-state fusion is real. It is that for an important subset of these experiments, the central measurement is a power measurement — and it has too often been made below the standard the discipline takes for granted. The most useful thing you can carry to the bench is not belief in the effect. It is a wideband analyzer, a proper grounding scheme, and a flat refusal to let voltage-times-current stand in for real power. Apply those principles across enough well-characterized cells and one of two things follows. Either the residual survives honest electrical accounting, and the anomaly becomes a problem worth every instrument in the building. Or it does not, and one of the debate's easy excuses is retired for good. Neither outcome settles the nuclear question, and both are worth the effort. The subtraction has two terms, and one of them has been waiting for someone who measures power carefully.
Editorial note: This primer presents a scholarly synthesis of solid-state fusion's relationship to grid integration and power systems, written for a college-level audience. The underlying nuclear claims of SSF/LENR remain scientifically contested. Evidence claims are tiered as established, contested, reported, or speculative as noted inline. Readers are directed to primary experimental literature for empirical evaluation.
References & Footnotes
- IEEE, IEEE Standard Definitions for the Measurement of Electric Power Quantities Under Sinusoidal, Nonsinusoidal, Balanced, or Unbalanced Conditions, IEEE Std 1459-2010 (New York: IEEE, 2010); revised as IEEE Std 1459-2025. Real power is the time average of instantaneous voltage times instantaneous current; under nonsinusoidal conditions it is not recovered by multiplying separately averaged or RMS quantities. Established. ↩
- M. C. H. McKubre et al., “Using Resistivity to Measure H/Pd and D/Pd Loading: Method and Significance,” in Condensed Matter Nuclear Science: ICCF-12 Proceedings (Singapore: World Scientific, 2006). The resistance ratio is the standard non-destructive loading diagnostic; excess-heat reports cluster at high loading, above roughly a D/Pd ratio of 0.85. Single-source method reference; the 0.85 figure is approximate. ↩
- M. Fleischmann and S. Pons, “Calorimetry of the Palladium–Deuterium–Heavy Water System,” Journal of Electroanalytical Chemistry 287 (1990): 293–348. Cited for the electrochemical energy-balance method: the thermoneutral (enthalpy) potential for heavy-water electrolysis is about 1.54 V, and only input power above it, adjusted for the fraction of evolved gas that recombines in-cell, is available as heat. Presented as method, not as confirmation of excess heat. ↩
- U.S. Department of Energy, Report of the Review of Low Energy Nuclear Reactions (Washington, DC, December 2004). Reviewers divided on the excess-heat evidence; the completeness of the energy-input accounting was among the central calorimetric questions. Institutional. ↩
- Arithmetic: 1 W ÷ (23.8 MeV = 3.8×10⁻¹² J per D→⁴He event) ≈ 2.6×10¹¹ events per second. Included to fix the scale at which the input measurement must be trusted. Established. ↩
- C. P. Berlinguette et al., “Revisiting the Cold Case of Cold Fusion,” Nature 570 (2019): 45–51. The Google-convened effort did not reproduce excess heat but identified the materials science of extreme hydrogen loading as under-explored. Peer-reviewed. ↩
- A. Züttel, “Materials for Hydrogen Storage,” Materials Today 6, no. 9 (2003): 24–33. Review of metal-hydride and related hydrogen-storage materials. Cited for the established storage application, not for any SSF claim. ↩
- IEEE, IEEE Standard for Interconnection and Interoperability of Distributed Energy Resources with Associated Electric Power Systems Interfaces, IEEE Std 1547-2018 (New York: IEEE, 2018). Cited only to anchor the conditional scale-up discussion; the application to an SSF heat source is speculative. ↩
- U.S. Department of Energy, Report of the Review of Low Energy Nuclear Reactions (2004); Edmund Storms, The Science of Low Energy Nuclear Reaction (Singapore: World Scientific, 2007). The objection is that a deuteron-deuteron fusion rate high enough to produce the reported heat and helium-4 should also produce a commensurate flux of energetic neutrons and gamma rays, which has not reliably appeared; this is independent of the heat measurement (general absence, well supported). Specific positive product reports are single-community and contested. ↩
- Advanced Research Projects Agency–Energy (ARPA-E), “U.S. Department of Energy Announces $10 Million in Funding to Projects Studying Low-Energy Nuclear Reactions,” February 2023. Eight teams, including groups at MIT, Stanford, and Lawrence Berkeley National Laboratory. Institutional. ↩
