The supercapacitor gap: why lab-tested carbon rarely delivers its promised energy in real devices

Research led by Professor Qiulong Wei, State Key Laboratory of Physical Chemistry of Solid Surface, Xiamen University, with collaborators at Imperial College London.

The persistent translation problem in energy storage

Supercapacitor research has a credibility problem. Thousands of papers report impressive capacitance numbers for new carbon materials, yet those numbers almost never translate directly into device-level energy density. The reason is straightforward but persistently ignored: measuring how much charge a material can store per gram in a laboratory half-cell tells you very little about how much energy a finished device - complete with electrolyte, separator, current collectors, and packaging - will actually deliver.

The gap between what a material promises and what a device delivers has frustrated engineers for years. Now a team led by Professor Qiulong Wei at Xiamen University's State Key Laboratory of Physical Chemistry of Solid Surface, working with collaborators at Imperial College London, has done something deceptively simple but long overdue: they built a systematic framework that connects the dots between activated carbon properties and assembled supercapacitor performance.

A conversion factor the field has been missing

The core finding is a number: approximately 0.35. That is the conversion factor between material-level performance metrics and device-level energy density for supercapacitors. Roughly two-thirds of the theoretical energy density disappears once you account for everything else inside a real device. This is not a surprise to anyone who has built supercapacitors, but having a validated, quantitative benchmark changes how the field can evaluate new materials.

The team arrived at this number by assembling practical pouch cells - not coin cells or three-electrode setups, but the kind of packaged devices that resemble commercial products. They systematically varied the amount of electrolyte and studied how it interacted with activated carbon materials of different porosities and capacitances.

What emerged was a clear relationship. The optimal electrolyte volume is not simply "as much as possible" or "enough to wet the electrodes." It corresponds precisely to the volume needed to fill all available pores: those within the activated carbon particles, the gaps between stacked particles, and the pores of the separator. Add more electrolyte than that, and you gain nothing in performance while adding dead weight. Use less, and ionic transport suffers.

The descriptor that links porosity to energy

To formalize this relationship, the researchers introduced a new parameter they call eta, which integrates both the specific capacitance and the porosity of an activated carbon electrode. The elegant aspect of eta is that it displays a linear relationship with device energy density. Plot one against the other for a range of materials, and the data points fall on a line.

This linearity is what makes the approach practical. If a researcher synthesizes a new activated carbon and measures its capacitance and porosity - both routine laboratory measurements - they can use eta to estimate what device-level energy density that material will achieve before ever assembling a full pouch cell. That saves weeks of fabrication and testing.

The team went further by building a computational tool they call the E-tool, which automates this prediction. Feed it material-level electrochemical data, and it outputs a projected device energy density. Validated against 43 different activated carbon materials drawn from published literature, the E-tool's predictions landed within 1% of experimentally measured device performance.

Electrolyte concentration and the 2Q sweet spot

Among the practical insights buried in the data, one stands out for device engineers. The researchers found that an electrolyte concentration of 1.0 molar - corresponding to what they term a "2Q" charge quantity, or twice the charge needed to balance the electrode capacity - delivered the best combination of rate capability and energy density. Higher concentrations introduced unnecessary ion redundancy without improving performance. Lower concentrations limited power output.

This finding matters because electrolyte optimization in supercapacitors has historically been somewhat ad hoc. Having a quantitative guideline that connects electrode properties to optimal electrolyte formulation is a practical contribution that could accelerate device development.

A pouch cell that beats commercial devices

As a demonstration, the team assembled a pouch cell using their optimized approach. It achieved 7.80 watt-hours per kilogram at the device level - significantly higher than the 4.73 Wh/kg delivered by commercial 100-farad cylindrical supercapacitors. The improvement came primarily from two sources: better electrolyte management (neither too much nor too little) and lower packaging mass relative to active material.

That 65% improvement over a commercial benchmark is notable, though context matters. Commercial cylindrical supercapacitors prioritize durability, safety, and manufacturability over raw energy density. A laboratory pouch cell optimized for a single metric does not face those constraints. Still, the result validates the framework's utility.

What the framework does not address

Several important caveats apply. The study focuses exclusively on activated carbon in organic electrolyte systems. Whether the eta descriptor and E-tool predictions hold for other electrode materials - metal oxides, conducting polymers, MXenes - or for aqueous or ionic liquid electrolytes remains untested. The linear relationship between eta and device energy density may not generalize.

Long-term cycling stability, self-discharge rates, and calendar aging - critical metrics for commercial supercapacitors - are not addressed. A material that looks optimal by the eta metric could still fail in practice if its pore structure degrades over thousands of cycles or if it promotes electrolyte decomposition.

The 43-material validation set, while reasonable, draws entirely from published literature. Publication bias toward high-performing materials could inflate the apparent accuracy of the E-tool. Testing against deliberately mediocre or problematic carbons would strengthen confidence in its generality.

The team suggests their theoretical model could serve as a foundation for AI-driven materials discovery. That is plausible in principle, but the training data for such an approach would need to include device-level measurements, not just material properties - and device data is far scarcer than material data in the literature.