Ionworks

Show what your material does in a real cell

Electrolyte, anode, and cathode developers are asked the same question by every cell maker: what does this do in our cell? Ionworks Studio turns your material data into physics-based cell models, so you can answer it with a simulation before anyone builds a prototype.

Book a demoSee how we work with materials teams
A material property set in Ionworks Studio: ionic conductivity, diffusion coefficient, transference number, and thermodynamic factor vs salt concentration for an electrolyte candidateDischarge capacity vs cathode areal loading for a 21700 DFN model, with capacity peaks at 2C and 3C

Trusted by leading materials and battery teams

Sila logoAnthro Energy logoNOVONIX logoAvrion Battery Labs logoFAAM logo

A coin cell is not the cell your partner builds

Coin cells are designed to screen materials, and they do that well. But the conditions that make them good screens also hide the constraints a commercial cell runs into: limited lithium inventory, lean electrolyte, thick electrodes, and long transport paths.

A material that looks strong at the electrode level can deliver less at the cell level than the incumbent it was meant to replace. Cell makers know this, which is why half-cell capacity alone rarely moves a qualification forward.

Read more on material properties vs cell metrics

Counter electrode
Coin: Excess lithium metal
Commercial: Graphite or silicon blend, N/P near 1.1
Electrolyte
Coin: Flooded
Commercial: Lean, set by the cell design
Areal loading
Coin: Often below 1 mAh/cm²
Commercial: 3 to 5 mAh/cm²
Format
Coin: CR2032, a few cm²
Commercial: 21700 jellyroll or large pouch
Schematic. Layer heights roughly to scale.

Where a thicker electrode stops paying off

Discharge capacity vs cathode areal loading for a 21700 DFN model at fixed can volume. At 0.5C and 1C capacity keeps rising with loading. At 2C it peaks near 4.9 mAh/cm² and at 3C near 3.9 mAh/cm², then falls as electrolyte transport limits thick electrodes.
Constant-current loading sweep, 21700 DFN at fixed can volume

Take the loading row from the table. In a 21700 at fixed can volume, the sweep re-derives electrode width from the winding geometry at each loading, so a thicker coating trades separator and foil for active material. At 0.5C and 1C that trade keeps paying: capacity rises across the whole range.

At higher rates it stops. The 2C capacity peaks near 4.9 mAh/cm², and the 3C capacity near 3.9 mAh/cm². Past that, lithium ions can’t cross the thick electrode fast enough, the electrolyte depletes deep in the coating, and capacity falls away.

For an electrolyte developer, that peak is the number to move. Rerun the sweep with your own conductivity and transference number and see how far it shifts, before a single pouch cell is built.

How we work with materials teams

  1. Sealed amber vials of candidate electrolytes and a micropipette on a lab bench
    Material property set in Ionworks Studio: ionic conductivity and diffusion coefficient vs salt concentration at 0, 25, and 45 °C, transference number, and thermodynamic factor

    01Start from your material data

    Measured properties or the output of your own property models: ionic conductivity, diffusivity, transference number, open-circuit potential, particle size. Coin-cell and half-cell tests fill in what property data alone cannot.

  2. Model fit in Ionworks: simulated voltage over measured data for a 1C rate-capability discharge
    Measured (black) vs fitted DFN (blue), 1C discharge

    02Parameterize a physics-based model

    Fit a DFN or SPMe model against your test data, with every parameter traced back to the measurement it came from. When a new batch or formulation is tested, the model is refit against it.

  3. Cell design optimization in Ionworks: discharge capacity over optimizer iterations for four multistart runs, converging on one design
    Discharge capacity over optimizer iterations, four multistart runs

    03Put the material in a real cell

    Place the parameterized material in a target design, from a 21700 jellyroll to a large-format pouch, at realistic loading and N/P ratio. Run the protocols a cell maker cares about: fast charge, rate capability, drive cycles.

  4. A stack of pouch cells next to a tray of 21700 cylindrical cells on a pilot-line table
    The exported cell model as a JSON parameter file, loaded into open-source PyBaMM in a Jupyter notebook and run as a 1C discharge, with the voltage curve below

    04Hand partners a model they can run

    Hand a partner’s engineering team the parameterized cell model in an open-source format that runs in PyBaMM. They can run it in their own tools and test your material in their own design space before they build anything.

What teams use the model to answer

  • How much does a higher transference number shorten a 10 to 80% fast charge in a 21700?
  • What blend ratio of silicon to graphite maximizes energy while staying inside the swelling budget?
  • Does a fast-charge protocol developed for one chemistry stay plating-free on ours?
  • What does our material add in Wh/kg at the cell level, not the electrode level?
Anthro Energy logo

At Anthro Energy, we develop novel polymer electrolyte materials and the models that predict their properties, from ion transport to mechanical and electrochemical behavior. Translating those material-level insights into confident predictions of how a full cell will perform is a different challenge, and that’s where Ionworks has been an exceptional partner.

Their team helped us build modeling and simulation workflows that take our material property predictions and show how they play out at the cell level. That has meaningfully sped up how quickly we can evaluate and integrate our new electrolytes into existing lithium-ion cell designs.

Ionworks brings deep expertise, responsiveness, and a genuinely collaborative approach, and they’ve become an important part of how we move from materials innovation to real-world cell performance.

Dr. Michael McEldrew
Electrolyte Scientist, Anthro Energy
Sila logo

We are working with Sila Nanotechnologies on the same workflow for silicon-carbon anodes: screening blend ratios, projecting swelling, and transferring fast-charge protocols across chemistries.

Frequently asked questions

Yes. Your material data parameterizes a physics-based DFN or SPMe model, which then runs in a target cell design, such as a 21700 jellyroll or a large-format pouch, at realistic loading and N/P ratio. The loading sweep above is one example: it shows where a thicker electrode stops paying off at each C-rate.
Measured properties or the output of your own property models: ionic conductivity, diffusivity, transference number, open-circuit potential, particle size. Coin-cell and half-cell tests fill in what property data alone cannot.
Yes. The parameterized cell model is exported in an open-source format that runs in PyBaMM, so a partner’s engineering team can run it in their own tools and test your material in their own designs.
Yes. The workflow is the same for any electrode or electrolyte material. Anthro Energy uses it for polymer electrolytes, and Sila for silicon-carbon anodes: screening blend ratios, projecting swelling, and transferring fast-charge protocols.
Yes. Ionworks is SOC 2 compliant and follows industry standard practices for data protection, access control, and deployment isolation. Run it in our managed cloud, in your own cloud or network, or fully air-gapped. Reports are available on request.

See your material in a 21700 before anyone builds one

Bring property data or a set of coin-cell tests. We will walk through building a cell-level model from them and running it in the format your partners use.

Book a demo