CASE STUDY 02 / BATTERY MATERIALS

Better battery cathodes through interface-aware and build-ready screening.

FluxMateria ran a complete local battery workflow in 26.8 seconds and showed that bulk ranking, interface readiness, battery-native scoring, and immediate build handoff do not lead to the same engineering answer.

26.8s local runtime 10-material curated pool Four distinct decision winners Prototype handoff included White paper included
Download full white paper (PDF) Download public data packet
26.8s
End-to-end runtime
10
Curated cathodes screened
4
Different decision winners
1.0
Holdout family accuracy
0.9429
Energy-rank Spearman

Why this study matters

This is not a one-metric leaderboard. FluxMateria used one workflow to separate the bulk answer, the interface answer, the battery-native answer, and the build-now answer.

Bulk-only was incomplete

LiNiO2 would have won too early

If the workflow had stopped at bulk properties, the conclusion would have collapsed into a conventional nickel-rich story. The rest of the stack showed why that would have been too simple.

Interface evidence mattered

LiMnPO4 came back into the picture

Interface-aware ranking reopened a manganese/phosphate direction that bulk ranking had buried. That is the clearest sign that the workflow changed the engineering answer rather than just decorating it.

Build-ready is different

Li4Ti5O12 is the first build, not the highest upside

The final handoff layer did its job: it separated the strongest immediate build package from the chemistry with the highest modeled upside.

Calibrated credibility

The battery-native layer was benchmarked before it was used as a public case-study decision engine. The holdout set preserved both family separation and ranking order.

1.0
Family accuracy
The calibrated stack kept the known cathode families separated correctly on the holdout set.
0.812
Capacity MAE (mAh/g)
Capacity estimation stayed close enough to support triage and ranking instead of only descriptive analytics.
0.149
Voltage MAE (V)
Voltage accuracy is strong enough for shortlist interpretation and build-package planning.
0.1427
Transport MAE
Transport and rate-performance proxies are now calibrated instead of being left as unbounded heuristics.
0.0642
Cycle MAE
Cycle-life heuristics are now good enough to influence ranking rather than merely annotate it.
0.9429
Energy-rank Spearman
The calibrated engine preserved the correct top-energy ordering across the reference set.

How FluxMateria got here in 26.8 seconds

This section shows the workflow structure at a glance, from the first candidate framing step to the final build handoff.

01

Curated Candidate Framing

Start from an engineering-relevant lithium-cathode pool so the workflow stays interpretable and decision-grade.

02

Bulk Engineering Screen

Establish the bulk-only answer first, before interface and battery-native correction.

Bulk answer: LiNiO2
03

Interface And Contact Pass

Re-rank for interface readiness instead of assuming bulk strength is enough.

Interface answer: LiMnPO4
04

Battery-Native Electrochemistry

Add transport, degradation, cycle-life, electrolyte/coating, and manufacturing logic.

Battery answer: LiMnO2
05

Calibration And Uncertainty

Carry benchmark support, confidence, and recommended next experiments forward with the shortlist.

06

Prototype Handoff

Finish at a build decision instead of a flat ranked list.

Build answer: Li4Ti5O12
Bulk Winner LiNiO2
Interface Winner LiMnPO4
Battery Winner LiMnO2
Build Winner Li4Ti5O12
Step 01

Curated candidate framing

Start from an engineering-relevant lithium-cathode pool instead of a random chemistry universe, so the result stays interpretable and actionable.

Step 02

Bulk engineering screen

Run the first ranking on bulk properties. In this study that layer alone would have pointed toward LiNiO2.

Step 03

Interface and contact pass

Re-rank for interface readiness instead of assuming bulk strength is enough. That is where LiMnPO4 re-entered the story.

Step 04

Battery-native electrochemistry

Add voltage, transport, degradation, cycle-life, electrolyte/coating fit, and manufacturing logic. That is where LiMnO2 rose to the top.

Step 05

Calibration and uncertainty

Carry benchmark context, confidence, and follow-up experiments forward so the output is more than a flat score table.

Step 06

Prototype handoff

End at a build decision rather than a generic shortlist. That is why Li4Ti5O12 became the best immediate build package.

Why 26.8 Seconds Matters

FluxMateria did not just score one material in 26.8 seconds. It completed a multi-stage decision workflow locally. As timing context, configuration-heavy DFT surface relaxations can take about 24 hours each and many-configuration searches can stretch into days or weeks, while battery lifetime testing remains time-consuming and can still run for years under accelerated protocols. FluxMateria compressed the computational decision layer, not the lab step.

Read the public pipeline note →

The four-winner result

The same workflow produced four different leaders because it was answering four different battery questions. That is the core signal of the study.

Decision Layer 01
LiNiO2
Bulk winner
274.51
Specific capacity (mAh/g)
3.85
Voltage surrogate (V)

If the study had stopped at bulk screening, this would have looked like the obvious answer. The rest of the workflow proved that would have been incomplete.

Decision Layer 02
LiMnPO4
Interface winner
170.87
Specific capacity (mAh/g)
4.10
Voltage surrogate (V)

This was the clearest correction in the whole campaign. Interface analysis moved LiMnPO4 from bulk rank 9 to interface rank 1.

Decision Layer 03
LiMn2O4
Hybrid winner
148.22
Specific capacity (mAh/g)
4.117
Voltage surrogate (V)

Once bulk and interface are treated together, the best compromise is not the same as either pure winner. That is why the hybrid lane matters.

Decision Layer 04
LiMnO2
Battery-native winner
77.7
Battery-readiness score
79.5
Prototype handoff priority

Transport, degradation, cycle life, manufacturability, and uncertainty together elevated LiMnO2 above the simpler bulk-only answer.

Prototype Handoff

Li4Ti5O12 is the best immediate build candidate.

This is the final distinction that makes the workflow useful in practice. Li4Ti5O12 is not the highest-energy candidate, but it is the cleanest build-now package because its degradation, cycle-life, and support profile are unusually favorable.

Build stage Immediate prototype candidate
Anode family Graphite
Electrolyte family Standard carbonate
Surface coating TiO2

Public shortlist

This shortlist focuses on the disclosed materials, their headline metrics, dominant risks, and the most useful next validation steps.

LiMnO2

Top battery-native candidate
285.49 mAh/g 3.95 V Li3PO4 coating

Highest-upside battery-native candidate in the current public shortlist.

  • Dominant risks: chemo-mechanical cracking, transition-metal dissolution, voltage-window stress
  • Next work: crack screening, dissolution screen, electrolyte/coating A/B screen

Li4Ti5O12

Top immediate build candidate
233.52 mAh/g 1.55 V TiO2 coating

The strongest first-build package even though it is not the highest-energy option.

  • Dominant risks: chemo-mechanical cracking, impedance growth, manageable interphase risk
  • Next work: half-cell voltage profile and pulse-power screening

LiMnPO4

Top interface-corrected candidate
170.87 mAh/g 4.10 V Li3PO4 coating

The clearest evidence that interface-aware analysis changed the engineering conclusion.

  • Dominant risks: cracking, dissolution, impedance growth
  • Next work: rate screen, dissolution screen, post-cycle structural checks

LiMn2O4

Top hybrid candidate
148.22 mAh/g 4.117 V AlPO4 coating

The best compromise once bulk and interface are treated together rather than separately.

  • Dominant risks: dissolution, cracking, voltage-window stress
  • Next work: dissolution screen, coating screen, crack mapping

LiNiO2

Top bulk-only candidate
274.51 mAh/g 3.85 V LiNbO3 coating

Still compelling on pure energy framing, but no longer the obvious all-around answer.

  • Dominant risks: cracking, upper-cutoff stress, lithium inventory loss
  • Next work: crack screen, coating screen, dissolution screen

LiCoO2

High-energy comparator
273.84 mAh/g 3.95 V LiNbO3 coating

A useful high-energy reference point for the public shortlist and the benchmarked ranking stack.

  • Dominant risks: cracking, voltage-window stress, lithium inventory loss
  • Next work: crack screen, coating screen, dissolution screen

Literature context

This is one of the strongest signals that the pipeline is solid. The disclosed shortlist converged on battery families that industry and academia have already taken seriously, not on chemically implausible outsiders.

Established commercial families

LiCoO2 and LiMn2O4 validate the baseline

LiCoO2 is the classic commercial cathode lineage, and LiMn2O4 is a long-established spinel family with known power advantages. That matters because FluxMateria landed on materials the field already knows are real, valuable, and industrially relevant.

Known upside, known tradeoffs

LiNiO2 and LiMnPO4 reflect real field constraints

LiNiO2 has long been attractive as a cobalt-light high-energy direction, but the literature repeatedly flags structural instability, thermal issues, and surface reactivity. LiMnPO4 is attractive because of its higher voltage, but the field has struggled with conductivity and diffusion limits. FluxMateria surfaced both of those tradeoffs in a coherent way.

Most interesting convergence

LiMnO2 has old roots and renewed momentum

LiMnO2 is not a new material, but it is a very interesting one. Older work documented structural and phase-stability problems, while newer literature has reopened it as a serious Ni/Co-free high-energy direction. That makes the FluxMateria result stronger, not weaker: it converged on a chemistry the field is actively revisiting.

Important caveat

Li4Ti5O12 should be read carefully

Li4Ti5O12 is best known in the literature as a zero-strain anode, not as a mainstream cathode. That does not make the result useless. It means the build-handoff lane is strongly rewarding stability, robustness, and immediate prototypeability. Publicly, that should be read as evidence that the handoff layer is finding the safest first-build package, not that the field somehow missed an obvious new cathode.

Why This Strengthens The Study

The shortlist mostly overlaps with material families that industry and academia have already explored deeply. That is a strong validation signal. FluxMateria converged on real electrochemical directions, then separated which ones look strongest in bulk, at the interface, in battery-native scoring, and in immediate build handoff.

Selected literature

Why this is novel, valuable, and faster

The novelty is not a claim that FluxMateria discovered a brand-new chemistry class. The novelty is that one coherent workflow changed the engineering answer several times before it reached the build decision.

Novel

One workflow, several decision layers

The same run produced a bulk answer, an interface answer, a battery-native answer, and a build-now answer. That is much closer to how real battery programs make decisions.

Valuable

Highest-upside is not the same as safest first build

FluxMateria made that distinction explicit. LiMnO2 is the highest-upside battery-native candidate, while Li4Ti5O12 is the cleanest immediate prototype package.

Faster

Under 30 seconds on local hardware

The study finished in 26.8 seconds on the local machine. That does not replace battery testing, but it changes what the experimental team does next.

What This Study Achieved

The remaining step is a real-world build and validation program. That is exactly the point of the study: not to replace physical battery testing, but to make that testing sharper, cheaper, and more defensible by separating the highest-upside chemistry from the best immediate build candidate.

What happens next

The computational work already did the narrowing. The next moves are practical validation steps, not more blind screening.

Immediate build lane

Build Li4Ti5O12 first

Use the build package as the first controlled prototype because it has the strongest immediate handoff profile in the current study.

Higher-upside lane

Stress-test LiMnO2 next

Run the highest-information validation package on LiMnO2 to see whether the modeled upside survives cell-level constraints.

Closed-loop lane

Feed the data back into FluxMateria

Use early cycling, impedance, and dissolution data to update the ranking engine rather than treating the first run as static.

Downloads

These downloads include the public white paper, the summary data packet, and the workflow note for the study. They focus on the key materials, headline metrics, and next validation steps.

Download full white paper (PDF) Download public data packet (JSON) Download public pipeline note
Research and engineering use only. This page and the downloadable public materials are not a manufacturing certification, safety sign-off, commercial qualification, or production release decision.

Need a battery workflow that ends at a build decision?

FluxMateria can combine ranking, interface analysis, battery-native screening, and prototype handoff inside one computational campaign.

Request pilot access →