mat001-sample / README.md
pradeep-xpert's picture
Upload folder using huggingface_hub
57512f0 verified
metadata
license: cc-by-nc-4.0
task_categories:
  - tabular-regression
  - tabular-classification
  - time-series-forecasting
tags:
  - materials-science
  - battery
  - lithium-ion
  - energy-storage
  - electrochemistry
  - synthetic-data
  - degradation-modeling
  - cycle-life
  - solid-state-battery
  - manufacturing
pretty_name: MAT-001  Synthetic Battery Materials Dataset (Sample)
size_categories:
  - 10K<n<100K

MAT-001 — Synthetic Battery Materials Dataset (Sample)

XpertSystems.ai Synthetic Scientific Discovery Platform · SKU: MAT001-SAMPLE · Version 1.0.0

This is a free preview of the full MAT-001 — Synthetic Battery Materials Dataset product. It contains roughly ~1% of the full dataset at identical schema, chemistry family distribution, and physics-aware calibration, so you can evaluate fit before licensing the full product.

File Rows (sample) Rows (full) Description
battery_materials_master.csv ~2,500 ~250,000 Cathode/anode/electrolyte compositions
charge_discharge_cycles.csv ~60,000 ~25,000,000 Per-cycle telemetry (capacity, voltage, IR)
degradation_profiles.csv ~20,000 ~3,750,000 Capacity-fade trajectories
thermal_stability_events.csv ~500 ~50,000 Thermal runaway / failure events
manufacturing_variations.csv ~5,000 ~250,000 Production lot variations
battery_pack_summary.csv ~500 ~25,000 Per-pack aggregate KPIs

Dataset Summary

MAT-001 simulates the full battery materials discovery lifecycle with physics-aware synthetic generation across 10 commercial and emerging battery chemistries, with:

  • 10 battery family archetypes: nmc811, nmc622, nca, lfp, solid_state, sodium_ion, lithium_sulfur, silicon_anode, graphene_enhanced, defense_high_temp — each with empirically-anchored cathode/anode/ electrolyte/separator/crystal structure profiles
  • Arrhenius degradation modeling with activation energy 0.42 eV and reference temperature 25 °C
  • Per-chemistry capacity fade multipliers: solid_state 0.72×, silicon_anode 1.35×, high-temperature 1.55×, fast-charge 1.28×
  • Dendrite formation modeling with chemistry-dependent base rates
  • 12 dopant types (Al, Mg, Zr, Ti, W, Mo, Nb, B, F, dual-dopants)
  • 6 binder systems: PVDF, CMC-SBR, PAA, alginate, PTFE, high-temp polyimide
  • Manufacturing yield variability with calibrated 93% baseline
  • Thermal runaway events with chemistry-specific onset temperatures (LFP 265 °C → defense_high_temp 310 °C → lithium_sulfur 175 °C)
  • Anomaly injection at 1.5% rate for outlier modeling

Calibrated Benchmark Targets

The full product is calibrated to 10 benchmark validation tests drawn from authoritative materials science and battery research sources (DOE Battery Targets, NMC/LFP production benchmarks, solid-state battery roadmap literature, EV cell cycle-life literature, UL9540A thermal safety literature, electrochemical cycling literature, battery degradation studies, fast-charge degradation literature, manufacturing quality benchmarks).

Sample benchmark results:

Test Target Observed Source Verdict
energy_density_nmc811_wh_kg 275.0 277.7 DOE battery targets; high-nickel NMC ✓ PASS
energy_density_lfp_wh_kg 165.0 167.1 DOE battery targets; LFP production ✓ PASS
energy_density_solid_state_wh_kg 365.0 370.9 solid-state battery roadmap literatu ✓ PASS
cycle_life_80pct_retention 1000.0 1555.6 EV cell cycle-life literature ✓ PASS
coulombic_efficiency 0.9950 0.9955 electrochemical cycling literature ✓ PASS
impedance_growth_1000_cycles 0.2800 0.2733 battery degradation studies ✓ PASS
fast_charge_capacity_loss_500_cycles 0.1200 0.1071 fast-charge degradation literature ✓ PASS
thermal_runaway_temperature_c 185.0 213.3 UL9540A and thermal safety literatur ✓ PASS
manufacturing_yield_rate 0.9300 0.8994 battery manufacturing quality benchm ✓ PASS
diffusion_coefficient_log10 -11.0000 -10.9654 materials diffusion coefficient lite ✓ PASS

Every benchmark in the sample lands within the same calibrated tolerance as the full product. Physics-aware generation means observed values are deterministic functions of the calibrated parameters, not stochastic artifacts that require large samples to converge.

Schema Highlights

battery_materials_master.csv (primary file)

Column Type Description
material_id string Unique material identifier
chemistry_family string 1 of 10 family archetypes
cathode_composition string Cathode chemistry (e.g. LiNi0.8Mn0.1Co0.1O2)
anode_material string Anode (graphite, silicon_graphite, lithium_metal, etc.)
electrolyte string Electrolyte system
separator string Separator material
crystal_structure string Crystal structure family
dopant string Dopant species (or 'none')
binder string Binder system
nominal_voltage_v float Nominal cell voltage
energy_density_wh_kg float Gravimetric energy density
power_density_w_kg float Power density
thermal_runaway_temp_c float Thermal runaway onset temperature
expected_cycle_life float Expected cycles to 80% capacity
diff_log10 float Log10 lithium diffusion coefficient
conductivity float Ionic conductivity (S/cm)
manufacturability_score float Manufacturability index (0–1)

charge_discharge_cycles.csv (per-cycle telemetry)

Column Type Description
material_id string FK to battery_materials_master.csv
cycle_number int Cycle index (1-N)
capacity_retention float Capacity as fraction of initial
coulombic_efficiency float CE for this cycle
voltage_hysteresis_mv float Charge/discharge voltage gap
internal_resistance_mohm float Cell internal resistance
temperature_c float Operating temperature
c_rate float Charge/discharge rate

degradation_profiles.csv (capacity fade trajectories)

Column Type Description
material_id string FK
timestep int Degradation timepoint
capacity_fade_pct float Cumulative capacity loss
impedance_growth_pct float Impedance growth
dendrite_formation_score float Dendrite formation risk (0–1)
sei_layer_growth_nm float SEI layer thickness growth

thermal_stability_events.csv (runaway events)

Column Type Description
material_id string FK
onset_temp_c float Runaway onset temperature
peak_temp_c float Peak temperature during event
heat_release_rate_w float Heat release rate
failure_mode string Failure mode classification

See manufacturing_variations.csv and battery_pack_summary.csv for the production-lot variation and per-pack aggregate schemas respectively.

Suggested Use Cases

  • Training battery chemistry classifiers — 10-class family identification
  • Cycle life prediction — regress to 80% retention from early-cycle data
  • Energy density forecasting from composition features
  • Thermal runaway detection — predict onset from operating signals
  • Dopant effect modeling — quantify impact of doping on cycle life
  • Solid-state battery design optimization — explore composition space
  • Manufacturing yield prediction from process variation features
  • Capacity fade trajectory modeling — sequence prediction tasks
  • Battery management system (BMS) algorithm training
  • Materials informatics / inverse design — find compositions hitting target energy density + cycle life
  • Battery digital twin training data
  • EV pack-level state-of-health (SoH) estimation

Loading the Data

import pandas as pd

materials = pd.read_csv("battery_materials_master.csv")
cycles    = pd.read_csv("charge_discharge_cycles.csv")
degrad    = pd.read_csv("degradation_profiles.csv")
thermal   = pd.read_csv("thermal_stability_events.csv")
mfg       = pd.read_csv("manufacturing_variations.csv")
packs     = pd.read_csv("battery_pack_summary.csv")

# Join cycle telemetry with material chemistry
enriched = cycles.merge(materials, on="material_id", how="left")

# Regression target: energy density from composition features
y_energy = materials["energy_density_wh_kg"]

# Classification target: chemistry family from observable telemetry
y_family = materials["chemistry_family"]

# Sequence prediction target: capacity retention curve per material
seq_data = cycles.groupby("material_id")["capacity_retention"].apply(list)

License

This sample is released under CC-BY-NC-4.0 (free for non-commercial research and evaluation). The full production dataset is licensed commercially — contact XpertSystems.ai for licensing terms.

Full Product

The full MAT-001 dataset includes ~29 million rows across all six files, with calibrated benchmark validation against 10 metrics drawn from authoritative materials science and battery research sources.

📧 pradeep@xpertsystems.ai 🌐 https://xpertsystems.ai

Citation

@dataset{xpertsystems_mat001_sample_2026,
  title  = {MAT-001: Synthetic Battery Materials Dataset (Sample)},
  author = {XpertSystems.ai},
  year   = {2026},
  url    = {https://huggingface.co/datasets/xpertsystems/mat001-sample}
}

Generation Details

  • Generator version : 1.0.0
  • Random seed : 42
  • Generated : 2026-05-16 15:47:32 UTC
  • Physics model : Arrhenius degradation, 0.42 eV activation energy
  • Overall benchmark : 100.0 / 100 (grade A+)