Theoretical Background¶
The scientific foundation of the ASCI framework.
The Catalyst Selection Problem¶
Computational catalysis databases contain thousands of DFT-calculated catalyst properties. The challenge is translating this data into actionable experimental priorities.
Traditional approaches:
- Single-objective ranking - Sort by one metric (e.g., activity)
- Manual multi-criteria - Ad-hoc weighting, poorly documented
- Pareto analysis - Non-dominated set, no ranking
ASCICat provides:
- Unified, interpretable metric
- Transparent, reproducible rankings
- Built-in sensitivity analysis
Three-Pillar Framework¶
Activity (Thermodynamic)¶
Based on the Sabatier principle and volcano relationships:
\[S_a(\Delta E) = \max\left(0, 1 - \frac{|\Delta E - \Delta E_{opt}|}{\sigma_a}\right)\]
Physical basis:
- DFT binding energies correlate with activation barriers
- Optimal binding balances adsorption and desorption
- Volcano plots validated experimentally
Stability (Thermodynamic)¶
Based on surface thermodynamics:
\[S_s(\gamma) = \frac{\gamma_{max} - \gamma}{\gamma_{max} - \gamma_{min}}\]
Physical basis:
- Lower surface energy → stronger surface bonding
- Correlates with dissolution resistance
- Validated for metal dissolution in electrochemistry
Cost (Economic)¶
Based on material economics:
\[S_c(C) = \frac{\log C_{max} - \log C}{\log C_{max} - \log C_{min}}\]
Physical basis:
- Raw material cost as proxy for catalyst cost
- Logarithmic scaling handles 5+ orders of magnitude
- Economic viability critical for deployment
Integration: The ASCI Score¶
Weighted linear combination:
\[\phi_{ASCI} = w_a \cdot S_a + w_s \cdot S_s + w_c \cdot S_c\]
Properties:
- Bounded: \(\phi \in [0, 1]\)
- Interpretable: Higher = better
- Customizable: Weights reflect priorities
- Constraint: \(w_a + w_s + w_c = 1\)
Assumptions¶
- DFT accuracy - Binding energies are meaningful
- Descriptor validity - Chosen descriptors capture key properties
- Linear aggregation - Trade-offs are linear
- Independence - Scores are approximately independent
Limitations¶
- Kinetic barriers not directly captured
- Support effects not included
- Electrolyte effects not modeled
- Deactivation mechanisms not predicted
Validation¶
ASCICat rankings are validated by:
- Pareto consistency - Top ASCI catalysts are mostly Pareto-optimal
- Known benchmarks - Pt-group metals rank highly for HER
- Sensitivity robustness - Rankings stable across weight ranges