Zeolites Screening and Design
Rank frameworks, Si/Al ratios, extra-framework cations and defect scenarios for molecular sieving, gas separation, drying and ion exchange.
Explore service →AI for Materials
AI for Porous Adsorbent Materials Screening and Design
Physics-aware molecular simulation and machine learning to prioritize porous adsorbents for separations, purification, storage, water treatment and controlled adsorption processes.
Start Your ProjectService overview
Screen adsorbents against the conditions that matter
Porous adsorbent selection is a coupled materials and process problem. Pore size, topology, accessible volume, surface chemistry, defects, particle form and operating conditions jointly determine uptake, selectivity, kinetics, working capacity and regeneration demand. We build decision-focused workflows that connect curated adsorption evidence with molecular simulation, physics-informed machine learning and multi-objective optimization.
Structure
Pore geometry, chemistry and accessibility
Performance
Equilibrium, kinetics and regenerability
Decision
Ranked candidates and validation plan
Specialized services
Material-family workflows with a common decision framework
Each pathway accounts for the structural representations, parameterization choices, synthesis variables and validation data appropriate to the adsorbent family.
Activated Carbons Screening and Design
Relate precursor, activation, pore-size distribution and surface functionality to uptake, selectivity, kinetics and regenerability.
Explore service →Porous Organic Polymers Screening and Design
Evaluate network chemistry, topology, functional groups and swelling-sensitive environments for targeted molecular capture.
Explore service →Porous Silica Screening and Design
Optimize pore order, diameter, surface silanols and grafted ligands for adsorption, purification and controlled loading.
Explore service →Hybrid and Composite Adsorbents Screening and Design
Model interfacial compatibility, accessible porosity and transport trade-offs in mixed-matrix, supported and hierarchical adsorbents.
Explore service →Integrated approach
From pore-scale interactions to validation-ready choices
The modeling stack is matched to the requested endpoint and the fidelity of available evidence.
✓
Evidence and structure audit
Normalize adsorption basis, temperature, pressure, activation history and sample form; repair structures and quantify accessible geometry.
✓
Physics-based simulation
Use GCMC for equilibrium loading, MD for diffusion, and targeted quantum calculations for adsorption sites or parameter refinement.
✓
AI-accelerated search
Train calibrated surrogates on descriptors, simulations and measurements; screen candidates under explicit chemical and process constraints.
✓
Scale-aware interpretation
Separate intrinsic predictions from pellet, binder, heat-transfer and bed-level effects, and define the experiments needed to close the gap.
Method selection
Computational methods tied to the decision endpoint
No single method resolves every adsorption question. We combine methods only where the underlying data and assumptions support the requested use.
| Project need | Typical methods | Key inputs | Decision output |
|---|---|---|---|
| Geometric prescreening | Pore-network analysis, accessible surface/volume, void fraction, pore-size distribution | Periodic structures or representative atomistic models; probe definition | Accessible candidates, bottlenecks and geometry-based exclusions |
| Pure-component adsorption | GCMC, force-field sensitivity, adsorption-site analysis | Framework, adsorbate model, charges, temperature and pressure range | Isotherms, Henry coefficients, working capacity and interaction maps |
| Mixture separation | Mixture GCMC, IAST where justified, selectivity and regenerability analysis | Feed composition, pressure swing, impurities, humidity scenario | Selectivity, deliverable capacity, Pareto-ranked candidates |
| Transport and kinetics | MD, free-energy barriers, kinetic models | Flexible/rigid framework choice, loading, temperature, diffusion path | Diffusivity, kinetic selectivity and likely transport limitations |
| Surface chemistry or ion effects | DFT, cluster/periodic calculations, charge and force-field refinement | Adsorption sites, defects, counterions and guest configurations | Binding mechanisms, site preference and parameter evidence |
| Large design-space search | Graph/descriptor ML, Gaussian processes, active learning, Bayesian optimization | Curated data, structures, constraints and validation budget | Predictions with uncertainty, ranked candidates and next evaluations |
| Process-relevant assessment | Surrogate isotherms, mass/energy balances, breakthrough or CFD models as scoped | Pellet and bed properties, cycle conditions, transport and thermal inputs | Process KPIs, sensitivity analysis and operating windows |
Model boundary: simulated crystal-level capacity is not treated as a direct guarantee of shaped-body or bed performance. Force-field choice, defects, water, impurities, activation state and inaccessible porosity are assessed as potential sources of discrepancy.
Project workflow
A transparent route from target profile to validation
Define the use case
Specify adsorbate or solute, feed composition, operating window, performance metrics and hard constraints.
Audit inputs
Review structures, isotherms, sample history, units, data coverage and parameterization risks.
Build the baseline
Establish geometric and physics-based calculations; benchmark against relevant measurements where available.
Screen and optimize
Use validated surrogates and constrained multi-objective search to prioritize candidates and modifications.
Plan validation
Report uncertainty and applicability, then recommend adsorption, kinetics, cycling or breakthrough tests.
Typical inputs
- Target molecule(s), mixture composition and competing species
- Temperature, pressure, concentration, pH or humidity range
- Required capacity, selectivity, kinetics, purity or recovery
- Candidate structures, pore data, synthesis variables and sample form
- Isotherm, calorimetry, breakthrough, cycling or kinetic data, if available
- Stability, regenerability, cost, precursor and manufacturing constraints
- Preferred validation standard and available experimental budget
Typical deliverables
- Curated dataset, provenance map and data-quality assessment
- Structure-ready model set and pore-geometry characterization
- Simulation protocol, parameter assumptions and benchmark results
- Adsorption, selectivity, diffusion or regeneration predictions with uncertainty
- Ranked candidate table and multi-objective trade-off analysis
- Design rules for pore architecture, chemistry or formulation variables
- Applicability-domain and sensitivity assessment
- Recommended experimental or higher-fidelity validation plan
- Technical report, figures and agreed reusable data/model outputs
Representative engagements
Projects framed around practical adsorption decisions
Humid-gas adsorbent down-selection
Goal: balance target-gas working capacity against competitive water uptake and regeneration demand. Workflow: structure audit, pure/mixture simulations, uncertainty-aware surrogate ranking and a humidity-validation matrix.
Activated-carbon design space
Goal: connect activation conditions and pore distributions to removal performance. Workflow: harmonize synthesis and adsorption data, build interpretable descriptors, identify Pareto regions and propose next experiments.
Composite adsorbent formulation
Goal: retain accessible porosity while improving shaping and transport. Workflow: combine component properties, interface descriptors, effective-medium or transport models and constrained formulation optimization.
FAQ
Frequently asked questions
Can you work with a small or incomplete proprietary dataset?
Yes, after a feasibility audit. Small-data workflows may use physically meaningful descriptors, Gaussian processes, transfer learning, simulation augmentation and active learning. If the endpoint is not supportable, we identify the minimum additional measurements needed rather than overstate prediction confidence.
How are water, impurities and framework flexibility handled?
They are scoped explicitly because they can change adsorption and transport. Depending on the system, we may run competitive adsorption scenarios, compare interaction models, evaluate selected flexible structures, or treat the effects through sensitivity bounds when reliable parameters are unavailable.
Can molecular simulation and machine learning be combined?
Yes. Molecular simulations can supply labels or physics-informed features, while calibrated ML surrogates accelerate screening. Hold-out-by-family validation, domain-distance checks and targeted high-fidelity calculations help distinguish interpolation from risky extrapolation.
Do you predict process performance from crystal structures alone?
Not without additional assumptions and data. Crystal-level adsorption is only one layer. Shaping, binders, particle size, heat and mass transfer, cycle design and bed hydrodynamics require separate inputs and, where requested, a process-scale work package.
Are experiments included?
The core service provides computational screening and validation recommendations. Customer-generated experimental data can be integrated iteratively; experimental testing or coordinated validation can be added as a separately scoped package.
How is confidential information handled?
Project data can be handled under an agreed confidentiality framework. Data access, reusable outputs, model transfer and retention expectations are defined during scoping.
Online Inquiry
Submit your project details below, and our team will respond within 24 hours.
x
Need help getting the data you need?
Talk to our technical team about your project!
I Want To Talk