Find task-fit solvents by jointly optimizing molecular performance, EHS profile, phase behavior, recovery demand, and plant feasibility.
Start a ProjectCD ComputaBio builds condition-aware solvent shortlists for new reactions, separations, crystallizations, and formulations. Unlike one-for-one solvent replacement, this service explores a broad candidate space and identifies Pareto-efficient options against a defined product, process, safety, environmental, and recovery objective.
Inputs may include solute, reactant, product, impurity, and solvent structures; measured solubility; SLE, LLE, and VLE data; temperature and composition ranges; reaction conversion/selectivity; extraction distribution; physical-property and SDS records; residual-solvent limits; recovery data; materials-of-construction constraints; and plant utilities. Units, measurement basis, composition, temperature, pressure, and source quality are normalized before comparison.
Assemble conventional, bio-derived, water-miscible, fluorine-free, low-volatility, and process-compatible candidates; define hard exclusions for hazard classes, regulation, materials compatibility, boiling range, supply, and allowable residues.
Combine Hansen descriptors, QSPR/ML, COSMO-based calculations where appropriate, activity-coefficient models, and experimental anchors to estimate solubility, miscibility, partitioning, SLE, LLE, VLE, and water sensitivity.
Rank media against conversion, selectivity, catalyst or enzyme compatibility, ion pairing, viscosity, dielectric response, donor/acceptor behavior, polymer swelling, coating quality, and formulation stability.
Integrate flammability, volatility, acute/chronic toxicity, CMR concern, aquatic hazard, persistence, renewability, waste route, and data confidence without allowing a single composite score to hide a critical hazard.
Evaluate boiling range, azeotropy, heat capacity, latent heat, phase split, solvent losses, distillation or membrane recovery, impurity buildup, corrosion, utility limits, and recycle drift.
Generate transparent Pareto fronts, uncertainty-aware rankings, sensitivity analyses, and a compact experiment plan that resolves the highest-value unknowns before scale-up.
Every score remains tied to the solvent identity, composition, operating condition, source, model version, and decision rule that produced it.
| Stage | Key Activities | Decision Output |
|---|---|---|
| 1. Task Definition | Specify solute or reaction, unit operation, product quality, temperature, pressure, composition, throughput, solvent-loss target, equipment, EHS exclusions, and success thresholds. | Use-case envelope and weighted objectives |
| 2. Data & Evidence Audit | Normalize structures, isomers, purity, units, temperature, pressure, composition basis, property provenance, SDS evidence, and measured performance; flag missing or conflicting values. | Curated evidence base and uncertainty register |
| 3. Candidate Generation | Search solvent databases and chemically feasible families; apply hard filters for regulation, hazard, residual limits, materials compatibility, availability, and plant constraints. | Traceable feasible candidate space |
| 4. Coupled Prediction | Link solubility/QSPR/ML and molecular calculations with activity coefficients, phase equilibria, reaction/formulation response, and uncertainty estimates. | Condition-specific performance map |
| 5. Process & Failure Stress Test | Assess precipitation or oiling, emulsions, third phases, inhibition or byproducts, azeotropes, viscosity, thermal instability, peroxide risk, corrosion, emissions, recovery energy, and recycle impurity buildup. | Pareto front, failure map, and process window |
| 6. Experimental Validation | Test a diverse shortlist with solubility, reaction, extraction, crystallization, stability, recovery, and recycle experiments; update models through active learning where useful. | Validated shortlist and scale-up gates |
Structures, synonyms, property and EHS records, provenance, condition tags, confidence grades, hard exclusions, and machine-readable files.
Transparent trade-offs among task performance, hazard, environmental criteria, recovery, cost, supply, and manufacturing feasibility.
Conditional solubility, miscibility, partitioning, SLE/LLE/VLE behavior, water tolerance, uncertainty, and applicability-domain flags.
Endpoint-level hazard evidence, critical red flags, volatility and emissions, waste routes, renewability indicators, data gaps, and rationale.
Boiling/phase-split strategy, azeotrope risk, energy demand, solvent losses, recycle drift, equipment compatibility, purity, supply, and QC ranges.
Selected candidates and controls, DOE ranges, analytical methods, replicates, recycle tests, acceptance gates, and model-update plan.
Balance rate, conversion, selectivity, catalyst or enzyme compatibility, heat removal, workup, impurity control, and solvent recycle.
Map solubility curves, supersaturation, polymorph risk, oiling out, particle attributes, mother-liquor recovery, and residual solvent.
Optimize distribution coefficient, selectivity, phase split, emulsion risk, water uptake, solvent loss, back-extraction, and regeneration.
Screen dissolution, swelling, viscosity, evaporation, film formation, porosity, additive compatibility, drying, and solvent retention.
Control solvation power, rheology, wetting, drying profile, stability, odor/VOC constraints, substrate compatibility, and performance.
Evaluate selective extraction, moisture tolerance, thermal sensitivity, downstream isolation, food/cosmetic suitability, and reuse.
Peer-reviewed open-access work supports three practical principles used in this service: solubility models should be evaluated on held-out data rather than fit alone; broad, curated solubility datasets enlarge the chemical and condition domain available to data-driven screening; and solvent selection should connect molecular or kinetic prediction to the real separation or recovery task. Accordingly, rankings include applicability and evidence flags, and high-impact uncertainties are converted into targeted experiments.
1 Jiang, L.; et al. Analysis of a nonsteroidal anti inflammatory drug solubility in green solvent via developing robust models based on machine learning technique. Scientific Reports 2025, 15, 19456. https://doi.org/10.1038/s41598-025-04596-y. Distributed under Open Access license CC BY 4.0, with modification.
2 Vermeire, F. H.; Green, W. H. Transfer learning for solvation free energies: from quantum chemistry to experiments. Chemical Engineering Journal 2021, 418, 129307. https://doi.org/10.1016/j.cej.2021.129307. Distributed under Open Access license CC BY 4.0.
3 Boobier, S.; et al. Machine learning with physicochemical relationships: solubility prediction in organic solvents and water. Nature Communications 2020, 11, 5753. https://doi.org/10.1038/s41467-020-19594-z. Distributed under Open Access license CC BY 4.0.
Our workflow preserves the link between solvent identity, formulation, temperature, pressure, assay, source, model, process assumption, and decision criterion. Computation narrows the search and explains trade-offs; final solvent selection requires use-case-specific experiments, safety review, regulatory assessment, equipment compatibility, and scale-relevant recovery testing. To discuss a molecule set, reaction, separation, formulation, or internal dataset, please Contact Us or submit the Online Inquiry below.
Submit your project details below, and our team will respond within 24 hours.
Talk to our technical team about your project!
I Want To Talk