Separation and Product Definition
Translate the feed and product specification into a molecular-screening target.
- C2 and C3 mixture composition
- Olefin or paraffin product strategy
- Purity, recovery, and trace-removal target
AI for Carbon Capture and Molecular Separation
Light Hydrocarbon Separation
Resolve close-boiling hydrocarbon separations with more than ideal selectivity. CD ComputaBio screens porous adsorbents and membranes against mixture composition, pore flexibility, diffusion, impurities, regeneration, product purity, and the cycle that defines the real separation.
Mixture specificKinetics awareProcess ranked
Service coverage
Light Hydrocarbon Separation Services
Mixture Adsorption Screening
Identify materials with useful affinity and working-capacity differences under the real feed.
- GCMC and mixture prediction
- Olefin- versus paraffin-selective routes
- Pressure and temperature dependence
Diffusion and Molecular-Sieving Analysis
Distinguish candidates whose dynamic pore response or transport rate controls separation.
- Molecular dynamics and energy barriers
- Gate-opening and framework flexibility
- Crystallite and membrane transport context
Multicomponent and Impurity Analysis
Test whether alkynes, CO2, water, or co-fed hydrocarbons change the credible shortlist.
- Ternary and multicomponent scenarios
- Trace contaminant removal
- Competitive adsorption risk
Cyclic and Process-Aware Ranking
Connect molecular behavior with product recovery, regeneration, productivity, and material form.
- PSA, TSA, SMB, or membrane context
- Pareto trade-off analysis
- Decision-ready validation plan
Separation material landscape
Materials for Molecular Separation
The library is grouped by how each material distinguishes near-size hydrocarbons: specific binding, rigid size exclusion, dynamic gating, or faster transport through a selective layer.
Flexible MOFs
Guest-responsive pockets can open at different pressures for an olefin and its paraffin partner.
Ultramicroporous Frameworks
Precisely sized channels target molecular recognition without relying only on strong binding.
Zeolite Molecular Sieves
Rigid apertures offer robust size and shape discrimination under cyclic operation.
Carbon Molecular Sieves
Tuned carbon slit pores create useful differences in hydrocarbon diffusion rates.
Selective Membranes
Continuous layers and mixed matrices translate molecular transport into steady product flow.
Separation decision model
How We Compare Separation Materials
Near-size hydrocarbons can reverse a ranking when framework flexibility, diffusion rate, loading, impurities, or regeneration is considered. We compare thermodynamic affinity and kinetic discrimination together, then connect them to product purity, recovery, cycle productivity, and material stability.
Illustrative comparison only; bar lengths are not experimental values.
Separation criteriaMixture weighted
Mixture selectivity
FEED
Working capacity
CYCLE
Diffusion contrast
KINETIC
Product recovery
PRODUCT
Regeneration duty
ENERGY
Stability and shaping
RISK
Separation project path
From Molecular Pair to Process Fit
Start with the molecules that must be split, identify the governing mechanism, then test whether it survives the intended cycle.
Define the Molecular Pair
Specify components, composition, pressure, temperature, purity, recovery, trace impurities, and preferred separation route.
Curate Pore Chemistries
Standardize structures and preserve diversity in aperture, flexibility, binding sites, topology, and material class.
Simulate Mixtures and Motion
Combine adsorption simulation, diffusion analysis, AI triage, and uncertainty checks for the target feed.
Select for the Cycle
Rank candidates by purity, recovery, productivity, regeneration, stability, and experimental validation value.
Project package
Mixture Data and Project Outputs
Separation Inputs
- Hydrocarbon components, feed composition, pressure, temperature, and flow context
- Target product, required purity, recovery, or trace-contaminant specification
- Preferred PSA, TSA, SMB, membrane, or equilibrium separation concept
- Material-family preferences, exclusions, and available adsorption or breakthrough data
Mechanism and Cycle Checks
- Structure provenance and accessible-pore checks
- Framework flexibility and diffusion-model documentation
- Multicomponent and impurity sensitivity analysis
- Regeneration, density, shaping, and stability flags
01 / DATACurated separation-material library
02 / RESULTSMixture and transport dataset
03 / DECISIONProcess-aware candidate shortlist
04 / NEXT STEPBreakthrough and cycling roadmap
Test the separation mechanism
Validate Selectivity Under Flow
For close-boiling hydrocarbons, a credible validation plan must distinguish equilibrium affinity from kinetic sieving and show that selectivity survives mixture loading, impurities, cycling, and material shaping.
Equilibrium, kinetic, and gate-opening hypotheses are tested separately before breakthrough and cycling data are used to judge product purity.
Plan Your Validation Strategy01
Confirm mixture affinity
Single- and Mixed-Gas Isotherms
Measure uptake and working capacity for the target olefin/paraffin or alkyne/olefin system.
- Matched pressure and temperature window
- Mixture adsorption where available
- Heat of adsorption and hysteresis
02
Resolve transport control
Kinetic Uptake and Diffusion Testing
Determine whether aperture dynamics, gate opening, or intracrystalline diffusion controls the observed separation.
- Time-resolved uptake
- Crystallite-size sensitivity
- Temperature-dependent transport
03
Challenge the product route
Dynamic Breakthrough Testing
Test product purity and recovery under representative composition, flow, and contaminant conditions.
- Binary and multicomponent feeds
- Trace alkyne or CO2 scenarios
- Dry and humid comparisons
04
Test practical operation
Cycling, Shaping, and Membrane Validation
Evaluate regeneration, pelletization, binders, pressure drop, or selective-layer integrity.
- Repeated adsorption–desorption cycles
- Pellet and binder effects
- Mixed-gas permeation where relevant
05
Update the decision
Model Calibration and Re-Ranking
Integrate measured equilibrium, kinetic, and cyclic results to refine the shortlist and operating assumptions.
- Mechanism confirmation
- Uncertainty reduction
- Updated validation priority
Published data
Research Behind Hydrocarbon Screening
Machine learning connected MOF screening to process modeling
1Prescreen the landscapeReduce a large experimental/hypothetical MOF space.
2Learn adsorption behaviorTrain from GCMC simulations and verify leading candidates.
3Test the cycleEvaluate top materials with VPSA process modeling.
Cheng and colleagues combined machine learning, molecular simulation, and vacuum pressure swing adsorption modeling for ethane/ethylene separation, demonstrating why process conditions belong in candidate ranking.[1]
View publicationDynamic molecular pockets separated propylene and propane
1Control local flexibilityCreate pockets with guest-dependent opening pressures.
2Resolve the mechanismCombine crystallography, adsorption, and computation.
3Validate dynamicallyMeasure mixture breakthrough, recovery, and recyclability.
Zeng and colleagues showed that a dynamically responsive MOF could distinguish propylene and propane through guest-dependent pocket opening, linking framework motion with breakthrough performance.[2]
View publication
Separation project guidance
Light Hydrocarbon Separation Questions
Model detail should follow the molecular pair, flexibility of the material, product specification, and operating route.
Which light-hydrocarbon separations can be studied?
Projects may address ethane/ethylene, propane/propylene, methane-containing mixtures, acetylene or propyne removal, and multicomponent feeds when suitable models and data are available.
Can you compare olefin-selective and paraffin-selective strategies?
Yes. The preferred direction depends on feed composition, product recovery, regeneration, and downstream purification requirements.
How do you account for flexible or gate-opening materials?
Projects can include framework-dependent isotherms, energy-barrier analysis, molecular dynamics, and sensitivity ranges; assumptions are documented because rigid-pore models may be insufficient.
Is ideal selectivity enough to rank materials?
No. Working capacity, diffusion rate, mixture loading, cyclic recovery, stability, and form factor can change the process-relevant ranking.
Can experimental breakthrough data be integrated?
Yes. Breakthrough curves, kinetic uptake, isotherms, and cycling data can calibrate the model and clarify whether equilibrium or transport controls the separation.
Start with the Hydrocarbon Mixture
Share the components, composition, operating window, product specification, and available material data. We will define a focused screening and validation plan for the separation decision.
Scientific References
- Cheng, M., Feng, M., Zhou, L., et al. A Machine Learning-Boosted High-Throughput Screening of Metal–Organic Frameworks for Ethane/Ethylene Separation: From Molecular Simulation to Process Modeling. Industrial & Engineering Chemistry Research 64, 17135–17146 (2025). https://doi.org/10.1021/acs.iecr.5c00726
- Zeng, H., Xie, M., Wang, T., et al. Orthogonal-array dynamic molecular sieving of propylene/propane mixtures. Nature 595, 542–548 (2021). https://doi.org/10.1038/s41586-021-03627-8
