TL;DR
This paper introduces a computational model that accurately captures how fluid dynamics, including buoyancy and shear, influence liquid-liquid phase separation in confined systems. Their findings reveal that neglecting fluid dynamics leads to incorrect predictions of droplet shapes, while including it accurately reproduces experimental results and suggests methods for accelerating microparticle manufacturing.
What Problem Does This Paper Address?
Liquid-liquid phase separation (LLPS) is a fundamental process in biology (e.g., forming cellular organelles) and a crucial technique in the pharmaceutical and medical industries for creating microparticles for drug delivery and cell analysis. The ability to control the precise shape and morphology of these separating droplets is key to their function. However, existing computational models often simplify LLPS by neglecting fluid dynamics, which can lead to inaccurate predictions of the final droplet shapes, particularly the energy-minimizing "crescent" shapes observed in experiments. This gap limits our ability to optimize manufacturing processes and understand LLPS in biologically relevant confined environments.
Methods
The authors developed a hierarchy of computational models to simulate LLPS. They started with a basic surface energy minimization model to predict equilibrium shapes. Then, they incorporated dynamics using a ternary Cahn-Hilliard model, which describes phase separation based on minimizing free energy. Crucially, they extended this to a Cahn-Hilliard-Stokes-Boussinesq (CHSB) model. This advanced model includes:
- Finite viscosity fluid dynamics: Accounts for how the fluid's resistance to flow affects droplet behavior.
- Surface tension: Models the forces at the interfaces between different liquid phases.
- Buoyancy: Incorporates the effect of gravity on density differences between phases, particularly important in confined settings.
- Shear flows: Simulates the impact of fluid motion driven by external forces (like pressure gradients in channels).
They used a spectral method (Dedalus) for numerical simulations, allowing for high-resolution 2D and 3D analyses.
Datasets
The study utilized experimental data from gelatin-polyethylene glycol (PEG) mixtures undergoing temperature-induced phase separation. While no new benchmark datasets were introduced, the experimental results served as a crucial validation point for their computational models.
Key Insights and Ideas
- Fluid dynamics is not optional: The most striking finding is that neglecting fluid dynamics (even buoyancy) leads to incorrect predictions of the minimum energy droplet shapes, failing to reproduce experimentally observed "crescent" morphologies.
- Buoyancy drives equilibrium: Buoyancy forces are critical for driving the system towards the desired energy-minimizing crescent shapes, breaking the symmetry that would otherwise lead to simpler shell structures.
- Shear accelerates formation: Shear flows, such as those found in microfluidic channels, can dramatically speed up the formation of these crescent shapes, offering a tenfold acceleration compared to models without significant flow.
- Model hierarchy is key: The paper demonstrates the value of building increasingly complex models, starting from simple energy minimization and progressively adding dynamics and fluid effects to capture real-world phenomena.
Key Results
- The Cahn-Hilliard model (without fluid dynamics) predicted incorrect core-shell morphologies from mixed initial conditions, failing to match experimental crescent shapes.
- The CHSB model, incorporating fluid dynamics and buoyancy, successfully reproduced the experimentally observed crescent-shaped droplets.
- Shear-driven flows in the CHSB model led to a tenfold speedup in the formation of crescent particles compared to buoyancy-driven flows alone.
- The study identified specific regimes where different initial concentrations lead to various final configurations (crescents, shells, or single-phase droplets).
Limitations and Future Work
The authors acknowledge that:
- Time scales: The computational simulations run on faster time scales than the experiments, requiring time rescaling for direct comparison.
- Parameter tuning: Certain parameters, like interfacial thickness and mobility, are constrained by computational resources rather than direct measurement.
- Complexity of biological systems: While the model is applicable to biological condensates, it doesn't yet account for more complex factors like surfactants or active processes.
Future work suggested includes:
- Investigating the role of fluid dynamics at even longer time scales and higher computational costs.
- Developing reduced asymptotic models for phase separation in low Reynolds number flows to improve computational efficiency.
- Extending the thermodynamic models to include non-constant mobilities, Flory-Huggins free energies, and Marangoni stresses for broader applicability.
- Exploring the impact of different container geometries and flow conditions on phase separation dynamics.
Summary updated