Entropy Scaling for Diffusion Coefficients in Fluid Mixtures

This paper presents a thermodynamically consistent entropy scaling framework that successfully predicts self- and mutual diffusion coefficients in fluid mixtures across a wide range of states, including non-ideal systems, by leveraging pure component self-diffusion data, infinite-dilution coefficients, and molecular-based equations of state.

The Gist

Imagine you are at a massive, chaotic party. You want to understand how people move around the room.

In the world of chemistry, this "party" is a fluid mixture (like alcohol and water, or oil and gas), and the "people" are molecules. Scientists have long been trying to predict exactly how fast these molecules drift and mix. This is called diffusion.

For a long time, predicting this movement in a mix of different molecules has been like trying to guess the outcome of a complex dance without knowing the music or the steps. We had good rules for when people are dancing alone (pure substances), but as soon as different groups started mixing, the math got messy, and our predictions often failed.

This paper introduces a brilliant new way to solve this puzzle using a concept called Entropy Scaling. Here is the breakdown in simple terms:

1. The Problem: The "Party" is Too Complicated

In a mixture, you have two types of movement:

• Self-Diffusion: Imagine a single guest wandering around the room, bumping into others, just to see where they end up. This is easy to predict if the room is full of just one type of person.
• Mutual Diffusion: This is the actual mixing. If you pour red dye into blue water, how fast do they blend? This is much harder because the molecules interact, push, and pull on each other in complicated ways.

Previous models tried to guess this mixing speed using simple formulas, but they often failed when the "guests" (molecules) were very different from each other (like a heavy oil molecule mixing with a light gas molecule).

2. The Solution: The "Universal Dance Floor" (Entropy Scaling)

The authors discovered a hidden pattern. They realized that if you look at the disorder (entropy) of the room, the speed of the dancers follows a single, simple rule.

Think of Entropy as the "crowdedness" or "chaos" of the party.

• When the room is empty (low chaos), molecules move fast and freely.
• When the room is packed (high chaos), molecules move slowly and struggle to get through.

The authors found that if you plot the speed of the molecules against the level of chaos, all different types of fluids fall onto the same single curve. It's as if every fluid, from gas to super-hot liquid, follows the same "dance rhythm" if you measure it the right way.

3. The New Trick: Treating "Diluted" Guests as "Solo" Dancers

The biggest breakthrough in this paper is how they handled the tricky "mixing" part.

Usually, predicting how a drop of dye mixes into water is hard. But the authors realized: When a drop of dye is so tiny it's almost invisible (infinite dilution), it acts like it's dancing alone in a sea of water.

They treated these "tiny drop" scenarios as if they were solo performances (pseudo-pure components).

• Step 1: They figured out the "dance rhythm" (entropy scaling) for a molecule dancing alone.
• Step 2: They figured out the "dance rhythm" for a molecule dancing in a sea of other molecules (infinite dilution).
• Step 3: They used a simple "mixing rule" (like a recipe) to blend these two rhythms together to predict how the whole party mixes.

4. Why This is a Big Deal

Before this, if you wanted to know how fast two specific chemicals mix at high pressure or extreme temperatures, you often had to run expensive experiments or complex computer simulations.

This new method is like having a universal translator.

• No Guesswork: You don't need to tweak the math for every new mixture.
• Wide Range: It works for gases, liquids, supercritical fluids (like the stuff inside a fire extinguisher), and even unstable states where the fluid is about to boil or freeze.
• Accuracy: It correctly predicts tricky behaviors, like when a mixture separates into two layers (like oil and water) or when it hits a "critical point" where the distinction between liquid and gas disappears.

The Bottom Line

The authors built a "GPS for molecules." Instead of getting lost in the complex math of how molecules push and pull, they realized that chaos (entropy) is the map. By understanding how the "crowdedness" of the fluid changes, they can predict exactly how fast molecules will mix, whether they are in a gas cloud, a liquid fuel tank, or a deep-sea chemical reaction.

It turns a chaotic, unpredictable party into a dance with a predictable beat.

Technical Summary

1. Problem Statement

Diffusion coefficients are critical for designing fluid separation processes, reactors, and combustion systems. However, experimental data for diffusion in mixtures are scarce, particularly for non-ideal systems and extreme states (supercritical, metastable).

• The Gap: While entropy scaling (relating transport properties to configurational entropy) is well-established for pure components and mixture viscosities/thermal conductivities, it has not been successfully applied to mixture diffusion coefficients.
• The Challenge: Existing empirical models (e.g., Vignes, Darken) often fail for strongly non-ideal mixtures and cannot consistently describe the relationship between self-diffusion ($D_i$), Maxwell-Stefan diffusion ($\mathcal{D}_{ij}$), and Fickian diffusion ($D_{ij}$) across the entire fluid region (gas, liquid, supercritical, and metastable). Furthermore, no generally applicable relation connects self-diffusion and mutual diffusion coefficients.

2. Methodology

The authors propose a thermodynamically consistent entropy scaling framework that predicts all diffusion coefficients in binary mixtures without adjustable mixture parameters. The approach relies on three core concepts:

A. Pseudo-Pure Component Treatment

The framework treats infinite-dilution diffusion coefficients ($D^\infty_i$) as "pseudo-pure" components.

• It was discovered that $D^\infty_i$ exhibits a monovariate scaling behavior with respect to the reduced configurational entropy ($\tilde{s}_{conf}$), similar to pure component self-diffusion.
• This allows $D^\infty_i$ to be modeled using the same entropy scaling correlations developed for pure fluids, requiring only a few reference data points (or even none if global parameters are used).

B. Entropy Scaling Formulation

The model uses a modified Rosenfeld scaling approach combined with the Chapman-Enskog (CE) theory for the low-density limit.

• Scaling Variable: The reduced configurational entropy $\tilde{s}_{conf} = -s_{conf}/(R m)$, where $s_{conf}$ is derived from a molecular-based Equation of State (EOS).
• Limiting Cases: The model defines scaling functions for four limiting cases:
  1. Pure component self-diffusion ($D^{pure}_i$).
  2. Infinite-dilution self-diffusion ($D^\infty_i$).
  3. Pure component self-diffusion of the second species ($D^{pure}_j$).
  4. Infinite-dilution self-diffusion of the second species ($D^\infty_j$).
• Functional Form: The scaled diffusion coefficient $\hat{D}$ is described by a continuous function of $\tilde{s}_{conf}$ involving system-specific parameters ($\alpha_2, \alpha_3$) fitted to the limiting cases.

C. Mixing and Combination Rules

To predict diffusion at arbitrary compositions ($x_i$), the framework employs linear mixing rules to interpolate between the limiting cases:

• Self-Diffusion ($D_i, D_j$): Calculated by mixing the pure component and infinite-dilution scaling parameters based on mole fractions.
• Maxwell-Stefan Diffusion ($\mathcal{D}_{ij}$): Calculated by mixing the infinite-dilution parameters of both components.
• Fickian Diffusion ($D_{ij}$): Derived from the Maxwell-Stefan coefficient and the thermodynamic factor ($\Gamma_{ij}$), which is calculated directly from the EOS ($\Gamma_{ij} = \frac{x_i x_j}{RT} \frac{\partial^2 G}{\partial x_i^2}$).

Key Feature: The framework couples the entropy scaling model with a molecular-based EOS (e.g., PC-SAFT for real fluids, Kolafa-Nezbeda for Lennard-Jones). This ensures thermodynamic consistency, allowing the model to handle phase equilibria (VLE, LLE) and metastable states naturally.

3. Key Contributions

1. Unified Framework: First successful application of entropy scaling to predict all diffusion coefficients (self, Maxwell-Stefan, and Fickian) in mixtures consistently.
2. Pseudo-Pure Concept: Demonstrated that infinite-dilution diffusion coefficients follow a universal monovariate scaling law with configurational entropy, treating them as pseudo-pure components.
3. Parameter-Free Prediction: Once the limiting cases are characterized (using minimal data), the model predicts diffusion across the entire composition range and state space without any adjustable mixture parameters.
4. Broad Applicability: The model works for gases, liquids, supercritical fluids, and even metastable/unstable states (e.g., spinodal regions), which previous models could not handle.

4. Results

The authors validated the model using Molecular Dynamics (MD) simulations for Lennard-Jones (LJ) mixtures and experimental data for real substance systems.

• Lennard-Jones Mixtures:

  • The model accurately predicted self-diffusion and Maxwell-Stefan coefficients across the entire composition range, including systems with strong non-ideality, extrema, and liquid-liquid equilibrium (LLE) gaps.
  • It correctly captured the behavior at the critical point where the Fickian diffusion coefficient approaches zero.
  • It outperformed empirical models (Vignes and Darken), which failed to capture trends in non-ideal systems and LLE.

• Real Substance Systems:

  • Tested on mixtures like n-hexane + n-dodecane, acetone + chloroform, and toluene + n-hexane.
  • Accuracy: Mean relative deviations were generally low (e.g., ~1–5% for self-diffusion in non-polar mixtures).
  • Non-Ideality: The model successfully captured non-linear behaviors caused by large molar mass differences or strong interactions.
  • Limitations: Slight deviations were observed for associating systems (e.g., 2-propanol + n-heptane) and near extrema in Fickian diffusion, likely due to complex local structuring not fully captured by the current EOS.

• Metastable States:

  • The framework successfully predicted diffusion coefficients in metastable and unstable regions (e.g., within the spinodal), demonstrating capabilities far beyond current experimental or empirical modeling limits.

5. Significance

• Predictive Power: This method enables the prediction of diffusion coefficients in regions where experimental data is non-existent (e.g., high pressure, supercritical, metastable), which is crucial for process design in extreme conditions.
• Thermodynamic Consistency: By coupling with an EOS, the model guarantees that diffusion predictions are consistent with phase equilibria and thermodynamic stability, a feature lacking in empirical correlations.
• Efficiency: It reduces the need for extensive experimental data or complex molecular simulations for every new mixture condition; only limiting case data (pure and infinite dilution) is required to generate the full mixture profile.
• Foundation for Future Work: The methodology provides a robust physical basis for extending entropy scaling to multi-component systems and more complex associating fluids.

In summary, this work resolves a long-standing challenge in transport phenomena by establishing a universal, thermodynamically consistent framework for mixture diffusion based on entropy scaling, bridging the gap between pure component theory and complex mixture behavior.