🔬 condensed matter

Entropy maximization underlies topology and mechanical properties in dynamic covalent hydrogels

This study demonstrates that bond exchange in dynamic covalent hydrogels drives networks toward higher-entropy topologies, a mechanism that resolves discrepancies with classic theory and enables accurate prediction and control of gelation and mechanical properties without bond loss.

Original authors: Lucien Cousin, Pietro Miotti, Bruno Marco-Dufort, Igor V. Pivkin, Mark W. Tibbitt

Published 2026-03-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Lucien Cousin, Pietro Miotti, Bruno Marco-Dufort, Igor V. Pivkin, Mark W. Tibbitt

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Idea: It's Not Just About How Many Hands Are Holding On

Imagine you have a giant room full of people (these are the polymer chains). Everyone has four hands (these are the 4-arm stars).

In a standard, permanent plastic (like a hard water bottle), everyone grabs hands with their neighbors and holds on tight forever. Once they are holding hands, the structure is fixed. If you want to know how strong the room is, you just count how many hands are holding on. More hands = stronger room.

But in this paper, the scientists are studying Dynamic Covalent Hydrogels. Think of these as a room where people are holding hands, but they can let go and grab a different person's hand without letting go of the group entirely. They can swap partners. This is what makes the material "reprocessable" (you can melt it down and reshape it) and "self-healing."

The big question the scientists asked was: If these people are constantly swapping hands, does the strength of the room still depend only on how many hands are holding on?

The Surprise: The "Perfect" Structure is Actually Messy

The scientists expected that if 50% of the people were holding hands, the room would be half as strong as if 100% were holding hands. They used old math (classic theories) to predict this.

But they were wrong.

When they measured the strength of their "swapping" rooms, the results were totally different from the predictions.

The Gel Point: They expected the room to become a solid "gel" (a connected web) when about 33% of the hands were holding on. Instead, it took 60% of the hands to hold on before the room became solid.
The Stiffness: Once it did become solid, it got stiff much faster than predicted.

The Explanation: Entropy is the "Party Planner"

Why did this happen? The authors realized that because the people can swap hands, they aren't just randomly grabbing whoever is near. They are rearranging themselves to find the most comfortable, chaotic, and "free" arrangement possible.

In physics, we call this Maximizing Entropy.

Think of it like a crowded dance floor:

Permanent Network: People grab hands and freeze. They might get stuck in a weird, inefficient pattern (like a knot).
Dynamic Network: People are dancing and swapping partners. They naturally drift toward a pattern where everyone has the most room to move and the most options for who to dance with next.

The scientists discovered that this "dancing" causes the network to form loops.

Imagine two people grabbing hands, then grabbing each other's other hands, forming a tiny circle.
In a permanent network, these loops are rare mistakes (defects).
In a dynamic network, the system wants to form these loops because it increases the number of ways the people can arrange themselves (higher entropy).

The Analogy:
Imagine trying to organize a line of people.

Old Theory: "Just count how many people are in line."
New Reality: "The people are rearranging themselves into a complex dance formation. They are forming small circles (loops) and clusters because it's the most 'fun' (energetically favorable) way to be. This changes the shape of the line entirely."

Because the system spends so much time forming these loops and rearranging, it takes more people holding hands (a higher conversion rate) to actually tie the whole room together into a solid web. That's why the "Gel Point" was delayed.

The "Magic Switch" Experiment

To prove that it wasn't just about the number of hands holding on, but the arrangement of those hands, they did a clever experiment:

They built a network where the hands were glued together (permanent).
They added a "magic potion" (a chemical called DBU) that temporarily made the glue slippery, allowing the hands to swap partners.
They let them swap for a while, then removed the potion so the glue hardened again.

The Result:

If the network was "loose" (low number of hands holding on), making it dynamic caused it to turn into liquid. The rearrangement broke the weak connections.
If the network was "tight" (high number of hands holding on), making it dynamic caused it to get stiffer. The rearrangement allowed the people to form better, stronger loops and connections.

The Takeaway: Even though the number of hands holding on never changed, the structure changed, and the material's properties changed completely.

Why Does This Matter?

This paper changes how we design "smart" materials.

Recycling: If you melt down a dynamic plastic to recycle it, you aren't just melting it; you are letting the molecules rearrange into a new, potentially different structure. The recycled plastic might be weaker or stronger than the original, even if you didn't break any chemical bonds.
Design: Engineers can't just count the bonds anymore. They have to understand that the material will "self-organize" into a specific shape to maximize its freedom (entropy).
Prediction: The authors created a new math model that accounts for this "dancing" and "looping." Now, they can accurately predict how strong these materials will be, which is crucial for things like 3D printing, medical implants, and self-healing tires.

Summary in One Sentence

Dynamic networks are like a dance party where the guests constantly swap partners; this constant rearrangement forces them into a specific, loop-filled structure that makes the material much harder to solidify and much stiffer once it does, proving that how the pieces are arranged matters more than just how many pieces are connected.

1. Problem Statement

Dynamic covalent networks (DCvNs) and Covalent Adaptable Networks (CANs) are soft materials capable of reprocessing, self-healing, and shape-shifting due to reversible bonds. While extensive literature exists on the mechanics of permanent covalent networks, the behavior of DCvNs remains poorly understood, particularly in regimes below full conversion ( $p < 1$ ).

Key Gaps Identified:

Deviation from Theory: Classic gelation theories (e.g., Flory–Stockmayer) and elasticity models assume that reactive groups react independently. These models predict a gel point ( $p_{gel}$ ) of $1/(f-1)$ (where $f$ is functionality) and a specific modulus evolution. However, experimental data on DCvNs often show a significantly delayed gel point and a steeper increase in elastic modulus than predicted.
Role of Rearrangement: It is unclear how the ability of DCvNs to rearrange after gelation affects their internal topology and macroscopic mechanical properties.
Structure-Property Relationship: Current models often fail to predict mechanical behavior based solely on the fraction of formed bonds ( $p$ ) and precursor architecture, suggesting that the internal network topology of DCvNs differs fundamentally from permanent networks even at the same $p$ .

2. Methodology

The authors employed a combination of experimental synthesis, rheological characterization, theoretical modeling, and numerical simulations.

Experimental System:

Materials: Model DCvNs were synthesized using 4-arm PEG stars (Tetra-PEG) functionalized with either boronic acids (PBA or APBA) or diols (gluconolactone, GL).
Control of Conversion ( $p$ ): The fraction of formed bonds ( $p$ ) was systematically varied from $\approx 0.5$ to $\approx 1$ by adjusting temperature, precursor concentration, or adding competitive binders.
Characterization: Shear rheometry was used to measure the storage modulus ( $G'$ ) and mechanical relaxation time ( $\tau_M$ ) across a wide frequency range. The gel point was identified via the crossover of $G'$ and $G''$ and the divergence of $\tau_M$ .
Validation Experiment: To isolate the effect of network topology from bond density, the authors created networks with disulfide bridges. They triggered bond exchange using a base (DBU) in networks with fixed $p$ (0.55 and 0.85), allowed the system to rearrange, and then let the bonds become permanent again. This tested if the mechanical properties changed solely due to topological rearrangement.

Theoretical Framework:

Network Entropy Hypothesis: The authors hypothesized that DCvNs rearrange to maximize network entropy (the information entropy of the connectivity graph), rather than just the conformational entropy of individual chains.
Graph Theory Modeling: The network was modeled as a graph where stars are vertices and bonds are edges.
- Connectivity Distribution ( $P_i$ ): The probability distribution of arms bound per star ( $P_i$ ) was derived by maximizing the number of accessible microstates (graphs) subject to the constraint of fixed $p$ .
- Loop Formation: The model incorporated the formation of loops (defects where two stars are connected by multiple bonds) as a mechanism to increase entropy. The proportion of $i$ -loops was calculated to maximize the total entropy.
Elasticity Model: A modified Phantom Network Model was developed to account for the specific connectivity distribution ( $P_i$ ) and the presence of loops/dangling chains, deriving an effective strand length ( $N_\infty$ ) to predict the elastic modulus.

Simulation:

Dissipative Particle Dynamics (DPD) simulations were performed to validate the theoretical predictions regarding the distribution of bound arms and gel points.

3. Key Contributions

Identification of Network Entropy: The paper introduces "network entropy" as a distinct thermodynamic driver in DCvNs. Unlike permanent networks, DCvNs rearrange to adopt a topology that maximizes the number of microstates (connectivity configurations), leading to a non-random distribution of cross-links.
Revised Gelation Theory: The authors demonstrate that the classic assumption of independent reactive groups fails for DCvNs. By maximizing network entropy, the system naturally forms a higher proportion of loops and specific connectivity patterns, resulting in a delayed gel point ( $p_{gel} \approx 0.6$ for 4-arm stars, vs. the theoretical $0.33$).
Unified Predictive Model: A new theoretical framework was developed that accurately predicts both the gel point and the evolution of the elastic modulus ( $G'$ ) as a function of $p$ . This model successfully collapses experimental data from various conditions (temperature, concentration, different chemistries) onto a single master curve.
Topological Memory: The study proves that the mechanical properties of a network are determined by its internal topology, which can be permanently altered by a transient period of bond exchange, even if the final bond density ( $p$ ) remains unchanged.

4. Key Results

Deviation from Literature Models: Experimental data showed a gel point of $p_{gel} \approx 0.6$ , significantly higher than the Flory–Stockmayer prediction of $0.33$. The modulus increased much more steeply with $p$ than predicted by existing dynamic network models.
Entropy Maximization Predicts Topology: The theoretical model, which assumes the system maximizes network entropy, predicted a connectivity distribution ( $P_i$ ) and loop fraction that matched DPD simulations and experimental gel points perfectly.
Role of Loops: The model revealed that DCvNs spontaneously form a high proportion of loops (specifically 2-loops, where two stars share two bonds) to maximize entropy. These loops act as topological defects that delay gelation but stiffen the network once formed.
Triggered Rearrangement Experiment:
- Case $p=0.55$ : A network that was effectively permanent (below the new gel point) liquefied when bond exchange was triggered, then solidified into a much weaker gel (or remained liquid) after the exchange stopped.
- Case $p=0.85$ : A network that was stiff (above the gel point) became significantly stiffer after triggering bond exchange and re-solidifying.
- Conclusion: The mechanical properties changed drastically without breaking or forming new bonds (net $p$ was constant), proving that the topology (arrangement of bonds) dictates the mechanics, not just the bond density.

5. Significance

Paradigm Shift in Polymer Physics: The work challenges the long-standing assumption that polymer network properties are solely a function of conversion ( $p$ ) and precursor functionality. It establishes that for dynamic networks, the pathway to equilibrium and the resulting topological entropy are critical determinants of material properties.
Engineering Implications: For the design of reprocessable materials (CANs), this implies that processing conditions (temperature, time) that allow for bond exchange can irreversibly alter the final mechanical performance of a material, even if the chemical composition remains the same. A material designed to be strong might become weak (or vice versa) after reprocessing.
Predictive Capability: The provided mathematical framework allows engineers to accurately predict the gel point and modulus of dynamic networks, enabling the rational design of hydrogels for biomedical applications (e.g., 3D printing, drug delivery) where precise mechanical tuning is required.
Generalizability: While demonstrated on boronic ester and disulfide systems, the concept of network entropy maximization is applicable to all Covalent Adaptable Networks, suggesting a universal principle governing the mechanics of dynamic soft matter.