🔬 condensed matter

Structural Origin of Water Heat Capacity Anomaly from Classical and Quantum Simulations

Using classical and quantum simulations with machine-learning potentials, this study reveals that water's anomalous heat capacity arises from structural fluctuations between low- and high-density local structures, while nuclear quantum effects primarily suppress high-frequency vibrations.

Original authors: Kam-Tung Chan, Dylan A. Folkner, Margaret L. Berrens, Alexei A. Stuchebrukhov, Lee-Ping Wang, Davide Donadio

Published 2026-07-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kam-Tung Chan, Dylan A. Folkner, Margaret L. Berrens, Alexei A. Stuchebrukhov, Lee-Ping Wang, Davide Donadio

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

Imagine water not as a simple, boring puddle, but as a bustling, chaotic dance floor where millions of tiny water molecules are constantly shuffling, spinning, and bumping into each other. Scientists have long been puzzled by a weird trick this dance floor plays: water holds onto heat way better than almost any other liquid. If you try to warm up a cup of water, it resists, soaking up a massive amount of energy before it actually gets hotter. Even stranger, if you supercool it (chill it below freezing without it turning to ice), this resistance to heating spikes to a dramatic peak around 230 K.

For years, researchers have been trying to figure out why water does this. Some thought it was because the water molecules were vibrating wildly like tiny springs. But in this study, a team of scientists ran incredibly detailed computer simulations to test the theories, and they found a different story.

The Quantum "Mute Button"
First, let's talk about the vibrations. In the old-school computer models (called "classical simulations"), the water molecules were treated like tiny balls connected by springs that could vibrate at any speed. These models predicted that water would hold too much heat—way more than what we see in real life. It was like the dance floor was vibrating so hard it was absorbing all the energy in the universe.

The researchers then turned on a special "quantum mode" for their simulations. In the real world, atoms are fuzzy and quantum-mechanical, which means they can't vibrate as frantically as the old models suggested. When the scientists added these quantum effects, it was like hitting a "mute button" on the high-pitched, super-fast vibrations. Suddenly, the simulated heat capacity dropped and matched real-world experiments perfectly.

So, the first big discovery is that quantum effects are the reason water doesn't hold even more heat than it does. They suppress the crazy high-frequency jiggling. However, the paper makes it clear that this quantum "mute button" doesn't explain the weird shape of the curve. It doesn't explain why the heat capacity spikes so dramatically at 230 K. That mystery remained.

The "Intruder" Detective
To solve the second part of the puzzle, the team looked at the structure of the dance floor. They invented a new way to count the dancers, which they call the "Second Solvation Shell Intruder" (SSSI).

Imagine a water molecule in the center. It has a tight circle of four closest friends (the first shell) and a slightly larger circle of sixteen friends (the second shell). Usually, these friends stay in their own lanes. But sometimes, a molecule from the outer circle sneaks into the inner circle's personal space. These are the "intruders."

The researchers found that water exists in two main moods, or local structures:

The "Low-Density" Mood (LDL): A calm, organized dance where few intruders are allowed. The molecules are spread out.
The "High-Density" Mood (HDL): A chaotic, crowded dance where many intruders are pushing their way in.

As the temperature changes, the water molecules are constantly switching between these two moods. The paper suggests that the massive spike in heat capacity happens right when the water is flipping back and forth between these two states the fastest. It's like a crowd of people suddenly deciding to switch from a slow waltz to a mosh pit; that transition requires a huge amount of energy.

The Two-State Map
The team mapped this behavior onto a simple "two-state" model. They calculated that the energy difference between being in the "calm" mood and the "chaotic" mood is about 3–4 kJ mol⁻¹. To put that in perspective, the energy needed to melt ice is about 6 kJ mol⁻¹. So, the energy cost of water molecules switching their local structure is nearly half the energy needed to melt ice.

This structural switching explains the entire weird curve:

At supercooled temperatures (around 230 K): The water is right on the edge of the switch. It's flipping between the calm and chaotic moods rapidly, causing the heat capacity to hit its maximum.
At room temperature: The water is mostly in the chaotic (HDL) mood, but it still has enough energy to occasionally dip back into the calm (LDL) mood. This constant, slow switching is what keeps water's heat capacity unusually high even when it's not supercooled.

What This Paper Rules Out
It is important to note what this study says is not the main culprit. The paper explicitly argues against the idea that the strange heat capacity is caused by the water molecules vibrating in a specific, unusual way. While vibrations matter for the total amount of heat water holds (thanks to the quantum effects), the anomaly—the weird peak and the high values—are driven entirely by the structural shuffling, not the vibrations.

How Sure Are They?
The authors are very confident in their numbers, but with a caveat: these are results from simulations, not direct measurements of the invisible dance floor. They used two different, highly accurate computer models (one based on a complex quantum chemistry method called MB-pol and another based on a density functional theory) and both gave the same story.

They found that the structural order parameter (the intruder count) tracks the heat capacity perfectly. The peak in the rate of structural change lines up exactly with the peak in heat capacity at 230 K. They estimate the energy scale of 3–4 kJ mol⁻¹ with a small margin of error depending on exactly how they defined the "calm" and "chaotic" states.

In short, the paper suggests that water's heat capacity anomaly isn't a mystery of how fast the molecules vibrate, but a story of how they rearrange their social circles. The water molecules are constantly negotiating whether to be in a quiet, organized group or a crowded, chaotic one, and that constant negotiation is what makes water so good at holding onto heat.

Technical Summary: Structural Origin of Water's Heat Capacity Anomaly

Problem Statement
Liquid water exhibits a suite of thermodynamic anomalies, most notably an anomalously large isobaric heat capacity ( $C_P$ ) under ambient conditions and a sharp maximum upon supercooling to approximately 230 K. While the liquid-liquid phase transition (LLPT) scenario posits that these anomalies arise from the interconversion between low-density-like (LDL) and high-density-like (HDL) local structures, a quantitative microscopic explanation linking specific structural fluctuations to the temperature dependence of $C_P$ has remained elusive. Furthermore, classical molecular dynamics (MD) simulations typically overestimate $C_P$ because they treat all vibrational modes as thermally populated, whereas path-integral molecular dynamics (PIMD) simulations, which include nuclear quantum effects (NQEs), yield values closer to experiment. However, previous studies have not fully disentangled the role of NQEs in setting the absolute magnitude of $C_P$ from the structural mechanisms driving its anomalous temperature dependence.

Methodology
The authors employed a combination of classical MD and PIMD simulations within the isothermal-isobaric (NPT) ensemble. To ensure accuracy, they utilized two distinct machine-learning interatomic potentials (neuroevolution potentials, NEPs):

NEP3@MB-pol: Trained on an ab initio many-body potential energy function at the coupled-cluster level (MB-pol).
NEP3@revPBE0-D3: Trained on a van der Waals-corrected hybrid density functional.

The study focused primarily on the NEP3@MB-pol model. Simulations were conducted over a temperature range of 210 K to 360 K at 1 bar. The authors calculated $C_P$ directly from the temperature derivative of enthalpy ( $H$ ) using polynomial fits. Additionally, they estimated quantum corrections to classical $C_P$ using a vibrational density of states (vDOS) approach based on the Bose-Einstein distribution.

To probe the structural origin of the anomalies, the authors introduced and analyzed the "solvation-shell averaged second solvation shell intruders" order parameter, denoted as SSSI(3). This parameter quantifies the number of water molecules in the second solvation shell that are not contained within the union of the first solvation shells of the first-shell neighbors. The SSSI(3) value distinguishes between LDL-like environments (low values) and HDL-like environments (high values).

Key Contributions and Results

Separation of Vibrational and Structural Contributions:
The simulations demonstrate that NQEs and structural fluctuations affect $C_P$ through distinct mechanisms.
- NQEs: Path-integral simulations and vDOS-based quantum corrections show that NQEs primarily suppress high-frequency vibrational contributions. This suppression lowers the absolute magnitude of $C_P$ by approximately 4–5 $R$ (gas constant units) at ambient temperatures, bringing classical predictions into agreement with experimental values. Crucially, NQEs have a negligible effect on the temperature of the $C_P$ maximum ( $T_{max} \approx 230$ K).
- Structural Fluctuations: The anomalous temperature dependence of $C_P$ , including the supercooled maximum and the high ambient values, is captured by the temperature derivative of the mean SSSI(3) order parameter, $d\langle \text{SSSI}(3) \rangle / dT$ . The peak in this derivative coincides with the $C_P$ maximum, indicating that the anomaly is driven by the rapid change in the population of LDL-like versus HDL-like local structures.
Two-State Mapping and Enthalpy Scale:
The authors mapped the continuous evolution of the SSSI(3) order parameter onto a simple two-state model, assuming the system consists of a fraction $\lambda(T)$ of HDL-like structures and $(1-\lambda(T))$ of LDL-like structures.
- This mapping allowed the decomposition of the total enthalpy $H(T)$ into a physically constrained baseline ( $H_{base}$ ) and a structural contribution ( $H_{SSSI}$ ) arising from the interconversion of local structures.
- From this model, the authors extracted an effective enthalpy scale ( $\Delta H_{L/H}$ ) of approximately 3–4 kJ mol $^{-1}$ associated with the transition between LDL-like and HDL-like local structures. This value is roughly half the enthalpy of ice melting (~6 kJ mol $^{-1}$ ) and is consistent with energy differences between amorphous ice phases.
Decomposition of Heat Capacity:
By decomposing $C_P$ into a baseline term ( $C_{P,base}$ ) and a structural term ( $C_{P,SSSI}$ ), the authors showed that:
- $C_{P,base}$ (comprising vibrational and residual configurational contributions) accounts for the background heat capacity.
- $C_{P,SSSI}$ reproduces the supercooled maximum and the excess heat capacity at higher temperatures.
- The structural term correlates strongly with the variance of SSSI(3) and the enthalpy-SSSI(3) correlation coefficient, which are both enhanced near the $C_P$ maximum.

Significance
The paper provides a microscopic framework consistent with the LLPT scenario, linking local structural topology to water's thermodynamic anomalies. It establishes that while nuclear quantum effects are essential for determining the correct magnitude of the heat capacity by suppressing high-frequency vibrations, the anomalous temperature dependence is governed by structural fluctuations between LDL-like and HDL-like local environments. The study successfully quantifies the effective enthalpy scale of this structural interconversion and demonstrates that a simple two-state mapping, despite the continuous nature of the structural evolution, effectively captures the thermodynamic behavior of water from supercooled to ambient conditions. The findings are robust across two different machine-learning potentials (MB-pol and revPBE0-D3), suggesting that the separation between a quantum-renormalized baseline and a structurally controlled anomalous contribution is a fundamental feature of water's thermodynamics.

Technical Summary: Structural Origin of Water's Heat Capacity Anomaly

More like this