Fluctuation-driven multi-step charge density wave transition in monolayer TiSe$_2$

Using large-scale molecular dynamics simulations with machine-learning potentials, this study reveals that monolayer TiSe$_2$ undergoes a fluctuation-driven, two-step CDW melting process characterized by topological defects and anisotropic thermal fluctuations that stabilize a chiral $3Q$ order, demonstrating that its complex physics can be explained without invoking excitonic correlations.

The Gist

Imagine a crowded dance floor where everyone is trying to move in perfect, synchronized patterns. In the world of quantum physics, this dance floor is a single layer of a material called Titanium Diselenide (TiSe₂), and the dancers are atoms.

For decades, scientists have been arguing about how these atoms decide to dance. Specifically, they are trying to understand a phenomenon called a Charge Density Wave (CDW). Think of this as a moment when the atoms suddenly stop dancing randomly and lock into a rigid, wavy pattern, like a stadium wave.

The big question was: How does this pattern break down when the material gets hot?

Here is the story of what this new paper discovered, explained without the heavy jargon:

1. The Old Theory: The "Slow Fade"

For a long time, scientists thought the transition from the "wavy pattern" to "chaos" was like a slow, smooth fade-out. They believed that as you heated the material, the wave would gradually get weaker and weaker until it vanished at a specific temperature (around 200 K). They thought it was a simple, predictable slide from order to disorder.

2. The New Discovery: The "Chaotic Intermission"

This paper, using powerful computer simulations (like a super-accurate virtual reality for atoms), found that the old theory was wrong. The atoms don't just fade out; they go through a messy, chaotic middle stage.

Instead of a smooth slide, the melting process happens in two distinct steps:

• Step 1 (The "Chirality" Break): Around 150 K, the perfect symmetry of the dance breaks. The atoms start favoring one direction over another, creating a "handedness" (like a left-handed vs. right-handed spiral). This is called chirality.
• Step 2 (The "Fluctuation" Zone): Between 200 K and 250 K, things get wild. The material doesn't immediately become a normal liquid. Instead, it enters a fluctuation regime. Imagine the dance floor is now covered in different groups of dancers. Some groups are still doing the wave, others are doing a different wave, and they are separated by "walls" where the dancing stops. These groups are constantly changing, appearing, and disappearing.

3. The Culprit: Thermal "Jitters"

What causes this chaos? The authors found that it's not some mysterious new force, but simply heat.

Think of heat as thermal jitter. When the atoms get hot, they start shaking. In this material, that shaking isn't uniform; it's anisotropic, meaning it shakes more in some directions than others.

• The Analogy: Imagine a calm pond. If you drop a stone, you get ripples. In this material, the "heat" acts like a giant, uneven wind blowing across the pond. This wind creates large, long waves (acoustic phonons) that crash into the organized dance pattern, breaking it into smaller, competing islands of order.

4. The "Ghost" of the Pattern

One of the coolest findings is that even though the perfect pattern is gone, the "ghost" of the wave lingers.

• The Soft Mode: In physics, there is a specific vibration (a phonon) that usually tells the atoms when to lock into a pattern. As the material heats up, this vibration gets "soft" (sluggish) and "overdamped" (like a door hinge that is so rusty it won't move).
• The paper shows that this vibration becomes so sluggish between 200 K and 250 K that it essentially stops working as a clear signal, leading to that messy, fluctuating zone where the atoms can't decide what to do.

5. Why This Matters: No "Magic" Needed

For years, scientists thought this complex behavior required a "magic" ingredient called excitons (a special bond between electrons and holes, like a ghostly couple holding hands).

• The Twist: This paper shows you don't need the magic. You can explain the entire complex melting process just by looking at how the atoms shake and push against each other (electron-phonon coupling) and how they jiggle due to heat. It turns out the "dance floor" physics is complicated enough on its own.

Summary in a Nutshell

Imagine a group of soldiers marching in perfect formation.

1. Old View: As the sun gets hotter, they just get tired and slowly stop marching one by one until everyone is standing still.
2. New View: As the sun gets hotter, the soldiers start getting jittery. First, they all lean to the left (chirality). Then, the formation breaks into small, squabbling groups marching in different directions, separated by gaps where no one is marching (domains and defects). Finally, the heat gets so intense that they all just run around randomly.

This paper proves that the "squabbling groups" phase is real, caused by the natural jitters of heat, and that we don't need to invent invisible forces to explain it. It gives us a clearer map of how quantum materials behave when they get warm, which is crucial for designing future superconductors and quantum computers.

Technical Summary

1. Problem Statement

The charge density wave (CDW) phase in titanium diselenide (TiSe$_2$) has been a subject of intense debate for decades regarding its microscopic origin, structural symmetry, and thermal melting mechanism. Key unresolved issues include:

• Microscopic Origin: Whether the CDW is driven primarily by exciton condensation, electron-phonon coupling, or a combination of both.
• Symmetry and Chirality: The existence of a chiral structural order (breaking rotational and inversion symmetries) and whether it is an intrinsic ground state or induced by external perturbations.
• Melting Mechanism: Conventional theories suggest a second-order phase transition from a $P3c1$ structure to a high-symmetry $P3m1$ phase at $T_{CDW} \approx 200$ K. However, recent experiments (UED, STM, XRD) suggest a more complex, multi-step melting process involving topological defects, domain walls, and a "hexatic-like" intermediate phase, which standard mean-field theories fail to capture.
• Limitations of Existing Models: Phenomenological and downfolding models often neglect large-scale thermal fluctuations, while standard ab initio calculations (harmonic DFT) overestimate transition temperatures and miss the intermediate fluctuation regimes.

2. Methodology

To address these limitations, the authors employed large-scale Molecular Dynamics (MD) simulations driven by a highly accurate, first-principles-trained machine learning interatomic potential.

• Machine Learning Potential (MACE): They constructed a Moment-based Atomic Cluster Expansion (MACE) interatomic potential trained on Perdue-Burke-Ernzerhof (PBE) Density Functional Theory (DFT) data. The training dataset included 2,180 structures across various supercell sizes ($2\times2\times1$ to $8\times8\times1$) and temperatures (10–500 K).
• Simulation Scale: MD simulations were performed on a massive $56 \times 56 \times 1$ supercell (containing 9,408 atoms), allowing for the explicit modeling of long-wavelength fluctuations, domain formation, and topological defects that are inaccessible in smaller unit cells.
• Thermodynamic Conditions: Simulations were run in the NVT ensemble (Bussi thermostat) across a temperature range from 25 K to 450 K.
• Analysis Tools:
  • Structural Analysis: Calculation of average atomic displacements, radial distribution functions (RDF), and time-averaged structure factors.
  • Symmetry Analysis: Use of Spglib to determine space group symmetries of time-averaged structures.
  • Phonon Dynamics: Calculation of phonon dispersions and spectral functions using the DynaPhoPy framework (projecting atomic velocities onto harmonic modes) to identify soft modes and damping.
  • Comparison: Results were benchmarked against experimental XRD, Raman, Inelastic X-ray Scattering (IXS), and ultrafast electron diffraction (UED) data, as well as Stochastic Self-Consistent Harmonic Approximation (SSCHA) calculations.

3. Key Contributions

• Resolution of Multi-Step Melting: The study demonstrates that the CDW melting in monolayer TiSe$_2$ is not a conventional second-order transition. Instead, it undergoes a two-step process characterized by an extended fluctuation regime.
• Discovery of Chiral Order Origin: The authors reveal that the asymmetric 3Q chiral CDW order (with $C_2$ symmetry) is spontaneously stabilized by anisotropic long-wavelength thermal fluctuations (acoustic phonons), rather than requiring excitonic correlations or external perturbations.
• Mechanism of Fluctuation-Driven Transition: The paper establishes that thermal fluctuations (specifically acoustic phonons) are the primary drivers for the creation of topological defects and domain walls, leading to a hexatic-like melting process.
• Exclusion of Excitonic Necessity: By successfully reproducing experimental transition temperatures and dynamics using a potential trained solely on electron-phonon coupled DFT data (without explicit excitonic terms), the work suggests that non-perturbative electron-phonon and phonon-phonon interactions are sufficient to describe TiSe$_2$ CDW physics, challenging the necessity of exciton condensation theories.

4. Key Results

A. Two-Step Melting Process

The simulations identified two distinct critical temperatures:

1. $T^* \approx 200$ K: The onset of the fluctuation regime. Below this, a chiral 3Q order exists. Above this, long-range translational order breaks down, but short-range order persists.
2. $T_{CDW} \approx 250$ K: The complete melting of the CDW order into the normal $P3m1$ phase.

• Intermediate Regime ($200 < T < 250$ K): This region is characterized by the proliferation of topological defects and domain walls. The system exhibits a mix of symmetries ($P1$, $P\bar{1}$, $C2/m$) and fluctuating domain sizes, consistent with a hexatic-like phase.
• Discrepancy with Harmonic Models: Standard harmonic DFT and SSCHA calculations predicted a single second-order transition at much higher temperatures ($T_{CDW} \approx 400-1150$ K) and failed to capture the intermediate fluctuation regime, highlighting the necessity of large-scale anharmonic MD.

B. Chiral $C_2$ Symmetry and Anisotropy

• Low-Temperature Phase ($T < 150$ K): The system stabilizes in a chiral 3Q phase with $C_2$ symmetry. This phase features three non-equivalent CDW displacement vectors ($\delta_{Ti1} > \delta_{Ti2} > \delta_{Ti3}$), breaking both rotational and inversion symmetries.
• Origin of Chirality: The authors found that this chirality arises from anisotropic thermal fluctuations of long-wavelength acoustic phonons. When thermal fluctuations were removed (via relaxation of time-averaged structures), the system reverted to an achiral $P3c1$ phase.
• Chiral Transition: The chiral order breaks down around $T_{chiral} \approx 150$ K, preceding the full melting at 250 K. This aligns with recent non-linear optical and photocurrent experiments.

C. Phonon Dynamics and Softening

• Overdamped Soft Mode: The simulations revealed a completely overdamped soft optical phonon mode in the $200-250$ K range.
• Kohn Anomaly: The transverse optical (TO) phonon at the $q=M$ point shows a Kohn anomaly that softens to near-zero frequency at $T_{CDW} \approx 250$ K before hardening again.
• Correlation with Fluctuations: The region of maximum phonon softening and overdamping perfectly matches the fluctuation regime identified in the structural analysis, confirming that the soft mode is responsible for the CDW fluctuations.

D. Validation Against Experiment

• The simulated $T_{CDW} \approx 250$ K aligns with experimental values for monolayer TiSe$_2$.
• The time-averaged structure factors and atomic displacements match XRD and ARPES data, including the deviation from mean-field behavior below $T_{CDW}$.
• The phonon dispersion curves at 25 K and 200 K agree well with Raman and IXS experimental data.

5. Significance

• Unified Framework: This work provides a unified microscopic framework explaining the complex phase diagram of TiSe$_2$ without invoking excitonic correlations, attributing the phenomena to strong anharmonic lattice dynamics and thermal fluctuations.
• Generalizability: The findings suggest that similar fluctuation-driven, multi-step melting processes involving hexatic phases and chiral orders may be common in other 2D quantum materials (e.g., $1T$-TaS$_2$, $2H$-TaSe$_2$, kagome metals).
• Methodological Advance: It demonstrates the power of combining machine-learning potentials with large-scale MD simulations to capture emergent phenomena (like topological defects and domain walls) that are invisible to mean-field theories and small-cell ab initio methods.
• Implications for Superconductivity: Understanding the interplay between CDW fluctuations, domain walls, and lattice dynamics is crucial for unraveling the mechanisms of unconventional superconductivity in doped TiSe$_2$, where CDW fluctuations are believed to enhance superconducting pairing.