Finite-temperature formation of magnetic plateaus and simplex liquid states on the frustrated ruby lattice

Using infinite tensor network states optimized with belief propagation, this study reveals that the frustrated spin-1/2 Heisenberg antiferromagnet on the ruby lattice forms stable magnetic plateaus hosting a novel "simplex liquid state" with residual entropy at low temperatures, a process that occurs continuously without a phase transition as the system cools.

The Gist

Imagine a crowded dance floor where everyone is trying to find a partner, but the rules of the dance are tricky. This is the story of a new paper about a special kind of magnetic material called the "ruby lattice."

Here is the breakdown of what the scientists found, using simple analogies:

The Tricky Dance Floor (Geometric Frustration)

In normal magnets, tiny atomic spins act like tiny compass needles that all want to line up in the same direction (like soldiers marching). But in this specific "ruby lattice" structure, the geometry is so twisted that the spins can't all be happy at once. It's like a game of musical chairs where there are more chairs than people, but the chairs are arranged in a way that makes it impossible for everyone to sit comfortably without bumping into someone else. This is called geometric frustration.

Usually, when you cool these materials down, they get frustrated and eventually "snap" into a rigid, ordered pattern (like a crystal) to solve the problem. But the scientists wanted to see what happens if you cool them down very slowly and carefully.

The Magic of "Belief Propagation"

To figure this out, the researchers used a powerful computer method called Tensor Networks, specifically a technique called Belief Propagation (BP).

Think of Belief Propagation like a rumor spreading through a large crowd. Instead of asking every single person in the room what they are doing, you ask a few people, who tell their neighbors, who tell their neighbors, and so on. Eventually, everyone has a good idea of what the whole group is doing without needing to check every single person. The researchers used this "rumor-mongering" math to simulate how these magnetic spins behave at different temperatures, even when the system is infinitely large.

The Surprise: No "Snap," Just a "Liquid"

When they cooled the system down, they expected the spins to suddenly "snap" into a rigid, ordered crystal (a phase transition). Instead, they found something much more fluid.

As the temperature dropped, the spins didn't freeze into a single pattern. Instead, they formed a "Simplex Liquid State."

• The Analogy: Imagine a group of people at a party. Instead of everyone standing in a perfect grid (a crystal), they form small, tight-knit groups of three (called "simplices"). These groups dance together, but the arrangement of the groups keeps changing.
• The Result: Even at very low temperatures, the system remains disordered. It's a "liquid" of these dancing groups. Because there are so many different ways these groups can arrange themselves, the system retains a lot of "residual entropy" (a measure of disorder). It's like having a deck of cards that is shuffled perfectly every time you look at it, never settling on one specific order.

The Magnetic Plateaus

The researchers also turned on a magnetic field (like a strong wind blowing across the dance floor). As they increased the wind, the spins tried to align with it.

Instead of smoothly turning, the spins got stuck at specific "plateaus."

• The Analogy: Imagine a staircase. As you push the system harder, the magnetization (how much it aligns) jumps up, then stays flat for a while (a plateau), then jumps again.
• They found stable "steps" where the magnetization was exactly 1/3, 1/2, or 2/3 of the maximum possible.
• The Twist: Even on these flat "steps," the material didn't become a rigid crystal. It stayed in that "liquid" state, just with a specific average alignment.

The "Lambda" Peak and the Switch

There was one very interesting moment near the middle of the staircase (the 1/2 plateau).

• The Analogy: Imagine the dance floor is split in two. On one side, the groups are dancing one way; on the other, they are dancing another. At a specific temperature and wind speed, the whole floor suddenly switches from one style of dancing to the other.
• This switch wasn't smooth. It created a sharp spike in the "heat capacity" (how much energy the system absorbs), shaped like the Greek letter Lambda (λ). This suggests that at the very edge of these plateaus, the system is on the brink of a major change, driven by quantum fluctuations.

The Big Takeaway

The most important finding is that this system never actually "freezes" into a traditional crystal.

Even as it gets incredibly cold, it stays in a disordered, liquid-like state filled with many possible arrangements. The scientists proved this by showing that the "heat capacity" (a measure of how the system reacts to temperature changes) stays smooth and continuous. If the system had frozen into a crystal, there would have been a sharp, jagged spike indicating a phase transition. Instead, it flowed smoothly into this new, exotic state.

In short: The researchers used a clever "rumor-spreading" math trick to show that a frustrated magnetic material doesn't freeze into a solid crystal when cooled. Instead, it turns into a "liquid" of dancing spin-groups that stays disordered and full of possibilities, even at temperatures near absolute zero.

Technical Summary

Technical Summary: Finite-Temperature Formation of Magnetic Plateaus and Simplex Liquid States on the Frustrated Ruby Lattice

Problem Statement
Geometric frustration in quantum systems is known to stabilize unconventional phases, such as quantum spin liquids, which avoid traditional magnetic ordering at low temperatures. However, understanding the formation of these states at finite temperatures and their thermodynamic signatures remains a significant challenge. Specifically, the zero-temperature ($T=0$) ground state of the spin-1/2 Heisenberg antiferromagnet on the ruby lattice with next-nearest-neighbor interactions was previously shown to exhibit fractionalized magnetic plateaus ($m_z = 0, 1/3, 1/2, 2/3$). These plateaus were interpreted as discrete crystalline phases breaking lattice symmetries. Open questions remained regarding the exact ground state degeneracy, the finite-temperature stability of these plateaus, and whether the system undergoes conventional phase transitions to reach them. Resolving these issues requires simulations that bridge computational results with experimental realizations where thermal and quantum fluctuations compete.

Methodology
The authors employ a purified infinite tensor network state (iTNS) ansatz to represent the unnormalized square root of the equilibrium density matrix, $\rho^{1/2}(\beta) = \exp(-\beta H/2)$, where $\beta = 1/T$. The study focuses on the isotropic case ($J_t = J_h = J_d = J$) of the Heisenberg Hamiltonian on the ruby lattice.

• Lattice Representation: The vertices of the triangular units are grouped into triangles (simplices), mapping the ruby lattice to a coarse-grained honeycomb geometry with coordination number $z=3$. The iTNS utilizes a two-site unit cell, where each tensor possesses six physical indices (representing the three spins in each triangle) and three virtual indices.
• Algorithm: The system is initialized at infinite temperature ($\rho^{1/2}(0) = I$) and cooled via imaginary time evolution using a second-order Trotterization of the propagator.
• Observables: Thermodynamic quantities are calculated using two approaches:
  1. Belief Propagation (BP): Standard BP contraction is applied to the norm network to compute expectation values.
  2. Loop-Corrected BP: To improve accuracy beyond the mean-field-like BP approximation, the authors apply first-order cluster corrections (loop corrections) to the free energy density. This accounts for correlations around the hexagons of the lattice.
• Scale: The simulations utilize GPU hardware to reach bond dimensions up to $\chi \sim 200$, allowing for accurate measurement of thermodynamic observables in the thermodynamic limit across a wide temperature range.

Key Results

1. Simplex Liquid State at Zero Field:

   • Upon cooling from infinite temperature, the system does not undergo a phase transition to a crystalline order. Instead, it forms a disordered "simplex liquid state" characterized by a non-zero residual entropy ($s_0 \approx \frac{1}{3}\lambda$, where $\lambda$ is the residual entropy of dimer coverings on a honeycomb lattice).
   • The heat capacity ($C_v$) remains finite and continuous at all temperatures, with no diverging peaks indicative of a phase transition.
   • Analysis of the reduced density matrix reveals a leading eigenvalue $\lambda_1 \approx 1/3$, corresponding to a state where two simplices are strongly paired. The remaining eigenvalues correspond to weakly correlated states. The system is a perfect disordered mixture of an exponentially large number of crystalline configurations $|\psi_d\rangle$ involving paired spin simplices.
   • The energy density of this liquid state ($e/J \approx -0.498$) matches the ground state energy of specific crystalline configurations found in previous $T=0$ studies, but the system remains in a translationally invariant liquid state rather than selecting a single crystal.

2. Magnetic Plateaus and Field-Induced Behavior:

   • In the presence of an external magnetic field, the system exhibits stable magnetic plateaus at magnetization densities $m_z = 0, 1/3, 1/2,$ and $2/3$.
   • Thermodynamic Signatures: The formation of these plateaus occurs smoothly without conventional phase transitions. The heat capacity remains continuous, though shallow peaks are observed.
   • Residual Entropy: The plateaus host "simplex liquid states" with extensive degeneracy. The $m_z = 1/2$ plateau, in particular, exhibits a large residual entropy (approximately double that of the $m_z=0$ plateau).
   • Stability: The $m_z = 0$ plateau is the most stable against thermal fluctuations. The $m_z = 1/3$ plateau is less stable, appearing only in a narrow field range at very low temperatures ($T/J \le 10^{-2}$). The $m_z = 2/3$ plateau is observed in ground state calculations but requires temperatures lower than currently accessible in the finite-temperature simulation to be fully resolved.
   • Phase Switching: At the interface between the $m_z = 1/2$ and $m_z = 1/3$ plateaus, the system exhibits a thermally driven switching between liquid states. This is marked by an asymmetric "lambda" peak in the heat capacity and a sharp increase in magnetic susceptibility, likely reflecting critical fluctuations near a zero-temperature quantum phase transition.

Significance and Claims
The paper demonstrates that tensor network techniques, specifically infinite tensor network states optimized with belief propagation and loop corrections, provide a powerful and scalable route to understanding frustrated quantum magnets at finite temperatures.

• Validation of Method: The work validates the BP-based tensor network approach as capable of accurately capturing thermodynamic properties over many orders of magnitude in temperature, including the detection of energy gaps and residual entropies.
• Nature of the Ground State: The study clarifies that the magnetic plateaus on the ruby lattice are not simple crystalline orders but rather "simplex liquid states"—disordered mixtures of exponentially many crystalline configurations that retain non-zero residual entropy and do not break lattice symmetries.
• Absence of Phase Transitions: A key finding is that the system reaches these stable magnetic phases without undergoing finite-temperature phase transitions, as evidenced by the continuous and finite heat capacity.
• Future Directions: The authors suggest that their methodology can be extended to study three-dimensional frustrated systems, such as the pyrochlore lattice, and to investigate perturbations that might break the ground state degeneracy in favor of topologically ordered superpositions.