⚛️ high-energy theory

Global phase diagram of two-dimensional dirty hyperbolic Dirac liquids

This study establishes a global phase diagram for two-dimensional dirty hyperbolic Dirac liquids, demonstrating that negative spatial curvature stabilizes massless Dirac excitations against weak disorder before driving the system through a continuous transition into a metallic state and finally into an Anderson insulator at strong disorder.

Original authors: Christopher A. Leong, Daniel J. Salib, Bitan Roy

Published 2026-08-11

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Christopher A. Leong, Daniel J. Salib, Bitan Roy

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 a world where the rules of the road are written not in flat, straight lines, but on a surface that curves away from you forever, like the inside of a saddle or a Pringles chip. This is the realm of "hyperbolic" geometry, a strange mathematical landscape that physicists are starting to explore to see how tiny particles behave when space itself is bent. In this world, particles called "Dirac fermions" act like massless, super-fast messengers, zipping around without getting stuck. Usually, scientists believe that if you throw enough random obstacles—like dirty spots or impurities—into a 2D world, these fast particles will eventually get lost, stop moving, and turn into a stuck, insulating block of matter. This is a well-known rule for flat worlds, like the honeycomb pattern of a beehive. But what happens if the floor beneath them is constantly curving? Does the bend in space change the rules of the game?

This paper dives into that exact question, simulating a massive, curved grid of atoms to see how "dirty" it can get before the particles stop dancing. The researchers found that on this curved stage, the particles are surprisingly tough. They can handle a surprising amount of messiness before they give up, passing through a "metallic" phase where they flow freely, before finally getting stuck. This discovery challenges the old idea that any amount of dirt in a 2D system will immediately freeze the particles, suggesting that the curvature of space itself acts like a shield, keeping the quantum dance going for a while longer than expected.

The Curved Dance Floor

To understand what the authors did, picture a giant, infinite dance floor made of tiles. In our normal, flat world, a honeycomb pattern (like a beehive) is a perfect fit. But on this special dance floor, the tiles are arranged in a way that creates a constant, negative curve—imagine trying to lay a flat sheet of paper over a saddle; it would wrinkle and buckle. The scientists built a digital version of this curved floor, specifically a pattern called a {10, 3} lattice, where every point connects to three neighbors, and the shapes are ten-sided polygons.

On this floor, they placed "spinless fermions," which are like tiny, invisible dancers who don't spin but hop from one tile to the next. In a perfect, clean world, these dancers move in a special way that creates a "Dirac fluid," a state where they zip around without mass, and the number of available spots for them to stand (called the Density of States, or DOS) drops to zero right in the middle of their energy range.

The Experiment: Throwing in the Chaos

The researchers then started throwing "impurities" onto the floor. Think of these as random, sticky spots or potholes that appear at random locations, making it harder for the dancers to hop. They wanted to see: How much dirt can this curved dance floor take before the dancers get stuck?

They ran massive computer simulations on grids containing more than 100 million sites (for the average behavior) and over a million sites (for the typical behavior). They used a clever mathematical trick called the "kernel polynomial method" to calculate how the dancers moved as they increased the strength of the dirt, denoted by a number $W$ .

The Three Acts of the Story

The simulation revealed a fascinating three-act play that happens as the dirt gets thicker:

The Clean, Curved Semimetal (Weak Dirt): When the dirt is very light (up to a strength of about $W = 0.50$ ), the dancers are unbothered. Even with some random bumps, the "average" and "typical" number of spots available for them at zero energy remains zero. The system stays a "semimetal," a stable state where the particles remain massless and free. This is the first surprise: in a flat world, even a tiny bit of dirt would usually ruin this delicate state, but on the curved floor, the dancers hold their ground.
The Diffusive Metal (Moderate Dirt): As the dirt gets stronger, between $W = 0.50$ and roughly $W = 10.0$ , something magical happens. The system undergoes a smooth transition into a "metallic" state. Now, the number of available spots becomes finite. The dancers are no longer massless; they have gained some "weight" and are moving in a diffusive, hopping manner, but they are still flowing freely. They haven't stopped; they've just changed their dance style.
The Anderson Insulator (Heavy Dirt): Finally, when the dirt gets very heavy (around $W = 10.0$ ), the dancers finally get stuck. This is the "Anderson insulator" phase. Here, the "typical" number of available spots drops back to zero, meaning the dancers are localized, trapped in place by the overwhelming mess. The flow stops completely.

Why Curvature Matters

The most exciting part of the story is what happens when you compare this to a flat world. The researchers also simulated a flat honeycomb lattice (the standard beehive pattern). On this flat floor, even the tiniest amount of dirt caused the dancers to get stuck immediately, turning the system into an insulator right away. There was no "metallic" middle ground.

This proves that the unique behavior on the curved floor isn't a fluke; it's caused entirely by the negative curvature of the space itself. The bending of the universe acts as a protective shield, allowing the particles to survive a moderate amount of chaos that would destroy them on a flat surface.

The Numbers and the Certainty

The authors are very specific about their findings. They found a critical point for the first transition (from semimetal to metal) at $W_{c,1} = 0.50 \pm 0.05$ . The second critical point (from metal to insulator) occurs at $W_{c,2} = 10.0 \pm 1.0$ . These numbers come from simulations on grids with over $10^8$ sites, which is huge for this kind of calculation.

They also calculated how the system behaves right at the tipping points, finding specific "exponents" that describe the transition. For the jump from semimetal to metal, the exponent for the typical density of states was found to be $\beta_t = 1.014 \pm 0.028$ . For the jump to the insulator, the exponent was $\beta_A = 1.873 \pm 0.093$ . These numbers suggest that the way the particles behave during these transitions is unique to this curved world and different from what we see in 3D flat systems.

The Bigger Picture

While the paper doesn't claim to have built a physical machine yet, it suggests that this behavior could be tested in "metamaterials"—artificial structures made of light or sound rather than atoms. Imagine a photonic lattice (a grid of light) shaped like a hyperbolic surface. If you shine light through it and add some disorder, you might see these same transitions: light flowing freely, then getting stuck, all because the space it travels through is curved.

In short, this paper suggests that if you live on a curved surface, the rules of physics are more forgiving. You can handle a lot more mess before everything comes to a halt. It's a reminder that the shape of the stage matters just as much as the actors on it.

Technical Summary: Global Phase Diagram of Two-Dimensional Dirty Hyperbolic Dirac Liquids

Problem Statement
The study investigates the stability and phase behavior of massless Dirac fermions living on two-dimensional (2D) bipartite hyperbolic lattices in the presence of random pointlike charge impurities. While the nearest-neighbor tight-binding model (NN-TBM) on Euclidean lattices (e.g., honeycomb) is known to yield massless Dirac fermions with a linearly vanishing density of states (DOS), the behavior of such "hyperbolic Dirac liquids" under disorder remains an open question. Specifically, the authors aim to determine whether the background negative spatial curvature of hyperbolic space alters the conventional wisdom regarding disorder in 2D orthogonal symmetry class systems, which typically predicts that any amount of disorder leads to an Anderson insulator.

Methodology
The authors employ the canonical NN-TBM for spinless fermions on {10, 3} hyperbolic lattices (Schläfli symbol, where $p=10$ and $q=3$ ) with open boundary conditions (OBC). The system is modeled with a real nearest-neighbor hopping amplitude $t$ (set to unity) and a random on-site potential $V(r_i)$ drawn from a box distribution $[-W/2, W/2]$ , where $W$ represents the disorder strength.

To analyze the spectral properties, the authors utilize the Kernel Polynomial Method (KPM). They compute:

Average DOS ( $\rho_a(E)$ ): Defined as the disorder-averaged sum of delta functions over eigenenergies.
Typical DOS ( $\rho_t(E)$ ): Defined as the geometric mean of the local DOS over a $p$ -gon, which serves as a more sensitive indicator of localization than the average DOS.

Numerical simulations are performed on sufficiently large systems to approach the thermodynamic limit, with the largest {10, 3} lattices containing over $10^8$ sites for average DOS calculations and over $10^6$ sites for typical DOS calculations. The study also includes comparative calculations on honeycomb lattices ({6, 3}) with periodic boundary conditions (PBC) to isolate the effects of curvature. The system size is characterized by the total generation number ( $n_{tot}$ ), where the central polygon is generation 1.

Key Contributions and Results
The paper establishes a global phase diagram for dirty hyperbolic Dirac fluids, identifying three distinct stable phases separated by two continuous quantum phase transitions (QPTs):

Disordered Semimetal Phase (Weak Disorder):
For disorder strengths $W \lesssim 0.50$ , the system remains a stable semimetal. Both the average DOS $\rho_a(0)$ and the typical DOS $\rho_t(0)$ vanish at zero energy. This indicates that the ballistic quasiparticles are stable against sufficiently weak disorder, a finding that contrasts with the behavior of 2D Euclidean Dirac systems.
Diffusive Metal Phase (Moderate Disorder):
At a critical disorder strength $W_{c,1} = 0.50 \pm 0.05$ , the system undergoes a continuous QPT into a diffusive metallic state. In this phase, both $\rho_a(0)$ and $\rho_t(0)$ become finite. This transition is unique to hyperbolic Dirac systems; it is not observed in their flatland (Euclidean) counterparts.
Anderson Insulator Phase (Strong Disorder):
As disorder increases further, a second QPT occurs at $W_{c,2} = 10.0 \pm 1.0$ . Beyond this point, the system enters an Anderson insulating phase. Here, $\rho_t(0) \to 0$ while $\rho_a(0)$ remains finite (non-critical), signifying the localization of wavefunctions.

Critical Exponents and Scaling
The authors compute universal critical exponents near the transition points:

Semimetal-to-Metal Transition ( $W_{c,1}$ ):
- DOS exponent: $\alpha_a \approx 1.028$ .
- Order parameter exponents: $\beta_a \approx 1.628$ (for $\rho_a$ ) and $\beta_t \approx 1.014$ (for $\rho_t$ ). The discrepancy between $\beta_a$ and $\beta_t$ suggests wavefunction multifractality across this transition.
Metal-to-Insulator Transition ( $W_{c,2}$ ):
- Order parameter exponent: $\beta_A \approx 1.873$ (derived from the scaling of $\rho_t(0)$ ).

Significance and Claims
The authors assert that their findings are solely attributable to the background negative spatial curvature of the hyperbolic plane. This is supported by control calculations on honeycomb lattices, where $\rho_t(0)$ vanishes for arbitrarily weak disorder as the system size approaches the thermodynamic limit, confirming the standard Anderson insulating behavior for 2D Euclidean orthogonal class systems.

The study challenges traditional wisdom regarding 2D disordered systems by demonstrating that:

A stable disordered semimetal with vanishing DOS can exist in 2D.
A disorder-driven semimetal-to-metal QPT is possible in 2D.
The global phase diagram of 2D dirty hyperbolic Dirac liquids closely resembles that of 3D Euclidean disordered Dirac-Weyl systems, yet the critical exponents differ, indicating distinct universality classes driven by spatial curvature.

The authors note that while they provide a qualitative explanation based on quasiparticle lifetime scaling and self-consistent Born approximation, a rigorous field-theoretic justification for the existence of the stable dirty semimetal and the specific QPTs remains an open task for future work. They suggest that classical metamaterials, particularly photonic lattices capable of supporting hyperbolic tessellations, could serve as experimental testbeds for these theoretical predictions.