Superconductivity near two-dimensional Van Hove singularities: a determinant quantum Monte Carlo study

Using determinant quantum Monte Carlo simulations, this study reveals that while superconducting transition temperatures are modestly enhanced near Van Hove singularities in the weak-coupling regime, the global maximum $T_c$ for the two-dimensional attractive Hubbard model actually occurs at intermediate interaction strengths and electron densities unrelated to the non-interacting density of states features.

The Gist

The Big Question: How do we make superconductors better?

Imagine you are trying to get a huge crowd of people (electrons) to dance together perfectly in sync. In physics, this synchronized dancing is called superconductivity. When electrons dance in sync, they can move without any friction, meaning electricity flows with zero resistance. This is the "holy grail" of energy efficiency.

For decades, physicists have believed that the best way to get these electrons to dance is to find a specific spot in the city where the crowd naturally gets very dense. In physics, this spot is called a Van Hove Singularity (VHS).

Think of the Van Hove Singularity as a traffic jam.

• The Theory: If you can park your electrons right in the middle of a traffic jam, they are forced to be close together. The theory says, "If they are close enough, they will naturally pair up and start dancing."
• The Goal: Scientists want to build materials where this traffic jam happens naturally, hoping it will make superconductors work at higher temperatures (like room temperature, rather than needing near-absolute zero).

The Experiment: A Digital Simulation

The authors of this paper didn't build a physical lab. Instead, they built a massive digital simulation (using a method called Determinant Quantum Monte Carlo) to watch how electrons behave in a 2D grid.

They tested two types of "traffic jams":

1. The Ordinary Jam (Logarithmic VHS): A standard, heavy traffic jam.
2. The Super-Jam (Higher-Order VHS): An even worse, more chaotic traffic jam where the density of cars spikes even more violently.

They wanted to see: Does making the traffic jam worse (the Super-Jam) make the dancing (superconductivity) happen much better?

The Surprising Results

Here is what they found, broken down into three acts:

Act 1: The Weak Interaction (The "Light Traffic" Phase)

When the electrons don't push each other very hard (weak interaction), the theory holds up, but not as well as everyone hoped.

• The Finding: Being near the traffic jam does help the electrons pair up. It boosts the temperature at which they start dancing.
• The Twist: However, the boost is much weaker than the old math predicted. The "Super-Jam" (Higher-Order) didn't make the dancing significantly better than the "Ordinary Jam." It was like adding a few more cars to a traffic jam; it didn't change the flow much.

Act 2: The Medium Interaction (The "Pushy Crowd" Phase)

As the electrons start pushing against each other more (medium interaction), things get messy.

• The Finding: The "traffic jam" effect starts to fade. The electrons get so busy interacting with each other that they stop caring about the traffic jam.
• The Metaphor: Imagine a crowded dance floor. If the music is soft, people might cluster near the DJ booth (the traffic jam). But if the music gets loud and people start shoving each other, they stop looking at the DJ and just start bumping into whoever is closest. The specific location of the DJ no longer matters.

Act 3: The Strong Interaction (The "BEC" Phase)

When the electrons push very hard (strong interaction), the traffic jam becomes irrelevant.

• The Finding: The best place for the electrons to dance is nowhere near the traffic jam.
• The Surprise: The simulation showed that the absolute best superconductivity happens at a specific density of electrons that has nothing to do with the band structure or the traffic jam. It happens at a "Goldilocks" zone of interaction strength (not too weak, not too strong) and a specific crowd size that the old theories never predicted.

The "Aha!" Moment

The most important takeaway from this paper is a reality check for scientists trying to engineer better superconductors.

For a long time, the strategy was: "Let's build a material with a perfect, sharp traffic jam (Van Hove Singularity) and tune the electrons to sit right on top of it."

This paper says: "That strategy only works if the electrons are very polite (weak interaction)."

As soon as the electrons get a bit rowdy (stronger interaction), the traffic jam stops being the main factor. In fact, trying to force the electrons into that traffic jam might actually be a waste of time. The best superconductivity comes from finding the right balance of how hard the electrons push each other, regardless of where the "traffic jam" is.

Summary Analogy

Imagine you are trying to get a group of strangers to form a conga line.

• Old Idea: You think the best way is to stand them all in a narrow hallway (the Van Hove Singularity) so they are forced to hold hands.
• This Paper's Discovery:
  • If the people are shy (weak interaction), the hallway works okay.
  • But if the people are energetic and want to dance (strong interaction), putting them in a narrow hallway actually makes them trip over each other.
  • The best conga line forms in a wide-open room, but only if the music is at the perfect volume—not too quiet, not too loud.

Conclusion: To build better superconductors, we shouldn't just obsess over creating "traffic jams" in the material's design. We need to focus on the "music volume" (interaction strength) and the "room size" (electron density), because that's where the real magic happens.

Technical Summary

1. Problem Statement

The paper investigates the role of Van Hove singularities (VHS) in enhancing the superconducting transition temperature ($T_c$) in two-dimensional (2D) systems.

• Context: In weak-coupling Bardeen-Cooper-Schrieffer (BCS) theory, a logarithmic divergence in the density of states (DOS) at an ordinary VHS is predicted to significantly boost $T_c$ (scaling as $T_c \sim \exp(-1/\sqrt{\lambda})$). A "higher-order" VHS (HOVHS), characterized by a power-law divergence ($\sim |\epsilon|^{-1/4}$), is theoretically predicted to yield even stronger enhancements ($T_c \sim \lambda^4$).
• The Gap: While the effect of VHS is well-understood in the weak-coupling limit, its relevance in the intermediate-to-strong coupling regime remains unclear. In strong coupling, finite quasiparticle lifetimes smear the DOS singularities, and interaction effects renormalize the pairing vertex. It is unknown whether engineering VHSs is an effective strategy for achieving high-$T_c$ in strongly correlated materials (e.g., kagome superconductors, twisted bilayer graphene, ruthenates) where interactions are significant.
• Goal: To systematically track the evolution of $T_c$ from weak to strong coupling in the presence of both ordinary and higher-order VHSs using numerically exact methods.

2. Methodology

The authors employ Determinant Quantum Monte Carlo (DQMC) simulations on the 2D attractive Hubbard model.

• Model Hamiltonian:
  $$H = \sum_{k\sigma} \epsilon_k c^\dagger_{k\sigma} c_{k\sigma} + U \sum_i n_{i\uparrow} n_{i\downarrow}$$
  where $U < 0$ is the attractive interaction. The electronic dispersion $\epsilon_k$ includes first ($t$), second ($t'$), and third ($t''$) neighbor hoppings on a square lattice.
• Band Structure Tuning:
  • Ordinary VHS: Achieved with $t' = -0.3t$ and $t'' = 0$, producing a logarithmic DOS divergence.
  • Higher-Order VHS (HOVHS): Achieved with $t' = -0.3t$ and $t'' = 0.1t$, flattening the dispersion to produce a power-law divergence ($\sim |\epsilon|^{-1/4}$).
• Simulation Parameters:
  • System sizes up to $L=22$ ($N=484$ sites).
  • Temperatures down to $T/t \approx 0.03$.
  • Interaction strengths $|U|$ ranging from weak ($|U| \lesssim W/3$) to strong ($|U| \approx W$, where $W=8t$ is the bandwidth).
• Key Advantages: The attractive Hubbard model avoids the fermion sign problem, allowing for unbiased, numerically exact simulations over a wide range of densities and temperatures.
• Analysis Techniques:
  • $T_c$ Extraction: Determined via the Berezinskii-Kosterlitz-Thouless (BKT) criterion using the superfluid stiffness $\rho_s$ ($\rho_s(T_c^-) = \frac{2}{\pi}T_c$).
  • DOS Proxy: Used the imaginary-time Green's function $G(r=0, \tau=\beta/2)$ to estimate the interacting DOS.
  • Perturbation Theory: Performed second-order perturbation calculations (including self-energy and vertex corrections) to benchmark the DQMC results in the weak-to-intermediate coupling regime.

3. Key Contributions and Results

A. Weak-to-Intermediate Coupling ($|U| \lesssim W/3$)

• Enhancement is Real but Limited: $T_c$ is indeed enhanced near the VHS density compared to off-VHS densities. However, the enhancement is significantly weaker than predicted by unrenormalized weak-coupling BCS theory.
• Role of Corrections: The discrepancy is explained by self-energy and vertex corrections.
  • Self-energy: Smears the DOS, broadening the $T_c$ peak.
  • Vertex corrections: Reduce the effective pairing interaction strength.
  • The DQMC results align well with second-order perturbation theory that includes both corrections.
• Ordinary vs. Higher-Order: Transitioning from a logarithmic VHS to a power-law HOVHS yields only a minor additional enhancement of $T_c$. The "magic" of the HOVHS is largely suppressed by interaction effects even at moderate coupling.

B. Strong Coupling ($|U| \gtrsim W/3$)

• Rapid Crossover: As interaction strength increases beyond $|U| \approx W/3$, the influence of the VHS is rapidly washed out.
• Shift in Optimal Density: The density $n$ at which $T_c$ is maximized shifts away from the VHS density.
  • For $|U| \approx 6t$, the global maximum of $T_c$ occurs at $n \approx 1.35$, a density unrelated to any feature in the non-interacting band structure.
  • At very strong coupling ($|U| \gg W$), the system enters a preformed pair (BEC) regime where $T_c$ scales as $t^2/|U|$.
• Mechanism: The shift is consistent with a strong-coupling expansion where the system is described by an effective hard-core Bose Hubbard model. The optimal density in this regime is determined by the interplay of kinetic energy and interaction, not the DOS singularity.

C. Global Maximum

The study identifies that the absolute maximum $T_c$ for the model is achieved at intermediate coupling ($U \approx -6t$) and at a density far from the VHS ($n \approx 1.35$). This contradicts the naive strategy of simply tuning the Fermi level to a VHS to maximize $T_c$ in strongly correlated systems.

4. Significance and Implications

• Limitations of VHS Engineering: The paper demonstrates that the effectiveness of VHSs in enhancing superconductivity is essentially a weak-coupling phenomenon. In the intermediate and strong coupling regimes relevant to many high-$T_c$ materials, the benefits of DOS singularities are severely limited by interaction-induced renormalization.
• Re-evaluating Experimental Systems: The findings suggest that for materials like Sr$_2$RuO$_4$, kagome superconductors, or twisted bilayer graphene, the observed high $T_c$ near VHSs may not be solely due to the DOS divergence. Instead, it may arise from other mechanisms or the systems may lie just on the weak-coupling side of the crossover.
• Theoretical Insight: The work bridges the gap between weak-coupling BCS theory and strong-coupling BEC physics, showing that the transition from "Fermi-surface pairing" (VHS driven) to "local pairing" (interaction driven) is abrupt rather than gradual.
• Guidance for Material Design: Strategies to achieve higher $T_c$ in correlated 2D systems should not rely solely on engineering DOS singularities. Instead, optimizing the interplay between interaction strength and band structure parameters (hopping terms) is crucial.

Conclusion

Using numerically exact DQMC simulations, the authors conclude that while Van Hove singularities do enhance superconductivity, this effect is strongly suppressed by electron-electron interactions. The optimal conditions for superconductivity in this model are found at intermediate coupling strengths and densities distinct from the VHS, challenging the paradigm that maximizing the DOS at the Fermi level is the universal route to high-$T_c$ superconductivity in strongly correlated materials.