Angle evolution of the superconducting phase diagram in twisted bilayer WSe2

This study resolves the apparent discrepancy between superconducting phase diagrams in twisted bilayer WSe$_2$ at different twist angles by demonstrating a smooth evolution of superconductivity across the range, revealing its consistent proximity to Fermi surface reconstruction and antiferromagnetic ordering rather than specific singularities, thereby establishing the system as a versatile platform for investigating correlated phases.

The Gist

Imagine you have two pieces of a very special, ultra-thin fabric made of a material called Tungsten Diselenide (WSe₂). If you stack them perfectly on top of each other, they act like a normal sheet. But, if you twist one piece slightly relative to the other, something magical happens: the atoms line up to create a giant, repeating pattern called a Moiré pattern (think of the rippling effect you see when you hold two window screens slightly out of alignment).

This paper is about what happens when you twist these two sheets at different angles and turn the "knobs" on a machine to squeeze or stretch the electrons inside.

The Big Mystery: Two Different Maps?

Recently, scientists found that these twisted sheets could conduct electricity with zero resistance (superconductivity), which is like a train moving on a track with no friction at all. However, there was a problem.

• Team A looked at a twist of 3.65 degrees and found superconductivity in one specific spot, surrounded by an insulator (a material that stops electricity).
• Team B looked at a twist of 5.0 degrees and found superconductivity in a completely different spot, surrounded by a magnetic metal.

It was like looking at two different maps of the same city and seeing two different downtowns. Scientists were confused: Are these two different cities, or just the same city seen from different angles?

The Experiment: Filling in the Gaps

The authors of this paper decided to build a whole series of these twisted sheets, with angles ranging smoothly from 5.0 degrees down to 3.8 degrees. They wanted to see if the "downtown" (the superconducting zone) moved smoothly from one location to the other, or if it jumped around.

They used a dual-gate system, which is like having two hands that can squeeze the material from the top and bottom simultaneously. By squeezing (changing the electric field) and adding more electrons (changing the density), they could map out the entire landscape.

The Discovery: A Smooth Journey

Here is what they found, explained with a few analogies:

1. The "Magic Zone" Moves Smoothly
Imagine the superconducting state is a hotspot on a map.

• At 5.0 degrees, the hotspot is near a "Mountain Peak" (called a Van Hove Singularity) where electrons crowd together.
• As they twisted the angle down to 3.8 degrees, that hotspot didn't jump; it slid smoothly across the map to a new location.
• This proved that the superconductivity in the 3.65-degree sample and the 5.0-degree sample are actually the same phenomenon, just shifted by the angle of the twist. They are the same "city," just viewed from different perspectives.

2. The "Magnetic Neighbor"
In every single sample, no matter the angle, the superconducting hotspot was always right next to a Magnetic Zone (specifically, an Antiferromagnetic state where electron spins point up and down like a checkerboard).

• Analogy: Think of superconductivity as a dance party. The paper suggests that the "music" that gets the electrons dancing (pairing up to flow without resistance) is provided by the magnetic neighbors next door. Even though the location of the party changes with the angle, the DJ (the magnetic fluctuations) is always the same.

3. The "Insulator Trap"
At the steeper angles (like 5.0 degrees), the magnetic zone was "leaky"—electrons could still move through it, so it acted like a metal.
But as they twisted the angle tighter (to 3.8 degrees), the magnetic zone became a solid wall. It turned into a perfect insulator at the halfway point of electron filling.

• Analogy: Imagine a crowd of people. At a wide angle, they are jostling but moving. At a tight angle, they lock arms and freeze completely, blocking anyone from passing. This "locking" happens exactly when the electrons fill half the available spots.

4. The "Weak vs. Strong" Connection
The researchers also checked how "strongly" the electrons were interacting.

• At 5.0 degrees, the electrons were like polite strangers at a party, barely interacting (Weak Coupling). The superconductivity here is similar to traditional superconductors.
• At 3.8 degrees, the electrons got closer and started interacting more (Intermediate Coupling), but they didn't become a chaotic, strongly correlated mess (like a Mott insulator). They stayed in a "Goldilocks" zone—strong enough to be interesting, but not so strong that the physics breaks down.

Why Does This Matter?

This paper is a huge step forward because it unifies two confusing observations into one clear story.

• It's a Universal Rulebook: It shows that you can tune these materials like a radio dial. By changing the twist angle, you can smoothly transition the material from one type of behavior to another.
• The Origin of Superconductivity: It strongly suggests that in these materials, magnetism is the key to superconductivity. The electrons pair up because of magnetic fluctuations, not because of some exotic, unknown force.
• A New Playground: It establishes twisted WSe₂ as a perfect laboratory for scientists to study how materials behave when you change the balance between how much electrons want to move (bandwidth) and how much they want to push each other away (interaction).

In a nutshell: The scientists took two confusing maps of a superconducting city, filled in the streets between them, and realized it's all one continuous city. They found that the "magic" of zero-resistance electricity is always dancing with its magnetic neighbor, and by simply twisting the fabric of the material, you can control exactly where and how this dance happens.

Technical Summary

1. Problem Statement

Recent discoveries of superconductivity in twisted bilayer tungsten diselenide (tWSe2) have sparked significant interest, yet two initial studies reported seemingly distinct phase diagrams at different twist angles ($3.65^\circ$ and $5.0^\circ$):

• At $3.65^\circ$: Superconductivity appeared at half-filling ($\nu=1$) in low displacement fields, resembling high-$T_c$ cuprates, with a correlated insulator (Mott state) emerging at higher fields.
• At $5.0^\circ$: Superconductivity emerged at large displacement fields and densities greater than $\nu=1$, near a Van Hove singularity (VHs), adjacent to a metallic antiferromagnetic (AFM) state, suggesting a BCS-type pairing mechanism.

The central question addressed by this work is whether these two superconducting states share a common physical origin or represent distinct, unconnected phases. The authors aim to map the continuous evolution of the phase diagram across the twist angle range ($3.8^\circ$ to $5.0^\circ$) to determine the relationship between twist angle, correlation strength, and the mechanism of superconductivity.

2. Methodology

The study employs a combination of advanced experimental transport measurements and sophisticated theoretical modeling:

• Device Fabrication: The authors fabricated dual-gated tWSe2 devices using a standard "tear-and-stack" technique. They utilized high-quality hBN encapsulation and graphite/hBN gates to independently control carrier density ($n$) and displacement field ($D$). Devices were created with twist angles spanning $5.0^\circ, 4.8^\circ, 4.3^\circ, 4.2^\circ, 4.1^\circ,$ and $3.8^\circ$.
• Experimental Measurements:
  • Transport: Longitudinal resistance ($R$) was measured as a function of density and displacement field at base temperatures (dilution fridge, $\sim$20 mK).
  • Superconducting Characterization: Critical temperature ($T_c$) was defined at the 80% resistance drop. Berezinskii-Kosterlitz-Thouless (BKT) transition temperatures ($T_{BKT}$) were identified via the $V \propto I^3$ power law in I-V characteristics. Critical current density ($J_c$) was extracted from coherence peaks in differential resistance ($dV/dI$).
  • Gap Analysis: Activation energies for insulating states were determined via Arrhenius plots of conductivity.
• Theoretical Modeling:
  • Multi-orbital Wannier Model: A realistic three-orbital tight-binding model was constructed to capture the twist-angle-dependent band structure and quantum geometry.
  • Functional Renormalization Group (FRG): Used to identify leading instabilities (superconducting and magnetic) in the weak-to-moderate coupling regime, accounting for mutual feedback between interaction channels.
  • Hartree-Fock (HF) Calculations: Performed to estimate the size of the insulating gap at half-filling and to confirm the nature of the Fermi surface reconstruction.

3. Key Contributions and Results

A. Smooth Evolution of the Phase Diagram

The most significant finding is that the phase diagrams at $3.65^\circ$ and $5.0^\circ$ are not distinct but are part of a smooth, continuous evolution driven by the twist angle.

• Shift of Features: As the twist angle decreases from $5.0^\circ$ to $3.8^\circ$, the region of Fermi surface reconstruction (associated with AFM ordering) and the superconducting pocket continuously shift toward lower carrier densities and lower displacement fields.
• Convergence to Half-Filling: At larger angles ($5.0^\circ$), the AFM state occurs at incommensurate densities ($\nu < -1$) with a partially gapped Fermi surface. As the angle decreases, this region moves closer to $\nu = -1$ (half-filling). At $3.8^\circ$, the AFM region fully overlaps with half-filling, resulting in a fully gapped insulating state.

B. Universal Proximity to Antiferromagnetism

Superconductivity is observed in all devices exclusively at the boundary of the inter-valley coherent antiferromagnetic (IVC-AFM) phase.

• The superconducting pocket appears whether the AFM state is metallic (partially gapped) or insulating (fully gapped).
• The superconducting pocket is not strictly tied to the Van Hove singularity (VHs). At smaller twist angles, the superconducting region moves away from the VHs, yet superconductivity persists, indicating that the VHs is not the primary driver.
• This universal proximity to the AFM phase strongly suggests that spin fluctuations mediate the pairing mechanism across the entire angle range.

C. Correlation Strength and Coupling Regime

• Correlation Evolution: The ratio of the critical temperature to the Fermi temperature ($T_c/T_F$) increases as the twist angle decreases, indicating a move toward stronger correlations. However, the values remain relatively small ($T_c/T_F < 1\%$), suggesting the system remains in a weak-to-moderate coupling regime (BCS-like) rather than entering a strongly correlated Mott regime, even at the smallest angles measured.
• Superfluid Stiffness: Analysis of the superfluid stiffness places the material on the "BCS side" of the BCS-BEC crossover for all angles, contradicting hypotheses that the small-angle limit represents a BEC-like state.

D. The Half-Band Insulator

• At $5.0^\circ$, no insulating gap is observed at half-filling.
• At $3.8^\circ$, a clear metal-to-insulator transition (MIT) is observed at $\nu = -1$ within a specific displacement field range.
• Theoretical calculations confirm that this insulating state arises from the commensurability of the AFM order with the half-filled lattice, leading to a fully gapped Fermi surface. This behavior differs from a standard Mott insulator, as it is highly dependent on the displacement field and only appears when the AFM order overlaps with half-filling.

4. Significance

• Unification of Moiré Superconductivity: The paper resolves the apparent contradiction between previous studies of tWSe2, demonstrating that the "cuprate-like" and "BCS-like" behaviors are simply different points on a continuous phase diagram controlled by twist angle and displacement field.
• Mechanism Identification: By showing that superconductivity persists away from the Van Hove singularity but remains strictly adjacent to the AFM phase, the authors provide strong evidence that spin-fluctuation-mediated pairing is the dominant mechanism in tWSe2, similar to findings in twisted graphene systems.
• Tunability Platform: The study establishes twisted TMDs as a unique platform where the ratio of interaction strength to bandwidth can be continuously tuned via twist angle and electric fields, allowing for the systematic exploration of the crossover between weak and intermediate coupling regimes.
• Theoretical Validation: The close agreement between the experimental phase diagrams and FRG/Hartree-Fock calculations validates the theoretical framework used to describe correlated states in moiré materials, specifically the role of inter-valley coherence and Fermi surface reconstruction.

In conclusion, this work provides a comprehensive map of the electronic phases in twisted bilayer WSe2, revealing a unified origin for superconductivity driven by spin fluctuations and mediated by the evolution of the Fermi surface geometry as the twist angle is varied.