Tracing Flumen\u2019s Topological Phases in Twisted Bilayer Graphene

Twisted bilayer graphene (TBG) at small twist angles exhibits a dazzling array of correlated and topological phases. For researchers already comfortable with the basics of moiré physics, the challenge is no longer whether topological phases exist, but how to reliably identify and characterize them in experiment and simulation. This guide distills the practical workflow—from band structure diagnostics to transport signatures—used to trace topological phases in TBG, with an emphasis on the pitfalls and decisions that separate a robust assignment from a false positive.

Why Topological Phase Assignment Fails Without a Systematic Approach

Assigning a topological phase in TBG is not a single measurement or calculation; it is a chain of inferences that can break at multiple points. The most common failure mode is conflating a trivial band inversion with a topological transition. In TBG, the narrow bandwidth and strong correlations can produce gap openings that mimic topology but originate from spontaneous symmetry breaking or strain. Without a systematic protocol, teams often report a quantum anomalous Hall (QAH) effect that later turns out to be a ferromagnetic insulator with edge currents from magnetic domains rather than intrinsic topology.

Another frequent issue is misidentifying the Chern number from transport data. The Hall conductance in TBG can be quantized even in trivial phases due to disorder-induced percolation or multi-carrier effects. We have seen published results where a plateau at σ_xy = ±e²/h was attributed to a Chern insulator, but subsequent thermal transport or local probe measurements revealed a trivial origin. The cost of a false positive is not just a retraction—it misdirects the field and wastes months of follow-up work.

This guide is for experimentalists and theorists who need a robust, step-by-step protocol to avoid these traps. We assume you have a working TBG device or a tight-binding model and want to answer: Is this a topological phase, and if so, what is its Chern number? The approach we outline combines band structure analysis, symmetry indicators, and transport diagnostics into a coherent workflow that has been refined across multiple research groups.

Prerequisites: What You Need Before Starting the Trace

Before you can trace a topological phase, you need a well-characterized TBG sample or model. For experimentalists, this means a device with a known twist angle (within ±0.1° of the target), minimal strain (less than 0.3% uniaxial), and clean interfaces. The twist angle is the single most critical parameter: even a 0.2° deviation from the magic angle (θ ≈ 1.1°) can suppress the flat bands and destroy the topological phases. We recommend using a combination of Raman spectroscopy, atomic force microscopy (AFM), and transport measurements to confirm the angle. For theorists, the prerequisite is a tight-binding model that accurately reproduces the band structure at the relevant twist angle, including the effects of relaxation and strain. The continuum model of Bistritzer and MacDonald is a good starting point, but it neglects lattice relaxation, which can shift the phase boundaries by up to 10%.

You also need a clear definition of what constitutes a topological phase in TBG. The most studied phases are the Chern insulator (C = ±1, ±2), the quantum spin Hall (QSH) phase (Z₂ invariant), and the fractional Chern insulator (FCI). Each requires different diagnostics. For Chern insulators, the key signature is a quantized Hall conductance at zero magnetic field, accompanied by edge states in scanning tunneling microscopy (STM) or nonlocal transport. For QSH phases, you need to confirm helical edge states that are protected by time-reversal symmetry. For FCI, the hallmark is a fractional Hall conductance (e.g., σ_xy = (1/3)e²/h) at zero field, which is extremely rare and requires careful temperature and density tuning.

Finally, ensure your measurement setup can access the relevant parameter space: back-gate voltage (or doping) to sweep the chemical potential through the flat bands, temperature down to at least 1.5 K (ideally 300 mK for FCI), and magnetic fields up to a few tesla to break time-reversal symmetry when needed. Many topological phases in TBG appear only in a narrow window of doping and twist angle, so high-resolution control is non-negotiable.

Core Workflow: From Band Structure to Topological Invariant

The workflow for tracing a topological phase in TBG proceeds through four stages: (1) band structure calculation or measurement, (2) identification of band inversions and gap closings, (3) computation or measurement of topological invariants, and (4) consistency checks with transport and local probes.

Stage 1: Obtain the Band Structure

For theorists, this means diagonalizing the tight-binding or continuum Hamiltonian at the target twist angle. Pay attention to the moiré Brillouin zone (mBZ) and the band ordering at the Γ, K, and K' points. In the magic-angle regime, the flat bands are isolated from the remote bands by a gap of about 30–40 meV. If this gap is absent or smaller than 10 meV, the system is likely in a metallic or semimetallic state, and topological phases are unlikely. For experimentalists, the band structure is inferred from the density of states (DOS) measured by STM or from the quantum oscillations in transport. A clear gap in the DOS at charge neutrality (CNP) is a prerequisite for a topological insulator phase.

Stage 2: Identify Band Inversions

Topological phases in TBG arise from band inversions at the mBZ high-symmetry points. In the continuum model, the inversion at the Γ point between the two flat bands is what drives the Chern insulator phase at integer fillings. To detect an inversion, plot the band parity eigenvalues or the Berry curvature across the mBZ. A sign change in the Berry curvature at a high-symmetry point indicates a band inversion. For experimentalists, band inversions can be inferred from the sign of the Hall conductance as a function of gate voltage: a sign change near a gap opening suggests a topological transition.

Stage 3: Compute the Topological Invariant

For Chern insulators, the Chern number is computed by integrating the Berry curvature over the mBZ. In tight-binding models, this is straightforward using the Fukui-Hatsugai-Suzuki method. For experimentalists, the Chern number is extracted from the Hall conductance plateau: σ_xy = C × e²/h. However, beware of trivial plateau contributions from disorder or multi-band transport. A robust check is to measure the Hall conductance as a function of magnetic field: a true Chern insulator will show a quantized plateau that is independent of field (except at very low fields where the edge states may be gapped). For QSH phases, the Z₂ invariant can be computed from the parity of the Kramers pairs at the time-reversal invariant momenta. Experimentally, QSH phases are identified by a quantized longitudinal conductance of 2e²/h in a two-terminal measurement, combined with a vanishing Hall conductance.

Stage 4: Consistency Checks

No single measurement or calculation is sufficient. Always cross-check with at least two independent techniques. For example, if transport suggests a Chern number of 1, confirm with nonlocal transport (which should show a large nonlocal signal due to edge states) and with STM (which should reveal edge states at the sample boundaries). For theorists, cross-check the Chern number with the Wilson loop method and verify that the gap remains open under small perturbations (e.g., 1% strain). A topological phase that disappears under tiny strain is likely fragile and may not be observable in real samples.

Tools and Setup: What You Actually Need in the Lab or Code

For experimentalists, the essential tools are a low-temperature transport setup with a high-impedance voltmeter (to avoid current leakage through edge states) and a scanning tunneling microscope with energy resolution below 1 meV. The transport setup must allow for both two-terminal and four-terminal measurements, with the ability to switch between longitudinal and Hall configurations without breaking vacuum. We recommend using a lock-in amplifier with a frequency below 100 Hz to minimize capacitive coupling. For STM, the tip must be stable within 1 pm to resolve the moiré pattern and edge states. A common mistake is using a tip that is too blunt, which averages over multiple moiré cells and washes out the edge state signature.

For theorists, the primary tool is a tight-binding code (e.g., PythTB, TBmodels, or Wannier90) that can handle large unit cells. TBG at the magic angle requires a moiré unit cell with thousands of atoms, so efficient diagonalization is critical. We recommend using a continuum model for initial screening (it is faster) and then switching to a tight-binding model for accurate topological invariants. The continuum model often misses the effect of lattice relaxation, which can open small gaps that change the topological phase. A good practice is to include a relaxation correction using the method of Carr et al. (2019) or to use a fully relaxed tight-binding model from the literature.

One tool that is often overlooked is symmetry analysis software, such as the Bilbao Crystallographic Server or the Python library symmetry-representation. These tools can compute the symmetry indicators of the band structure from first principles, which can predict the topological phase without computing the full Chern number. For TBG, the symmetry indicators at the Γ and K points are particularly informative: a nontrivial indicator at Γ often signals a Chern insulator, while a nontrivial indicator at K suggests a QSH phase. Using symmetry indicators as a first pass can save hours of computation.

Variations for Different Constraints: Twist Angle, Strain, and Substrate Effects

The topological phase diagram of TBG is highly sensitive to the twist angle. At the magic angle (θ ≈ 1.1°), the flat bands are narrowest, and the interaction-driven phases (Chern insulator, FCI) are most robust. As you move away from the magic angle by ±0.2°, the bandwidth increases, and the topological phases become weaker or disappear entirely. For example, at θ = 1.3°, the Chern insulator phase at filling ν = ±2 (relative to CNP) is replaced by a trivial insulator. If your sample has a twist angle outside the magic-angle window, consider applying uniaxial strain to tune the bands back into the topological regime. Uniaxial strain of 0.5–1% can effectively shift the magic angle by up to 0.2°, but it also breaks the C₃ symmetry, which may destroy the topological protection. We have seen cases where strain-induced topological phases are actually trivial due to the loss of symmetry.

Substrate effects are another major variable. TBG on hexagonal boron nitride (hBN) is the standard, but the alignment between the hBN and the TBG can introduce a moiré potential that gaps the Dirac points and modifies the topological phases. If the hBN is aligned within 1° of the TBG, a secondary moiré pattern appears with a period of about 15 nm, which can induce a gap of up to 20 meV at the CNP. This gap is trivial (it comes from the substrate potential) and can be mistaken for a topological gap. To distinguish, measure the gap as a function of magnetic field: a topological gap will show edge states, while a substrate-induced gap will not. Alternatively, use a device with hBN that is intentionally misaligned by more than 5° to suppress the secondary moiré.

For devices with a dual-gate geometry, the displacement field (D) is an additional tuning knob. Applying a perpendicular electric field breaks the layer symmetry and can induce a transition from a trivial insulator to a Chern insulator at integer fillings. The critical displacement field for this transition is typically around 0.5 V/nm. If your device cannot reach this field, you may not see the topological phase. We recommend designing devices with a thin hBN dielectric (20–30 nm) to maximize the displacement field for a given gate voltage.

Finally, consider the role of disorder. In TBG, disorder from twist angle inhomogeneity or charge puddles can destroy the topological phase even if the average parameters are correct. A sample with twist angle variations of more than 0.05° across the device will likely show a smeared Hall plateau. Use STM to map the twist angle variation before investing time in transport measurements. If the variation is too large, consider using a different device or applying a global back-gate to compensate for local doping variations.

Pitfalls and Debugging: What to Check When the Phase Assignment Goes Wrong

The most common pitfall is a false positive from a trivial gap. As mentioned, a gap at the CNP can arise from substrate alignment, strain, or even from the formation of a charge density wave (CDW). To debug, measure the gap as a function of temperature: a topological gap will typically close at a higher temperature (above 10 K for Chern insulators) than a CDW gap (which often closes below 5 K). Also, check the gap's response to a magnetic field: a topological gap is usually insensitive to in-plane fields, while a CDW gap may be suppressed by a field of a few tesla.

Another frequent issue is misidentifying the Chern number from a Hall plateau that is not perfectly quantized. A plateau at σ_xy = 0.95 e²/h may still be a Chern insulator if the deviation is due to contact resistance or a parallel conduction channel. To rule out trivial contributions, measure the Hall conductance in a four-terminal configuration with the voltage probes placed away from the current contacts. If the plateau improves to 0.99 e²/h, the deviation was likely from contact effects. If it remains at 0.95, there may be a trivial bulk contribution. In that case, try reducing the temperature or applying a small magnetic field to suppress the trivial channel.

For theorists, a common bug is a miscomputed Chern number due to a gauge singularity in the Berry curvature integration. The Fukui-Hatsugai-Suzuki method is robust, but it requires a fine k-point grid (at least 100×100 for the mBZ). If your Chern number is not an integer, increase the grid density and check for band crossings near the zone boundaries. Also, ensure that the bands are well separated: a gap smaller than 1 meV can lead to numerical errors. If the gap is that small, the phase may be topological but fragile—consider whether it is observable in real samples.

Finally, do not overlook the role of the Fermi level. Even if the band structure has a nontrivial topology, the phase will not be observed if the chemical potential is not inside the topological gap. Always check the doping level relative to the flat bands. In transport, this means measuring the longitudinal resistance as a function of gate voltage: a topological phase should coincide with a peak in the resistance (due to the gap) and a plateau in the Hall conductance. If the Hall plateau appears but the longitudinal resistance is low, the system may be in a metallic state with a trivial Hall effect.

Frequently Asked Questions and Next Steps

How do I distinguish a Chern insulator from a quantum anomalous Hall effect from magnetic impurities?

Magnetic impurities can also produce a QAH effect, but the signature is different. In TBG, the Chern insulator is intrinsic and does not require magnetic doping. To distinguish, measure the Hall conductance as a function of temperature: the intrinsic Chern insulator will have a plateau up to a higher temperature (10–20 K) compared to a magnetic impurity-driven QAH (often below 5 K). Also, check for hysteresis in the longitudinal resistance: magnetic impurities often show a hysteretic magnetoresistance, while the intrinsic Chern insulator does not.

What is the minimum twist angle accuracy needed to observe topological phases?

For the magic-angle regime, you need accuracy within ±0.1° to see the flat bands and the associated topological phases. For larger twist angles (θ > 1.5°), the topological phases are weaker, and you may need ±0.05° accuracy. If your device has a twist angle gradient, the phase may only appear in a small region of the sample; use local probes like STM to map the phase.

Can I observe fractional Chern insulators in TBG?

Yes, but only in the cleanest samples with twist angles very close to the magic angle (within 0.05°) and at temperatures below 300 mK. The fractional Chern insulator has been observed at filling ν = ±2/3 and ν = ±1/3. The signature is a fractional Hall conductance plateau (e.g., σ_xy = (2/3)e²/h) that is robust to small magnetic fields. If you suspect an FCI, confirm with thermal transport: the thermal Hall conductance should also be quantized with a fractional value.

What are the next moves after identifying a topological phase?

First, publish your results with the full set of consistency checks. Include the raw data for the Hall conductance, longitudinal resistance, and STM edge state images. Second, explore the phase diagram by varying the twist angle, doping, and displacement field to map the boundaries of the topological phase. Third, consider device applications: a Chern insulator can be used for low-power electronics or as a platform for Majorana fermions when proximitized with a superconductor. Finally, compare your results with theoretical predictions to refine the models. If your phase does not match any existing theory, it may be a new topological phase worth further investigation.