Twist angle is the moiré field’s central knob. Choose it, and you choose the superlattice period, the depth of the potential wells, and where excitons sit inside each cell. What the field has been slower to map is what happens to that choice when the sample is heated — and heating is not optional. Annealing is routine in van der Waals fabrication: it drives out trapped contaminants and improves interlayer contact. Every twistronic device is, in practice, a device that has been baked.
A paper published in ACS Applied Nano Materials on October 30, 2026, maps it in three landmarks, measured on a single hBN-encapsulated bilayer twisted to 59.5° by a team led by Andreij C. Gadelha at the Universidade Federal do Ceará and the Centro Brasileiro de Pesquisas Físicas, with collaborators at the Universidade Federal de Minas Gerais and Japan’s National Institute for Materials Science. At 100 °C, nothing changes. At 200 °C, the moiré lattice reconstructs into nanoscale domains separated by domain walls, and trapped blisters are largely removed. At 300 °C, the charge–strain state turns. At 400 °C, the sample oxidises and degrades. It is manufacturing science rather than a device demonstration, and it arrives with a method attached: the whole map is drawn with fast far-field Raman and photoluminescence mapping, a protocol that could serve as a quality check on real process lines.
Why the near-60° corner matters
Twisted bilayers of transition metal dichalcogenides attract attention because the moiré superlattice confines excitons on the nanometre scale, opening the way to tunable light emitters and quantum-light sources. That is the paper’s framing, and it is the reason the near-60° configuration deserves its own stability map.
The physics of twisted bilayers is most sensitive near the high-symmetry limits: parallel stacking at 0° and antiparallel stacking at 60°. Near 60°, the two MoS₂ layers sit close to the commensurate 2H stacking of the bulk crystal, and small deviations from that limit produce large changes in the moiré period, the interlayer coupling, and the excitonic landscape. The sample here sits half a degree from the limit — 59.5°, measured by polarisation-resolved second-harmonic generation — giving a moiré period of about 36 nanometres.
That sensitivity cuts both ways. At marginal twist angles, the lattice is prone to atomic reconstruction: rather than maintaining a smooth sinusoidal modulation of the stacking, the atoms shift to form large domains of the preferred commensurate stacking, separated by narrow domain walls where the registry changes abruptly. Those walls — strain solitons — are not defects in the usual sense; they are the structure’s way of accommodating the twist, and they reshape the potential an exciton experiences. Whether and when heating triggers that reconstruction is a first-order question for anyone building a device at this angle.
One flake, five temperatures
The sample was made by the all-dry tear-and-stack method: a MoS₂ monolayer is picked up in two halves with a polymer stamp, the stage rotated by 58° between the halves, and the halves stacked to form the twisted bilayer, encapsulated top and bottom in hexagonal boron nitride. The resulting twist, measured rather than assumed, is 59.5°.
Annealing was performed on a hot plate in ambient air, ramping at 10 °C per minute to the target temperature, holding for one hour, then cooling to room temperature. The steps were 25, 100, 200, 300, and 400 °C. Crucially, the measurements were made after each step, at room temperature: this is an ex-situ experiment, designed to isolate the irreversible, processing-relevant changes from the reversible temperature dependence of the spectra, which the same community has characterised separately in earlier work.
The probes are Raman and photoluminescence spectroscopy, acquired with a confocal spectrometer at 2.33 eV excitation, laser power kept below 1 mW to avoid local heating, and mapped spatially across the flake. Photoluminescence is the more sensitive probe of the moiré-averaged optical response, while Raman adds structural and electronic information — strain, doping, and the domain-wall network.
200 °C: reconstruction arrives
The first landmark is the quiet one. Annealing at 100 °C produces no noticeable change in either the photoluminescence or the Raman maps — the gentle anneals commonly used to clean interfaces sit below the threshold at which the moiré structure begins to move.
At 200 °C, it moves. In the photoluminescence maps, the bubbles visible in the lower region of the as-fabricated sample are significantly reduced, and reconstructed nanoscale domains become more prominent: the B exciton, the higher-energy of the two principal excitonic features, shows the strongest spatial contrast in both intensity and linewidth, and its maps now trace the new domain structure.
The Raman maps tell the same story from the structural side. After annealing at 200 °C, nanoscale line-like features appear in the A′ phonon mode’s intensity and linewidth maps, and the authors assign them to domain-boundary or soliton-network structures formed by atomic reconstruction. Three checks rule out folds or wrinkles: the lines are absent at 25 and 100 °C and appear only upon annealing; along them, the A′–E′ phonon frequency separation remains at the bilayer value, whereas a fold would shift it toward multilayer values; and they have no topographic counterpart in atomic force microscopy, as expected for domain walls buried at the twisted interface.
The picture is that annealing drives the lattice toward the commensurate 2H-like stacking, accommodated by a stable network of domain walls. The global twist angle cannot be erased thermally; what changes is how the twist is distributed — enlarged domains of preferred stacking plus a soliton network, rather than a smooth moiré modulation.
That reorganisation is not uniform. Local spectra at three points on the flake show different regions responding differently to the same treatment: at one point a new low-energy emission feature emerges after the 300 °C step, with increased local electron doping; at a second, the spectra remain nearly unchanged; at a third, a new well-separated low-energy peak emerges and the B exciton intensity falls.
300 °C: the charge–strain turn
Between 200 and 300 °C, the Raman maps show no further changes in the domain structure. But the correlation analysis of the two phonon frequencies tells a different story: plotting the out-of-plane A′ frequency against the in-plane E′ frequency separates the effects of strain and charge doping, and the distributions trace a clear transition at 300 °C and above — less tensile strain, more electrons.
The mean effective tensile strain falls from about 0.28% at 25 °C to about 0.20% at 300 °C, while the mean electron density rises from roughly 0.49 × 10¹³ cm⁻² to 0.87 × 10¹³ cm⁻². At 200 °C the values sit between the two — a drift that becomes unmistakable at 300 °C, consistent with the photoluminescence changes at the same step.
At 400 °C, the extracted values shift back partway — strain to about 0.27%, electron density to about 0.75 × 10¹³ cm⁻² — but the paper is careful not to read this as recovery: at that temperature the Raman response is dominated by degradation, and the apparent shift reflects a damaged sample, not a restored charge–strain state.
400 °C: oxidation, not physics
The third landmark is the hard limit. After the 400 °C step, the photoluminescence response becomes highly irregular and fragmented, indicating severe optical degradation. The mechanism is named in the paper: annealing in ambient air means that at 400 °C the MoS₂ undergoes thermal oxidation — an irreversible, material-consuming reaction with oxygen that forms volatile molybdenum oxides with sulfur loss. The hBN encapsulation delays it — no degradation up to 300 °C — but at 400 °C oxidation sets in at edges, defects, and blister boundaries, fragmenting the surviving area.
The 400 °C limit is a limit of the processing environment as much as of the material: it is oxidation in air, not a thermal instability of the moiré lattice itself. An anneal in inert atmosphere or vacuum might push the ceiling higher, and the paper does not test that. What it establishes is the ceiling for the most common processing condition — a hot plate in air: 300 °C for safe operation, 400 °C as the point of irreversible loss. The 200 °C reconstruction, by contrast, is intrinsic to the moiré system and will happen in any atmosphere.
The method is the second result
The paper’s second claim is about how the map was drawn. Probing nanoscale moiré structure has historically meant scanning probe microscopy or transport measurements — detailed but slow, invasive, or limited in spatial coverage. Optical spectroscopy offers a non-destructive, spatially resolved alternative, and this work argues it is sufficient for the fabrication question: has the moiré structure changed, and in which direction?
The argument rests on the complementary sensitivities the authors demonstrate. The photoluminescence maps — the B exciton’s intensity and linewidth — are the most sensitive to the moiré-averaged optical response, capturing the growth of reconstructed domains, the reduction of blisters around 200 °C, and the irreversible degradation at 400 °C. The Raman maps add the structural and electronic layer: the A′ intensity and linewidth reveal the domain-wall and soliton features, while the A′–E′ correlation quantifies local strain and doping and reveals the transition at 300 °C and above. Together they yield a practical optical protocol linking structural reconstruction, the strain and doping state, and the onset of degradation — without resorting to slower scanning-probe techniques. A far-field optical map takes minutes across a whole flake; a scanning-probe image of the same area takes far longer and risks disturbing the sample. If the optical signatures are reliable markers, they become a quality-control tool: anneal, map, and know whether the moiré survived the process step.
What it is not
Three cautions, stated plainly because the framing invites overreading. First, this is a map of one configuration: a near-60° homobilayer of MoS₂, made by tear-and-stack from exfoliated crystals, encapsulated in hBN, annealed in air. The authors test reproducibility on an additional near-60° bilayer at 56°, and the characteristic evolution is partially reproduced there — evidence that the behaviour is a robust feature of near-60° twisted bilayer MoS₂ rather than specific to a single flake. But two samples is a beginning, not a survey. Whether the same thresholds hold for other twist angles, heterobilayers, CVD-grown material, or wafer-scale fabrication is untested, and the 200 °C figure should not be quoted as a universal property of twisted MoS₂.
Second, the experiment is ex-situ by design. The maps are acquired at room temperature after each annealing-and-cooling step, isolating the permanent, processing-relevant changes. It does not capture what the spectra look like while the sample is hot, nor the reversible temperature dependence of the excitonic response — characterised in a related 2025 study of twisted bilayer MoS₂ over 93 to 573 K. One question asks what a device looks like after it has been baked, the other what it looks like while it is hot. This paper answers the first.
Third, the 200 °C reconstruction is not necessarily bad news. Reconstruction reorganises the moiré landscape, but it does not destroy it, and the removal of blisters at the same temperature improves interlayer contact and narrows the emission. A device annealed at 200 °C may be a better device than the as-fabricated one, with a different — but stable — domain structure. The paper’s message is not “do not heat past 100 °C.” It is that heating past 200 °C changes the structure you designed, and heating past 300 °C changes the electronic state as well, so any process step in that range should be treated as a design decision, not a cleaning step. The practical window: below 200 °C, the moiré is preserved as fabricated; at 200 °C, expect reconstruction; at 300 °C, expect the charge–strain state to shift; at 400 °C in air, expect to lose the sample.
What to watch
The most immediate follow-up: apply the same optical protocol to other twist angles and other materials, and find out whether the 200 °C threshold moves. Near-0° twisted bilayers are the obvious next case, as are heterobilayers such as MoS₂/WSe₂, where the interlayer exciton dominates the optical response.
A second follow-up is the atmosphere. The 400 °C degradation is oxidation, and oxidation is a property of the environment. Repeating the anneal series in argon, nitrogen, or vacuum would separate the intrinsic thermal limit of the moiré structure from the chemical limit of the process, and tell fabricators whether an inert-atmosphere bake can safely exceed 300 °C.
A third is the connection to device performance. This paper maps structure and spectra; it does not build an emitter and measure its output after each anneal. The natural next step is to close that loop: fabricate a near-60° emitter, anneal it through the same steps, and measure the emission linewidth, the single-photon purity, and the spatial distribution of emitters at each stage. The moiré field’s promise is programmable arrays of quantum emitters, and whether the array survives the process can only be answered by watching the emitters themselves.
Twist angle has always been the moiré field’s most powerful design variable. This paper adds the constraint that makes the variable usable: the temperature at which the design you chose stops being the design you have.
Sources and further reading
Gadelha, A. C.; Tavares, I. F.; Almeida, L. F. D.; Oliveira, G. G. A.; Rojas-Lopez, R. R.; Seniuk, T.; Caram, R. E. D.; Silveira, I. S.; Borsoi, G.; Araújo, K. A. S.; Watanabe, K.; Taniguchi, T.; Righi, A.; Malard, L. M.; Alencar, R. S.; Filho, A. G. S. “Thermal Stability of Near-60° Twisted MoS₂ Bilayers with Implications for Twistronic Optoelectronic Devices.” ACS Applied Nano Materials (2026), published online October 30, 2026. DOI: 10.1021/acsanm.6c04396. Received September 15, 2026; revised October 22, 2026; accepted October 26, 2026. Open access under CC-BY 4.0.
For the reversible temperature dependence of twisted bilayer MoS₂ optics, see Santos, F. S.; et al. “Enhanced Exciton Thermal Stability and Phonon Anharmonicity in Twisted Bilayer MoS₂: Implications for Nano-Optoelectronic Devices.” ACS Applied Nano Materials 2025, 8, 9268–9275. For atomic reconstruction in twisted TMD bilayers, see Weston, A.; et al. “Atomic reconstruction in twisted bilayers of transition metal dichalcogenides.” Nature Nanotechnology 2020, 15, 592–597. For the wider moiré context, see this site’s explainer on moiré excitons and its report on the first real-space imaging of moiré-confined excitons in twisted bilayer MoS₂. This article is analysis of published research, not investment advice.