It is a familiar fact that the surface of an object can differ significantly from its interior. The inside of a banana is tasty whereas the peel is not. Graphene, discovered at the beginning of the twenty-first century, is the first known example of a material made entirely of surface, lacking any interior or bulk. Graphene and similar two-dimensional (2D) materials defy physical theories developed in the twentieth century to explain the properties of conventional three-dimensional solids.
Our study, published in Nature, draws attention to twisted bilayer 2D materials, engineered by sandwiching together two nanometer-thin layers that are intentionally misaligned by a twist angle. A small twist angle creates a quasi-crystalline moiré pattern in the arrangement of atoms, significantly affecting the material’s optical and electronic properties. The prevailing assumption is that the twist angle is fixed during the one-off process of synthesizing the bilayer, after which time it cannot be modified.
Our results provide the first experimental evidence that a bilayer twist angle can be dynamically tuned with ultrashort pulses of laser light. The tuning mechanism derives from the bilayer’s normal modes of oscillation, analogous to the ringing of a bell struck by a mallet. In our experiment, the ringing consists of a repeating pattern of vortex-like motions that locally add or subtract from the static twist angle, twisting and untwisting the material. The laser pulse plays the role of the mallet that gets the vortices moving.
The experimental challenge
Electron diffraction (ED) is the technique of choice for precisely measuring small twist angles in bilayer 2D materials. Scattering in ED reveals repeating patterns in the coordinates of atoms. The specific signature that indicates the formation of moiré stacking domains is the appearance of subtle satellite peaks dressing the main Bragg peaks.
The principal challenge in our experiment is that the bilayer twist-untwist motion is much too rapid — five hundred billion cycles per second — to measure in a standard electron microscope. Freeze-framing the atomic motion requires an ultrafast electron diffraction (UED) beamline, which can deliver precisely synchronized electron pulses. Further compounding the experimental difficulty, the typical configuration of a UED beamline lacks the angular magnification and transverse beam coherence necessary to resolve moiré satellite diffraction peaks. Performed at Cornell University's Micro Electron Diffraction for Ultrafast Structural Analysis (MEDUSA) beamline, what made our experiment technically feasible is the novel pairing of a high-brightness photoelectron source with custom-made electron optics.
The physical mechanism
The specific material we study in our work, prepared by Fang Liu's group at Stanford University, consists of two semiconducting layers: WSe2 and MoSe2 . The layers are held together by relatively week van der Waals forces, accentuating the anisotropic response of the material to laser excitation.
As a semiconductor, twisted WSe2/MoSe2 responds to light in an especially interesting way: photons can be absorbed in the creation of long-lived, bound electron-hole pairs (excitons). Photoexcited negatively charged electrons prefer to migrate to the MoSe2 layer and positively charged holes to WSe2. This charge separation is the source of an electrostatic force that tends to pull opposite layers together.
Ultrafast charge separation in WSe2/MoSe2 provides the channel through which energy delivered by the pump laser pulse is transferred to a twisting motion in the material. Due to the moiré pattern, the van der Waals forces between layers are inclined in a spiral shape, balanced by in-plane elastic stress. As the layers are pressed together by the pressure of layer-separated electric charges, the van der Waals forces push atoms along a cork-screw-like chiral path.
A surprising signal
Our time-series experimental data covers an interval of four-trillionths of a second. We compare the change in intensity of the main Bragg peaks to the intensity of satellite peaks. Pumping a material with laser light typically raises the material’s temperature, which adds some randomness to its structure and thus causes diffraction intensities to fall. Consistent with typical behavior, our data shows that the main Bragg peak intensities fall as the bilayer absorbs laser light.
The surprise is that in our data the intensity of the moiré satellite peaks immediately and strongly increases following the arrival of the laser pulse. No similarly strong increase can be seen at any other scattering angle. Thus, the pump laser must be amplifying the same structural feature, the bilayer twist, which gives rise to the static satellite diffraction peaks. Shortly after this transient increase, the satellite peak intensities fall below the level implied by heating effects, indicating that the twisting motion reverses direction. The twisting-untwisting oscillations do not continue beyond a single cycle due to the damping effect of the substrate that supports the specimen.
Outlook
The twisting response demonstrated in our data points forward to the development of optical control over the functionality of twisted bilayers. Applications we consider include modulating the wavelength of excitonic lasers and optically switching twisted 2D materials from normal conducting to superconducting states. Further experimental work could explore more direct excitation schemes that match the frequency of the driving light to the frequency of the twisting motion, thereby extending the applicability of optical twist tuning to a wider variety of bilayer materials. We have shown that ultrafast electron diffraction provides the capability to experimentally investigate the rich dynamics of moiré patterns in 2D materials.