Twisted graphene stores memory by combining strain-induced hysteresis and nonlinear response, enabling reprogrammable, low power memory in a pure carbon system.
(Nanowerk Spotlight) Electronic memory has always depended on order. From the earliest vacuum tubes to the dense flash architectures inside today’s devices, engineers have pursued control over structure, purity, and symmetry. But the materials now reshaping condensed matter physics are governed by a different logic.
In moiré superlattices, layered systems formed by stacking two-dimensional crystals with a slight rotational misalignment, the disorder introduced during fabrication does not degrade performance. It defines it. Imperfections once treated as manufacturing errors now enable access to new phases of matter.
This inversion of assumptions has transformed twisted graphene systems into a rich experimental platform. When two graphene bilayers are rotated relative to each other by a small angle, the resulting interference between their atomic lattices produces large-scale geometric patterns known as moiré superlattices. These patterns reshape the material’s electronic behavior in ways that are highly sensitive to twist angle, strain, and symmetry. Researchers have used them to explore superconductivity, magnetism, and strongly correlated insulators. But their potential as active components in electronic devices remains largely unexplored.
Theoretical work has suggested that breaking inversion symmetry in such systems could produce nonlinear electrical responses, where output is not proportional to input, without relying on junctions or interface layers. At the same time, mechanical strain in these twisted structures generates internal electric fields through the flexoelectric effect, offering a possible route to hysteresis and memory.
The idea that a monolithic carbon system, free from polar materials or charge traps, could mimic biological learning by storing states electrically has remained speculative.
A new study published in Advanced Materials (“Second-Order Synaptic Memory using Inherent Plasticity of Moiré Superlattices”) demonstrates that this is no longer a theoretical possibility. The team from CIC nanoGUNE in Spain and National Institute for Materials Science in Japan, reports that twisted double bilayer graphene can combine nonlinear response and strain-induced hysteresis to produce a functional, reconfigurable memory element. The work marks a new direction in the design of neuromorphic devices from single-element materials.
This figure shows how twisting two bilayer graphene sheets creates a special pattern called a moiré superlattice, which changes how electricity moves through the material. Panel (a) illustrates the layered structure of the device, where two graphene bilayers are rotated at a small angle and sandwiched between insulating layers. Panel (b) shows the electrical measurement setup, with voltage and current probes used to detect signals in two directions. Panel (c) presents how electrical resistance varies with the number of charge carriers in three different samples. Sharp resistance peaks appear at specific carrier concentrations, revealing the effects of the moiré pattern. Panel (d) zooms in on these peaks and shows that each one splits into two, a sign that the twist angle varies slightly across the material due to strain. Panels (e) and (f) display how resistance depends on both carrier concentration and vertical electric field. These measurements reveal a clear hysteresis loop—meaning the resistance depends on the direction in which the electric field is changed. This is an essential feature of memory. Panel (g) highlights this hysteresis more clearly, showing that the resistance does not return to the same value when the field is swept up and down. Together, these results demonstrate that strained, twisted graphene can “remember” past electric field changes, a behavior central to memory devices. (Image: Reprinted with permission by Wiley-VCH Verlag) (click on image to enlarge)
The researchers fabricated several devices using twisted double bilayer graphene, with twist angles ranging from approximately 0.7 to 1.9 degrees. A larger-angle control sample with a 10 degree twist was also studied. The samples were encapsulated between layers of hexagonal boron nitride to provide stability and allow control over electric fields.
Using a field effect transistor configuration, the team independently varied the carrier density and vertical displacement field while measuring resistance and voltage at both the primary and second harmonic frequencies.
One of the most significant findings was the appearance of hysteresis in the material’s resistance as the vertical displacement field was swept back and forth. This hysteresis, a lag in the response of the system that depends on its previous state, was observed near the charge neutrality point and at other critical fillings of the moiré bands. It was absent in the large-angle control sample, confirming its connection to the twist-induced strain.
The researchers showed that this hysteresis is not the result of interfacial traps or ferroelectric switching. Instead, it reflects a change in the material’s electronic structure that is retained even after the field is removed.
By measuring how resistance varied with temperature and field, the team extracted the size of the energy gap in the electronic band structure. This gap also showed hysteretic behavior, supporting the idea that the entire band structure is reconfigured in a history-dependent way. The researchers proposed that this arises from flexoelectric coupling, in which strain gradients produce polarization that responds nonlinearly to external fields. Inhomogeneous strains near the domain walls of the moiré pattern, amplified by twist angle disorder, provide the necessary conditions for this effect.
In parallel, the researchers investigated the material’s second-order nonlinear electrical response. This kind of response occurs when a material without inversion symmetry generates signals at twice the frequency of an applied voltage. In the twisted graphene system, the broken symmetry of the moiré superlattice enables such effects to emerge from the bulk of the material. The second-order voltages varied with both field and carrier concentration and changed sign at integer fillings of the moiré band structure. This behavior is consistent with a disorder-driven extrinsic mechanism, where local symmetry breaking interacts with the underlying band geometry.
Crucially, these two effects—hysteresis and second-order nonlinearity—are not independent. By combining them, the researchers constructed a memory element whose output could be tuned by field and carrier density. The magnitude and sign of the nonlinear voltage response could be deterministically controlled based on how far the displacement field was swept in previous cycles. This enabled a memory system with multiple stable levels, each associated with a distinct second-order electrical response. The memory state could be changed using pulsed electric fields, and the levels were retained without continuous power.
The team demonstrated a full synaptic memory function. A sequence of electric field pulses was used to either strengthen or weaken the output, mimicking potentiation and depression in biological synapses. The number of stable levels exceeded sixteen, and the memory retained its state over many cycles with minimal degradation. Retention was stable over days, with signal decay of less than one percent.
The energy cost per memory update was estimated between 0.5 and 0.8 picojoules, on par with advanced artificial synapse designs that use more complex materials. Because the second-order voltage can be generated without continuous current, the system could also be adapted for low power or passive operation.
These results show that memory and learning behavior can be achieved in a single-element material using only strain and symmetry. No polar molecules, charge trapping layers, or engineered heterostructures are required. Instead, the twisted geometry and natural strain of the moiré pattern provide the necessary internal degrees of freedom.
The fact that both the sign and magnitude of the memory state can be tuned electrostatically adds further flexibility. The memory operates at cryogenic temperatures in this study, but the underlying mechanism could potentially be extended to higher temperature systems.
This work demonstrates that carbon alone, when configured at the right geometric scale, can express electronic plasticity. By turning disorder into a functional feature, the researchers have created a new type of memory device that is defined by its architecture rather than its composition. It offers a direct route to embedding synaptic behavior into quantum materials without added complexity.
As neuromorphic computing moves toward systems that more closely resemble biological networks, such reconfigurable, compact, and efficient memory elements could help shape the foundations of future computing architectures.
If this article was useful, support our independent nanotechnology reporting with any amount.
Your contribution funds the next explainer and keeps Nanowerk open for everyone.
For authors and communications departmentsclick to open
Lay summary
Prefilled posts
Nanowerk Newsletter
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
