In 2010, a team led by Michael Crommie at Lawrence Berkeley National Laboratory discovered that triangular "nanobubbles" spontaneously form in graphene grown on a platinum substrate. Examining them with a scanning tunneling microscope, they found pseudo-magnetic fields exceeding 300 tesla inside these bulges. No magnet had been brought near the sample from outside. The stretching of the graphene lattice alone caused electrons to behave as if they were in an extremely strong magnetic field (Levy et al., Science, 2010).
This discovery of "strain engineering" showed for the first time that graphene's electrical properties could be manipulated through physical deformation alone, without chemical alteration. Until then, the field of electrical control in graphene had been dominated by a choice between chemical doping and electrostatic gating. This finding raised the possibility of a third option: treating shape itself as a design variable.
However, the 2010 discovery concerned only pseudo-magnetic fields—that is, changes in the electrons' motional state. This is distinct from "polarization," the phenomenon in which positive and negative charges separate within a material to generate a voltage. Whether bending graphene could produce true electric polarization—the flexoelectric effect—had been predicted theoretically but remained experimentally unresolved.
A 1986 Theory, a 2008 Prediction, and 18 Years of Silence
The flexoelectric effect is a phenomenon in which positive and negative charges separate to produce polarization when a material experiences a strain gradient—a state in which the degree of stretching or compression varies from place to place. Whereas the piezoelectric effect requires asymmetry in the crystal structure, the flexoelectric effect is, in principle, permitted in any material. In 1986, A. K. Tagantsev established a theoretical framework for this effect in crystalline dielectrics, but in bulk materials the strain gradients are so small that the effect is negligibly weak.
The situation changed in 2008. Sergei Kalinin and Vincent Meunier, then at Oak Ridge National Laboratory, proposed the concept of "quantum flexoelectricity" (electronic flexoelectricity) in a paper published in Physical Review B. In low-dimensional systems like graphene, they argued, bending causes the electron distribution in the orbitals to become asymmetric, generating a local dipole moment. Using density functional theory (DFT) calculations, they showed that polarization strengthens as the radius of curvature shrinks, and predicted that this effect could become dominant at the nanoscale.
Yet for 18 years after this prediction, no one had captured this effect experimentally. The reason was simple: measurement was extraordinarily difficult. Polarization arises only in regions a few atoms wide, and conventional electrical measurement techniques could not distinguish the signal from noise.
A "Ghost Signal" Hidden in Old Data
The breakthrough came by chance. Sathvik Ajay Iyengar, a doctoral student at Rice University, was reviewing old conductive atomic force microscopy (c-AFM) data he had collected together with collaborator Manoj Tripathi (then at the University of Sussex, now at South Dakota Mines). Repeatedly, an unexplained spike in electrical signal appeared precisely at the locations corresponding to the sharpest folds in the graphene sample.
Iyengar brought this data to Meunier, who was also his co-advisor during his doctoral studies—and the very person who had proposed the theory in 2008. To verify whether this signal truly originated from the flexoelectric effect, the research team combined three approaches: local electrical measurements using a specialized scanning probe microscope, laser Raman spectroscopy to examine the stretching state of atoms, and computer simulations based on first-principles calculations.
The sample consisted of CVD-grown graphene transferred onto a molybdenum disulfide () substrate. Because graphene's lattice constant (0.24 nm) differs from that of (0.31 nm), the lattice mismatch at their interface causes graphene to spontaneously buckle, forming wrinkles (nanowrinkles). The height of these wrinkles reaches 6–8 nm, roughly six times greater than wrinkles formed on a silicon oxide substrate.
A Ridge with a 0.5 nm Radius of Curvature Pushes Electrons to One Side
Detailed examination of the wrinkle tips revealed a radius of curvature reaching approximately 0.5 nm (5 Å)—a value approaching the 3.4 Å radius of a fullerene, and near the physical lower limit of curvature that carbon materials can sustain.
What does this extreme curvature cause? Graphene is a single atomic layer of carbon atoms arranged in a hexagonal lattice. Of each carbon atom's four valence electrons, three form in-plane bonds, while the remaining one occupies a orbital perpendicular to the plane. In flat graphene, the orbitals are distributed symmetrically above and below the sheet. But when a ridge bends with a sub-nanometer radius, the outer orbitals stretch while the inner ones compress. This asymmetry skews the electron density toward one face, producing a state of separated positive and negative charges—that is, polarization.
The research team named this mechanism "quantum orbital flexoelectricity." What distinguishes it from the conventional understanding of the flexoelectric effect is that the origin of polarization lies not only in classical lattice strain but in quantum-mechanical rehybridization at the orbital level.
The measurement results showed strong agreement with theory. When a bias voltage of approximately 1 V was applied via c-AFM, a reproducible flexoelectric current was detected at the wrinkle locations. This threshold voltage (approximately 1.01 V) corresponds to the band offset between the wrinkled and flat regions predicted by DFT calculations (approximately 1.2 V). The interface between wrinkled and flat graphene effectively forms a type-I heterojunction.
Not "Height" but "Sharpness" That Determines the Response
The most practically significant implication of this discovery is that the electrical response depends not on the height of the wrinkle but on the sharpness of its curvature. Iyengar noted, "The sharpness of the wrinkle mattered far more than its overall size. This suggests that precisely controlling curvature at the nanoscale could allow us to tune electrical behavior."
This property is advantageous for device scalability. Even if the height varies, as long as the radius of curvature at the ridge is consistent, the same electrical response is obtained. Indeed, the research team confirmed through molecular dynamics simulations that wrinkles spontaneously nucleate within an extremely short timescale of 0.1 nanoseconds, then glide across the substrate to aggregate.
| Metric | Conventional Mesoscale Flexoelectric Systems | Graphene Nanowrinkles in This Study |
|---|---|---|
| Radius of curvature | Several nm to several μm | Approx. 0.5 nm (5 Å) |
| Polarization density (theoretical) | Range of several to $10^{-4}$ C/m² | Approx. 4 C/m² |
| Polarization density (experimental) | Same as above | Approx. 1 C/m² |
| Amplification factor of the effect | Baseline | 5–7 orders of magnitude larger |
| Driving condition | Requires external stress or electric field | Current generated by applying ~1 V to naturally formed wrinkles |
| Control variable | Chemical composition of the material | Geometric curvature (no chemical modification needed) |
A note on the polarization density estimates: both the theoretical value of approximately 4 C/m² and the experimental value of approximately 1 C/m² were calculated by the research team within the paper itself. These are reported as being 5–7 orders of magnitude larger than the typical polarization density of mesoscale flexoelectric systems (on the order of $10^{-6}$ to $10^{-4}$ C/m²). However, this comparison is an order-of-magnitude estimate, and the conditions of the mesoscale systems used for comparison (material type, curvature) are not uniform.
From Chemistry to Geometry—But the Path Has Only Just Begun
Ajayan stated, "At the atomic scale, we've shown that an ordinary wrinkle can become an extraordinary electronic function. By demonstrating that graphene's electrical behavior can be redesigned through geometry alone, this opens a new path in materials design—one that controls properties through structure rather than chemistry."
This direction holds the potential to circumvent a fundamental constraint that graphene electronics has long faced. Chemical doping introduces defects into the lattice and reduces mobility. Electrostatic gating requires external circuitry, which hinders device miniaturization. If shape control becomes established as a third option, applications in non-volatile memory, strain-gated transistors, and energy-harvesting devices come into view.
That said, many challenges remain unresolved at this stage. First, no method has yet been established for deliberately designing and positioning the location and curvature of wrinkles. In this study, the observed wrinkles formed spontaneously due to lattice mismatch with the substrate; techniques for creating wrinkles of a desired curvature at a desired location remain a task for the future. Second, no architecture has been proposed for integrating wrinkles into functional electronic circuits. Third, expansion into composite flexoelectric systems through stacking with other two-dimensional materials is mentioned in the paper only as a future direction.
An effect predicted on paper 18 years ago was, indeed, alive within the wrinkle of a single atomic layer. The question that remains is whether this "electricity of shape" can be arranged, amplified, and integrated into circuits according to human design.