Strain and Gate Voltage Tune Light Response in Graphene–MXene Model

Researchers report a fully computational study predicting that stretching a graphene–Ti₃C₂O₂ MXene heterostructure by just one percent more than doubles its electronic bandgap, from 0.18 to 0.43 electronvolts, while simultaneously reshaping the material’s optical response in the visible and near-infrared range.

The system under study is a van der Waals stack of graphene on oxygen-terminated Ti₃C₂O₂, chosen because this MXene termination offers demonstrated ambient stability and an intermediate work function that favors charge transfer with graphene. The team restricted the analysis to this well-defined composition to keep the calculations internally consistent, and they caution that extrapolation to fluorine- or hydroxyl-terminated MXene surfaces is not implied and would require separate calculations. Throughout the paper, the general term graphene–MXene refers specifically to this oxygen-terminated system.

The computational strategy addresses a known gap in multiscale modeling. Density functional theory describes how electronic structure responds to atomic-level deformation but cannot handle wavelength-scale wave propagation, while classical electromagnetic solvers handle propagation but need material-level inputs such as refractive index and absorption coefficient. Previous sequential-coupling studies treated material properties as static, leaving the interplay between strain, electronic structure, and optics unaddressed. In the new workflow, strain is applied first within DFT, strain-dependent optical constants are derived through the Kramers–Kronig relations, and those constants are then fed into FDTD simulations as static inputs, with separate runs for each strain level and no optical feedback into the electronic-structure step.

The calculations were performed with the Vienna Ab initio Simulation Package using the PBE functional with D3 dispersion corrections, a 500 electronvolt plane-wave cutoff, and a 12 by 12 by 1 k-point mesh, both validated by internal convergence tests. The heterostructure was modeled as a periodic supercell combining a 5 by 5 graphene unit cell with a 4 by 4 Ti₃C₂O₂ unit cell, keeping lattice mismatch below two percent, with a 25 angstrom vacuum region. Selected single-point hybrid HSE06 calculations on the relaxed geometries yielded a rigidly upshifted gap of roughly 0.08 to 0.11 electronvolts without altering the curvature of the bandgap–strain response, which the authors cite as evidence that the qualitative trends are robust.

At the unstrained interface, the calculations reveal substantial electronic reorganization. The graphene Dirac cone shifts downward by 0.25 electronvolts relative to the Fermi level, and Bader charge analysis quantifies a net transfer of 0.084 electrons per unit cell from the MXene layer to graphene, consistent with n-type doping and an estimated interfacial dipole moment of 1.2 debye. The computed work-function difference of 0.42 electronvolts is interpreted as forming a Schottky-type barrier. Control calculations, including band unfolding onto the primitive graphene Brillouin zone, indicate that the opened gap arises primarily from an asymmetric on-site potential of about 65 millielectronvolts imposed on graphene’s two sublattices by the terminating oxygen atoms, combined with weak hybridization between graphene pz orbitals and titanium d states confined to roughly half a nanometer from the interface.

Under biaxial tensile strain from zero to one percent, applied in 0.5 percent steps, the bandgap follows a quadratic rather than linear dependence, described by the relation Eg equals 0.18 plus 0.25 times the strain squared, with a regression fit above 0.99. The authors attribute the quadratic form to strain-induced breaking of graphene’s sublattice symmetry and nonlinear interlayer orbital hybridization, and they note that because only tensile strain was sampled, the data do not establish an even-in-strain law or a universal strain coefficient. The strain also shifts the joint density of states: the main absorption feature moves by 120 millielectronvolts and gains 12 percent in oscillator strength near 670 nanometers at one percent strain. A Drude-model cross-check estimates a carrier-density modulation of about 2.3 times 10 to the 12 per square centimeter per percent strain, corresponding to a refractive-index sensitivity of roughly minus 14, about twice the value the same model gives for pristine graphene.

The FDTD stage converts the DFT-derived dielectric response into a vacuum-independent two-dimensional sheet conductivity to eliminate supercell artifacts, cross-verifying a thin-layer approximation against surface-current boundary conditions with less than 1.2 percent discrepancy across the near-infrared telecommunication window. Simulations at 670 nanometers show pronounced field localization at the heterointerface, with an enhancement factor of 2.83 relative to monolayer graphene on silicon dioxide and an estimated 68 percent of the electromagnetic energy confined within 3 nanometers of the interface. The authors interpret this as constructive interplay between graphene’s two-dimensional plasmons and the MXene’s localized surface plasmons. The absorption spectrum peaks at 670 nanometers, with 38 percent integrated absorption across the 400 to 900 nanometer range. Electrostatic gating from 0 to 3 volts is predicted to shift the Fermi level by 0.3 electronvolts, blueshift the reflection minimum by 26 nanometers, and raise integrated absorption by 38 percent.

Combining one percent strain with 3 volts of gating produces, within the coupled model, a nonlinear enhancement in photoresponsivity: a peak of 1.22 amperes per watt at 670 nanometers, a 44 percent improvement over the unstrained, ungated reference case and over values reported for graphene–silicon waveguide photodetectors. The optomechanical sensitivity, defined as relative intensity change per unit strain, reaches 4.2 times 10 to the minus 2, exceeding values the model computes for graphene–MoS₂ and graphene–hBN references, and converts to an effective gauge factor of about 375, roughly 2.5 times the commonly cited figure for resistive graphene strain sensors. One percent strain shifts the absorption peak by 16 nanometers, suggesting, under the model’s assumptions, the possibility of optically detecting nanometer-scale deformations while avoiding contact-resistance and Joule-heating artifacts of electrical readout. Estimated response times of 2.1 milliseconds for the optical channel and 2.3 milliseconds for the mechanical channel derive from a simplified treatment without explicit damping or energy dissipation and are presented as upper bounds rather than validated transduction speeds.

The paper is unusually explicit about its limits. The detectivity and noise-equivalent-power figures, the latter estimated at 8.2 times 10 to the minus 13 watts per root hertz, come from a simplified shot-noise-limited dark-current model that excludes flicker noise and contact resistance. No phonon dispersion, elastic-constant, or ab initio molecular dynamics calculations were performed, so no claim of dynamical or thermal stability is supported. The heterostructure was treated as a freestanding bilayer with no substrate model, and no systematic FDTD mesh-convergence sweep was carried out. A phenomenological cyclic-loading analysis predicts degradation below 3 percent over 50,000 strain cycles, and a thermal-broadening model predicts a 16 percent photoresponsivity reduction between 100 and 300 kelvin, alongside an estimated interfacial strain of about 0.17 percent per 100 kelvin from thermal-expansion mismatch, which the authors flag as a calibration concern for any future device. All of these figures are model-based estimates pending experimental validation.

The authors position the graphene–Ti₃C₂O₂ heterostructure as a computationally grounded, strain-tunable platform for coupled optoelectronic and optomechanical sensing, with the idealized interface and semiclassical optics acknowledged as a tractable starting point rather than a device demonstration. Future work, they suggest, should address defect-mediated scattering, finite-temperature quantum corrections, explicit substrate and damping models, and the experimental measurements needed to test whether the predicted quadratic bandgap opening, plasmonic field confinement, and optical strain readout survive contact with real materials.

Subject of Research: Computational prediction of strain-tunable electronic and optical properties of graphene–Ti₃C₂O₂ MXene heterostructures using a sequential DFT-to-FDTD framework

Article Title: Strain-dependent electronic and optical response of graphene–Ti₃C₂O₂ MXene heterostructures: A sequential DFT-to-FDTD computational study

Article References: Vaghef-Koodehi, A., Nikoufard, M., Karamitaheri, H., & Samadpour, E. (2026). Strain-dependent electronic and optical response of graphene–Ti₃C₂O₂ MXene heterostructures: A sequential DFT-to-FDTD computational study. Results in Optics, Article 101197. https://doi.org/10.1016/j.rio.2026.101197

Image Credits: AI Generated

DOI: 10.1016/j.rio.2026.101197

Keywords: graphene, MXene, Ti3C2O2, heterostructure, strain engineering, DFT, FDTD, bandgap tuning, plasmonics, optomechanical sensing, photodetector, computational study