Raman strain–shift measurements and prediction from first-principles in highly strained silicon

Abstract

This work presents how first-principles simulations validated through experimental measurements lead to a new accurate prediction of the expected Raman shift as a function of strain in silicon (Si). Structural relaxation of a strained primitive cell is first performed to tackle the relative displacement of the Si atoms for each strain level. Density functional perturbation theory is then used to compute the energy of the optical phonon modes in highly strained Si and retrieve the strain–shift trend. The simulations are validated by experimental characterization, using scanning electron microscopy (SEM) coupled with backscattering Raman spectroscopy, of silicon microbeams fabricated using a top-down approach. The beams are strained up to 2% thanks to the internal tensile stress of silicon nitride (Si3N4) actuators, allowing a validation of the perturbation theory in high-strain conditions. The results are compared with the phonon deformation potentials theory, and the uncertainty caused by the various parameters found in the literature is discussed. The simulated strain–shift coefficients of $$-175.77\,{\text {cm}}^{-1}$$ (resp. $$-400.85\,{\text {cm}}^{-1}$$ ) and the experimental one of $$-160.99\,{\text {cm}}^{-1}$$ (resp. $$-414.97\,{\text {cm}}^{-1}$$ ) are found for the longitudinal optical LO (resp. transverse optical TO $$_{1}$$ ) mode, showing good agreement.