Structure-property mapping and inverse design of an auxetic architected screw under mechanical loading

Auxetic architected solids provide an effective route for tailoring stiffness and transverse deformation through geometry rather than material composition alone. In screw-like structures, this capability is attractive because axial loading can be converted into lateral expansion, which modifies interfacial load transfer and resistance to pullout. This work develops a mechanics-centered framework for the structure-property mapping and inverse design of an auxetic architected screw. A re-entrant hexagonal architecture was embedded in the screw core, and finite element simulations were used to quantify the dependence of global effective elastic modulus, Poisson's ratio, stiffness, and porosity on geometric parameters. The resulting dataset was used to train a multilayer perceptron surrogate model, which achieved R^2>0.98 for all predicted mechanical outputs. A tandem inverse design strategy was then used to recover geometric parameters from prescribed mechanical targets. The optimized screw was manufactured by selective laser melting and assessed by uniaxial tensile testing, digital image correlation, axial pullout experiments, multiaxial finite element loading, and a representative spine-screw boundary-value model. The designed architecture exhibited a negative Poisson's ratio, an 11.5% increase in average pullout force compared with a solid screw of identical external geometry, and more distributed stress transfer in the surrounding solid domain. The results show that data-driven inverse design can provide a practical route for controlling the effective response of auxetic architected screws and for translating target mechanical properties into manufacturable structural geometries.