In this study, three-dimensional functionally graded NiTi bulk materials were fabricated using laser powder bed fusion (LPBF) by in-situ adding Ni powder into equiatomic NiTi powder. The gradient zone exhibited a Ni composition ranging from approximately 49.6 to 52.4 at.% over a distance of about 2.75 mm. The functionalities along the compositional gradient were examined through differential scanning calorimetry analysis and spherical indentation. This unique gradient resulted in location-specific functionalities, including superelasticity characterized by wide and narrow hysteresis loops, shape memory effect, and various phase transformation temperatures. The rapid cooling rate during fabrication led to the presence of excess Ni in the solid-solute state within NiTi. This unique solid-solute compositional gradient in NiTi resulted in varying lattice parameters, influencing the compatibility between martensite and austenite and allowing for tailored hysteresis. This discovery presents new avenues for designing multifunctional materials through in-situ additive manufacturing.

Superelastic metamaterials have attracted significant attention recently, but achieving such functionality remains challenging due to partial superelasticity and premature fracture in additively manufactured components. To address these issues, this study investigates the premature fracture in Ni-rich NiTi metamaterials fabricated by laser powder bed fusion. A comparative analysis of two structures (Gyroid network and Diamond shell) reveals that the structural stability of bending- and stretching-dominated structures is reversed compared to typical elastic-plastic response, due to the tension-compression asymmetry of base NiTi. The premature fracture and partial superelasticity of these as-fabricated samples are attributed to low deformation ability for accommodating tensile stress. Based on these findings, a heat treatment introducing Ni₄Ti₃ precipitates was employed, successfully achieving macroscopic superelasticity in the NiTi metamaterials, with consistency between model prediction and experiments.