In this work, a new method, "Nanobonding" [1,2] is conceived and researched to bond Si-based surfaces, via nucleation and growth of a 2 D silicon oxide SiOxHx interphase connecting the surfaces at the nanoscale across macroscopic domains. Nanobonding cross-bridges two smooth surfaces put into mechanical contact in an O2/H2O mixed ambient below T <200 °C via arrays of SiOxHx molecules connecting into a continuous macroscopic bonding interphase. Nano-scale surface planarization via wet chemical processing and new spin technology are compared via Tapping Mode Atomic Force Microscopy (TMAFM) , before and after nano-bonding. Nanobonding uses precursor phases, 2D nano-films of beta-cristobalite (beta-c) SiO2, nucleated on Si(100) via the Herbots-Atluri (H-A) method [1]. beta-c SiO2 on Si(100) is ordered and flat with atomic terraces over 20 nm wide, well above 2 nm found in native oxides. When contacted with SiO2 this ultra-smooth nanophase can nucleate and grow domains with cross-bridging molecular strands of hydroxylated SiOx, instead of point contacts. The high density of molecular bonds across extended terraces forms a strong bond between Si-based substrates, nano- bonding [2] the Si and silica. A new model of beta-cristobalite SiO2 with its <110> axis aligned along Si[100] direction is simulated via ab-initio methods in a nano-bonded stack with beta-c SiO2 in contact with amorphous SiO2 (a-SiO2), modelling cross-bridging molecular bonds between beta-c SiO2 on Si(100) and a-SiO2 as during nanobonding. Computed total energies are compared with those found for Si(100) and a-SiO2 and show that the presence of two lattice cells of !-c SiO2 on Si(100) and a-SiO2 lowers energy when compared to Si(100)/ a-SiO2 Shadow cone calculations on three models of beta-c SiO2 on Si(100) are compared with Ion Beam Analysis of H-A processed Si(100). Total surface energy measurements via 3 liquid contact angle analysis of Si(100) after H-A method processing are also compared. By combining nanobonding experiments, TMAFM results, surface energy data, and ab-initio calculations, an atomistic model is derived and nanobonding is optimized. [1] US Patent 6,613,677 (9/2/03), 7,851,365 (12/14/10), [2] Patent Filed: 4/30/09, 10/1/2011