Researchers at Wenzhou University, MIT, the Shanghai Institute of Technical Physics (Chinese Academy of Sciences), the University of Chinese Academy of Sciences, Caltech, the Hong Kong University of Science and Technology, and Fuzhou University have developed a general method for growing large-scale, single-crystalline films of two-dimensional metal oxides that are directly compatible with silicon technology - a longstanding barrier to building spintronic, ferroelectric, and neuromorphic components directly onto conventional chips.
Hybrid integration of 2D single-crystal metal oxides with silicon circuits is central to the "More-than-Moore" roadmap for extending chip functionality beyond simple transistor scaling, but growing such oxides as true single crystals has proven difficult. Conventional epitaxy relies on strong chemical bonding to lock a growing film into registry with its substrate, which causes defects when integrating chemically dissimilar materials like oxides and silicon. Van der Waals epitaxy on layered substrates such as mica avoids that problem by using weak, non-directional interactions instead, but the resulting surfaces are typically too chemically inert to guide film growth in a single orientation, so films tend to form as a mosaic of randomly oriented crystallites rather than a true single crystal. The team's solution was to reconstruct the oxygen atomic plane exposed at the surface of cleaved mica. Quantum mechanics calculations showed that this reconstruction creates a periodic potential landscape that energetically favors one crystallographic orientation for incoming oxide atoms, without requiring strong chemical bonds - effectively letting correctly oriented crystal nuclei grow faster and dominate while misoriented ones are starved of material.
Using this reconstructed mica template, the researchers grew single-crystal 2D films of four transition metal monoxides - cobalt oxide, iron oxide, nickel oxide, and manganese oxide - along with doped variants, each coalescing seamlessly into continuous, unidirectionally oriented single-crystalline films. As a flagship demonstration, the team grew a centimeter-scale iron-doped cobalt oxide (Fe-doped CoO) film exhibiting room-temperature ferromagnetic semiconductor behavior, with a Curie temperature of up to 430 K (about 157°C), comfortably above room temperature, addressing a key limitation of most dilute magnetic semiconductors, which typically lose their magnetism well below the temperatures needed for practical devices.
The authors point to applications across spintronics (spin filters, magnetic tunnel junction components, and spin-polarized current sources integrated directly onto silicon), neuromorphic computing (magnetic semiconductors capable of synaptic-like tunable resistance and magnetization switching), and quantum information, where cleanly interfacing correlated 2D oxides with dielectric substrates could enable hybrid quantum devices controlled with ordinary semiconductor electronics. Because mica is inexpensive, naturally abundant, and available in large high-quality sheets, the approach also sidesteps the cost and scarcity issues associated with some other van der Waals growth substrates. The authors note that transferring these films onto working silicon circuitry, controlling defect densities, and engineering p-type and n-type doping for transistor structures remain open engineering challenges ahead of commercial use.