Ferroelectric Nitride Heterostructures on CMOS Compatible Molybdenum for Synaptic Memristors

Area of Science:

• Materials Science
• Solid-State Physics
• Semiconductor Technology

Background:

• Ferroelectricity in III-nitride (III-N) semiconductors, achieved through alloying with elements like scandium, is crucial for advanced electronic, photonic, and quantum systems.
• Current methods for growing single-crystal III-N semiconductors typically rely on substrates like sapphire, silicon, or silicon carbide, hindering integration with complementary metal-oxide-semiconductor (CMOS) technology.

Purpose of the Study:

• To demonstrate the growth of single-crystalline ferroelectric nitride semiconductors on a CMOS-compatible metal substrate.
• To investigate the epitaxial relationship between wurtzite and body-centered cubic crystal structures for novel semiconductor growth.
• To explore the potential of these materials in neuromorphic computing through the development of synaptic memristors.

Main Methods:

• Epitaxial growth of single-crystalline ferroelectric nitride semiconductors on a metal-molybdenum substrate.
• Characterization of the crystal structure and epitaxial relationship.
• Measurement of ferroelectric properties, including wake-up-free behavior.
• Fabrication and testing of a ferroelectric GaN/ScAlN heterostructure for synaptic memristor applications.

Main Results:

• Successfully demonstrated single-crystalline ferroelectric nitride semiconductors grown on a metal-molybdenum substrate, compatible with CMOS technology.
• Established a unique epitaxial relationship between wurtzite and body-centered cubic structures, enabling growth on polycrystalline molybdenum.
• Observed robust and wake-up-free ferroelectricity in epitaxially grown scandium aluminum nitride (ScAlN) directly on metal.
• Developed a ferroelectric GaN/ScAlN heterostructure memristor capable of emulating spike-time-dependent plasticity.

Conclusions:

• This work presents a viable pathway for integrating III-N semiconductor architectures with mature CMOS technology.
• The developed ferroelectric nitride memristors show significant promise for applications in neuromorphic computing.
• The ability to grow high-quality ferroelectric nitrides on metal substrates opens new avenues for next-generation devices.