Â鶹ÒùÔº

Two-dimensional (2D) materials, thin crystalline substances only a few atoms thick, have numerous advantageous properties compared to their three-dimensional (3D) bulk counterparts. Most notably, many of these materials allow electricity to flow through them more easily than bulk materials, have tunable bandgaps, are often also more flexible and better suited for fabricating small, compact devices.

Past studies have highlighted the promise of 2D materials for creating advanced systems, including devices that perform computations emulating the functioning of the brain (i.e., neuromorphic computing systems) and chips that can both process and store information (i.e., in-memory computing systems). One material that has been found to be particularly promising is hexagonal boron nitride (hBN), which is made up of boron and nitrogen atoms arranged in a honeycomb lattice resembling that of graphene.

This material is an excellent insulator, has a wide bandgap that makes it transparent to visible light, a good mechanical strength, and retains its performance at high temperatures. Past studies have demonstrated the potential of hBN for fabricating memristors, electronic components that can both store and process information, acting both as memories and as resistors (i.e., components that control the flow of electrical current in electronic devices).

While hBN-based memristors can perform remarkably well, integrating them with existing silicon-based complementary-metal-oxide-semiconductor (CMOS) technology has proved challenging. Most proposed integration methods are either too expensive to be implemented on a large scale or fail to reliably yield defect-free devices.

Researchers at Arizona State University, King Abdullah University of Science and Technology (KAUST) and other institutes recently devised a new scalable strategy to grow hBN films at CMOS-compatible temperatures and realize memristors that are compatible with existing electronics. Their proposed approach, in Nature Nanotechnology, relies on a technique known as electron cyclotron resonance plasma-enhanced chemical vapor deposition (ECR-PECVD).

"Hexagonal boron nitride (hBN) is particularly attractive for non-volatile resistive-switching devices (that is, memristors) due to its outstanding electronic, mechanical and chemical stability," wrote Jing Xie, Ali Ebadi Yekta and their colleagues in their paper. "However, integrating hBN memristors with Si-CMOS electronics faces challenges as it requires either high-temperature synthesis (exceeding thermal budgets) or transfer methods that introduce defects, impacting device performance and reliability.

"We introduce the synthesis of hBN films at CMOS-compatible temperatures (<380°C) using electron cyclotron resonance plasma-enhanced chemical vapor deposition to realize transfer-free, CMOS-compatible hBN memristors with outstanding electrical characteristics."

Using their proposed strategy, the researchers could deposit hBN onto wafers directly, without having to transfer them. The films they created were highly uniform across large areas and were made up of many crystals (i.e., polycrystalline).

"Our studies indicate a polycrystalline structure with turbostratic features in as-deposited hBN films and good wafer-level uniformity in morphology (size, shape and orientation)," wrote Xie, Yekta and their colleagues. "We demonstrate a large array of hBN memristors achieving high yield (~90%), stability (endurance, retention and repeatability), programming precision for multistate operation (>16 states) and low-frequency noise performance with minimal random telegraph noise."

Using the hBN-based films they created, the researchers created a large array of memristors that were found to perform remarkably well, retaining their performance over time. In addition, they successfully integrated the memristors with existing CMOS technology.

"We directly integrate memristive devices on industrial CMOS test vehicles to demonstrate excellent endurance, achieving millions of programming cycles with a high technology readiness level," wrote the authors. "This represents an important step towards the wafer-scale CMOS integration of hBN-memristor-based electronics."

In the future, this proposed strategy could contribute to the integration of memristors based on hBN or potentially also other 2D materials with CMOS-based electronics. Moreover, their work could inspire other research groups to devise similar strategies for the reliable synthesis of quality 2D films.