SEMATECH Technologists Demonstrate Breakthrough Novel solutions for high mobility channel CMOS devices, FinFETs, RRAMs and technologies beyond CMOS devices

SEMATECH experts reported on innovative approaches to realize advanced CMOS logic and memory device technologies and 3D through-silicon via (TSV) manufacturing at the International VLSI Technology, System and Applications Symposium (VLSI-TSA) on April 23-25, 2012.

Today nearly all electronic devices are built on complementary metal-oxide semiconductor (CMOS) technology. For over half a century, silicon-based materials have been the basic layers used in the manufacturing of CMOS transistors; however these staple materials, as well as materials derived from silicon such as insulators and contact metals, are reaching their limits as the industry looks to lower power dissipation in CMOS devices and as scaling approaches the physical limits of silicon transistors.

In a series of nine research papers, an international team of SEMATECH researchers reported on innovative materials and new transistor structures to address key aspects of transistor performance, power and cost. The papers, selected from hundreds of submissions, outlined leading-edge research in high-k/metal gate (HKMG) materials, resistive RAM (RRAM) memory and planar and non-planar CMOS technologies.

“Through intense research and development efforts, SEMATECH is developing manufacturable solutions and practical implementation approaches for innovative materials in future transistor structures,” said Raj Jammy, vice president of Emerging Technologies. “The research that was presented at VLSI TSA demonstrates SEMATECH’s leadership in developing new materials, processes and concepts that will pave the way for emerging technologies.”

One potentially industry-changing technology, a direct metal bonding interconnect approach, was introduced by Sitaram Arkalgud, director of SEMATECH’s 3D Interconnect program. In order for 2.5D and 3D integration to reach its full potential, chip-to-interposer and chip-to-chip interfaces have to support a very large number of power and signal connections. Today most solder-based interconnect schemes will not scale sufficiently due to mechanical, electrical, thermal and reliability limitations.

Arkalgud revealed SEMATECH’s copper-to-copper direct bonding (CuDB) technology as a promising technology to aggressively scale chip-to-chip interconnects and keep pace with advances in TSV. He also discussed recent progress and remaining technical and economic hurdles in moving toward high-volume manufacturing of CuDB interconnects.

SEMATECH technologists also reported technical advances in the following areas:

Silicon Channel Devices

Evaluating stress-induced leakage current (SILC) in full gate-last (FGL) high-k/metal gate devices to address sources of SILC and propose possible process options for improvement. A high-quality interlayer during gate stack formation was found to be critical to improving FGL device performance and reliability.
Modeling positive bias temperature instability (PBTI) degradation in Zr-doped HfO2 gate stacks by considering fast and slow electron trapping processes. PBTI was found to improve when the fast trapping component was suppressed.

Non-Silicon Channel Devices

Using different ALD oxidizers to study the effects of III-V oxides on device performance. With a O3 precursor, As-As, AsOx, GaO, and In2O3 were found to be the main native oxides/byproducts. H2O-based precursors remain stable with no III-V oxide detected throughout a low temperature flow. Electrical performance also improved with H2O-based high-k, suggesting that H2O-based ALD is the key process for III-V CMOS.
Exploring alternative high-k gate dielectrics for III-V, Ge and Si MOSFETs. High-field carrier mobility and MOSFET parameter characteristics were improved by atomic layer deposition (ALD) of a thin beryllium oxide layer to passivate the interface between the Si channel and high-k gate dielectric.

Non-Planar Devices

Studying FinFET Vt tuning. Both performance and the electrical properties of the gate stack were improved by an Al implantation, representing progress towards realizing multi threshold voltage FinFET device architectures for the 14 nm node and beyond.
Studying the impact of fin doping on high-k/midgap metal gate SOI FinFETs. Threshold voltage can be effectively modulated with doping in ~25 nm wide fins. For sub-10 nm fin widths, however, the active dopant atoms must be precisely placed inside the fins, which ion implantation cannot do. A conformal doping technique with perfect dose control, such as monolayer doping, was discussed which may be the solution for future planar and non-planar devices.
Evaluating the parasitic capacitance of planar FETs and double-gated (DG) FinFETs. Optimization with a fixed fin-to-height ratio significantly reduces parasitic capacitance, which renders DG FinFETs comparable to planar FETs. Fin width and height must be controlled in the DG FinFETs, otherwise the parasitic capacitance uniformity will degrade.
Investigating the impact of source/drain (S/D) activation anneal on GAA pFETs. Low-temperature pFETs were fabricated and benchmarked against devices with a S/D activation anneal. When S/D is implanted before the gate spacer, the un-annealed devices exhibited higher peak transconductance and drain current but have a higher off-current than their annealed counterparts. Pre- and post-spacer S/D implant schemes were also explored.

Advanced Non-Volatile Memory

RRAM switching performance up to 1×108 cycles at low power and a 100x reduction of the high-resistance-state current was achieved by identification and utilization of key parameters for establishing superior control of the RRAM conductive filament formation.

SEMATECH’s Front End Processes program, located at the College of Nanoscale Science and Engineering (CNSE) of the University at Albany, is exploring innovative materials, new transistor structures, and alternative non-volatile memories to address key aspects of system-level performance, power, variability and cost, and to help accelerate innovation in the continued scaling of logic and memory applications.

If you liked this article, please give it a quick review on ycombinator or StumbleUpon. Thanks