Extreme scalability of MRAM cells and CMOS integration for hybrid circuits – EXCALYB

E-beam lithography, reactive ion etching (RIE), ion beam etching (IBE)

The project reached it's first milestone number with the nanofabricaton of memory elements with electrical diameter sub-30 nm, which corresponds to the diameter of the stud diameter calculated from the measured resistance value.

Achieving functional perpendicular MRAM cells with TMR signals around 70-80% and switching critical currents of the order of 3MA / cm² for pulses of 100ns. The thermal stability calculated from measurements obtained is between 30-50kBT.

Cells to greater stability between 40 and 110kBT for cells with MgO coverage and stability values vary between 100 and 300kBT for cells intended for writing termally assisted anisotropy re-orientation.

1) Nanofabrication of sub-20 nm magnetic cells with different RIE and IBE etching strategies.
2) Increase TMR, reducing the critical current STT switching, scalability validation perpendicular and planar self-referencing junctions.
3) Integration of CMOS circuit junctions with 20nm diameter

Self-referenced multi-bit thermally assisted magnetic random access memories
Stainer, Q. and Lombard, L. and Mackay, K. and Lee, D. and Bandiera, S. and Portemont, C. and Creuzet, C. and Sousa, R. C. and Dieny, B.
Applied Physics Letters, 105, 032405 (2014)

Influence of magnetic electrodes thicknesses on the transport properties of magnetic tunnel junctions with perpendicular anisotropy
Cuchet, L., B. Rodmacq, S. Auffret, R.C. Sousa and B. Dieny
Applied Physics Letters 105 (2014) 052408

Effects of the heating current polarity on the writing of thermally assisted switching-MRAM
Chavent, A., J. Alvarez-Hérault, C. Portemont, C. Creuzet, J. Pereira, J. Vidal, K. Mackay, R.C. Sousa, I.L. Prejbeanu and B. Dieny
IEEE Transactions on Magnetics 50 (2014) 3401504

Field dependence of spin transfer torque switching current in perpendicular magnetic tunnel junctions
Cuchet, L., R.C. Sousa, L. Vila, S. Auffret, B. Rodmacq and B. Dieny
IEEE Transactions on Magnetics 50 (2014) 4401404

Influence of cooling rate in planar thermally assisted magnetic random access memory: Improved writeability due to spin-transfer-torque influence
A. Chavent, C. Ducruet, C. Portemont, C. Creuzet, L. Vila, J. Alvarez-Hérault, R. C. Sousa, I. L. Prejbeanu and B. Dieny
Appl. Phys. Lett. 107, 112403 (2015)

Respective influence of in-plane and out-of-plane spin-transfer torques in magnetization switching of perpendicular magnetic tunnel junctions
A. A. Timopheev, R. Sousa, M. Chshiev, L. D. Buda-Prejbeanu, and B. Dieny
Phys. Rev. B 92, 104430 – Published 28 September 2015

The aim of the EXCALYB project is to develop the processes, material stacks and CMOS integration for new types of high density, highly scalable non-volatile memories. This will allow for high density (feature size <20nm corresponding to 1 Tbit/in²) with high data rates, as well as a very significant reduction of power consumption of electronic circuits in standby mode.

EXCALYB project aims to establish an innovative sub-20nm technology platform to be used in the evaluation of magnetic tunnel junction based spintronic devices at nanoscale dimensions. The developed process flow will not be specific only to MRAM cells, and could be used to evaluate any device based on magnetic tunnel junctions, such as spin-torque oscillators, magnetic field sensors and generally any electrically connected pillar device having magnetic materials. A successful integration will provide an alternative approach for the evaluation of integrated hybrid circuits, using CMOS and magnetic elements. This low-cost platform unique in France is complementary with ongoing efforts to create 200/300mm magnetic backend lines. The proposed alternative significantly reduces the costs associated with evaluating circuit designs that include magnetic tunnel junctions. The cost reduction comes from the fact that the 3 to 4 backend mask levels are not required each time a new design needs to be evaluated. Mask reticles for 200/300mm magnetic backend process can account for significant initial costs, becoming the major roadblock to use emergent technology used in new applications.

The EXCALYB project aims to provide the technology missing for these applications, requiring the availability of high performance non-volatile memory with associated non-volatility for very low-power consumption. The project will allow demonstrate these devices and explore their scalability in future generations, by reducing the cell size for extreme integration low power operation. EXCALYB will leverage the existing knowledge of each partner to achieve a significant scientific and technological breakthrough. Magnetic memory cell concept breakthroughs pursued in this project could be used directly in novel MRAM memory implementations by Crocus Technology. The explored concepts provide a clear technology roadmap for MRAM below 20nm cell sizes. This will allow strengthening the intellectual property and know-how enabling the deposition of devices with important industrial applications. It will strengthen the position of Crocus Technology as a French/European contender in the high performance non-volatile arena.

The research involves in particular bringing to maturity a perpendicular MTJ stack for sub-20 dimensions using thermal assistance for the spin torque writing process. Our own initial results have shown the world best figure of merit between thermal stability and writing current. There is considerable improvement to expect from the first demonstration, especially in terms of TMR amplitude, leading to even higher values of thermal stability to STT switching current ratio. Establishing a process flow to sub-20nm dimensions will allow experimental validation of the expected scalability. It will be a significant breakthrough to validate ultimate density perpendicular cells, in both standalone cells and in integrated hybrid circuits. The concept of hybrid non-volatile circuits is extremely appealing due to the stand-by power savings that it can achieve. The in-plane self-reference cells represent a low-density high value application for embedded high temperature applications MRAM that could quickly find market acceptance. The project will test the scalability of current cells, and provide the material research necessary to scale these cells down to 45nm, corresponding to 3 additional technology generations from the current 130nm cell size.