Sputtered Epitaxial Perpendicular MTJs for Next-Generation MRAM

Background

Magnetic material-based memories are an emerging class of devices promising non-volatility, high-speed and low power. The first generation of memory devices, known as Spin Transfer Torque (STT) magnetic random-access memory (MRAM), are in large scale production for embedded memory replacement. However, slow STT-MRAM write speeds are limiting its application beyond the eFlash space. With a view towards expanding the application space, our team is working on alternative switching mechanisms of which Spin-Orbit-Torque (SOT) and Voltage Control of Magnetic Anisotropy (VCMA) are the most promising. However, the state of the art is limited to the polycrystalline CoFeB/MgO system, where defects, grain boundaries and pinholes are common failure points. To solve this, a high risk, high reward, approach is to enable epitaxial MRAM.

State-of-the-art and challenges

At imec, the world record for VCMA-MRAM has been demonstrated on imec’s 300mm platform [1]. Despite this demonstration, the VCMA coefficient of 100 fJ/Vm is far less than the minimum of 300fJ/Vm, and ideally 1000fJ/Vm, required for a true stand-alone device. In the literature the major gap in VCMA coefficients is due to the material quality, where values over 300 fJ/Vm have been shown in epitaxial systems. However, the substrates typically used, for example single crystal MgO, are not industrially feasible.

An in-plane epitaxial MTJ has been demonstrated using a NiAl seed layer deposited via magnetron sputtering on a Si substrate [2]. This process has been reproduced in imec and fabricated up to the MgO, including a Fe layer with perpendicular anisotropy. Whilst this is an important step, the remainder of the MTJ is yet to be made. Typically, this would be done using Ru as an RKKY spacer with CoPt as the ‘Hard Layer’, both critical for pinning of the fixed magnetic reference. However, both materials, and their typical alternatives, are not structurally compatible with a MgO epitaxial system. Thus, the goal of this work will be to develop and combine MgO compatible, epitaxial, layers to build a full MTJ. Some examples of these materials are FePt, NdFeB (Hard-Layers) and RuAl (RKKY).

Experimental details & Methodology

For the material studies in this work, the student will use a brand-new deposition system recently installed in the imec lab. This will allow for rapid testing, ‘failing fast’, of these exploratory materials. These will be analyzed using imec’s broad set of magnetic and structural characterization equipment. The final goal of the work will be to transfer the knowledge and design of the final work to imec’s 300mm deposition tool with the aid of researchers in the team.

Key to the Ph.D research is to:

1. Development of in-depth knowledge of magnetron sputtering processes and complex alloy formation in thin films
2. In-depth study and expertise of crystallographic, composition and interface studies of sputtered thin films
3. Detailed understanding of magnetic characterization, and linking of these to the structural properties