New Kind of Magnetoresistance Has Implications for Semiconductor Industry

Scientists at the University of Minnesota, in collaboration with researchers from Pennsylvania State University have discovered the existence of magnetoresistance in topological insulator-ferromagnetic bilayers. This discovery has significant implications for the semiconductor industry, and opens up the door to enabling low power computing, brain-like computing, and chips for robots in the near future.

The details of their research are published in Nature Communications under the title “Unidirectional spin-Hall and Rashba−Edelstein magnetoresistance in topological insulator-ferromagnet layer heterostructures.” The study confirms the existence of such unidirectional magnetoresistance and reveals that the adoption of TI, compared to heavy metals, improves the magnetoresistance performance by about twice at a temperature of 150 Kelvin (-123.15 Celsius).

TIs have been recently found to hold promise in improving the writing energy efficiency of magnetic memory cells. But the device geometry demanded a new magnetoresistance phenomenon to accomplish the read function of the memory cell. From an application perspective, this discovery is exciting; it resolves the problem of the missing read functionality of TI-based memory devices.

This discovery has significant implications for the semiconductor industry, and opens up the door to enabling low power computing, brain-like computing, and chips for robots in the near future.

As a joint project, the paper is authored by scientists from ECE, Robert F. Hartmann Chair Prof. Jian-Ping Wang, Yang Lv, Delin Zhang and Mahdi Jamali from the University of Minnesota, and James Kally, Joon Sue Lee and Nitin Samarth from Department of Physics at Pennsylvania State University.

What is magnetoresistance and how does an MRAM work?

Magnetoresistance is the tendency of a material to change its electrical resistance when an externally-applied magnetic field or its own magnetization is changed. The phenomenon has found its success in hard disk drive read heads, magnetic field sensors, and magnetoresistive random access memory (MRAM), a rising star among memory technologies.

While magnetic recording currently dominates data storage applications, MRAM is slowly gaining a foothold in the memory market. Externally, unlike hard disk drives which have mechanically spinning disks and swinging heads, the MRAM looks like a chip you would find soldered on a circuit board in a computer or mobile device. They are used for information storage because some ‘hard’ magnetic materials can hold their orientation of magnetization for a long time, before it is reset by an externally applied magnetic field.

Inside an MRAM chip, there are millions of nanometer-sized pillars called ‘cells’. Typically, each cell contains one bit of binary information, a ‘0’ or ‘1’. An MRAM cell can be very sophisticated, but it contains three key layers, two magnetic thin film layers sandwiching an oxide thin film layer. They are about a fraction of a nanometer to a few nanometers thick. And laterally the pillar is about tens of nanometers. Such a structure, or device is called magnetic tunnel junction (MTJ).

For an MTJ to function as a memory cell, it must fulfill the ‘write’ and ‘read’ functionality. This means a ‘0’ or ‘1’ can be written by an electrical current on to the MTJ and later can be read back from it electrically as well. To write a different bit of information into an MTJ, the magnetization of one of the magnetic layers is switched to the opposite direction by electrons whose spins are set or polarized by the other magnetic layer.

The spin of an electron is an intrinsic property, like mass and charge. It is a form of angular momentum; we can picture it as if the electron is actually spinning and its direction of the axis of spinning is the ‘spin’ of that electron (like the North of the earth is the spin of the earth). When sufficient number of electrons with spins pointing in the same direction interact with the magnetic layer, the magnetization will be rotated to the direction of spins. That is how a new bit of information is written electrically and stored magnetically.

The read function of an MTJ is realized by a phenomenon called ‘tunneling magnetoresistance (TMR)’. This effect also originates from interactions between electron spins and magnetic layers. It results in higher resistance across the MTJ when the magnetization of two magnetic layers are opposite to each other compared to the resistance when they are parallel to each other. Therefore, the magnetization of the magnetic layer, which represents the information stored, can be read out by electrical means.

The MRAM as it currently stands is not energy efficient when it comes to writing at certain speeds, and more efficient ways to generate spin-polarized electrons have to be devised. TI materials show promise in resolving this problem and the collaborative research conducted by scientists at the University of Minnesota and Pennsylvania State University is an answer. Their discovery of the existence of magnetoresistance in TI-ferromagnet bilayer structures points to the possibility of an MRAM that can not only achieve its full read functionality, but also provide the energy benefit offered by topological insulators. 

Learn more about MRAM, the device developed by Prof. JP Wang, and Yang Lv, and its impact on stochastic computing in IEEE Spectrum.

This research was funded by the Center for Spintronic Materials, Interfaces and Novel Architectures (C-SPIN) at the University of Minnesota, a Semiconductor Research Corporation program sponsored by the Microelectronics Advanced Research Corp. (MARCO) and the Defense Advanced Research Projects Agency (DARPA).