 |
|
Biography
Thomas Windbacher was born in Mödling, Austria, in 1979. He studied physics at the Technische Universität Wien, where he received the degree of Diplomingenieur in October 2006. He joined the Institute for Microelectronics in October 2006 and finished his doctoral degree on engineering gate stacks for field-effect transistors in 2010. From 2010 until the beginning of 2012 he worked as a patent attorney candidate in Leoben. In March 2012 he rejoined the Institute for Microelectronics, where he currently works on the modeling and simulation of magnetic device structures.
Towards a Fully Non-Volatile Information Processing System: Buffered Non-Volatile Magnetic Logic Gate Grid
The ostensibly unbroken demand for cheap and powerful electronics has driven the scaling efforts of the semiconductor industry for many decades. Today the dimensions of CMOS devices have attained a level where physical limits and the steep increase in factory costs for each subsequent technology node will soon bring an end to scaling. In parallel, ensuring qualities like power consumption, interconnection delay, durability, fast operation, and long retention times are of utmost importance for each technology generation.
Therefore, in order to ensure future progress, it is necessary to investigate alternative materials, devices, and computational principles.
The exploitation of spin as a degree of freedom - instead of charge - is very appealing due to its non-volatility, high endurance, and fast operation. However, the transition to non-volatile information processing systems involves a reevaluation and redesign of all basic building blocks and the way in which they interface to achieve the maximum benefit.
For example, non-volatile flip flops based on CMOS/magnetic tunnel junction hybrids already perform quite well with respect to power and speed in competition with pure CMOS devices, but can not challenge pure CMOS with respect to integration density. The intrinsic signal mismatch between the CMOS (voltage, charge) and the micro magnetic (spin transfer torque, resistance) domain causes the need for additional transistors to bridge the gap every time information is read, written or processed. Thus leading rather to an integration density decrease than a densification.
Therefore, we proposed a non-volatile magnetic flip flop which is capable of performing everything in the magnetic domain without restricting its functionality to mere memory. The resulting flip flop offers the advantageous features of spintronics and a very small footprint at the same time. Furthermore, it is possible to create a novel buffered magnetic logic grid with the aid of spin transfer torque majority gates (cf. Fig. 1).
Fig. 1: Illustration of the proposed buffered magnetic logic gate grid and an example of an concatenable 1-bit full adder realized with a single spin transfer torque majority gate and three flip flops acting as buffers.
