Demonstration of Nanosecond Operation in Stochastic Magnetic Tunnel Junctions

Abstract

Magnetic tunnel junctions operating in the superparamagnetic regime are promising devices in the field of probabilistic computing, which is suitable for applications like high-dimensional optimization or sampling problems. Further, random number generation is of interest in the field of cryptography. For such applications, a device’s uncorrelated fluctuation time-scale can determine the effective system speed. It has been theoretically proposed that a magnetic tunnel junction designed to have only easy-plane anisotropy provides fluctuation rates determined by its easy-plane anisotropy field and can perform on a nanosecond or faster time-scale as measured by its magnetoresistance’s autocorrelation in time. Here, we provide experimental evidence of nanosecond scale fluctuations in a circular-shaped easy-plane magnetic tunnel junction, consistent with finite-temperature coupled macrospin simulation results and prior theoretical expectations. We further assess the degree of stochasticity of such a signal.

Voltage-controlled superparamagnetic ensembles for low-power reservoir computing

“…While this figure is not precise, it indicates the potential for ultra-low-energy consumption. Furthermore, Safranski et al 34 have verified that uncorrelated thermally-driven switching can be observed in CoFeB MTJs for dwell times of 5 ns and above, which we can take as a floor for the operational speed of these nanodots. Longer dwell times can easily be induced by scaling up the size of the magnetic elements, thus increas-ing the energy barrier between states.…”

Probabilistic computing with p-bits

“…It is placed in series with a transistor, and the drain voltage is thresholded by an inverter 3 to obtain a random binary output bit whose average value can be tuned through the gate voltage V IN . It has been shown both theoretically 24,25 and experimentally 26,27 that s-MTJ-based p-bits can be designed to generate new random numbers in times ∼ nanoseconds. The same circuit could also be used with other fluctuating resistors 28 , but one advantage of s-MTJ's is that they can be built by modifying magnetoresistive random access memory (MRAM) technology that has already reached gigabit levels of integration 29 .…”

Benchmarking a Probabilistic Coprocessor

“…The overall power consumption of one hardware p-bit is around 20µW which is orders of magnitude better than CMOS alternatives 10 . Another advantage of this compact RNG-source is that it is based on stochastic MTJs which can operate at nanosecond speeds which has been shown theoretically as well as experimentally [16][17][18][19] . The random number generation of the p-bit design can be tuned with the input voltage at the gate of the NMOS-transistor V IN and has a bipolar output voltage V OU T = ±V DD /2.…”