Low-Energy Truly Random Number Generation with Superparamagnetic Tunnel Junctions for Unconventional Computing

Abstract

Low-energy random number generation is critical for many emerging computing schemes proposed to complement or replace von Neumann architectures. However, current random number generators are always associated with an energy cost that is prohibitive for these computing schemes. In this paper, we introduce random number bit generation based on specific nanodevices: superparamagnetic tunnel junctions. We experimentally demonstrate high quality random bit generation that represents orders-of-magnitude improvements in energy efficiency compared to current solutions. We show that the random generation speed improves with nanodevice scaling, and investigate the impact of temperature, magnetic field and crosstalk. Finally, we show how alternative computing schemes can be implemented using superparamagentic tunnel junctions as random number generators. These results open the way for fabricating efficient hardware computing devices leveraging stochasticity, and highlight a novel use for emerging nanodevices.

Stochastic Memory Devices for Security and Computing

“…Mater. [21] However, such postprocessing circuits generally occupy additional area and consume extra energy. First, the circuit has a relatively large area, namely 2400 F 2 in 65 nm technology, i.e., 10 µm 2 .…”

“…The device structure is similar to the MTJ stack of Figure 2e, as it displays two stable magnetic states, P and AP, corresponding to low and high resistance, respectively. [21,100] On the other hand, the FL lateral dimension in a superparamagnetic tunnel junction is relatively small, thus resulting in the energy barrier being of the order of kT. In a nonvolatile MRAM, the FL geometry and its interface with the MgO are engineered in order to guarantee high perpendicular magnetic anisotropy, resulting in a large ∆E >> kT.…”

“…These two stable states are usually interpreted within the energy potential profile of Figure 8b, where P and AP states are metastable states separated by an energy barrier of amplitude ∆E dictated by the magnetic anisotropy. [21] As a result, the FL magnetization spontaneously switches between the two stable states due to the enhanced sensitivity to thermal fluctuations. [100] As a result, the FL magnetic state is stable at room temperature, and can switch to the opposite magnetization only under an external perturbation, such as a current pulse for STT-MRAM.…”

“…[140,141] Figure 8c shows the time evolution of the MTJ resistance under a sensing current of 10 µA, which induces a negligible perturbation on the FL magnetization, [108] while maximizing the junction lifetime thanks to the reduced stress on the dielectric. [21] Note that in order to obtain an unbiased random bitstream with 50% "0"/"1" probability, the distributions of the transition times should be exactly the same. Figure 8c also shows a digitalized version of the same signal, which is obtained with a comparator.…”

“…[41] The concept of voltage-based TRNG is illustrated in Figure 9a, showing the measured I-V characteristics for a bipolar RRAM device during six set/reset consecutive cycles. [21] Copyright 2017, American Physical Society. [69] The characterization of the random set transition is performed with the sequence of triangular pulses in Figure 9b, including: (1) a positive set pulse to deterministically initialize the device in LRS, (2) a negative reset pulse with a stop voltage V stop to induce transition to the HRS, (3) a positive random set pulse, with amplitude V A = 〈V set 〉 to stochastically induce a set event, and finally (4) a read pulse to measure the cell resistance after the random set.…”