Computers presently generate random numbers algorithmically, i.e. from a seed. Therefore, they are not truly random, but pseudorandom, good enough for some purposes but not simulations that require large quantities of random numbers and cryptography. True random numbers are needed for neuromorphic systems, and Monte Carlo simulations, which are extensively used to model and solve complex problems like accurate climate models, biological processes and particle production in high-energy colliders.
The aim of this project is to explore the potential of magnetic materials and devices to generate large numbers of truly random numbers efficiently, with small amounts of energy.
Magnetic noise represents a new opportunity in this regard as small ferromagnetic elements can be two-state systems, with their magnetization “up” and “down” states separated by an energy barrier Eb. If the energy barrier Eb is comparable to the thermal energy kT, where k is the Boltzmann constant and T is the device operating temperature, the magnetization fluctuates between the two states, which is known as superparamagnetism. A magnetic tunnel junction (MTJ) device can convert these thermally driven magnetization fluctuations into two-level electrical signals that are easily read out and, further, MTJs are readily integrated with complementary metal-oxide semiconductor (CMOS) technology.
For more information see:
Return to research projects