Skip to Main content Skip to Navigation
New interface

Artificial spin ice : from statistical physics to stochastic computing

Abstract : Since their introduction in 2006, artificial spin ice systems have sparked an interest for their similarity with frustrated systems such as water ice or some pyrochlores. These networks of interacting nanomagnets can be observed using simple microscopy techniques and be tailored at will. In this work, we have studied customised geometries in order to observe the evolution of dipolar interactions and develop a device. First, we have studied the tuning of the dipolar interactions induced by rotation of the elements of the square geometry. For an as-grown system that the ground state evolution from antiferromagnetic to ferromagnetic orders observed is consistent with dipolar computations. Taking into account the long-range interactions in our Monte Carlo simulations, we have described the total dipolar energy along with the vertex populations of our system with all its tilted networks. This has led to a study of the thermodynamics of our system and a unique effective could be ascribed. Secondly, we have taken an interest into the possibility of tailoring these artificial spin systems for applications in random number generation. We have seen that these systems are a better fit exploiting domain wall propagation rather than reversals by coupling. We have then established that the main challenge to overcome in large connected structures is the trade-off between nucleation field from the outputs and depinning field of the upper part of the nanostructure. Indeed, it is necessary to ensure than the domain wall propagating in the structure is indeed the injected one and that no parasitic nucleation occurs which is overcome this issue. The domain wall propagation has been characterised in terms of propagation fields and types of reversals. We have shown that our optimised system exhibits reversal induced by the injected domain wall mainly in unidimensional fashion with clear stochastic behaviour. The characterisation of this behaviour has shown that a mean weight value describes the choices observed which do not exhibit any correlation. Our artificial spin system therefore behaves as a Galton board, the prime device for random number generation. This PhD work has shown the versatility of artificial spin systems as spin ice for understanding the influence of tuned interactions in their thermodynamics. This work has also demonstrated their versatility for an application with the characterisation of their stochastic behaviour through an optimisation of the geometry.
Complete list of metadata
Contributor : Thèses UL Connect in order to contact the contributor
Submitted on : Monday, November 8, 2021 - 9:15:08 AM
Last modification on : Sunday, June 26, 2022 - 3:18:46 AM
Long-term archiving on: : Wednesday, February 9, 2022 - 7:45:39 PM


Files produced by the author(s)


  • HAL Id : tel-03418689, version 1



Maryam Massouras. Artificial spin ice : from statistical physics to stochastic computing. Physics [physics]. Université de Lorraine, 2021. English. ⟨NNT : 2021LORR0143⟩. ⟨tel-03418689⟩



Record views


Files downloads