PHYSICAL REVIEW APPLIED 24, L011002 (2025)
Letter
Entropy-assisted nanosecond stochastic operation in perpendicular
superparamagnetic tunnel junctions
Lucile Soumah ,
1,*
Louise Desplat ,
1,2,*,†
Nhat-Tan Phan ,
1
Ahmed Sidi El Valli ,
1
Advait Madhavan ,
3,4
Florian Disdier ,
1
Stéphane Auffret,
1
Ricardo C. Sousa ,
1
Ursula Ebels ,
1
Mark D. Stiles ,
3
and Philippe Talatchian
1,‡
1
Univ. Grenoble Alpes, CEA, CNRS, Grenoble INP, SPINTEC, 38000 Grenoble, France
2
Nanomat group, Q-MAT Research Center and European Theoretical Spectroscopy Facility, Complex and
Entangled Systems from Atoms to Materials (CESAM), Université de Liège, Sart Tilman B-4000, Belgium
3
Physical Measurement Laboratory, National Institute of Standards and Technology, Gaithersburg,
Maryland 20899, USA
4
Institute for Research in Electronics and Applied Physics, University of Maryland, College Park,
Maryland 20742, USA
(Received 5 March 2024; revised 6 May 2025; accepted 22 May 2025; published 2 July 2025)
We demonstrate good agreement between mean dwell times measured in 50-nm-diameter perpendic-
ularly magnetized superparamagnetic tunnel junctions (SMTJs) and theoretical calculations based on
Langer’s theory. Due to a large entropic contribution, the theory yields Arrhenius prefactors in the fem-
tosecond range for the measured junctions, in stark contrast to the typically assumed value of 1 ns. Due
to the low prefactors, and fine tuning of the perpendicular magnetic anisotropy, we report measured mean
dwell times as low as 2.7 ns under an in-plane applied field at negligible bias voltage. Under a perpendicu-
lar applied field, we predict a Meyer-Neldel compensation phenomenon, whereby the prefactor scales like
an exponential of the activation energy, in line with the exponential dependence of the measured dwell time
on the field. We further predict the occurrence of (sub)nanosecond dwell times as a function of effective
anisotropy and junction diameter at zero bias voltage. These findings hopefully pave the way toward the
development of ultrafast low-power unconventional computing schemes operating by leveraging thermal
noise in perpendicular SMTJs, which can be scaled down below 20 nm.
DOI: 10.1103/v53b-mz5b
In magnetism, mastering thermal activation is essential
for understanding the complex interplay of temperature
and dynamics that governs the behavior of spintronic
nanostructures. While enhancing thermal stability is cru-
cial for preventing information loss in nonvolatile memory
[1–3], reduced stability facilitates rapid state switchings,
enabling energy-efficient cognitive computing schemes
[4,5].
Magnetic tunnel junctions (MTJs) are a prime exam-
ple of devices in which the study of thermal activation is
important. MTJs consist of two ferromagnetic layers sep-
arated by an insulating oxide and exhibit two metastable
states, corresponding to the relative orientations of the
magnetization in the two layers, namely, parallel (P) and
antiparallel (AP). These states, readable through the tun-
neling magnetoresistance (TMR), have distinct resistance
levels and are separated by an energy barrier. In particular,
we refer to junctions in which thermal fluctuations induce
*
Authors L.S. and L.D. contributed equally to this work.
†
Contact author: louise.desplat@cea.fr
‡
Contact author: philippe.talatchian@cea.fr
random switchings between the two states at a scale of a
few seconds and below [6–8] as superparamagnetic tun-
nel junctions (SMTJs). Despite their inherently stochastic
resistance fluctuations, the corresponding state probabil-
ities can be controlled deterministically, either through
current-induced spin-transfer torques (STTs) [9–11] or via
an external magnetic field. This tunability, coupled with
their energy efficiency, has made SMTJs highly appealing
for cognitive applications, including stochastic implemen-
tations of artificial neural networks [12], brain-inspired
[13], and probabilistic schemes [14,15].
A strategy to further reduce energy consumption in these
schemes is to lower the mean dwell times between magne-
tization reversals by reducing the energy barrier between
states to a few kT
RT
[16], where k is Boltzmann’s con-
stant and T
RT
is the room temperature. Macrospin models
suggest that the mean dwell times cannot be lower than a
characteristic attempt time, τ
0
≈ 1 ns, even with negligi-
ble energy barriers, implying a significant speed limitation
[17]. Such models imply [6,17] that in-plane MTJs have
larger limiting speeds than perpendicular ones, focusing
research on in-plane MTJs, which are reported to reach
dwell times of a few nanoseconds [6,18,19].
2331-7019/25/24(1)/L011002(7) L011002-1 © 2025 American Physical Society
