Quantum Transport at very low temperatures in resistive memories
Post-doctoral project at the PHELIQS/INAC – LETI/DCOS, Cea 17 rue des Martyrs, 38042 Grenoble
Cedex 9, France
Contact : Jean-Pascal Brison, jean-pascal.brison@cea.fr
Tel : +33 (0)4 38 78 52 48
General framework: Quantum effects observed at low temperature in low dimensional conductors are
now well understood: conductance quantum, universal quantum fluctuations, weak localization effects,
Coulomb blockade, multiple Andreev reflexions at normal/superconducting interfaces... They can even
be used as the signature of physical singularities controlling electrical transport (tunnel or point contact,
1D filament...). This idea is the cornerstone of a new approach we want to use in order to shed light on
the question of variability observed in nano-devices used as resistive memories (RRAM). This will be
done both in "standard" dielectric RRAM, and in a new generation based on strongly correlated electron
materials: Mott insulators.
Indeed, resistive memories (RRAM) are in principle highly promising candidates for future memories
combining non-volatility, speed and low consumption. They store information by changing their
resistance state (low or high), which is obtained by the formation or breach of a conducting filament,
induced by a voltage pulse. The LETI/DCOS is expert in the fabrication of such devices, realized with
thin layers (between 5 and 30nm) of dielectric materials, driven in a low resistance state by the formation
of an oxygen-vacancy filament (so-called OXRAM), or by the formation of a metallic filament (CBRAM,
with diffusion of metallic ions from one of the electrodes). However, the major challenge (worldwide) for
these RRAM is to control their variability, notably in the high resistance (off) state.
The common wisdom is that this variability is directly connected to the filamentary conduction regime,
common to both OXRAM and CBRAM. The main target of this project, which is “basic research
oriented”, is to shed some light on the mechanisms driving the variability, in order to find a way to lower
the variability, or to replace the dielectric layer by new materials where the problem could be handled
differently. The “new materials” of interest are Mott insulators, which present the same “memory effect”
as the dielectrics, but have an intrinsic thermodynamically stable insulating or conducting state,
controlled by pressure, doping, stress, temperature… Ideally, succeeding to induce the change of the
resistive state in the bulk of the Mott insulator would certainly suppress the main cause of variability, but
even if we stick to a filamentary mechanism, the nature of these filaments could be very different from
those in the dielectrics.
Project: it is mainly focused on the realization and the analysis of transport measurements at very low
temperature (down to 10mK), and under magnetic field (test of the quantum coherent effects like weak
localization or universal conductance fluctuations, appearing in low dimensional systems), on two kinds
of devices:
• On state of the art CBRAM and OxRAM, selected by the DCOS, where we suspect anomalies
in the conducting filaments.
• Single crystals and later thin films of Mott insulators, produced at PHELIQS/INAC and
SPINTEC/INAC, where we hope to produce the conducting filaments by a purely electronic
mechanism (without ion displacements like in the dielectrics).
These low temperature measurements will be performed at PHELIQS, whose teams master both
quantum transport in nanoscale devices and the physics of strongly correlated electron systems. We
have targeted a first family of Mott insulators: spinels of type AM
4
X
8
(GeV
4
S
8
, GaV
4
S
8
…), for which the
metal-insulator transition happens without structural changes (it remains cubic), with a small Mott gap
(200-400meV), and for which both single crystals and thin films can be grown. Owing to the small Mott
gap, filamentary transitions are induced with low electric fields (1-10kV/cm)
4
.
The first task of the project is to study low temperature transport in the CB-RAM and OXRAM devices
from the LETI. The simple idea, mentioned above, is that the variability of the devices may arise from
residual conduction path of very low dimensions (1D or 0D), which should be revealed by
signatures on the quantum transport at low temperature: Coulomb blockade, tunneling resistance,
4Janod et al., Resistive Switching in Mott Insulators and Correlated Systems
Advanced Functional Materials, 2015, 25, 6287-6305