Keywords:
Hyperspectral images, Data processing, Unmixing algorithm, STEM-EELS, Electrical biasing, OxRRAMcliquer pour voir la liste des membres du jury/clic here for the jury members
Abstract
As storage capacity requirements are ever greater, research on emerging memory device technologies and their development is booming. Among the emerging memory devices, this thesis is interested in resistive oxide-based random-access memories (OxRRAM). The movement of oxygen ions during the electrical switching of this type of memory is still very poorly understood and knowing it would help to improve and optimize these devices. An operando TEM analysis protocol has allowed the development of an experimental method dedicated to the characterization of variations in oxygen ions distributions during the electrical polarization of this kind of stack. The sample holders, on which the sample preparation used depends, as well as the artefacts to be taken into account during operando switching, have their own specificities. For this, three TEM sample holders dedicated to electrical polarizations in operando TEM were presented, that of NanoFactory, Hummingbird, and finally Protochips.
An EELS hyperspectral image processing protocol based on the VCA algorithm has been developed and applied to two types of memory stacks. The first one is a reference memory stack based on SrTiO3 on which two studies have been made. A first study of this type of stacking was carried out on previously acquired data, the results of which have already been published with a stack based on crystalline SrTiO3. This first analysis confirmed the correct working of the hyperspectral image processing protocol. A second analysis was carried out on a memory stack based on polycrystalline SrTiO3, little known. The STEM-EELS operando analysis of this second sample was carried out through the use of the Protochips microarray associated with the data processing developed, allowing to learn more. This second analysis shows that the use of this data processing based on the VCA algorithm can provide additional information to conventional processing. The second type of stack studied is a La2NiO4-based memory device, designed for applications in the field of neuromorphic applications due to its volatile behavior. The characterization and data processing protocols developed during this thesis can thus serve as supports for the studies of other micrometric-sized memory devices.
Membres du jury/ Jury members :
Dr. |
G. PATRIARCHE |
CNRS, C2N, Université Paris-Saclay, Palaiseau (FRANCE) |
Rapporteur |
Prof. |
N. DOBIGEON |
INP-ENSEEIHT, Université de Toulouse, Toulouse (FRANCE) |
Rapporteur |
Dr. |
M.-P. BESLAND |
CNRS, IMN, Nantes (FRANCE) |
Examiner |
Prof. |
M. VERON |
Grenoble INP, Saint-Martin-d’Hères (FRANCE) |
Examiner |
Dr. |
M. MOUIS |
CNRS, IMEP-LaHC, Grenoble INP, Grenoble, (FRANCE) |
Thesis Director |
Dr. |
D. COOPER |
CEA, LETI-SCMC, Grenoble (FRANCE) |
Thesis Co-director |
Dr. |
C. JIMENEZ |
CNRS, LMGP, Grenoble INP, Grenoble (FRANCE |
Thesis Co-director |
3 parvis Louis Néel - 38000 Grenoble
Accès : TRAM B arrêt Cité internationale
Free entrance - No registration