Electrical characterization of resistive switching memories and development of automated scalpel-SPM methodologies

Drilakis, Christos; Δριλάκης, Χρήστος

dc.contributor.author	Drilakis, Christos	en
dc.contributor.author	Δριλάκης, Χρήστος	el
dc.date.accessioned	2021-01-08T07:48:50Z
dc.date.available	2021-01-08T07:48:50Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/52741
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.20439
dc.description	Εθνικό Μετσόβιο Πολυτεχνείο--Μεταπτυχιακή Εργασία. Διεπιστημονικό-Διατμηματικό Πρόγραμμα Μεταπτυχιακών Σπουδών (Δ.Π.Μ.Σ.) “Μικροσυστήματα και Νανοδιατάξεις”	el
dc.rights	Default License
dc.subject	Resistive switching memories	en
dc.subject	Titanium oxide	en
dc.subject	Nanoparticles	en
dc.subject	Conductive filament observation	en
dc.subject	AFM	en
dc.subject	Μνήμες εναλλαγής αντίστασης	el
dc.subject	Οξείδιο του τιτανίου	el
dc.subject	Νανοσωματίδια	el
dc.subject	Απεικόνιση αγώγιμου δρόμου	el
dc.subject	Μικροσκόπιο ατομικής δύναμης	el
dc.title	Electrical characterization of resistive switching memories and development of automated scalpel-SPM methodologies	en
heal.type	masterThesis
heal.secondaryTitle	Χαρακτηρισμός διατάξεων εναλλαγής αντίστασης και ανάπτυξη αυτοματοποιημένου συστήματος	el
heal.classification	Physics	en
heal.classification	Φυσική	el
heal.language	en
heal.access	campus
heal.recordProvider	ntua	el
heal.publicationDate	2020-11-04
heal.abstract	Filamentary-based resistive switching memories represent an intriguing class of emerging electronic devices. They rely on ion migration in order to create or dissolve nanoscale conductive ﬁlaments (CF) within insulating materials. Depending on the formation or rupture of the CF, the device switches its resistance state and stores a logical bit. The characterization and understanding of these devices have been a challenge for the scientific community for many years. Conductive Atomic Force Microscopy (C-AFM) is a powerful technique that enables the physical and electrical characterization of any type of surface, including thin films for resistive switching memories applications. Particularly, a recently advanced technique called SPM tomography or Scalpel SPM provides depth profiling by tip-induced material removal, while also probing the local conductivity. This can generate multiple planar conductive maps of a resistive memory cell that can be used for a 3D tomogram production. In this work, we studied Valence Change Memories that can be used as nonvolatile storage class devices. Specifically, our devices were based on a MIM structure (Metal-Insulator-Metal) with Au and Ti as top electrodes, a sub-stoichiometric titanium oxide bilayer configuration as the active switching layer and Au as bottom electrode. We tested two kinds of devices, the reference sample and an identical one with embedded Pt nanoparticles (NPs) in the middle of the resistive layer. The NPs incorporation provided enhanced performance in terms of memory window and DC endurance over consecutive cycles as well as wider CF, mainly due to enhancement of the electric field in the vicinity of the NPs during the set process. The reference sample had a memory window of 102 and the NPs sample had 103. Moreover, the NPs sample maintained the same memory window for 10 DC consecutive cycles, whereas the reference sample window started degrading immediately after the first cycle. Moreover, we utilized the Scalpel-SPM technique in order to perform a 3D imaging of the CF for our two devices. After unsuccessful attempts on removing the top electrode (TE), we used a Pt tip directly in the oxide region as a TE to set the device in the ON state. Afterwards, we were able to produce a 3D tomogram of the CF for the two samples. The CF in the reference sample showed a relatively constant size, whereas in the nanoparticles-embedded sample it resembled a cone. The larger size of the filament in the NPs vicinity is related to enhanced oxygen vacancy generation probability during the set process. Additionally, we observed a more localized and standard-shaped filament in the NPs sample than the one in the reference sample which was randomly shaped and positioned, indicating the imposition of the CF formation/rupture into specific locations by the NPs. The maximum current for the CF and another conductive spot in the current map disclosed similar values in the second oxide layer, indicating that the resistive switching process takes place mostly in the first layer. The area calculation for the CF showed a conical shape for both samples, with the NPs-embedded sample filament size increasing in a faster rate as a function of the depth. The CF in both cases is smaller close to the surface and bigger in the bottom, caused by the accumulation of oxygen anions in the surface (less oxygen vacancies) during the set process. The theoretical resistance of the filament showed similar evolution with the corresponding experimental one during scalpeling, while having significant value difference, mostly due to high tip-sample contact resistance. Finally, we provide insights regarding the difficulty of analyzing a Scalpel SPM dataset which consists of numerus images and requires a lot of processing time. This effect lead to the development of automated routine, by creating scripts in Python that could speed up the editing of our experiments. For that reason, we created algorithms that could automate many steps of a Scalpel-SPM analysis, like the leveling of read-out images, depth and removal rate calculation and current map auto scale setting. In that way, we can have a complete and rapid understanding of our experiments result.	en
heal.advisorName	Tsoukalas, Dimitris	en
heal.advisorName	Τσουκαλάς, Δημήτριος	el
heal.committeeMemberName	Raptis, Yannis	en
heal.committeeMemberName	Ράπτης, Ιωάννης	el
heal.committeeMemberName	Giannakopoulos, Konstantinos	en
heal.committeeMemberName	Γιαννακόπουλος, Κωνσταντίνος	el
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Εφαρμοσμένων Μαθηματικών και Φυσικών Επιστημών	el
heal.academicPublisherID	ntua
heal.numberOfPages	71 σ.	el
heal.fullTextAvailability	false