In-Gap States of HfO2 Nanoislands Driven by Crystal Nucleation: Implications for Resistive Random-Access Memory Devices

Envisioned extremely scaled, high-performance memory devices request to conduct the step from thin semiconductor films to nanoscale structures and the use of promising high-k materials such as hafnium oxide (HfO2). HfO2 is well suited for use in resistive random-access memory (ReRAM) devices based o...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:ACS applied nano materials 2023-01, Vol.6 (1), p.148-159
Hauptverfasser: Schmidt, Niclas, Rushchanskii, Konstantin Z., Trstenjak, Urška, Dittmann, Regina, Karthäuser, Silvia
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Envisioned extremely scaled, high-performance memory devices request to conduct the step from thin semiconductor films to nanoscale structures and the use of promising high-k materials such as hafnium oxide (HfO2). HfO2 is well suited for use in resistive random-access memory (ReRAM) devices based on the valence change mechanism. Here, we provide a decidedly scaled system, namely, HfO2 nanoislands that are grown by van der Waals epitaxy on highly oriented pyrolytic graphite (HOPG). The electronic and structural properties of these well-separated, crystalline HfO2 nanoislands are investigated by scanning probe methods as well as ab initio methods. The topography reveals homogeneously formed HfO2 nanoislands with areas down to 7 nm2 and a thickness of one unit cell. They exhibit several acceptor- and donor-like in-gap states in addition to the bulk band gap, implying bulk properties. X-ray photoelectron spectroscopy indicates hafnium carbide formation as one possible origin for defect states. Going further to the crystal nucleation of HfO2, nanocrystals with a diameter of 2.7–4.5 Å are identified next to carbon vacancies in the topmost HOPG layer, indicating that carbon is incorporated into the islands at early nucleation stages. A precise description of these nuclei is accomplished by the simulation of small Hf m O n (:C) clusters (m = 3 to 10; n = 3 to 22) with and without carbon incorporation using ab initio methods. The comparison of the theoretically determined lowest-energy clusters and electronic states with the experimental results allows us to identify the structure of the most relevant HfO2 sub-nanometer crystals formed during the first nucleation steps and the nature of the in-gap states found at the surfaces of HfO2 nanoislands. That way, a model system is derived that consists of distinct structural units, related to surface states or defect states, respectively, that will promote the tailoring of in-gap states of smallest HfO2 structures and thus the scalability of memory devices.
ISSN:2574-0970
2574-0970
DOI:10.1021/acsanm.2c04165