Germanium, antimony, tellurium, their binary and ternary alloys and the impact of nitrogen: An X-ray photoelectron study

Abstract

The structural features and electrical properties of thin materials developed for phase change memory (PCRAM11GlossaryBE: binding energy; DC: direct current; EXAFS: extended X-ray absorption fine structure; IMFP: inelastic mean free path; FWHM: full-width half maximum; GST: GeSbTe; GSTN: nitrogen-doped GST; LEIS: low energy ion scattering NRA: nuclear reaction analysis; OTS: ovonic threshold switching selector; PCRAM: phase-change random access memory; PCM: phase change material; PP-TOFMS: plasma profiling time-of-flight mass spectrometry; PVD: physical vapor deposition; Q-time: queue-time between process and XPS steps, during which the samples are stored in the air; RBS: Rutherford backscattering spectrometry; RF: radio frequency; TOF-SIMS: time-of-flight secondary ion mass spectrometry; VBM: valence band maximum; WDXRF: wavelength-dispersive X-ray fluorescence; XPS: X-ray photoelectron spectroscopy; XRR: X-ray reflectometry) are highly influenced by the stoichiometry and the binding states. Chemical state X-ray photoelectron spectroscopic analysis (XPS) of materials such as germanium telluride (GeTe), antimony telluride (Sb2Te3, Sb2Te) and germanium antimony telluride (GST) is challenging due to the complexity of their 3d and 4d spectra resulting from overlapping binding energies (BE) when samples are exposed to air. The complexity increases when nitrogen is added to the phase change material (PCM) in order to tune its properties, such as the crystallization temperature. We have used quasi in situ XPS to resolve the binding states of PCM materials, starting from elementary materials (Ge, Sb, Te) and their nitrides (GeN, SbN, TeN), then addressing binaries (GeTe, SbTe, GeSb) and related nitrides (GeTeN, SbTeN, GeSbN), in order to finally characterize GST and N-doped GST with various stoichiometry. On top of oxygen-free quasi in situ analysis and for most sample systems, we have investigated the influence of air exposure on the binding states, first after limited queue-time (few minutes), then after longer periods (from 1 week to 4 months). We did not use the (arbitrary) BE correction based on the C 1s spectra of adventitious carbon layer as it led to a large spread in BE values for elements even present in the same chemical state. Finally, we found that BE differences such as the ones we present here between Ge 3d5/2 and Te 4d5/2 and between Sb 4d5/2 and Te 4d5/2, are highly relevant metrics to monitor the main structural features and their evolution in PCM thin films.