AVS 60th International Symposium and Exhibition
    Synchrotron Analysis Focus Topic Tuesday Sessions
       Session SA+AS+MG+SS-TuA

Paper SA+AS+MG+SS-TuA9
HAXPES Study of Full High-κ /Metal Gate Stacks Deposited on Ge Substrates

Tuesday, October 29, 2013, 4:40 pm, Room 203 C

Session: HAXPES Studies on Interfaces and Buried Layers
Presenter: C. Fleischmann, IMEC, Belgium
Authors: C. Fleischmann, IMEC, Belgium
I. Kalpyris, IMEC, Belgium
T. Conard, IMEC, Belgium
C. Adelmann, IMEC, Belgium
S. Sioncke, IMEC, Belgium
J.P. Rueff, Synchrotron SOLEIL, France
J. Ablett, Synchrotron SOLEIL, France
W. Vandervorst, IMEC, KU Leuven, Belgium
Correspondent: Click to Email

The introduction of Ge in CMOS devices beyond the 14 nm technology node requires effective passivation of the Ge gate stack. Besides the interface passivation, a highly scaled gate stack is needed for the next generation of CMOS devices. Scaling of the gate stack can be achieved by several means, for instance by changing process conditions. In this work, we investigate the influence of both the high‑κ stack and the metal gate on the properties of a GeOxSy interfacial passivation layer by Hard X-ray Photoelectron Spectroscopy (HAXPES). Using high energy x-rays (4 to 8 keV), we are able to probe the buried interface between the high-κ layer and the Ge substrate, and hence to reveal direct information on the chemistry and the thickness of the GeOxSy passivation layer. Note that such a buried interface is not accessible using a “standard” XPS tool relying on Al Kα x-ray radiation.

In this study, we considered three high-κ materials (Al2O3, HfO2 and an Al2O3/HfO2 bi-layer) and three metal gates (TiN, TiW and Pt). Samples have been measured both directly after atomic layer deposition and after forming gas anneal, to investigate the effect of a thermal treatment on the interfacial properties. To disentangle the impact of the particular metal gate, comparison is made to a high-κ stack sample without metal gate.

We first demonstrate the importance of analyzing full stacks, as no effect of annealing was observed on the stacks without metal gates, while clear modification of the Ge/high-κ interfacial layer thickness is observed when a gate is present. We than show that this effect on interfacial layer thickness depends on both the high‑κ and the metal gate material used. This can lead for example to an increase (i.e. HfO2/TiN) or to a decrease (i.e. Al2O3/TiN) of the interfacial layer thickness after annealing. As a global trend, the thinnest interfacial layers are obtained for pure Al2O3. However, the interfacial layer thickness appears to be more sensitive to variations in the metal gate rather than the high-κ material. We also show that the introduction of a very thin Al2O3 layer (~2 Å) between the Ge substrate and the HfO2 layer strongly influences this observed sensitivity of the interfacial layer properties to the metal gate and forming gas anneal. Aside from quantifying the interfacial layer thickness, we will also analyze changes in the interfacial layer from a chemical point of view. As a conclusion, the final layer structure (hence, the Equivalent Oxide Thickness of the gate stack) is thus a complex interplay between the initial GeOxSy thickness before forming gas anneal and the chemistry of the high-κ and metal gate materials.