841
(Invited) Characterization of Interfacial Dipoles at Dielectric Stacks by XPS Analysis

Wednesday, 4 October 2017: 08:30
Chesapeake D (Gaylord National Resort and Convention Center)
S. Miyazaki (Nagoya University), A. Ohta (Graduate School of Engineering, Nagoya University), and N. Fujimura (Nagoya University)
In metal/high-k dielectric gate-stack technology, a clear understanding of inner potential changes at heterointerfaces is one of crucial issues from a viewpoint of precise tuning of threshold voltage for advanced MIS FETs. So far, charge transfer due to bond hybridization at metal/high-k dielectric interfaces [1], areal density difference of oxygen atoms between dielectrics in stacked dielectrics [2, 3], and oxygen vacancies in high-k dielectrics [4] have been discussed as possible origins of the potential change mostly based on capacitance-voltage (C-V) characteristics. However, in MIS structures with stacked dielectrics, quantitative evaluation of inner potential changes caused by electrical dipoles at individual interfaces has not been always confirmed from the C-V analysis and is still matter of research. In this work, to overcome uncertainty or limitation in the C-V analysis, we have focused on direct evaluation of interface dipoles by means of photoemission measurements and characterized vacuum level difference caused by electrical dipole formation in the region near the interfaces between SiO2 and high-k dielectrics.

After wet-chemical cleaning steps of p-Si(100), a ~200 nm-thick SiO2 layer was grown at 1000°C in pure O2. Subsequently, high-k dielectrics such as HfO2, Al2O3, and Y2O3 with a layer thickness of 0.6-0.8 nm were deposited on thermally-grown SiO2/Si by the magnetron sputtering, in which the Ar/O2 gas flow ratio and the power density kept constant at unity and 1.52 W/cm2, respectively. Very uniform coverage with ultrathin high-k dielectrics was also confirm by AFM measurements. Then, post deposition anneal (PDA) was performed at 600°C in dry-N2 for 5 min to densify the high-k dielectric layers.

To evaluate inner potential changes in dielectric stacks so prepared, the cut-off energy of secondary photoelectrons (SEs) were measured in high resolution XPS. As shown in Fig. 1, an abrupt potential change due to electrical dipoles at the interface between dielectrics, resultant abrupt potential change can be measured the change in the measured cut-off energy of SEs when the measured kinetic energy of photoelectrons was calibrated by the kinetic energy of core-line signals from the underlying layer. The observation of cut-off energy provides us an advantage in simple and precise evaluation of the potential change due to electrical dipoles as compared to a discussion on dipole formation based on the energy shift of core-line signals which reflects not only the potential change due to dipoles but also the chemical shift, that is, change in the chemical bonding features.

After the energy calibration of photoelectrons and intensity normalization was made by core-line signals originated from thermally-grown SiO2 to eliminate the energy shift due to the charge up effect during the photoemission measurements, Si 2p signals show a slight increase in the lower binding energy side (Fig. 2(a)) with the formation of the top high-k dielectric layer. Such sifted signals are attributable to coordination of less electronegative M (Hf, Al, or Y) ions to the second nearest neighbor of Si atom through O ion at the interface between SiO2 and high-k dielectrics. From the Si 2p intensity ratio of Si-O-M to SiO2 signals, observed Si-O-M bonding units was estimated to be 0.3 ~ 0.6 nm in average thickness, indicating that the high-k dielectric/SiO2 interface is compositionally abrupt. The cut-off energy of SEs taken for the sample after the formation of Al2O3 and HfO2 on SiO2 was shifted by 0.38 and 0.08 eV toward the higher kinetic energy side, respectively (Fig. 2(b)). On the other hands, for the sample after Y2O3 formation, an opposite energy shift of 0.05 eV was detected. The observed energy shift in each high-k dielectric/SiO2 stack is attributable to the abrupt potential change due to the presence of interface dipoles as predicted in Fig. 1. Using the O anion density of individual high-k layers on SiO2 crudely estimated from the analysis of O1s signals, we have found that there seems to be a positive correlation between the observed potential changes at the interfaces and the ratios in O atomic density between SiO2 and high-k dielectrics (Fig. 3).

In summary, the evaluation of the potential change due to the interface dipoles from the energy shift of SE cut-off has been demonstrated, and a positive correlation between the density differences in O atoms at high-k dielectric/SiO2 interfaces and measured potential changes was also clarified.

References

[1] T. Nakayama et al., ECS Trans., 3 (2006) 129.

[2] K. Iwamoto et al., Tech. Dig. VLSI Symp., 2007, 70.

[3] K. Kita et al., Appl. Phys. Lett., 94(2009)132902.

[4] Y. Akasaka et al., JJAP, 45 (2006) L1289.