AVS 65th International Symposium & Exhibition
    Thin Films Division Wednesday Sessions
       Session TF+EM+MI-WeM

Paper TF+EM+MI-WeM5
Effects of Process Gases and Gate TiN Electrode during the Post Deposition Anneal to ALD-Al2O3 Dielectric Film

Wednesday, October 24, 2018, 9:20 am, Room 102A

Session: Thin Film Processes for Electronics and Optics I
Presenter: Masaya Saito, Tohoku University, Japan
Authors: M. Saito, Tohoku University, Japan
A. Teramoto, Tohoku University, Japan
T. Suwa, Tohoku University, Japan
K. Nagumo, Tohoku University, Japan
Y. Shiba, Tohoku University, Japan
R. Kuroda, Tohoku University, Japan
S. Sugawa, Tohoku University, Japan
Correspondent: Click to Email

We investigated effects of a post deposition anneal(PDA) to Al2O3 film which was formed by the low temperature atomic layer deposition(ALD)(1). Al2O3 films were formed at 75oC by the ALD process using Al(CH3)3 as a precursor and H2O as an oxidant, and these thicknesses were 40 nm. TiN films were formed as gate electrodes of MIS capacitors. We applied three kinds of PDAs. N2 and O2 annealing at 400oC were applied just after the ALD(PDA Ⅰ(a), (b)), and N2 annealing at 400oC was applied after the gate electrode formation (PDA Ⅱ). We measured C-V and I-V characteristics.

The negative VFB compared to the ideal one(+0.54 V) was observed in the C-V curve without PDAs, and the fixed charge density calculated from VFB was ~1013 cm-2. In contrast, positive VFB compared to the ideal one was observed in the C-V curve with PDA Ⅰ(b), and the fixed charge density was ~1011 cm-2. These results indicate that PDA Ⅰ(b) is effective to decrease the fixed charges. Moreover, the hysteresis in the C-V curve decreased by PDA Ⅰ(b), and the leakage current also decreased within the voltage range of C-V measurement. It is considered that the decrease of hysteresis was caused by decreasing the electrons that were trapped in the Al2O3 film(2). However, a few samples broke down at low voltage after PDA Ⅰ(b). In contrast, we couldn’t measure the C-V characteristics of almost capacitors with PDA Ⅰ(a) because the capacitors broke down at low voltage. In the case of PDA Ⅱ, which means N2 annealing after the gate electrode formation, the leakage current decreased same as PDA Ⅰ(b) and the production yield was still good even after PDA Ⅱ. This was different tendency from PDA Ⅰ(a) even as the same annealing. However, VFB shifted to the positive voltage by the PDA Ⅱ, and the fixed charge density was ~1012 cm-2. This was the same phenomenon as the PDA Ⅰ(b).

We considered these phenomena as follows; the oxidizing species were desorbed by PDA just after the ALD, and then the film quality was bad after the N2 annealing. The oxygen for improving the film was supplied by the O2 annealing with the desorption. When PDA was carried out after the gate electrode formation, the desorption did not occur because of capping by the gate electrode, and then the oxidizing species were sufficient for improving the film by the annealing.

Reference:

(1) Y. Koda, et al., ECS Trans., 72(4), 91-100 (2016)

(2) L. Sambuco Salomone, et al., J. Appl, Phys., 123, 085304 (2018)

Acknowledgement:

This work was carried out at fluctuation free facility of New Industry Creation Hatchery Center, Tohoku University.