Saturday, March 6, 2021

Thermal ALE of germanium rich SiGe by CU Boulder and ASM Microchemistry

Epitaxially grown SiGe is an important material for CMOS Logic. It is integrated as the channel material and by inserting a higher concentration of germanium the mobility of the transistor can be improved. The industry calls it Epi, but what is really referred to a thermal CVD process producing an epitaxially grown layer of silicon or silicon-germanium onto a single crystalline silicon wafer.

As CMOS scaling has progressed the IDMs and Foundries have moved from the planar field-effect transistor (FET) architecture to a narrow fin-based transition the FinFET. The next evolutionary step on the horizon will be the transition to a nanowire-based architecture forming a gate-all-around FET (GAA-FET). At some point in time beyond the 2 nm node, the lateral scaling possibility will hit a wall and it is foreseen that the CMOS scaling will gup upwards like other technologies in order to cram in more devices per unit area. In a first approach, it may be that the NMOS and PMOS transistors are rearranged from being processed next to each other to put one of them on top of the other. Intel recently presented this at IEDM2020 (LINK). Having done that you can foresee continuing on a vertical scaling path also for CMOS just like 3DNAND and start to build those skyscrapers.

When going vertical, you will need highly conformal deposition processes as provided by ALD and in high volume production since the event of 90 nm DRAM (Samsung) and 45 nm Logic (Intel), however, etch is a problem since the reactive ion etching process are typically directional with the plasma under low-pressure processing conditions used. Also, the Argon plasma ALE processes to etch Silicon, silicon Germanin gallium nitride, and III/V materials are directional or anisotropic as the etch guys say or non-conformal as we ALD people say.

Typically the best way to achieve isotropic etch conditions, meaning you remove material at the same rate or as for ALE the same amount per cycle (etch per cycle EPC), is to skip the plasma that causes the anisotropic etch. Here Dr Abdulgatov and co-workers in the famous SM George Lab, CU Boulder together with Varun Sharma and friends from ASM Microchemistry, one of Dresden's best shining ALD-Stars, publish a paper on Thermal ALE of germanium rich SiGe that is quite clever. Here using PVD Si0.15Ge0.85 samples, which are difficult to make by Epi due to the high Ge content. I think we will see more of this for also GaN, SiC and III/V materials coming up.


AI Abdulagatov, V Sharma, JA Murdzek, AS Cavanagh, SM George
Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films

Abstract: The thermal atomic layer etching (ALE) of germanium-rich SiGe was demonstrated using an oxidation and “conversion-etch” mechanism with oxygen (O2) or ozone (O3), hydrofluoric acid (HF), and trimethylaluminum [TMA, Al(CH3)3] as the reactants. The crystalline germanium-rich SiGe film was prepared using physical vapor deposition and had a composition of Si0.15Ge0.85. In situ spectroscopic ellipsometry was employed to monitor the thickness of both the SiGe film and the surface oxide layer on the SiGe film during thermal ALE. Using a reactant sequence of O2-HF-TMA, the etch rate of the SiGe film increased progressively with temperatures from 225 to 290 °C. At 290 °C, the SiGe film thickness decreased linearly at a rate of 0.57 Å/cycle with a surface oxide thickness of 18–19 Å. This etch rate was obtained using reactant pressures of 25, 0.2, and 0.4 Torr and doses of 1.5, 1.0, and 1.0 s for O2, HF, and TMA, respectively. The TMA and HF reactions were self-limiting and the O2 reaction was reasonably self-limiting at 290 °C. Using an O3-HF-TMA reaction sequence, the SiGe ALE etch rate was 0.42 Å/cycle at 290 °C. This etch rate was obtained using reactant pressures of 15, 0.2, and 0.4 Torr and dose times of 0.5, 1.0, and 1.0 s for O3, HF, and TMA, respectively. The O3, TMA, and HF reactions were all self-limiting at 290 °C. Atomic force microscopy images revealed that thermal ALE with the O2-HF-TMA or O3-HF-TMA reaction sequences did not roughen the surface of the SiGe film. The SiGe film was etched selectively compared with Si or Si3N4 at 290 °C using an O2-HF-TMA reaction sequence. The etch rate for the SiGe film was >10 times faster than Si(100) or Si3N4 that was prepared using low-pressure chemical vapor deposition. This selectivity for the SiGe film will be useful to fabricate Si nanowires and nanosheets using SiGe as the sacrificial layer.

Full text open source: LINK


Figure from Journal of Vacuum Science & Technology A 39, 022602 (2021); https://doi.org/10.1116/6.0000834

No comments:

Post a Comment