Impact of substrate temperature and film thickness on the interfacial evolution during atomic layer deposition of HfO₂ on InP

Abstract

Minimizing the low-quality interfacial layer (IL) growth during high-k film deposition on III-V compound substrates has been a major research area, in addition to the development of various interface passivation techniques [1]. On GaAs substrates, Hinkle et al. [2] reported a pioneering work on the metal-organic (MO) precursoractivated self-cleaning effect, a spontaneous clean-up of the surface oxides during the atomic layer deposition (ALD) of HfO 2 and Al 2 O 3 . A similar cleaning effect was also observed on InGaAs [3,4]. Among many contributing parameters, the ALD temperature is known to strongly affect the self-cleaning mechanism; for example, Suri et al.[5] reported that it is insignificant at 200 °C and remarkably enhanced at 300 °C when tetrakis(dimethylamino)hafnium/H 2 O were used for HfO 2 deposition on GaAs. In conjunction with affecting the defective IL removal efficiency, the ALD temperature is also one of the important criteria in determining the final microstructural and electrical properties of the high-k films.Together with GaAs and InGaAs, InP is also a promising candidate as an alternative n-channel material, however, fewer works have been reported on the self-cleaning effect during the ALD high-k process on InP. Recently, an interesting IL evolution behavior showing a large dependence on the HfO 2 thickness was reported when tetrakis-(ethylmethylamino)hafnium (TEMAHf) and H 2 O were used on InP at an ALD temperature equal to or higher than 300 °C [6,7]. In these reports, largely insufficient selfcleaning was observed at an early stage and a continuous decrease of the IL occurred as the film thickness was increased, which is a quite different trend to that on GaAs or InGaAs, which shows a nearly complete removal of IL after only a few ALD cycles [2,4,5]. However, a detailed mechanism for this phenomenon was not given.On these grounds, we investigated the effect of the ALD temperature on the IL evolution during the HfO 2 formation on InP and tried to rationalize it by focusing on the temperature-dependent self-cleaning effect and possible simultaneous substrate oxidation during the ALD process. In addition, the film-thickness-dependent IL thickness variation was accounted for by other possible contributing factors.As substrates, (100)-oriented InP wafers with an n-type carrier concentration of (4−5) × 10 15 cm -3 were used. The substrates were successively cleaned by 1% HF and 21% (NH 4 ) 2 S diluted solutions. The ALD HfO 2 process was carried out in a cross-flow type ALD system at various substrate temperatures: 200, 250, and 300 °C. During each deposition cycle, TEMAHf and H 2 O were sequentially delivered with N 2 purging in between, and the total number The effects of both the deposition temperature and the HfO 2 film thickness on the interfacial layer (IL) evolution were studied when tetrakis(ethylmethylamino)hafnium and H 2 O based atomic layer deposition (ALD) was performed on InP substrates. While the self-cleaning effect resulted in an IL-free structure after formation of ~2 nm thick ...

In situ study of the role of substrate temperature during atomic layer deposition of HfO2 on InP

“…11,13 However, another in situ XPS report of ALD HfO 2 "half cycle" at a substrate temperature of 250 C has shown consistent results with this study, where the In-oxide concentration is seen to decrease and the P-oxide concentration remains constant during the initial ALD stages. 15 This contradiction between ex situ and in situ results from the thin oxide films (less than 2 nm) is possibly due to the interfacial oxide regrowth, and this oxidation is seen to be enhanced by increased substrate temperature upon removal from the ALD reactor.…”

“…8,9 Suri et al have shown, using tetrakis(dimethylamido)hafnium (TDMA-Hf) and H 2 O precursors, that the "self cleaning" effect on GaAs is temperature dependent, with the removal of oxides being more pronounced at 300 C than at the 250 C or 200 C. 10 However, quite different trends for the "self cleaning" effect have been reported on GaAs in comparison to those for acid etched and S-passivated InP. 9,11,12 A less efficient "self cleaning" effect has been observed during thin film deposition (2 nm) of ALD HfO 2 , while the "self cleaning" effect has been reported to be enhanced significantly following thick (8 nm) ALD using tetrakis(ethylmethylamino)hafnium (TEMA-Hf) and water as precursors at a substrate temperature of 300 C. 7 Recently, An et al 13 reported the impact of ALD temperature on HfO 2 deposition on S-passivated InP substrates using TEMA-Hf and water, with a significant "self cleaning" effect occurring at 200 C and 250 C but substantial InP substrate oxidation occurring at 300 C. 13 This apparent variation during ALD HfO 2 on different III-V substrate materials therefore warrants further study. In addition, most previous studies on InP have been carried out using ex situ ALD, and spurious reactions may occur due to atmospheric exposure.…”

In situ study of e-beam Al and Hf metal deposition on native oxide InP (100)

“…In contrast to the In-In bond, both before and after annealing, the decrease in intensity for both the In-P and the P-oxide peaks is consistent with equal attenuation by the Al/Al-oxide overlayer, suggesting that no P-oxide diffusion is detected by ARXPS. These results are also consistent with indium out-diffusion 11,27 and no detectable P-oxide diffusion through the high-k dielectrics from the high-k/InP stacks. 12 Figures 3(a)-3(d) show the In 3d 5/2 , P 2p, In 4d/Hf 4f, and O 1s spectra, respectively, from the initial InP surface, after 2 nm e-beam Hf deposition (based on the attenuation of In 3d 5/2 spectra) and following UHV annealing at 250 C. For the In 3d 5/2 spectra, the peak with a binding energy separation to the InP bulk (at 444.8 eV) of À1.09 eV is assigned to the In-Hf peak, and the peak with a binding energy separation of À0.67 eV is assigned to the In-In peak.…”

Indium diffusion through high-k dielectrics in high-k/InP stacks

“…Hence, this may reduce the In outdiffusion upon annealing. While both In and PO 2 diffusion through HfO 2 were observed in previous reports using TOF-SIMS, [9][10][11] the detection limit of TOF-SIMS (7 Â 10 7 /cm 2 ) is significantly lower than LEIS (3.3 Â 10 11 /cm 2 ) and ARXPS (majority of distribution), 18,30 so it is assumed that the concentration of diffused P is below the limit of detection for LEIS and XPS but sufficient to be detected by TOF-SIMS.…”

“…8 An et al and Kang et al have reported diffusion of In atoms and P-oxides through a thick (>6 nm) HfO 2 layer on InP by Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS). [9][10][11] However, detailed analysis of this substrate elemental diffusion through other high-k dielectrics as well as interfacial chemistry upon annealing has not been reported.…”