Department of Materials Science & Engineering, The University of Texas at Dallas , Richardson, Texas 75080, United States.
Versum Materials, Inc. , 1969 Palomar Oaks Way, Carlsbad, California 92011, United States.
Langmuir. 2018 Feb 27;34(8):2619-2629. doi: 10.1021/acs.langmuir.7b03522. Epub 2018 Feb 12.
Despite the success of plasma-enhanced atomic layer deposition (PEALD) in depositing quality silicon nitride films, a fundamental understanding of the growth mechanism has been difficult to obtain because of lack of in situ characterization to probe the surface reactions noninvasively and the complexity of reactions induced/enhanced by the plasma. These challenges have hindered the direct observation of intermediate species formed during the reactions. We address this challenge by examining the interaction of Ar plasma using atomically flat, monohydride-terminated Si(111) as a well-defined model surface and focusing on the initial PEALD with aminosilanes. In situ infrared and X-ray photoelectron spectroscopy reveals that an Ar plasma induces desorption of H atoms from H-Si(111) surfaces, leaving Si dangling bonds, and that the reaction of di-sec-butylaminosilane (DSBAS) with Ar plasma-treated surfaces requires the presence of both active sites (Si dangling bonds) and Si-H; there is no reaction on fully H-terminated or activated surfaces. By contrast, high-quality hydrofluoric acid-etched SiN surfaces readily react with DSBAS, resulting in the formation of O-SiH. However, the presence of back-bonded oxygen in O-SiH inhibits H desorption by Ar or N plasma, presumably because of stabilization of H against ion-induced desorption. Consequently, there is no reaction of adsorbed aminosilanes even after extensive Ar or N plasma treatments; a thermal process is necessary to partially remove H, thereby promoting the formation of active sites. These observations are consistent with a mechanism requiring the presence of both undercoordinated nitrogen and/or dangling bonds and unreacted surface hydrogen. Because active sites are involved, the PEALD process is found to be sensitive to the duration of the plasma exposure treatment and the purge time, during which passivation of these sites can occur.
尽管等离子体增强原子层沉积(PEALD)在沉积高质量氮化硅薄膜方面取得了成功,但由于缺乏原位表征来非侵入性地探测表面反应,以及等离子体诱导/增强的反应复杂性,因此很难获得对生长机制的基本理解。这些挑战阻碍了对反应过程中形成的中间物种的直接观察。我们通过使用原子级平坦的、单氢化终止的 Si(111)作为定义明确的模型表面来检查 Ar 等离子体的相互作用,并将重点放在最初的与氨硅烷的 PEALD 上,从而解决了这一挑战。原位红外和 X 射线光电子能谱揭示,Ar 等离子体诱导 H 原子从 H-Si(111)表面解吸,留下 Si 悬空键,而二仲丁基氨硅烷(DSBAS)与 Ar 等离子体处理表面的反应需要活性位(Si 悬空键)和 Si-H 的存在;在完全氢化或活化的表面上没有反应。相比之下,高质量的氢氟酸刻蚀的 SiN 表面容易与 DSBAS 反应,生成 O-SiH。然而,O-SiH 中背键合氧的存在抑制了 Ar 或 N 等离子体中的 H 解吸,这可能是因为 H 被离子诱导解吸稳定化。因此,即使经过广泛的 Ar 或 N 等离子体处理,吸附的氨硅烷也没有反应;需要进行热过程来部分去除 H,从而促进活性位的形成。这些观察结果与需要存在配位不足的氮和/或悬空键以及未反应的表面氢的机制一致。由于涉及活性位,因此 PEALD 过程对等离子体暴露处理的持续时间和吹扫时间敏感,在此期间这些位可能会发生钝化。