Mechanisms for undesired nucleation on H-terminated Si and dimethylamino-trimethylsilane passivated SiO2 during TiO2 area-selective atomic layer deposition

Abstract

During TiO2 atomic layer deposition (ALD) using TiCl4 and H2O at ∼150 °C, nucleation proceeds rapidly on hydroxylated SiO2 but is inherently delayed on passivated surfaces such as H-terminated silicon (Si-H) and trimethylsilyl-passivated SiO2 (SiO2-TMS) formed using dimethylamino-trimethylsilane (DMA-TMS) as a small molecule inhibitor. In this work, we explore details of TiO2 nucleation on both Si-H and SiO2-TMS and show that the mechanisms leading to unwanted nuclei depend strongly on the passivation mechanism. Initial growth is observed as a function of ALD cycles using scanning electron microscopy to obtain average particle size, density, and overall surface coverage fraction. Also, average film thickness vs cycle is estimated using ellipsometry or Rutherford backscattering spectrometry. Data are compared to an analytical model that considers that either nucleation sites are present on the starting non-growth surface or sites are generated during the ALD process. On the Si-H surface, data and modeling indicate that nucleation occurs predominantly from a fixed number of nucleation sites present on the starting growth surface that start to immediately grow. However, on TMS-passivated SiO2, nucleation sites are predominantly generated during the growth process so that the density of nucleation sites increases as growth proceeds. Results indicate that nucleation sites are created when adsorbed ALD reactants become kinetically trapped on the SiO2-TMS surface. This demonstrates that mechanisms associated with unwanted nucleation during area-selective deposition (ASD) can depend on details of the surface passivation scheme, thereby providing insight to help to improve ASD strategies for advanced applications.

Enhancing Performance and Function of Polymethacrylate Extreme Ultraviolet Resists Using Area-Selective Deposition

“…For each polymer + PAG material, exposure to EUV leads to a decrease in film thickness from 35 to ∼26 nm. For the PtBuMA + PAG sample, the surface became more hydrophilic after EUV treatment, and XPS results (Figure S2 and Table S1) show an increase in CO bonds relative to C–O, consistent with conversion from tBu to hydrophilic −OH groups (i.e., deprotection). , We note that for the polymers without PAG, EUV exposure did not significantly affect WCA (data not shown), as without PAG the polymer deprotection reaction occurs to a much lesser extent, leaving hydrophobic tBu protecting groups as the predominant surface groups. , These trends in surface hydrophobicity (determined by the polymer protecting groups) can indicate the potential for nucleation inhibition during TiO 2 ALD, where more hydrophobic surfaces are expected to have a longer nucleation delay. ,, …”

“…Lines are drawn as guides to the eye. Data points corresponding to TiO 2 growth rate on SiO 2 in (a) and (b) are reproduced from refs , .…”

“…Close examination of the trends in Figure reveals that TiO 2 nucleation does not necessarily correlate directly with surface hydrophobicity but rather appears to depend on the functionality of the pendant group (WCA is sometimes used as a convenient initial indicator for selectivity as most ASD processes nucleate more slowly on hydrophobic surfaces compared to hydrophilic hydroxyl terminated surfaces). , For example, comparing nucleation on PMCPMA and PCHMA homopolymers, growth proceeds readily on PMCPMA but not on PCHMA, even though the initial WCA (85–86°) and the size of the protecting group (i.e., number of C atoms) is similar for both materials. A key difference between these polymers is that on PCHMA, the ester group is linked to the cyclic protective moiety via a secondary carbon, whereas for PMCPMA, the link is through a tertiary carbon.…”

“…As shown in Figure , the inhibition for both tone inversion and resist hardening could be achieved via an inherent delay on the clean surface or by using a passivating molecule such as a small-molecule inhibitor (SMI) that selectively adsorbs on the desired nongrowth substrate. An SMI of interest in this work is N , N -dimethylamino-trimethylsilane (DMA-TMS), which can chemisorb onto available surface −OH sites to form passivating −O–Si(CH 3 ) 3 groups: , − Si − normalO normalH surface + normalN false( normalC normalH 3 false) 2 − Si false( normalC normalH 3 false) 3 → prefix− Si − normalO − Si false( normalC normalH 3 false) 3 , surface + normalN false( normalC normalH 3 false) 2 normalH …”

“…20 As shown in Figure 1, the inhibition for both tone inversion and resist hardening could be achieved via an inherent delay on the clean surface or by using a passivating molecule such as a small-molecule inhibitor (SMI) that selectively adsorbs on the desired nongrowth substrate. An SMI of interest in this work is N,N-dimethylamino-trimethylsilane (DMA-TMS), which can chemisorb onto available surface −OH sites to form passivating −O−Si(CH 3 ) 3 groups: 38,39 (1) DMA-TMS could enhance resist hardening by passivating and blocking ALD on exposed SiO 2 while allowing coating of or infiltration into the patterned resist. Likewise, DMA-TMS could promote tone inversion by passivating the resist surface, allowing selective growth on the exposed substrate region.…”