STUDY OF AREA-SELECTIVE DEPOSITION USING ATOMIC LAYER DEPOSITION AND CHEMICAL VAPOR DEPOSITION: COMPETITIVE ADSORPTION AS AN ALTERNATIVE ROUTE

Abstract

Area-selective deposition (ASD) has gained much interest over the past few years from the semiconductor industry as the downscaling of everyday electronics and the feature sizes of their structures continues. The current standard method of ASD involves numerous steps including, but not limited to, photolithography, image transfer, development, etch, deposition, and lift-off. Once the bottom-up ASD process is achieved, the industry may be capable of omitting these unnecessary steps. ASD can reduce the cost and time of production in addition to other benefits like edge-placement error that could happen due to misalignment during multiple patterning. This dissertation focuses on different approaches to achieve ASD and study of thin films that could potentially be used for ASD. We present the concept of competitive adsorption using a third species called co-adsorbate, other than the precursor for atomic layer deposition (ALD) and co-reactant. First, we develop an ASD process using chemical vapor deposition (CVD) and selectively deposit up to ~ 30 nm of smooth and uniform ZrO2 film with Zr[N(C2H5CH3)]4 (TEMAZ), O2 and 4-octyne on SiO2, the iv growth surface, in presence of Cu, the non-growth surface. Post-deposition characterization includes in situ X-ray photoelectron spectroscopy (XPS), angleresolved XPS, ex situ spectroscopic ellipsometry (SE), scanning electron microscopy (SEM), and atomic force microscopy (AFM). Density functional theory (DFT) calculations conducted by the DiStasio research group in the Chemistry Department at Cornell University support the selective nature of 4-octyne towards SiO2 vs. Cu. The binding energy of the co-adsorbate species to Cu is determined to be greater by a factor of 3 compared to that on SiO2. In addition, the bond length of the central C≡C bond and the C–C≡C bond angle of the adsorbed 4- octyne suggest that this alkyne undergoes rehybridization to have alkene-like structure and form a covalent bond to the metal surface. We examine experimentally a process space in two directions: the partial pressure of the co-adsorbate and substrate temperature. We find a selective window in which AS-CVD is achieved and a clear boundary with the process space that separates selective and non-selective region. Ny varying the dosing sequence, we also demonstrate that the adsorption 4-octyne on the Cu surface is most likely reversible. In the second part of this presented work, we investigate very similar yet different system for ASD. We employ ALD as the method of deposition and 3-hexyne which has shorter alkyl chain on each side of the internal triple bond than 4-octyne. The DFT calculations show similar results for 3-hexyne and 4-octyne: rehybridization on Cu and stronger binding energetics towards Cu vs. SiO2. With this intrinsically selective inhibitor, we were able to deposit ~ 1.5 nm of ZrO2 from 10 cycles of ALD via competitive adsorption, again using TEMAZ and O2 as the thin film precursor and co- v reactant. After 10 cycles, we observe nucleation of Zr on Cu, the non-growth surface, by XPS and find evidence of the formation of cuprous oxide (Cu2O). The calculated Auger parameter from XPS shows oxidation of the Cu surface after 10 cycles with 3- hexyne when no such evidence is found in pristine 10 cycles (without 3-hexyne). AFM also supports this idea by showing a significant increase in the roughness of the Cu substrate after 10 cycles of AS-ALD. Lastly, we explore a different approach than previous reported ones to deposit silicatene, a perfect 2D bilayer SiO2 without a single dangling bond. Plasma-enhanced ALD with Si[N(CH3)2]3H (TDMAS) and O2 plasma shows continuous growth of SiO2 film as a function of the number of ALD cycles (up to 100 cycles) on Pt, Ru, and Pd surfaces, indicating that silicatene was not formed. Subsequent annealing in an oxygen rich environment inside a furnace did not help in producing the bilayer; rather, the thermal energy disturbed the internal structure and formed metal silicide as evidenced by ARXPS. We also investigate thermal ALD with SiH3N[CH(CH3)(C2CH3)]2 (DSBAS) and SiH2[NC(CH3)3]2 (BTBAS) as aminosilane precursors and H2O, O2, H2O2 as the coreactants. In situ XPS shows inherent selectivity ~ 1.0×1015 Si atoms-cm-2 on Cu minimal amount on Al2O3 regardless of the number of cycles or the co-reactant used. As DSBAS loses its one and only amine group to dissociatively chemisorb, surfaces will be left with -SiH3 termination which was not capable of oxidation to regenerate the active sites on the surface to proceed to the next ALD cycle under the conditions and co-reactants we investigated. The adsorption of BTBAS, which has two initial amine groups, resulted in greater amount of lingering N groups on the surface as determined vi by XPS. However, this -SiH2NRX surface termination does not enhance the nucleation significantly as 5 cycles of BTBAS and H2O2 at 180 °C showed similar amount of Si in comparison to ½ cycle BTBAS on both Cu and Al2O3 substrates.