Activation and Deactivation of Semiconductor Surfaces

Abstract

The development of surface modification techniques is an integral part of advancing semiconductor device fabrication beyond subtractive methods. Chemically altering surfaces to be more or less reactive can eliminate steps in production to form devices quicker and with less material waste. This doctoral dissertation presents a series of work dedicated to improving our understanding of various techniques of surface modification and developing processes that enhance the utility of existing methods. Self-assembled monolayers (SAMs) made from octadecyltrichlorosilane (OTS) are implemented for surface deactivation and defects are studied. Defective areas in SAMs limit their utility in device production and are notoriously difficult to eliminate. However, the addition of intermediate cleaning steps are shown to prevent defects to deactivate the growth of 10 nm TiO2 films by atomic layer deposition (ALD). Spectroscopic ellipsometry, goniometry, and x-ray photoelectron spectroscopy (XPS) are used to examine SAMs deposited using different cleaning steps and assess their ability to act as an ALD resist. Polar and non-polar agglomerates adsorbed to the SAM are removed by solvent extraction and aqueous cleaning in series, and defect free layers are deposited by a second liquid phase immersion. Atomic force microscopy (AFM) confirmed the removal of agglomerates so that OTS coated SiO2 surfaces were as smooth as clean SiO2, producing deactivated surfaces suitable for prototype devices. Intermediate cleaning steps were applied to OTS deposition to reduce immersion times from 24 h to 1 h and SAMs were analyzed with XPS. Peaks shown in O 1s regions indicating the presence of surface hydroxyls showed that immersion in SC-1 (NH4OH:H2O2:H2O) doubled hydroxyl concentration on the substrate while OTS surface coverage was left relatively unchanged. In addition to the removal of polar agglomerates from the SAM surface, SC-1 hydroxylates pinhole defects in the SAM to re-activate the underlying substrate for additional deposition. A second OTS deposition improved surface coverage to cover nearly every active site on the substrate to form defect-free layers. Exposure to 200 cycles of TiO2 ALD using TiCl4 and H2O confirmed that no defects were present, as afterwards Ti was not detected on the surface by XPS. Layers were patterned with conductive mode AFM to form open trenches approximately 160 nm wide within the SAM for area selective deposition. Approximately 8 nm of TiO2 was selectively deposited within trenches while no significant deposition was noted outside trenches to demonstrate an area-selective ALD process. Indirect surface modification is also demonstrated by altering ALD precursors and implementing in-situ cleaning steps. Aluminum oxide was grown using both water and peroxide as oxidants and resulting films were analyzed and compared using XPS. Peaks in the O 1s and Al 2p regions monitored the formation of the Al-O bond and found that film nucleation occurred quicker with a higher coverage of Al when using peroxide as an oxidant. Ellipsometry, however, showed no change in growth rate. The improvement of Al coverage with no increase in thickness indicates that surfaces exposed to peroxide ALD are more reactive during each growth cycle and can form denser films. Using XeF2 in-situ to clean Si wafers could also act as a surface deactivator during ALD. Heated XeF2 exposure removed native SiO2 either directly or by surface disruption for lift off with an additional heating step. Controllable oxide removal was achieved by introducing an Al-O bond on the surface to slow down native oxide etching. Hence, indirect methods of both surface activation and deactivation are possible by altering process chemistries.