AVS 49th International Symposium
    Thin Films Tuesday Sessions
       Session TF-TuA

Paper TF-TuA7
Trimethylaluminum-initiated ALD Growth of Al@sub 2@O@sub 3@ on Si: An In-situ Infrared Study

Tuesday, November 5, 2002, 4:00 pm, Room C-101

Session: Atomic Layer Deposition - Oxides
Presenter: M. Frank, Agere Systems and Rutgers University
Authors: M. Frank, Agere Systems and Rutgers University
Y.J. Chabal, Agere Systems
G.D. Wilk, Agere Systems
Correspondent: Click to Email
The initial phase of alternative gate dielectric growth on Si by atomic layer deposition (ALD) is critical for the final Si-dielectric interface properties. The ubiquitous and deleterious presence of an interfacial SiO@sub 2@ layer has motivated deposition on H-terminated Si surfaces. However, even for these surfaces a thin SiO@sub 2@ layer is formed during growth. It is believed to arise from the initial reaction of water (forming OH and Si-O-Si), and to be necessary for subsequent metal precursor reaction. We have designed a model reactor to perform an in-situ infrared absorption study of Al@sub 2@O@sub 3@ growth on both HF-etched and oxidized Si from trimethylaluminum (TMA, Al(CH@sub 3@)@sub 3@) and water at 300°C. We have thus observed all relevant species (CH@sub 3@, OH, Si-H, oxides) present at the surface up to 16 ALD cycles. We find that H-terminated Si(100) surfaces are neither oxidized nor hydroxylated by water, in contrast to what is expected (yet, D@sub 2@O experiments show H-D isotopic exchange and similar O@sub 2@ exposures lead to oxidation of the H-terminated surface). Instead, TMA exposure leads to the formation of both Al-CH@sub 3@ and Si-CH@sub 3@ groups, with varying kinetics. Transfer of CH@sub 3@ to Si predominantly occurs at Si step sites, and is absent on atomically smooth H/Si(111)-(1x1). Once Al is deposited, three reactions take place upon water exposure, each with different kinetics: replacement of Al-bonded CH@sub 3@ by OH; transfer of CH@sub 3@ from Al to Si; and the catalytic oxidation of Si. In contrast, Si-CH@sub 3@ remains unaffected by water, and may therefore be responsible for C contamination at the Si/oxide interface. During subsequent TMA-water cycles, more interfacial SiO@sub 2@ is formed, while Al@sub 2@O@sub 3@ growth proceeds according to the well-known self-terminated mechanism on all Si substrates. This leads to the formation of Al@sub 2@O@sub 3@ films equivalent to what is grown in commercial ALD reactors.