(167e) Spatial Atomic Layer Deposition By "Air Hockey" Design for Dielectric Multilayer Optical Films | AIChE

(167e) Spatial Atomic Layer Deposition By "Air Hockey" Design for Dielectric Multilayer Optical Films

Authors 

Grasso, J. A. - Presenter, University of Connecticut
Oliveira, N., University of Connecticut
Willis, B., University of Connecticut

Spatial Atomic Layer Deposition by
“Air Hockey” Design for Dielectric Multilayer Optical Films

J.
Grasso, N. Oliveira, B. Willis

University of
Connecticut

            Dielectric multilayer structures
exhibit exceptional specular properties for employment in various optical
applications; such as, filters, antireflection and high reflection coatings. Specular
properties can be tailored for specific applications by altering the
composition of the materials in the stack. Atomic layer deposition (ALD) is one
proven method used for constructing multilayer stacks. The conformity, atomic-level
thickness control, and freedom to choose countless materials associated with
films deposited by ALD are advantageous to the performance of sophisticated
multilayer optical devices. Dielectric multilayer optical films can prevent
radiation lost from objects at high temperatures by altering the optical
properties of the cold-side; leading to the resurrection of the incandescent
source. These complex films are considerably thick and are unreasonable to
fabricate with current technology. Slow deposition rates for conventional ALD
contribute to the impracticality of utilization of this technique for high
throughput processing. Previous work reports that 20 layer stacks take upwards
of 16 hours to complete. Spatial ALD operates by continuously supplying the
precursors while keeping them physically separated in different zones;
eliminating the essential time intensive purging step in conventional ALD. The
deposition rate is drastically improved in spatial ALD due to the enhancement
of the mass transfer of precursors to the substrate, transforming the process
to being kinetically limited. The magnitude of this effect may be correlated
with the distance separating the precursor inlets and the substrate. Advances
in the deposition rate improvement requires an extensive understanding of the
fluid dynamics of the gases and its dependence of the process parameters.

            We
present a case study for the well understood ALD of alumina by
trimethylaluminum (TMA) and water using a novel spatial ALD system operating
analogous to an air hockey table. Dispersed nitrogen inlets float the substrate
allowing for virtually frictionless motion amplifying its linear velocity. Discrete
introduction of the precursors is achieved via a nitrogen barrier which
prevents undesirable mixing of the precursors as the substrate travels across
the injector region. Each injector vent is surrounded by two vacuum pumps to
eliminate excess gas not utilized in the reaction. Exploratory experiments have
been conducted in attempt to understand the affect that precursor ratios and
gap height have on performance. Spectroscopic ellipsometry measurements are
used to study the optical properties and measure the thickness of the thin
epitaxial films. 1D laser displacement measurements illustrate the ability to
control the gap height by adjusting the flow rate of the table gas; height
control within several microns can be achieved. A full factorial design of
experiments is being implemented to determine the key factors that contribute
to film uniformity.

The fluid dynamics under the
substrate are not well understood and computational efforts are being made in
COMSOL Multiphysics to gain insight to the underlying physics. It is believed
that the gap height may be the most significant factor in determining the
quality of film deposited. In comparison to the size of substrate, the gap
height is significantly smaller in magnitude and therefore the underlying fluid
dynamics can be determined by the thin gap approximation, or lubrication
theory. Lubrication theory can be used to correlate the pressure distribution
under the substrate with the gap height. The gap height has a substantial
influence on the behavior of the fluid motion. The fluid flow between an inlet
and exhaust vent under the moving substrate can be simulated as Couette flow
with a pressure gradient. When the desired direction of flow is opposite the
motion of the substrate an adverse pressure gradient is present and backflow can
occur. This phenomenon is strongly dependent on the gap height and the vent
spacing; it was found to occur when the gap height is small and the vent
spacing is large. If operating the system in the backflow regime, undesired
mixing of the precursors would result and chemical vapor deposition behavior would
be observed. This places a constraint on the system to ensure this operation
does not occur. The effectiveness of the nitrogen curtain as a diffusional
barrier can also be analyzed in COMSOL and will affect the design of the system
and operating conditions. Further understanding of the system will allow for optimization
of the design.