More than 20 years ago Movchan and Demchishin published their structure zone model (SZM)
of thin film morphology as a function of a single macroscopic parameter, the normalized substrate
temperature. In all its simplicity, this model already reflects many experimental observations
of thin film growth with surprising accuracy. Later modifications of the model which
included the influence ofresidual gas pressure and electrical bias potential in sputtering extended
its validity to this family of deposition processes. Evolutionary features of thin film growth
and the fractal nature of thin film structures were also discussed. None of these previous
models, however, providedfor a vitreous, fully dense structure as observed with some ion and
plasma assisted deposition processes. In 1988, the author proposed an extension of the original
Movchan-Demchishin SZM with afourth zone reflecting the vitreousphase. In thispaper, we
will discuss the validity ofthis extension and generalize the 4-zone model.
A two-dimensional hard-disk model of thin-film deposition is described; the
model is of the type originally introduced by Henderson et al.1 We have implemented a
simple (and necessarily approximate) way to incorporate the effects of surface diffusion
in our model, and a means to connect the input parameters of the computer algorithm to
the evaporation parameters of substrate temperature and evaporation rate. The effects
of surface diffusion and vapor-incidence angle on film structure are explored; the
effects of surface diffusion predicted by the model correlate with the transition in film
structure (Zone 1 to Zone 2) first described by Movchan and Demchishin.2
In this paper we extend our previous fluorination work presenting an unusual
fluorination effect in growing device quality superconducting film directly on
sapphire. The superiority of the film quality is attributed to the fact that
fluorine plays a significant role in the control of nucleation and in the enhancement of the growth rate of the superconducting film in the basal plane, therefore,
an "epitaxial" film is obtained. Furthermore, the high quality fluorinated film
can be grown at a lower temperature. As a result, the grain boundary weak link
effect and the interface diffusion between the superconducting film and the substrate are minimized. The superconducting film with a high critical current
density and a very smooth surface is achieved. With our technique, we believe
that device quality superconducting film could be grown not only on sapphire but
on other flexible inexpensive continuous substrates for high field applications
which could lead to a major technological advancement.
A method to obtain accurate thickness data to characterize the emission patterns of
evaporation sources is described. Thickness data is obtained through digital image processing
algorithms applied to the monochromatic transmission bands digitized from a set of multilayer
Fabry-Perot filters deposited on large flat circular substrates. These computer image-processed
taper-thickness patterns are reduced to orthonormal polynomial series expansions in two steps,
using Tschebyshev and associated Legendre polynomials. The circular glass substrates employed
to characterize each type of evaporation source are kept stationary during the evaporation process
of evaporation of each layer to obtain the specific thickness distribution for each type of source.
The theory of thin film optics is well developed for the spectral analysis of a given optical coating.
The inverse synthesis - designing an optical coating for a certain spectral performance - is more
complicated. Usually a multitude of theoretical designs is feasible because most design problems
are over-determined with the number of layers possible with three variables each (n, k, t). The
expertise of a good thin film designer comes in at this point with a mostly intuitive selection of
certain designs based on previous experience and current manufacturing capabilities.
Manufacturing a designed coating poses yet another subset of multiple solutions, as thin if in
deposition technology has evolved over the years with a vast variety of different processes. The
abundance of published literature may often be more confusing than helpful to the practicing thin
film engineer, even if he has time and opportunity to read it. The choice of the right process is
also severely limited by the given manufacturing hardware and cost considerations which may not
easily allow for the adaption of a new manufacturing approach, even if it promises to be better
technically (it ought to be also cheaper). On the user end of the thin film coating business, the
typical optical designer or engineer who needs an optical coating may have limited or no knowledge
at all about the theoretical and manufacturing criteria for the optimum selection of what he needs.
This can be sensed frequently by overly tight tolerances and requirements for optical performance
which sometimes stretch the limits of mother nature. We introduce here a know1edge-based system
(KBS) intended to assist expert designers and manufacturers in their task of maximizing results and
minimizing errors, trial runs, and unproductive time. It will help the experts to manipulate
parameters which are largely determined through heuristic reasoning by employing artificial
intelligence techniques. In a later state, the KBS will include a module allowing the layman user
of coatings to make the right choice.
Experimental and computer procedures are described to calculate in advance the optimal
monitoring wave lengths for each layer for any all-dielectric multilayer stack on any substrate
holder configuration. Thickness ratios [tooling factors] between monitoring and working
substrates are obtained experimentally to compute the spectral behavior of the irregular stack that
has to be deposited on the monitoring substrate to obtain the desired, correct all-dielectric thin film
stack on the working substrates. The computational algorithm takes into account the overall
monitoring beam's spectral response. The results obtained are a set of possible turning-point
monitoring wave lengths for each layer to be deposited and plots the transmittance or reflectance
behavior of these wave lengths as a function of layer thickness growth, to allow the user to select
the most convenient among them.
Ellipsometry is widely used for investigating the optical properties of thin films on planar substrates,
including films of adsorbed proteins or polymers. The average thickness and effective refractive index of the
adsorbed layer are calculated by measuring the A and 'P ellipsometry parameters. Unfortunately the thickness
of the adsorbed protein layers is often too thin to significantly affect the and 'I' parameters. However, using a
substructure consisting of an additional sublayer placed between the substrate and the adsorbed layer, we can
improve the sensitivities of both and 'P to changes in the adsorbed layer, provided that the thickness of the
sublayer is optimized. We show that for a Si02 layer on a Si wafer, the optimum Si02 thickness is about 1350
A when the incident angle is 70 degrees and the wavelength is 6328 A. The materials of the sublayer can be
metal, semiconductor and/or dielectric.
Spectral ellipsometric data taken on thin-film samples are often used to
obtain optical properties of the thin-film materials. A homogeneous model is
usually used for the films and a dispersive refractive index is varied to fit
the data. Often the data do not fit a homogeneous model. In this case the film
is modeled by one or more layers to approximate an inhomogeneous film. A common
procedure is to fix the dispersion curve for the film material and to model the
indices of the various layers with density or porosity as the adjustable
parameter. This paper demonstrates that dispersion and film inhomogeneity can
affect ellipsometric data in a similar fashion in certain spectral regions.
Thus, the dispersion curve derived using a homogeneous film model may be in
error due to inhomogeneities in the film. An example is presented for a silica
(SiO2) film on a silicon (Si) substrate. Ellipsometric data are calculated
using handbook values for the refractive indices of the materials, but for an
inhomogeneous silica film. These data are analyzed using a homogeneous film
with a variable Cauchy dispersion equation. The best-fit dispersion curve is
found to deviate significantly from the handbook data for silica.
A study has been made of the nucleation and growth of thin microcrystalline
films of Cu,Au,Pt,Ni prepared by ion-beam sputtering. In-situ measurements of
the film resistance during deposition have been analysed in terms of percolation
theory for the early, discontinuous, phase and thin film, grain boundary,
scattering for the quasi-continuous phase. This analysis yields values of the
percolation threshold, fractional coverage and lateral grain dimensions, for
different deposition rates, which are compared with the corresponding values
obtained from direct observations and soft X-ray multilayers reflection spectra.
The minimum thickness of the metallic nuclei is shown to be four atom layers.
Computer simulations of the structure of 13-atom metallic clusters formed by deposition on
crystalline and non-crystalline substrates were carried out. The formation of discontinuous thin
films during vapor deposition and the growth of bidimensional columns were simulated.
The use of SIMBAD, a two dimensional ballistic deposition simulation of the growth of thin films, is
suggested for investigation of refractive index inhomogeneities in integrated optics devices. Refractive
index variation as a function of packing fraction is obtained experimentally for evaporated MgF and
sputtered SiO2 by depositing films at angles. These relationships are used to translate SIMBAD density
predictions to refractive index predictions for films deposited on integrated optics topographies.
We simulate the growth of a thin film in two dimensions with a computer implementation of the molecular dynamics
(MD) method. The system consists of a krypton substrate maintained at a temperature of about 10 degrees Kelvin,
toward which argon atoms are periodically directed (with a velocity corresponding to 120 degrees Kelvin). The resulting
argon film follows the (horizontal) spacing of the krypton lattice until the thickness of the film approaches an average
thickness of about 10 monolayers. As deposition proceeds, the configuration of the film changes to incorporate an edge
misfit dislocation at the film-substrate interface; this relieves the interfacial stress. We also apply the MD method to study the relaxation of thin-film structures predicted by a hard-disk growth model.
We consider two variations of the growth model; the first is similar to that described by Henderson et al.,6 the second
is a variation which incorporates the effect of surface diffusion. The voids in the relatively open microstructure predicted
by the Henderson model are very effective in relieving interfacial stress. The numerous lattice defects (grain boundaries,
dislocations, and vacancies) in the denser microstructure predicted by the second type of hard-disk model result in a film
with high stress.
Molecular dynamics simulations of the structure of silica glass surfaces formed in a perfect vacuum as well
as in the presence of a water vapor show the type, location, and concentration of specific features formed in the
surface. A bond rupture mechanism which causes silanol formation far removed from the original reaction site is
observed. The 3-membered ring is proposed as a site for H adsorption in the glass.
According to the tangent rule, which relates the deposition angle to the growth direction of
the columns, which comprise many vapor-deposited thin films, and according to the relative
geometric and trigonometric relations between the evaporation source and the substrate, the
theoretical distribution formulas describing the uniformity of the colunmar structure are given for
four kinds of general fixtures. These distributions have been computed for various conditions.
We also analyze and discuss a structural thin-film model, and have concluded that the structure
obtained with the rotary spherical fixture is the best of several fixture alternatives.
A simple procedure to evidence inhomogeneity in the refractive index of a single
dielectric thin film included in a AR coating is developed. It requires only the
use of standard measurement equipment: a spectrophotometer and a three wavelength
e 1 lipsometer.
The results show a good agreement between different characterization methods:
reflectance, ellipsometry and single layer spectrophotometric analysis.
The cylindrical model of optical thin films microstructure is assumed for the derivation of the
equations expressing the structure-dependent anisotropic optical parameters in this paper. The
equations are given for various values of cylinder obliquity and for several incident plane
orientations. The equations are based on a capacitive model in which two index components are
given and on the dielectric ellipsoid theory. They are applied to modeling ZnS thin films and the
computed results have been given. We discuss the results, especially noting consistency of our
model-based computations with measurements.
A method based on continuously modulating the vapor deposition angle, δ, in the range from
0° < δ < 180° at an appropriate rate is proposed, by which the typical colunmar structure of vapordeposited
thin films can be prevented in favor of a more compact structure, similar to a bulk
material. The growth rates of such films are quantifed, and the their microstructure is simulated by
computer. At the same time, an experimental study has been performed successfully. Lastly, the
mechanism underlying these improved properties is analyzed and discussed.
In-situ measuring the intensity and characteristics stresses in optical coatings during
deposition can get a lot of important information for improving the properties of optical thin
films, which is more important for some devices uesd in modern science and technology. This
paper gives a new method which uses a kind of most in-fashion sensing technology, a
single-mode optical fiber sensing system developed in the middle of 1900s to test the
stresses in optical coatings during deposition with the aid of a single-chip micro-computer
and a micro-printer. The experimental results show that this method takes advantages of high
sensitivity wide dynamic range, small volume, available in-situ measurement and easily
interfaced to any vacuum system.
Aluminum nitride films were deposited on fused silica by reactive dc magnetron sputtering from an
Al-target in an Ar/N2 atmosphere.
In-situ measurements during the deposition provide data concerning mechanical stresses inherent to
the growing thin films. By variation of both the composition of the sputtering gas (Ar,N2) and the total
gas flow in the vacuum chamber, the occuring intrinsic stresses could be shifted in magnitude and
direction. Stress values of the thin films ranged from -1.2GPa (compressive) to +1.2GPa (tensile) when
the Ar/N2 ratio was varied between 3:1 and 1:3 for the different total gas flows of 5Osccm, lOOsccm,
and 200sccm (corresponding to total gas pressures of approximately 2x101Pa, 4x101Pa, and 8x101Pa
Investigation of optical film properties, such as refractive index, as well as of structural properties
were carried out and the results were related to the state of stress the films were in. The optical
characterization (n,k) was achieved by photospectrometry. Structure and chemical composition were
analysed by electron diffraction,transmission electron microscopy (ThM) and Auger electron
spectroscopy (AES) respectively.
The refractive index dependence on temperature and electric field can cause drastic modifications on
the optical properties of optical filters. A simulation has been developed which handles electric field and
temperature dependent refractive indices in multilayer stacks. In this paper we have restricted ourselves to
an electric field dependence and results obtained for single line rugate filters designed by the modified
Fourier transform technique, are presented.
The anisotropic and inhomogeneous dispersion ofevaporated thin films are considered in
this paper. We present possible models of both based on the basic dispersion equation, and
discuss methods to measure dispersion of both anisotropic and inhomogeneous indices, n(, 0)
and n(X, t), respectively . These methods are applied to a series of ZnS thin films deposited with
a range of incident angles  and thicknesses . The coefficients in the proposed dispersion
models have been computed according to the experimental results for 0 nm < t < 150 nm for n(A,
t) and over 0° < 0 < 500 for n(0). We conclude with an analysis and discussion of both indices.
Thin films are used as coatings and filters on solid substrates in optics,
condensed-matter physics, and many other fields. Ballistic-aggregation
simulations have served well to explain the properties of such films and to
suggest methods for their improvement. Thin films are also widely used
without backings, as self-supporting structures. We have used simulations to
advantage in the modeling of such components. To explain some of their
properties, 3-D modeling is essential and a simple modeling program is
adequate. Our studies using both 2-D and 3-D methods explain qualitatively
the tendency of such films to curl, the directional character of that curling,
the differing x-ray reflectance from the two sides of such a film, and the
variation of mechanical and other properties upon position in a large film.
We have performed extensive model calculations in order to understand the effect that roughness
has on the X-ray diffraction from multilayers. We have developed models to calculate the low and
the high angle X-ray diffraction spectra, including kinematic and dynamical simulations. These
model calculations were used to understand a variety of systems including crystalline/crystalline,
crystalline/amorphous and amorphous/amorphous multilayers. These model calculations were
compared with the diffraction spectra of multilayered systems prepared using sputtering and
Molecular Beam Epitaxy (MBE) techniques.
Using the experience acquired from these model calculations we have recently developed a
comprehensive nonlinear optimization program to refine the structure of multilayers from X-ray
diffraction spectra. Kinematic formulation is used to refine the high angle data and dynamical
(Fresnel) formalism was used to fit the low angle spectra. A comparison of the results obtained
from the structural refinements with EXAFS and artificially prepared rough multilayers indicates
that this type of approach give a reliable and speedy determination of the roughness, interdiffusion
and lattice parameter variations in multilayers. This work has been described in a number of papers
in the last few years.
There are already a large and varied number of existing selective coatings which are used for photothermal energy conversion, but lately coatings based on thin films produced by sputtering have begun to be used in industry. A coating of this type is based on a multilayer stack consisting of both dielectric and absorbing layers. Included in the absorbing layers are layers which have a dispersion of metal particles in a ceramic matrix, and in general the metal concentration in these layers is not uniform but graded. Programs for the calculation of optical parameters of such a multilayer cannot currently be purchased commercially. Therefore it was decided to develop a program which would calculate the reflectance, solar absorbtivity, thermal emissivity, and photothermal efficiency of a selective coating. Throughout the use of this program, it is possible to calculate the highest photothermal efficiency for any temperature, to evaluate the influence of each individual layer on the resultant behaviour of the entire stack, and to calculate the angular dependence of the solar absorbtivity.