1-1 Motivation
Evaporation in vacuum is a significant process for the production of thin films.
Through knowledge of physical and applicative features of the films and the results of considerable technological efforts in the field of evaporation techniques have increased industrial application of these techniques in many fields. The advent of electron beam evaporation in vacuum coating has therefore exerted a strong impetus on this development.
Electron beam evaporation is used to form coatings of a wide variety of materials, from metals to ceramics to semiconductors, with many different applications. Metal lines in microelectronic devices are most often produced either by sputtering or electron beam evaporation [Wolf and Rauber, 1986]; ceramic thermal barrier coatings are evaporated onto turbine blades [DeMasi-Marcin and Gupta, 1994]; various protective coatings are evaporated onto steel strip [Bakish, 1995]; and a new class of titanium-matrix composites is being made by evaporating titanium alloys onto fibers and subsequently consolidating them into a dense composite with high fiber volume fraction [Storer, 1993; Storer, 1996].
For different applications, the desired coating thickness distribution varies as well.
In many cases, a uniform thickness distribution is desired, such as alloy; in others where yield is more important, such as metal-matrix composites, a narrow plume is desired.
This area of the research is therefore devoted to analysis of the vapor flux distribution as a function of source temperature distribution, in order to design beam patterns which give rise to desired coating thickness distributions.
In previous documents and theses, there are not so many reports discussed the
three-dimensional computation model about flow simulator in the field of electron beam physical vapor deposition (EBPVD). In order to understand the phenomena of evaporation kinetic transport, I take advantage of Direct Simulation Monte Carlo method [Bird, 1994] to simulate the production of alloy in the chamber of multiple sources. To find what process parameters in constant evaporant surface temperature that substrate will achieve film thickness uniformity and composition we desire.
1-2 Modeling of Electron Beam Physical Vapor Deposition Process 1-2-1 Overview
Basically, physical vapor deposition is a vacuum coating process in which a directed vapor stream propagates from the evaporator to the substrate. Since generation and guidance of the beam must also take place in a vacuum, evaporation in this design should be happened in vacuum environment. As a result, system requires highly vacuum equipment. Fig. 1.1 shows the principle of electron beam evaporation. A plant for electron beam evaporation consists of a work chamber with a vacuum pumping system, a crucible for the evaporant, an electron gun, and a substrate with its fixtures and heating appliances.
In contrast to conventional heating modes, the evaporant is heated by a beam that impinges directly onto its surface; the greatest portion of the kinetic energy in the beam is converted into heat. The surface is therefore brought to such a high temperature that it becomes the source of a vapor stream. After a series of collisions between vapor particles, the particles approach substrate quickly. The substrate to be coated is arranged in this vapor stream and part of the vapor condenses on it in the form of a thin film. Here, one design is important; we must focus on the position of electron gun in the chamber.
Because of the simultaneous phenomena of evaporation and deposition, collisions between evaporated atoms are isotropic. In order to provide steady electron beam and prevent from damaging due to deposition, a 270° gun is often used. It usually locates on the outside of chamber, depending on magnet system to change electron beam direction and impinge on the evaporant. The concept above can be understood by right hand rule.
An electron in motion in a magnetic field experiences an electromagnetic force perpendicular to its direction and to the magnetic field.
The electron beam physical vapor deposition process is composed of three sections that we need to study. First, inclusion dissolution and flotation behavior in the melting/refining hearth, heat transfer, fluid flow and melt interface shape in the hearth.
Such models are reported elsewhere in the literature [Bellot, et al., 1993; Bellot, et al., 1998]. Second, evaporation kinetics in a periodically-heated surface. It is shown that beam scan frequency has a significant effect on evaporation rate from a molten pool, which may be used for control of composition in electron beam melting, and in conjunction with power for somewhat independent control of evaporation rate and source temperature in electron beam evaporation. The third, rarefied gas dynamics in electron beam evaporation, and an evaluation of source geometry designs for exercising some control over coating thickness distribution on the substrate. The vapor interactions above the melt can have a tremendous effect on the vapor flux distribution, and also on the recondensation of evaporated atoms back into the melt due to collisions in the vapor phase. This recondensation is discussed as a possible source of error in the evaporation rate calculations. It is reported [Powell, 1997] that the recondensation fraction is a function of the ratio of source diameter to equivalent mean free path d/λ0, and can
climb as high as around 10%, even without background gas present.
For vaporization of a substance in a high vacuum, the specific evaporation rate , the amount evaporated per unit time per unit area, according to Langmuir’s equation:
J
where Rg is the ideal gas constant, T the absolute temperature, M the molecular mass of the evaporating species, and its vapor pressure. The vapor pressure of a pure species
pv
pv can in turn be estimated using the Clausius-Clapeyron equation DT evaporating species and listed in Ref. [Brandes, 1983].
C D
The strongly nonlinear nature of Eq. 1.1 and 1.2 make evaporation rate extremely sensitive to temperature fluctuations. These fluctuations are governed by the beam power, spot size, frequency, and ability of the molten material to dissipate heat from the surface by conduction, convection, radiation and evaporation. Heating is by electron impact, and the dominant flow drivers are Marangoni shear and buoyancy. Evaporation can be said to fall into four regimes, presented here in terms of scan frequency for a given pattern geometry:
1. At very high frequencies (> 400 Hz), the dwell time will be very short (< 30 µsec) and the temperature fluctuations relatively small (~250℃), so evaporation near the beam spot will not be a significant fraction of total evaporation and power can be considered uniformly distributed over the scan pattern.
2. At high frequencies (60-400 Hz), the beam will generate a hot spot temperature high enough to affect evaporation rates (250-650℃ above surroundings), though this will not cause significant transient fluid flow. The low Prandtl number of metals leads to longer time scales for fluid flow than heat transfer, making possible significant transient heating without significant transient flow at these frequencies.
3. At moderate frequencies (20-60 Hz), transient flow generated by Marangoni shear is sufficiently strong to significantly modify temperature fluctuations and affect evaporation rates. The onset of significant flow can be estimated using the Peclet number as described in appendix A.
4. At low frequencies (< 20 Hz), various other phenomena may affect surface temperature, such as turbulent fluid flow, ionized metal vapor interfering with the beam [Tripp and Mitchell, 1993], and depressions in the melt surface generated by large vapor pressure excursions at high temperatures [Gilbaud, 1995; Tran Kong and Bird, 1978]. The transition to this regime depends on which phenomenon is dominant.
Also, there are some process parameters that will influence the evaporation rate, including beam power, pattern length, spot size, electron accelerating voltage, contamination in the evaporant, background chamber pressure, and so on.
1-2-2 Vapor propagation
Thin films made by electron beam physical vapor deposition play an increasingly important role in a wide variety of products and fields. However, the mechanics of vapor
transport are poorly understood, in part because rarefied gas behaviors very different form that of more familiar fluids. In particular, the stream of atoms evaporating from a surface follows the well-known cosine distribution, but in high-flux processes such as electron beam evaporation, collisions between evaporated atoms actually lead to a focusing of the vapor plume toward the surface normal, with the resulting flux distributed as or even [Schiller, et al., 1982]. This result is completely different from the intuitive expectation that more collisions will lead to dispersing of the plume.
For this reason, it has never even been considered that one could exercise any control over deposition profile in evaporation processes.
θ
cos2 cos3θ
The vapor stream emerging from an evaporator is characterized by the vapor flux distribution Φ(α). One approach to describing the vapor flux distribution of real small area evaporators is via a cosine function of higher order:
α unity. It has been shown that the description of the vapor flux distribution of electron beam evaporators according to Eq. (1.3) is fully adequate in an angular range of up to about 30° if the evaporation rate is not too high [Schiller, et al., 1982]. With growing evaporation rates, a more pronounced directional dependence is to be expected. It has been speculated that the extent of focusing should depend only on the ratio of source diameter to equivalent mean free path
n
/λ0
d [Powell, 1997], that is, the inverse of what might be called the local Knudsen number. The equivalent mean free path here is that
pv
Now the film thickness distribution on any desired substrate arrangement will be calculated from the evaporator characteristic. One wants to know the film thickness distribution on a plane substrate or substrate arrangement parallel to the surface of a small area evaporator and finds out the relation described as:
2 the film thickness for
dS dS0 rS dS0
=0
α on the substrate. is the distance of the substrate plane from the evaporator and is the distance between the normal to the evaporator center and the substrate any point under consideration. The film thickness distribution
with corresponds to the case of point source evaporation.
hv
If alloys are to be deposited, uniform composition of the film must be obtained over the total substrate surface and film thickness. In effect, two basic principles are used for the deposition of alloys: depositions from single or multiple evaporation sources. In the case of multiple evaporation sources the constituents are separately evaporated from several crucibles, the number of which correspond to that of the alloying elements, and
jointly condensed on the substrate. Deposition from two crucibles should now be explained by using as an example a binary alloy AB, that is, an alloy made up of the constituents A and B. Separate vapor stream with evaporation rates as given by Eq. (1.1).
When the crucibles are separated by a distance l, which is short compared to the distance between the substrate and the crucibles, one obtains an extended range where the vapor stream contains both alloying elements. Owing to the directional dependence of the vapor stream, however, adequate alloying constancy can be obtained only within a restricted substrate area. The influence of the geometric array on the uniform composition of the film depends on the ratio .
hv
hv
l /
Since the evaporation rate shows a pronounced dependence on the temperature, the accuracy of the alloy composition is limited. Thus a highly constant evaporator temperature is a necessary condition for obtaining uniform evaporation rates and represents a basic requirement for producing films of adequate alloy constancy when using co-evaporation from several crucibles.
Co-deposition can be performed with the aid of various electron beam evaporators.
Another possibility is to use the beam of one gun to heat several crucibles. In this case the beam power is distributed among the individual crucibles by programmed deflection [Cron and Adams, 1969]. Beam power distribution to the crucibles take place by adjusting a defined duty cycle for the deflection currents of the beam guidance system. In this way it is possible to adjust the evaporation rates of the constituents and thus control alloy composition of the film.
Multiple-source evaporation is used in the manufacture of alloy films whenever the evaporant cannot be produced with the required composition and single-source
evaporation proves to be impossible. Mixing in the vapor phase is practical in cases where the vapor pressures of the constituents are vastly different, for example, differing by four or more orders of magnitude.
Simultaneous electron beam evaporation from two sources appeared to be a promising solution to the basic problem of controlled and reproducible deposition of alloy films. However, most alloy film deposition work seems to be still carried out by evaporation from one alloy source. Yet composition control is quite problematic and limited in this process. On the other hand, co-deposition from separate sources usually involves more complicated and expensive equipment. Difficulties are also encountered in controlling evaporation rates of the individual constituents. Employing one electron beam generated by a self-accelerated gun and oscillated in a controlled manner between two materials is a relatively simple scheme, considerably less expensive than a double-gun configuration. The dwelling time of the beam on each source determines the heat input and, therefore, the rate of evaporation of each constituent. Composition of co-deposited films can be controlled by varying the ratio of the two dwelling times.
1-3 Literature Survey
The development of electron beam technology into a special field of its own is closely related to the advances in vacuum engineering and electron optics. The history of this basic science has been extensively dealt with elsewhere. In 1905 Marcello von Pirani successfully carried out the first experiments on electron beam melting of refractory metals such as tantalum. But since vacuum engineering and electron optics were still in their infancy at that time there was no industrial demand for such a technique. In 1938
von Ardenne and Ruhle employed magnetic-lens systems for beam focusing to drill small bores and evaporate metals, respectively. Around 1950 Steigerwald wrote a paper on the technical possibilities of the beam as a tool for drilling and machining in the micron range.
In the long run, however, the development of nucleonics and space engineering called for new technological processes for, say, welding, melting, and evaporation. In the mid-1950s this situation stimulated the use of electron beams for technological purposes.
A characteristic example is Stohr’s work [Stohr, 1958] on the technical development of electron beam welding. In the following years electron beam evaporation was increasingly used for many coating jobs. Up to 1965 all these techniques were developed to maturity so that electron beam melting, welding, evaporation, and machining gained a secure position as production processes. After 1975, the industrial application of electron beam processes has been processing in microelectronics and radiation treatment of plastics and coatings were developed into full-fledged production techniques. In the 1990s, the instruments of electron beam technology had been well developed. Due to some needs in the field of space, semiconductor, many applications about alloy began to be studied. In 1991, Hiroshi [Hiroshi et al., 1991] used electron beam furnace to melt sponge titanium. Alec [Alec, 1992] took advantage of electron beam melting providing the incremental improvement for both titanium alloys and superalloys which we need.
Tomoo [Tomoo et al., 1992] investigated aluminum evaporation behavior in the electron beam cold hearth remelting process. In order to clarify the quantitative effect of the beam oscillation rate on the aluminum evaporation behavior, Hideo [Hideo and Alec, 1992]
melted Ti-6Al-4V alloy, developed a mathematical model on the basis of a small scale electron beam melting experiment. Schulz [Schulz, et al., 1995] experimented on rotating
cylindrical electron beam to deposit thermal barrier coatings, using ZrO2-based ingot sources with stabilizing oxides of 6.5 and 20 wt.% Y2O3 and 25/2.5 wt.% CeO2/Y2O3 respectively. In 2000, the direct simulation Monte Carlo method was used by Boyd [Boyd, 2000] to model the physical vapor deposition of titanium using electron beam evaporation. It is concluded that electronic energy is an important factor to consider in the modeling of flows of this nature. In the same year, the deposition of superconducting films of YBa2Cu3O7-δ was investigated both computationally and experimentally by Fan [Fan, et al., 2000]. The numerical analysis and experimental studies employ DSMC method and atomic absorption spectra taken in the evaporated yttrium plume and deposited film thickness profiles. Collisions between the atoms are found to have a significant effect on the film growth rate and area of uniform deposition as the evaporation rate of yttrium increases. Powell [Powell, et al., 2001] also used DSMC method to compare titanium evaporating from a disk surface with ring source. Beginning from 1998, a research team led by Prof. Wadley [Wadley and Groves, 1997] at the University of Virginia designed a new physical vapor deposition technique, named for Directed Vapor Deposition (DVD). Compared with conventional electron beam physical vapor deposition, it used low vacuum electron beam evaporation in combination with a carrier gas stream to transport and vapor spray. Recently, it is highly studied, with its high rate, efficient deposition of refractory elements, alloys, and compounds onto flat or curved surfaces. Hass [Hass, et al., 2004] took experiments and found that the coating thickness around the circumference of a stationary, non-rotated fiber placed perpendicular to the axis of a gas jet containing aluminum atoms is sensitively dependent upon the jet’s Mach number and the chamber pressure near the substrate. By employing gas jets having
low Mach numbers (< 0.1), highly uniform coating of aluminum on cylindrical fibers have been achieved without fiber rotation. DSMC simulations have been used to understand the fundamental phenomena.
1-4 Specific Objectives of the Thesis
Based on previous reviews, the current objectives of the thesis are summarized as follows:
(1) A DSMC code developed in MuST Lab. is used to model this rare flow field.
(2) To verify the flux distribution obeying cosine distribution.
(3) To simulate a simple case — one source in the chamber, observing the variation of density, temperature and velocity in the chamber.
(4) To contrast simulations with and without background pressure effect.
(5) To deposit the alloy Ti6Al4V and discuss the uniformity of composition and thickness on the substrate.
(6) To change some parameters to compare the uniformity with each other, making conclusions how we should adjust to obtain acceptable uniformity.
The organization of the thesis would be stated as follow: First is this introduction, and next is the numerical method. Then show the results and discussions. Finally summarize and recommend the future work.