Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
t
A l.O Introduction and General Discussion
EB Graper
Al.0.0 INTRODUCTION
Thin films, invisibly thin on edge, are at the core of integrated circuits, materials technology, optics and
communications. The process for deposition of thin films underlies the production of a vast range of
consumer products. This deposition technology is a mix of applied science and art with physics at its core.
Understanding, practical application and seldom published technology are the focus of this work.
The physical vapour deposition processes of interest in this handbook all take place in a vacuum. The
vacuum environment is a complex one and is neither a void nor inert. To be successful, thin film deposition
must be done with awareness of the effect of the environment upon the vapour flux and growing film.
A general understanding of vacuum is of importance and the literature is extensive, most finding its roots
in the classic work by Duschman and Lafferty [l]. The three most important aspects of the vacuum
environment to thin film deposition are: the pressure, expressed as the mean free path (MFP), the partial
pressure of reactive gasses in inert working gasse& .;,ind~ the fi_~in~ v.a,pou~. fil!Iv~(tp. ~~~S~~ve ~gas impfogement
rate ratio. These aspects of vacuum are tabulated against absolute pressure in table 1. Each of these aspects
of vacuum has a direct effect on thin film deposition. From table 1 it is seen that at about 1 x 10-5 Torr
the mean free path is 5 m or about ten times the usual source to substrate distance for evaporated thin film
deposition. This means that at about 1 x 10-5 Torr about 5% of evaporant atoms will undergo a collision
during their travel from the source to the substrate giving up energy, being scattered and perhaps reacting
with the residual gas, usually water vapour. The mean free path is a particularly significant consideration
when scaling up vacuum processes and should be at least ten times the source to substrate distance. When
source to substrate distance is increased, the pressure must be decreased proportionally or the evaporant
will have more interactions with the residual gas in the vacuum chamber.
@ 1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0: 1
Thermal Evaporation
The second significant pressure consideration is the arrival rate ratio: as a film is growing, it is also being
bombarded by residual gas. The ratio tabulated is the molecular arrival rate ratio of film vapour at 10
Als to that of residual gas molecules. When this ratio is unity each atom of film is accompanied by one
of residual gas. Though the reaction probability of this gas is on the order of 0.1 it is obvious that it is
desirable to have significantly more evaporant film vapour arriving at the substrate surface than potentially
contaminating gas. In evaporation this residual gas is the total chamber pressure and is predominantly
water vapour. In sputtering, the working gas represents nominally 99.9% (10- 3 Torr Ar, with a residual
contaminant pressure of 1o- 6 ) of the total pressure. The effect of the residual contaminant pressure on
films during sputtering is as severe or more so, as is the total chamber pressure during evaporation because
much of the gas is ionized, making it more reactive. Unfortunately, an expensive and complex residual gas
analyser with a UHV pressure reducing inlet is necessary to measure this extremely important pressure.
The arrival rate ratio can be improved by either reducing the pressure or increasing the film deposition
rate. Particular care must be taken when scaling up a process by increasing the source to substrate distance
because the deposition rate will fall as the square of the distance. This means increasing the source to
substrate distance from 16 in (40 cm) to 25 in (63 cm) requires doubling the deposition rate to maintain
the arrival rate ratio (or halving the pressure).
Using table 1 the maximum deposition pressure can be established for a thin film process based upon
its sensitivity to residual gas. Maintaining film purity during growth of films by MBE, typically done at
10-9 Torr or lower, requires an exceptionally low pressure because of the low (0.1 Als) deposition rate.
Maintaining high film purity when sputtering with Ar working gas at 10- 2 Torr and a arrival ratio of 10
requires the total reactive gas partial pressure to be below 10-6 Torr (table 1) during deposition.
Vacuum thin film deposition processes generate a source of vapour flux of the desired film material to be
condensed upon a substrate. Vaporization of a solid or liquid material is the process of changing its phase
to a vapour, driven by the input of energy. Thin film deposition sources are the means to bring about this
phase change.
The energy input required for vaporization is identical for sputtering and evaporation though the energy
sources are very different. The vaporization energy, delivered by the vaporization process consists of the
following components.
( 1) Latent heat, the heat necessary to raise the temperature (energy) of the material to that where phase
change can occur.
(2) Phase change heat, the heat of vaporization and the heat of fusion (for melting materials). This energy
is the dominant energy required for vaporization.
(3) Kinetic energy imparted to the vapour in excess of the minimum necessary for phase change.
The microscopic process of evaporation, to be rigorously considered, must be understood from a statistical
thermodynamic perspective. This is best done by referring to the extensive literature [1]. An important
intuitive understanding of vaporization can be developed from two different perspectives depending upon
the means of energy input. The first is evaporation and the second sputtering.
The ~Y.~po~,~ti9.n P.\:9~-e,~& , (~ the pbas1 .change, using heat, of a solid to a vapour, as occurs in boiling. The
difference between vacuum evaporah-on and boiling is that the boiling point is defined as the temperature
at which the phase of a material changes from a liquidto a gas at one atrnosp~ere9f press1:1r~. , In thin film
evaporation the pressure of vapour over the' evaporanis ~~rface 15 ori "file 'order of-6. i -.:1 T~rr depending
upon the evaporation rate when the phase change occurs. The temperature is therefore much reduced
from the boiling point (i.e. Al boils at 2300C and evaporates at 1080C at I Torr). From tabulated
vapour pressure tables and equations [2] the approximate evaporation temperature of many materials can
be determined. The thermal evaporation of a material requires the addition of the heat of fusion and heat
of vaporization, and latent heat (for Al heat of fusion = 270 cal/cm 3 , heat of vaporization = 78 kcal cm3 )
and the thermal kinetic energy supplied to the evaporant by the electron beam or hot refractory metal
A 1.0:2 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Introduction and General Discussion
source. The latent heat and kinetic energy are negligible. Upon condensation this energy is given up to
the substrate and represents the principal source of substrate heating [3]. The evaporation temperature
determines the median kinetic energy of the evaporant (1000C = 0.2 eV and 2000C = 0.4 eV). This
energy is a measure of the velocity of the evaporant.
In simple evaporation the kinetic energy of the evaporant is limited to thermal energy. Ion plating and
ion assisted deposition processes increase this energy by either ionizing a working gas (ion plating) and
accelerating it through 100-1000 V or bombarding the growing film from a low energy source using
an energetic neutralized ion beam (ion assisted deposition) during growth, adding energy to the growing
surface. Both precesses can increase the evaporant energy at the film surface by an order of magnitude
or more, sharply improving adhesion and film density (as well as the throw, the ability of the evaporant
to uniformly coat into holes and cracks). This added energy also significantly increases the heating of the
substrates during deposition.
The energy to vaporize a material being sputtered is delivered by momentum transfer to the target from
the energeti.-; working gas ions, usually Ar, rather than heat. These ions of the working gas are accelerated
in the sputtering discharge to a few lOOs of electron volts. They give up this energy upon collision with
the target surface causing a cascade of collisions within the target. Those surface target atoms receiving
sufficient momentum from the sum of the collision cascades will be sputtered, often with significant excess
kinetic energy. That energy of the working gas ions not imparted to the sputtered material or retained by
the working gas atom as it leaves the target heats the target and represents the majority of the discharge
energy. The sputtered atom, on leaving the target, with 10-100 eV of energy enters the working gas
plasma and drifts several cm to the substrate. During this drift the target atoms undergo thermalizing
collisions with the working gas (mean free path = 1 cm at 5 x 10-3 Torr) reducing their energy to 1-10
eV. Upon impacting the substrate the target atoms condense, giving up their energy as heat, and forming
more adherent films than those grown by evaporation.
The sputtering plasma additionally offers the opportunity, by properly biasing the substrates, for ion
bombardment of the growing film, driving reactions, improving density and adhesion. These energetic
deposition processes, as in ion assisted evaporation, sharply increase the energy input to the substrates
during film growth, making control of the substrate temperature difficult.
Alloys and mixtures of materials find wide technical application but are difficult to deposit as thin films. No
method easily deposits all alloys while preserving their composition. Sputtering is the deposition method
of choice for alloys but secondary effects and large expensive targets are a limitation. Using evaporation,
the constituents of alloys (with rare exceptions) evaporate at different rates, due to their different vapour
pressures, producing films of variable composition. Evaporation or sputtering of the constituents of alloys
from individual sources is versatile but difficult to control and, because the sources cannot be co-located,
films of varying or layered composition result. This can be used to advantage where a range of alloy
compositions is required for study.
Sputtering can, in theory, produce films identical in composition to those of the target. The sputtering
process vaporizes the alloy target quantitatively, atom by atom. Good practical results are obtained with
two limitations . .First, the sputtering yield (film material vaporized from the target per Ar ion impact-see
section A3) differs greatly between metals requiring long target conditioning to develop the target surface
composition to the inverse of the sputtering yield. It has also been measured that different metals leave
the target surface at different average angles causing composition variations across the substrate surface
[4]. Composite targets of sectors of each alloy component can give good results, using a rotating substrate
table.
Evaporation of alloys is only practical for those having similar vapour pressures. The interactions of the
components of alloys during evaporation are complex. Theories of alloy evaporation have been successfully
developed [5] but do not find direct application in film production. Alloys will virtually never evaporate
with exact composition. Only if, at evaporation temperature, the vapour pressures of the components
of an alloy are identical will the film composition be nearly that of the starting alloy. Practical alloy
@ 1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:3
Thermal Evaporation
evaporation requires using an evaporant alloy with the composition chosen to yield a film of the desired
composition. The composition of this 'special alloy', selected by trial and error to be one yielding films
of the correct composition, is approximately the inverse of the volatility of the alloy components. The
evaporant inventory in the source must also be kept constant, to maintain constant composition, by frequent
replenishment with alloy matching that of the film. For example, at 1mm vapour pressure Al and Cu have
temperatures of 1082C and 1142C respectively and Al-Cu alloys evaporate well, but are Al rich. A
'special alloy' of Al-4%Cu alloy will yield approximately Al- 2% Cu films. A film of 95% Pb (627C at 1
mm)-5% Sn (1092C at 1 mm) solder is much more difficult to evaporate. A 'special alloy' of about 95%
Sn is required due to the preferential evaporation of the Pb. The alloy composition will change rapidly as
the small amount of Pb in it is evaporated. This alloy must, therefore, be replenished continuously to a
constant volume with material matching the film composition. Alloys can also be flash evaporated from
a hot, essentially dry source. This technology is limited to alloys available as gas free wire that melt at a
high enough temperature to permit feeding onto the hot flash evaporation source.
Al.0.4 DISTRIBUTIONS
A principal requirement of all thin films is that they coat the substrate uniformly. Unfortunately, in
evaporation, the approximate point source of evaporant, is intrinsically non-uniform. To mitigate this
non-uniformity, one uses understanding -of the physics of distributions to design tooling that supports the
substrates to receive uniform coating or uses tooling that is distribution independent.
Evaporant distribution variations can be divided into two classes, those due to variation in source to
substrate geometry and those caused by the evaporant flux. At a microscopic level the flux from all
emitters can be treated as a series of cosine point sources (figure Al.0.1). The cosine emitter, rooted in
statistical thermodynamics, describes the flux from a small emitter (a pinhole) on a evaporating surface: ~"
This ideal ernittei:has-a. flux
~_,,.-----. ~':.'::'.~"'::; .
of
.. ... .
~
( b)
Figure Al.0.1. The distribution of a vapour flux as (a) a single cosine from a
point and (b) as the sum of the cosines over an extended source. Note that the
direction of the extended emitters is normal to the emitting surface.
In the simplest uniform isotropic case, a series of equal cosines covering a point source in space, the cosines
sum to a uniform evaporant flux in all directions. Taking a real example (figure Al.0.2) the cosines radiate
isotropically about the axis of the coil creating a cylindrically uniform source. The distribution parallel
to the axis of a coil, assuming uniform loading, is uniform over the length of the coil, diminishing at the
ends. One can see that, for a point source, a substrate holder that is a fixed spherical section, centered on
the source, will yield uniform films. It is also apparent that even for a simple real case, a coil, this solution
fails in one axis. As a result, non-uniformity must be tolerated or tooling that averages the evaporant flux
by moving the substrate must be used. Before presenting specific types of tooling, understanding a simple
isotropic, cosine and general case of a summed cosine distribution, as represented by electron beam guns
is important.
Al.0:4 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Introduction and General Discussion
The distribution describes the evaporant flux intensity and direction from the source. This flux is a function
of source design, evaporation rate, source fill, evaporant material and chamber pressure (vacuum). Three
effects combine to determine the film thickness on a substrate; the source to substrate distance,the evaporant
distribution from the source and the angle of incidence on the substrate. To receive. u~i fo~~ fiims a ~io~d-' :.
0
The effect of source to substrate distance on film thickness is accurately described by ~ &quare law-. The
film flux through -a--given-solid angle' is the same whether that flux is passing through a small area close to
--tl1e-source br a larger area farther away. The exact relation can be developed from the ratios of the areas
subtended by a constant solid angle of evaporant flux as follows.
Substrate
A I
(b)
Source Source
Incidence~
for a small solid
angle of radius a
I/
I
( c)
Source
Figure Al.0.3. The geometric relations of (a) a radiating source, (b) a source on a flat
substrate, and (c) the details of the incident flux on a flat substrate showing the circular and
elliptic area of incidence at points A and B.
@ 1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:5
Thermal Evaporation
The area of spherical zone centred upon the source (figure Al.0.3(a)) can be expressed as follows:
. "
/. A1 = 2rrd?O - case) = df r })
(Al.0.3)
A1 2rrdi(l - cos e) di ''
In equation (A 1.0.2) we see the area Js proportional to the square of the distances from the source. That
is, if one doubles the source to substrate distance, the area is four times as large.
The thickness of the film deposited by an evaporant flux is inversely prop.o~onal to the area upon which
it is deposited: i
\..____,_.~--'"
(Al.0.5)
Real films follow this square law accurately except where the pressure is high enough to cause scattering,
thu;s '(foubil.ri'g the aisfance yields a filrri. f/4'".as thick.
It is illustrative to develop the ratio of the flux at the centre (point A) and edge or corner (point B) of a
flat substrate (figure Al.0.3(b)) assuming uniform (isotropic) emission from the source, using the inverse
square law. From equation (Al.0.4) we know that the evaporant flux is a inverse function of the square
of the source to substrate distance. That is the flux (rate, R) at A and B are:
(Al.0.6)
Assuming an isotropic distribution the function f (e) is independent of(} and identical at A and B so the
relative flux at the edge of a flat plate is given by
Ra df 2
- =- =cos e. (Al.0.7)
RA di
Taking f (e) =cos e to describe the flux from a simple cosine point emitter and again taking the ratio of
the rate R at points A and B:
(Al.0.8)
Ra df case
(Al.0.9)
RA= di
From the isotropic case (equation Al.0.7), substituting d?f d'f. = cos 2 e we have
Ra 3
- =cos e. (Al.0.10)
RA
The relative flux (rate) at the edge versus centre of a flat substrate is of only limited interest. This is
because, for the foregoing to be useful, the rate must be equal to the thickness at points A and B. That
is, the film must impact the substrate nominal to its surface. This is, of course, not true at point B. To
expand this example for film thickness we must consider the effect of the angle of incidence of the film
flux at points A and B. If we examine a small circular area (solid angle) point, A, over the source and a
similar solid angle of flux at the edge of the substrate, B, we find that the one at the edge appears as an
ellipse on the substrate surface (figure Al.0.3(c)).
Al.0:6 Handbook of Thin Film Process Technology @ 1995 IOP Publishing Ltd
Introduction and General Discussion
Taking the ratio of the area coated at A, a circle diameter a, and the ellipse at B
AA (rr/4)a 2
As
= (rr /4)(a 2 /cos(}) = cose.
Combining this with equations (A!.0.7) and(Al.0.10) using T = Rate/Area = R/ A, we find thickness
variation across a flat substrate of a film from an isotropic source given by
Ts 3
- =cos 8 (Al.0.11)
TA
and for a cosine source by
Ts 4
- =cos 8. (Al.0.12)
TA
In the general case for evaporant distribution f (8) on a a flat substrate holder, the thickness distribution
is given by
. Ts dx
- = 2 cos8/(9) (Al.0.13)
TA dB
with 8 being the angle between the substrate and evaporant flux at B.
In a typical evaporation system (37 mm (15 in) source to substrate distance, flat substrate holder 25 mm
(10 in) diameter) the angle (} = 18 (cos 18 = 0.949) yielding a thickness at the edge (cos(} = 0.949) 3 of
0.85 times (cos 3 8) that in the centre for an isotropic source and 0.81 (cos4 (}) times that at the centre for
a cosine.
The distribution from an isotropic or cosine emitter represents that from ideal sources. Unfortunately, real
sources are much more complex. Each resistance heated source has a unique distribution (figure Al.0.4)
and the only published measurement of these, the Sloan notebook, is long out of print. The empirical
description of the flux from real electron beam heated sources [6] is given by:
where a is the isotropic component and n is the__be~ming exponent. The distribution function (equation
(Al.0.14)) contains two rate depCnoe'iitZ~cifi~i~nts ;; . .and~-.(ii"gure Al.0.5). These functions describe a
beaming or focused component (cosn 8) of the flux derived from the hot, slightly concave, central emitting
area of an electron beam heated source and an isotropic component, a, originating from the high pressure
A~tD~_isotropic
_, _ _ __case s
(low mean free path, viscous flow) gas cloud on the surface of the evaporant. This description reduces to
for \ ; and
_ ......--- f (.,to 'cosine emitter' for n = 1.
:::: to the often referred
(j
oO~
Taking the description from rea~source(equation (Al.0.14)) and substituting it for the angle independent
isotropic distribution in equation (Al.0.6)
1
RA= 2[(1 - a)cosn 8 +a] at A, (} = 0 (vertical)
dA
- d2 (Al.0.15)
A
@ 1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:7
CDs &
Thermal Evaporation
1
Ra= 2[(1- a)cosn ()+a].
da
RA and Rs are now general terms for the flux at points A and B on (figure Al.0.3) any flat substrate
holder.
15 (al 15 (bl
"O
c ;;
0
u
c
cu 0
u
VI cu
~ 10 VI
'-
10
Cl. cu
Cl.
VI VI
E E
- s
0
'-
VI
C1I
c
0
.!::
VI
O"
c 5
.E ..9
-d
cu
e::::
~
OJ
e::::
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 1'00
Degrees Degrees
15 (d)
15 ( c)
:0-
"O c
c 0
u
0
u QJ
QJ VI
VI
10 .,,. ,,- ....... '- 10
' ....
'- cu
cu
Cl.
"" ~
VI
E I
/ "'"' '' \
VI
E
0
..._
'-
VI
C1I I
I
I '
I
\
\ -
0
'-
VI
5
O"
5
c
.E '' I
I E
c
-
cu I \ QJ
~
er:: I
I
' \
l
\
d
e::::
' '
\
\
Figure Al.0.4. Evaporant distributions from (a) a dimple boat with a 40 mm x 10 mm dimple, (b) a dimple boat
with a 10 mm diameter x 3 mm deep dimple, (c) a coil 40 mm long with 6, 10 mm turns and (d) a crucible 10
mm inside diameter and 12 mm high (solid line-longitudinal axis).
From the foregoing, it is seen that the farther from the source (smaller angle B) a substrate is placed,
the more uniform the distribution. Unfortunately, the deposition rate falls as the square of the source to
Al.0:8 Handbook of Thin Film Process Technology @ 1995 IOP Publishing Ltd
Introduction and General Discussion
--
~
c
.~
u
/
~ 1
8
-s
c:
c
QI
4
c.
x
QI
10 100 1000
1
Deposition rate at 20 cm (A s- )
substrate distance. Long distances yield low rates and a long path for the evaporant. This increases the
opportunity for the evaporant vapour or the slower growing film to react with the vacuum environment.
Linear scaling requires constant rate at the substrate and a constant mean free path to source/substrate
distance ratio to maintain the deposition environment when increasing source to substrate distance.
The thickness distribution on any substrate holder can be calculated in the same manner as that for a flat
plate using the general distribution function (equation (A 1.0.14)). Though analysis using this function is
useful its value is limited by the variability of the distribution (figure Al .0.6). For this reason the design of
substrate holders to match specific distributions is uncommon except where one must coat large substrates
with unusual curvatures such as spacesuit visors. Instead, substrate holders are empirically designed to
make them distribution independent.
~ 6
::>-. 5
.c
E 4
VI
.:=
c
0
~
::::J
.a
~
Ci
10 20 30 40 50
Angle from normal (degrees)
The complex and variable evaporant distribution of all sources (equation (Al.0.14)) requires well thought
out substrate holders to obtain films of sufficient uniformity to meet normal functi~nal requirements. e
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:9
Thermal Evaporation
substrates must be positioned to receive a uniform film, or moved such that all receive a comparable
integrated evaporant flux. Uniform coating of c.urved substrates (lenses and mirrors) often depends upon
experience and cannot be achiev~ by" ~oper ci1o!ce_qf as_~ bst~~te ho_lder alone. Of particular importance is
the positfon ofthes ourc-erelative to the substrates as discussed in the following section on source/substrate
geometry~ The r an"ge ofs ubstrate holder designs is boundless but among the most common are shown in
figure Al.0.7 .
Source Source
(a) Flat plate {b} Dome
Source
( c) Flat planetary
Source
( d) 0 om e p I an eta r y
Source
f e) Drum
Figure Al.0.7. Common substrate holders.
Flat Plate: As a fixed or rotating holder, flat plates are inexpensive, and all very large substrates act as flat
plates. With a included evaporant angle of 30 and a cos 1 () (simple cosine) source distribution the best case
uniformity will be 10%. The techniques in the following section on source/substrate geometry enable
substantially better uniformity to be obtained using off-axis sources and rotated flat substrate holders.
Domes: The substrate holding dome is the most common simple substrate holder. Spherical domes (or
planetaries) with a radius equal to the source to substrate distance eliminate geometric distribution errors.
The evaporant flux is normal to the surface, improving pattern mask resolution. This leaves only the
source flux introduced thickness non-uniformity of 5% for a simple cosine (cos 1 0) source flux and a
36% included angle. The uniformity of the film deposited on a spherical dome can always be improved
Al.0:10 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Introduction and General Discussion
,/
by moving the dome farther from the source. Domes are a good compromise between the complexity of
a planetary and the poor uniformity of a fiat plate.
Planetaries: Where film uniformity of better than 3% is required a planetary substrate holder is the only
option. Though not geometrically perfect, planetaries regularly produce films as uniform as 1 % [7]. A
large amount of complexity, loss of reliability and cost must be justified to obtain this uniformity. Most
planetaries use 3 dome planets positioned on a spherical surface or 5 ft.at round planets in a plane over
the source. As the name implies, the individual planets rotate while the entire set is rotated above the
source. This rotation requires a powerful motor drive and numerous bearings within the hot part of the
coating chamber. Though a planetary will yield substrates with films of uniform thickness, the cyclic rate
variation due to the planetary motion during coating can be extremely large. Rate variations of a factor
of 9 (3-27 Als with an average of 9 Ais) have been measured for a common high capacity dome type
planetary with a included angle of 100. Planetaries also introduce a range of operational problems. They
are complex to handle for loading because of the required substrate retaining clips and are quite difficult
to heat. Vigorous cleaning often bends the planets causing jamming. If large substrates (more than 50%
of the planet diameter) are coated in a dome planetary, non uniformity will be seen across the substrates
to the extent that they flatten the dome.
Drums: These ferris wheel like holders offer extraordinary capacity while remaining mechanically fairly
simple. The uniformity of a drum with a centrally located source is similar to that of a fiat plate in one
axis and a dome in the other. This can be sharply improved by installing masks on both sides of the source
to reduce the thickness along the centerline. Good uniformity is obtained without masks in large multiple
source drums. Exceptional economy can be obtained, as in decorative coating, by centrally locating an
array of isotropic emitting resistance sources within a large drum.
In all systems with rotated substrate holders and having the source centred under the axis of rotation, the
film is thickest in the centre of the holder. This is because sources radiate in a beam (equation (Al.0.14))
subtending a narrow angle. The uniformity can be sharply improved by moving the source from beneath
the axis of rotation out toward the edge of the holder. Laboratory experience shows that a film uniform
to 3% can be deposited on a fiat plate holder from a electron beam heated source at approximately the
outer edge of the holder and one plate holder radius from the source. A similar optimal off centre source
position exists for domes and planetaries. The uniformity on a planetary with optimally located sources
can exceed the limits of measurement (0.5% ).
In addition to improved film uniformity, off centre location of evaporation sources significantly simplifies
the vacuum system layout. There is no longer only one source position but a annular ring yielding uniform
films. As a result several sources can be located on this ring. The thickness monitoring head located at
the edge of the substrate holder can now be positioned more nearly over the source, improving thickness
measurement repeatability. Off centre sources provide better step coverage on planetary substrate holders
but, conversely cannot be used for lift off or with pattern delineation masks because of the widely varying
evaporant angle of incidence. Off centre sources and reduced source to substrate distance can deposit
uniform films on convex optics. Unfortunately, at this time, no known software provides a good solution
to the distribution problem, so experience and trial and error must be used.
Off centre sources potentially introduce several process problems. When the source is not on the axis of
rotation the evaporant rate varies even more than for a axially positioned source. This can cause problems
with reactively evaporated or rate dependent films. In addition, this angle of incidence varies with each
rotation of the substrate holder across the source. In the case of large dome planetaries the angle of
incidence can vary from normal (90) to 25 for a source located half way between the rotation axis and
the edge of the planetary during each rotation. It is well established that the columnar growth direction of
films is toward the evaporation source. Recent measurements also show that significant changes in index
of refraction occur at beyond an angle of incidence of 30 [8]. In addition, the evaporant from off-centre
sources is less efficiently captured by the substrates, reducing the average deposition rate and increasing
evaporant cost. 1
'
/
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:11
Thermal Evaporation
The selection of evaporation material is the critical process choice as it will be the thin film. Two principal
factors govern evaporability:
( 1) Bound gas
A material, no matter how attractive as a thin film, is useless if it cannot be acquired in an evaporable
form. For electron beam guns this means large gas-free chunks or chips. Resistance heated sources require
wire or small chips. In general the lowest possible volume-to-surface area ratio is desired to minimize
surface absorbed gas (principally water vapour). Powder, without premelting, is almost always difficult or
impossible to evaporate, particularly in an electron beam gun.
Today a number of suppliers offer 'evaporation grade' materials for thin film deposition. These materials
are, ostensibly, tested for evaporability and should yield defect free films of the nominal composition. The
'evaporation grade' of the most common thin film metals, Al, Au and Cr which usually yield defect free
films, work very well. Unfortunately this is not true for many less common metals, such as Fe, Ni, Pt,
Si and Ti. These metals absorb gas during purification, mostly H and 0. Upon heating to evaporation
temperature, this gas is liberated driving droplets of metal from the evaporant. This 'spitting' is the
principal cause of film defects. Though most 'evaporation grade' metals are vacuum melted, this does
not heat them to as high a temperature as does evaporation and often does not begin to drive off the
spit causing absorbed gas. 'Conditioning' metal evaporants to drive off bound gas is seldom a problem
with resistance evaporation because of the small capacity of most sources and the limited range of metals
evaporated. Electron beam heated sources however may hold 10-30 cm3 of poorly degassed evaporant
in a water cooled hearth in which it is impossible to melt the entire charge. This combination of a large
evaporant inventory and limited conditioning ability melt temperature because of the hearth cooling makes
some electron beam melts into time bombs of unconditioned evaporant.
The deposition of films of compounds finds wide application, particularly in optics. When evaporating
a compound one is, unlike with chemical elements, not assured the film is of the same composition as
the evaporant. Developing the technology of compound deposition is the subject of intense continuing
research [9]. The 'evaporation grade' compounds of the major suppliers will usually perform better than
the metals. This is because simple air melting or furnace degassing of the common optical thin film
materials is satisfactory. To prepare a laboratory grade powder or granular material for evaporation, under
a hood, with eye protection, place the material on a 100 mm square of 0.2 mm (0.01 in) Mo plate and
gently melt it with a propane torch. Allow the molten evaporant to drip off one corner of the Mo plate
onto a cooled Cu collector or into a boat or electron beam gun hearth liner. Powders that will not melt
at red heat will often evaporate satisfactorily after heating. Unfortunately, many materials (compounds of
Pb, S, Se, Ti, Zn) are too volatile to condition with a torch, and most oxides melt at temperatures beyond
that of a propane torch.
The deposition of compounds assumes the compound is stable at evaporation temperature. Many materials
of interest (CaF, CeF, MgF) are quite stable. Unfortunately, the most useful optical materials, oxides, have
widely variable compositions which produce films of similarly variable properties. Suboxide formation is
A 1.0: 12 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
/
minimized by introducing, during coating, a partial pressure of oxygen (on the order of 2 x 10- 4 Torr) into
the coating chamber (excluding oxygen if the suboxide is desired as with SiO) or oxygen ion bombardment
of the films along with slow (1-5 A s- 1 ) well controlled evaporation and a suitable elevated substrate
temperature (usually -over 200C). Exact deposition processes are among the most proprietary technology
of succ~ssful optical thin film fabricators.
As a result of the difficulties evaporating some important materials, thin film materials suppliers have
developed proprietary new materials. These materials are usually made of the 'desired' material, i.e.
ZrO, and a stabilizing 'impurity', ZrTi04 . The only drawbacks of proprietary materials are their cost and
their specialty nature, limiting the applicability of some film design software and current film deposition
technology research.
The evaporation of alloys is complex and a complete treatment [5] is beyond the scope of this work. A
first order analysis shows each component of an alloy evaporates as if it were the only material present.
This means that two materials of similar vapour pressure (i.e. Sn and Cu) will evaporate approximately
in the ratio of the starting material. Unfortunately, most alloys of importance [NiCr, Inconel (Ni 72%,
Cr 16%, Fe 8%)] are made up of metals of significantly different vapour pressures. As a result, 80%Ni-
20%Cr, will start out evaporating very Cr rich and as the Cr is consumed slowly the composition will pass
through 80:20 to nearly pure Ni. Films of the desired composition can be deposited by:
(b) Depositing from a pool, continuously replenished with the desired film alloy by wire or rod feed, of
a suitably composed 'special alloy'. The starting 'special alloy' is empirically chosen to yield the film
desired composition.
(c) Evaporating a small per cent (thin film) of the contents of a large pool of 'special alloy' and maintaining
the pool volume constant by frequent small additions of the desired film alloy.
(d) Flash evaporating continuously fed wire or granules of the desired film alloy.
These techniques all add complexity or have limitations. All will work with metals differing in temperature
at a vapour pressure of 10-4 (a better visualized way to present vapour pressure) of less than 200C. For
alloys having great differences in vapour pressure only co-evaporation provides completely independent
control of the film constituents.
A final evaporation material consideration is when to replenish and when to replace evaporant charge. An
electron beam gun can hold in excess of 20 cm3 (15 gm of Al, 50 gm of Ni) worth $50-$100. Even a
small resistance boat holds in excess of $100 worth of gold. This makes charge replacement a significant
economic consideration. As this evaporant is consumed, the material is replenished, reconditioned and
used, concentrating low volatility impurities. The principal reason to discard an evaporant charge is concern
about film composition or purity. In addition, oxide impurities from reaction with the vacuum environment
and refractory metal dissolved from the boat or hearth liner contaminate the evaporant. The evaporation
of compounds introduces decomposition as a limit on evaporant charge life. This is a particular problem
with oxides where suboxide formation (Tii0 3 and Ti02) and residual fully oxidized material (Si0 2 in
SiO) cause gradual index of refraction shifts in optical films. Operating experience also shows that old
evaporant charges, though of low purity are well degassed, often yield films with fewer defects.
(2) Where minimum film defects are the critical process parameter, replace the charge very seldom.
(3) Where film functional properties (index of refraction, etch residue, conductivity) are critical, measure
the property with reference to evaporant charge age and replace the charge with one half the age showing
property deterioration.
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:13
Thermal Evaporation
REFERENCES
[1] Dushman Sand Lafferty J 1992 Scientific Foundations of Vacuum Technique (New York: Wiley)
O'Hanlon J F 1980A A User's Guide to Vacuum Technology (New York: Wiley-Interscience)
[2] Dushman S 1992 Scientific Foundations of Vacuum Technique (New York: Wiley) pp 691-737
[3] Breitweiser G, Varadarajan B N and Wafer J 1969 Influence of film condensation and source radiation on substrate
temperature J. Vac. Sci. Technol. 7 (1) 274-77
[4] Harper J M E, Berg S, Nender C et al 1992 Enhanced sputtering of one species in the processing of multielement
thin films J. Vac. Sci. Technol. A 10 (4) 1765-71
[5] Santalia T 1970 Kinetics and thermodynamics in continuous electron-beam evaporation of binary alloys J. Vac.
Sci. Tech. 7 (6) S22-S29
[6] Graper E B 1973 Distribution and apparent source geometry of electron-beam heated evaporation sources J. Vac.
Sci. Technol. 10 (1) 100-103
[7] Behmdt K H and Doughty R W 1966 Fabrication of multilayer dielectric films J. Vac. Sci. Technol. 3 (5) 264-72
[8] Flory F 1993 Anisotropy in thin films Appl. Opt. 32 (28) 5649-59
[9] Chow R, Falabella S, Loomis G E et al 1993 Reactive evaporation of low-defect density hafnia Appl. Opt. 32
(28) 5567-73
A 1.0: 141 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Thermal Evaporation
Acustic Temperature ( q
Name Symbol Melting Bulk Impedance t1,. ........ ,.. o .......... ro ,....,,, Index of
Point Density Ratio, lo lo Electron l-_JB.e..lii..1..4-t.a.r:~.,.---l Refraction Remarks
c a/cm 1 Beam Source Material @ microns
Torr Torr
Toxic . Wets w/o alloying,
Barium Ba 725 J . 51 545 7J5 fair Boat Mo . 85 . 50 reacts with ceramics.
Barium
Oxide Bao 192 J 5 . 72 "1300 Poor Boat Pt l.98 @ .5 9 Decomposes slightly .
Barium
Sulfide Bas 1200 4 . 25 1100 Boat Mo 2.16@ . 59 Sublimes.
Beryllium
Chloride 440 l . 90 "150 Boat Ho Very toxic.
Beryllium
fluoride 800 l. 99 "480 Good Boat Mo 1.33@ . 59 Very toxic, sublimes.
Bismuth Toxic.
Selenide 710 7 . 66 ::650 Good Sputter or co-evaporate .
Bismuth Toxic .
Telluride 585 6. 82 "600 Boat Mo Sputter or co-evaporate.
Bismuth
Sulfide 685 7. 39 Toxic.
Boron
Carbide 2350 2. 52 2500 2650 Xl nt . Similar to chromium.
Boron
Cx ide 460 2 . 46 "1400 Good Boat Mo l . 46
Boron
Sulfide 310 l. 55 800
Cadmium
Arsenide 7 21 6 . 21 Toxic.
Cadmium
Bromide 567 5. 19 :,300 Boat Mo Sublimes.
Cadmium
Chloride 960 4 . 05 "400 Sublimes .
~admium
Fluoride 1100 6 . 64 "600 l . 56 @ 58
Cadmium
Iodide )8 7 5. 67 "250
Cadmium
Oxide CdO 1430 8 . 15 2.49@ .67 Disproportionates .
Inficon Z-Ratio 0 = acoustic impedance ratio, Maxtec Inc. Acoustic Impedance (A . I.): 8.83.;. z
Z-Ratio 0 Leybold Inficon
Al.0:16 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Introduction and General Discussion
Cadmium
Silicide CdSi0 2 " 600 l . 69
Calcium
Silicate CaOSio, 1540 2. 90 Good 1. 61
Calcium
Sulfide CaS subl. 2.5 1100 Box Mo 2 . 14 @ 59 Decomposes .
Calcium
Tungstate cawo, 1620 6. 06 Good Boat w l.92 @ .59
Carbon 2 . 94 .19
(Diamond) 3727 3. 52 22 2.42 .66 Deposit by CVD.
Cerium Ce 795 6 . 67 . 86 970 1380 Good Boat w 1.91 @ .59 Films oxidhe easily.
Cesium
Chloride CsCl 646 J. 99 ::500 Boat HO 1. 64 Hygroscopic.
Cesium
Fluoride CsF 682 4 .11 ::500 Boat Mo
Cesium
Hydroxide CsOH 272 3. 67 550 Boat Pt
Cesium l.99 . 23
Iodide Cs I 621 4. 51 2. 95 ::soo Boat w l.62 53
Chromium
Bromide CrBr 2 842 4 . J6 550 Boat Mo
Chromium
Carbide cr1c 2 1890 6 . 68 ::2000 fair Boat
Chromium
Chloride crc1 2 824 2 . 88 550 . Box Mo Sublimes easily.
Chromium Disproportionates to
Oxide cr 2o 1 2435 5. 21 ::2000 Good Box Mo 2. 5 @I 59 lower oxides, reoxidizes
P 600C in air .
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:17
Thermal Evaporation
Chromium-
Silicon Cr-SiO influen c ed by composition Good Bar w Flash.
v .......... "' ,...,.
Boat .02 W 1.10 . 2J Alloys with refractory
Cobalt Co 1495 8 . 92 . J4 850 1200 Xlnt . Coi.l W 5.65 2.2 metals.
Cobalt
Bromide 678 4. 91 400 Box Mo Sublimes.
Cobalt
Chloride 724 J. J7 472 Box Mo 1. 51 @ . 6J Sublimes.
Cobalt
Oxide CoO 1935 6 . 45 Sputtering preferred.
Copper (II)
Sulfide cus 1113 6 . 75 . 82 ,,500 1. 45 Sublimes.
Dysprosium
Fluoride 1360 =BOO Good Box Ta Sublimes.
Dysprosium
oxide Dy 2o 1 2340 7 . 81 "1400 Boat w Loses oxygen.
Erbium
Fluoride 1350 7. 81 =7 50 Boat Mo JVST A3(6) ,2320 .
Erbium
Oxide 2400 8. 64 =1600 Boat w Loses oxygen.
Flammable, sublimes.
Europium Eu 826 5. 26 1. 62 280 480 Fair Boat Ta Low tantalum solubility .
Europium
Fluoride 1390 6. 5 =950 Box Mo
Europium
Sulfide Eus 5. 75 Good
High Ta solubility.
Gadolinium Gd 1312 7. 89 . 67 760 1175 Xlnt. Boat Ta Flammable .
Gadolinium
oxide Gd 20 1 2310 7 . 41 Fa i r Box w 1. 8 . 55 Loses oxygen.
Gallium w
Antimonide GaSb 710 5.6 Fair Boat Ta _3 . 80 2. 2 Flash evaporate.
Gallium 3 . J4 .78
Arsenide GaAs 1238 5. 31 1. 59 Good Boat w 2 . 12 23 Flash evaporate .
Gallium Sublimes .
Nitride GaN 800 6.1 "200 ~aporate Ga in io 1N 2
Gallium
Oxide(jl) 1900 5 . 88 Boat w Loses oxygen.
Gallium Decomposes .
Phosphide GaP 1348 4. l =520 920 Boat w 3 @ 2 . 15 Vapor mostly P .
Maxtec Inc . Acoustic Impedance (A.I.)= 8.83 + Inficon Z-Ratio 0 = acoustic impedance ratio,
Z-Ratio 0 Leybold Inficon
Al.0:18 Handbook of Thin Film Process Technology @ 1995 IOP Publishing Ltd
Introduction and General Discussion
Boat . 02- 4 . 20 2. 1
Germanium Ge 9J7 5. J2 . 51 B 12 1167 Xlnt. . 04 W 4 . 00 20
,.._.,,.., " ' Q c
Germctnium Sublimes.
Nitride 450 5. 2 5 :,550 Sputtering preferred.
German1um
Telluride Ge Te 725 6. 20 381 Box Ho
Hafnium
Boride HfB 2 J250 10. 5
Hafnium
Carbide HfC 3890 12. 20 "2600 Sublimes.
Hafnium
Nitride HfN JJ05 lJ. B
Hafnium
Silicide HfSi 2 1680 8. 02
Holnium
fluoride 114J 7. 64 =:900
Indium (I)
Sulfide 65J 5. 87 650
Iron
Chloride reel, 674 J . 16 JOO 1.57 @ .59 Sublimes.
Iron
Iodide 592 5. Jl 400
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:19
Thermal Evaporation
r ron (I I I) Disproportionates to
J xide 15)8 5 . 18 Good Boat 3. 0 fe 1o, at 1530c.
!ran
Sul fide feS 1195 4.74 Decomposes.
Bo3t . 02 W
Kantha 1 feCrAl 1500 7.1 =1150 Coil W 1.74@ . 58 JVST 7, 7)9 (1980) .
Lanthanium
Ber ide 2210 2. 61 Good Toxic.
Lead
Bromide PbB,- 2 )7 3 6 . 68 :::300 Boat Ho Toxic.
Lead
1 ~ I Toxic.
_Io_d_i_d_e---+--Pb_r_,_ _-+-_50_2__,.__6~----+----+-"-J_2_0--+-----+-c-ru_c_i_b1_e+-Q-u~ar~t-z---1--"2_._1_ _+--J--~o~p-t_.~s-o~c-._6_5_,_9_14~....,....--
Lead Pt Toxic . No decomposition.
oxide PbO 886 9.53 :::sso Boat Ho 2.51@ .59 J.Opt.Soc.Am. 52,161
/10..C:""I\
Lead Pt Toxic.
Stannate 1115 8.1 670 905 Poor Boat Ho Disproportionates.
Lead
Selenide PbSe 1065 8. 10 =500 Box Ho :::3. 5 @ 1. O Toxic, sublimes.
Lithium
Oxide 1427 2. 01 850 Boat w l. 64 @ 59
Lutetium
Oxide 2487 9. 41 1400 Boat w Decomposes.
"1agnes ium
Bromide 700 ) . 72 :::450 Boat Ho Decomposes.
Al.0:20 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Introduction and General Discussion
Basket
Manganese Mn 12U 7. )9 . 4) 508 6-18 Good Boat Ho 2.59 ~ .59 Flammable, sublimes.
Manganese
Bromide 695 4. 38 500 Boat Ho
Manganese
Chloride HnC1 1 650 2 . 98 450 Boat Ho
Manganese
Oxide 1705 4 . 86 Boat l. 7 3
Manganese
Sulfide HnS 1615 ) . 58 . 94 l )00 Boat Mo 2. 7 Decompo~es.
Mercury
Sulfide HgS 583 8. 10 250 Toxic, decomposes.
Molybdenum
Bo ride Mo 1 51 2200 7. 48 Poor
Molybdenum
Silicide MoSi 1 2050 6. )1 Boat Mo 1. 9 Slight o 1 loss.
Molybdenum
Sul fide MoSa 1185 4 . 80 ,.50 Boat w Decomposes.
Molybdenum
Oxide Moo 3 795 4 . 69 "'900
. 89 . 39 Flammable .
Neodynium Nd 1024 7 . 00 . 84 7)1 1062 Xlnt. Boat Ta . 30 .88 Low Ta solubility.
Nickel
Bromide 96) 4 . 64 362 Boat Mo Sublimes .
Nickel
Chloride NiC1 1 1001 3. 55 4 44 Boat Mo Sublimes.
Nickel
Oxide NiO 1990 6. 69 "'1480 2.18 @ . 48 Dissociates upon heating.
Niobium
{Columbium) Nb 2468 8. 57 . 47 1728 2287 Xlnt . Coil w 1. 80 ~ . 58 Attacks W source .
Niobium
Boride NbB 1 3050 6 . 97
Niobium
Carbide NbC ) 500 7. 82 Fair
Niobium
Oxide NbO 7. )0 1100 Boat w
@ 1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:21
Thermal Evaporation
Acust i c Temperature ( C)
Name Symbol Mel ting Bulk Impedance 1 ' "'_. .... _ n . ..... ........... - ..... Index of
Point Density Ratio , io io ' Electron Refraction Remarks
C a / cm 1 Torr Torr Beam Source Material @ microns
Niobium
oxide (V) 152 0 4. 4 7 Boat .02 W 2. 3
Niobium
Te lluride NbTe 5 7. 6 Composition variable.
Ni obium- Co-evaporate
Ti n Nb 1Sn Xlnt . from 2
Palladium
oxide PdO 870 8. 70 S7S Decomposes .
1. l
Platinum
o xide 4 50 10 . 2
Potassium
Fluoride KF 846 2 . 48 :::soo Poor Boat Mo l.JS @ 1.4 Melt in air to outgas .
Po tassium
Hydroxide KOH 360 2. 04 '=40C Boat Mo Melt in a~r to outgas .
Praseodymiun
Chloride Prcl 1 786 4. 02 ::: so c Boat Mo l. 86
Rhenium
Oxide 297 6.10 :::100 Boat Mo
Rhodium Rh 1966 12. 41 . 24 1272 1707 Good Coil w 1.62 . 55 EB gun preferred.
Rubidium
Chloride RbCl 71 S 2 . 80 '=SSO Boat MO 1. 49
Rubidium
Iodide RbI 64 l 3 . 59 "'40C Boat MO l . 68 @ S8
Spits violently in EB .
Ruthenium Ru 2SCO 12 . 4S . 2C 178C 2260 Poor Boat w Requires long degas.
Maxtec Inc . Acoustic Impedance (A . I.)= 8 . 83 + z Inf icon Z-Ratio 0 = acoustic impedance ratio,
Z-Ratio0 Leybold Inf icon
Al.0:22 Handbook of Thin Film Process Technology @ 1995 IOP Publishing Ltd
Introduction and General Discussion
Samarium L<:>ses 0 2
Oxide 2 3 50 8. 3 5 Good Boat Ir 1. 97 Films smooth, clear .
Scandium
Fluoride Scf 1 1550 2. 50 "1400 Good Boat . 02 w
Scandium
oxide sc,o 1 2300 ) . 86 Fair Boat Ta l. 88 @ 55 Loses o 2
Silicon 1610 2. 20 l. 47 . JO
Dioxide Si0 2 -1710 -2. 70 l. 00 "1025 Xlnt. 1. 45 2. 8 Quartz xlnt. in EB.
Silicon
Selenide Sise 550 Boat Mo Toxic.
Sil icon
Sul fide Sis subl . l. 85 450 Boat Mo
Silicon
Telluride SiTe 2 4. )9 550 Boat Mo Toxic.
Silver
Bromide AgBr 431 6. 49 l.18 Boat Mo 2.28@ .58
Silver
Iodide Ag I 558 6 . 01 Boat Ta 2.02 I' .59
Sodium
Hydroxide Na OH 318 2 .13 =470 Boat Pt l. 36 Melt in air to outgas.
nol ......... ,. ... ,.. .....
@ 1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:23
Thermal Evaporation
Acustic Temperature ( C)
Name Symbol Melting Bulk Impedance u,,,r-. ......... D1 . ..,,.~~ ., .... ..,,. ... " nn Index of
Po i nt Density Ratio, lo lo ' Electron 1---=~c.l..<IJWOS..,._--j Refraction Remarks
c o / cm 3 z Torr Torr Beam Source Material @ microns
Strontium Sublimes .
Oxide sro 2461 4.9 1500 Boat Ta 1.88 @ 58 Reacts with Mo and w.
Strontium
Sulfide SrS >2000 3.7 0 2 . 11 @ 59 Decomposes.
Toxic .
Sulfur 115 2 . 07 2. 29 1J 57 Po or Boat Mo Poisons va c uum system .
Sputtering preferred ;
Supermalloy Ni / Fe/ Mo 1410 8. 9 Good or co-evap . from 2
D~rm;11ollnv ;11.n~ Mo
Tantalum
Bo ride 3000 11. 15
Tantalum
Carbide Tac 3880 14. 65 " 2 500 JVST 12 , 811 ( 1975) .
Tellurium Te 452 6. 2 5 . SJ 157 2 77 Poor Boat Mo 4.9 @ 6.0 Toxic . Wets w/ o alloying.
Terbium
Fl uoride 1176 :::000 Boat Mo
Terbium
Oxide 2)87 7. 87 lJOO Boat Ta Partially decomposes.
Terbium
Oxide Tb 4 0 7 Boat Ta Films TbO.
Thallium Tl JOl 11. 89 1. 58 280 480 Poor Boat Mo Wets freely, very toxic .
Thallium 2 . 65 . 44
Bromide Tl Br 480 7. 56 1. 77 "200 Boat Mo 2.J2 24 Toxic, sublimes.
Thalium 2 . 20 .75
Chloride Tl Cl 430 7. 00 l. 21 ::: 150 Boat Mo 2. 6 12 Toxic, sublimes.
Thal 1 ium
Iodide (B) TlI 440 7. 09 "200 Boat Mo 2.78 @ .75 Toxic, sublimes.
Thorium
Bromide ThBr 4 5. 67 Boat Mo 2 . 47 5 Radioactive, sublimes.
Thorium
carbide ThC 2 277J 8 . 96 " 2J OO Radioactive.
Thorium 1. 8 55
Ox ide Tho 2 J050 9. 86 :: 2100 Good Boat . 02 w 1. 75 2.O Radioactive.
Thulium
Oxide Tm20l 8 . 90 1500 Boat Ir Decomposes.
f\.1.0:24 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Introduction and General Discussion
Tin
Selenide Sn Se 861 6. 18 "400 Good Boat Mo JVST 12, 110 (l'i75).
Tin
Sulfide sns 882 5. 22 "4 50 Boat Mo
Tin
Telluride SnTe 780 6. 44 "450 Boat Mo
Titanium
carbide TiC Jl40 4. 9) "2)00 JVST 12, 851 (1975).
2. 55 . )8
1840 4 . 26 . 40 "1)00 Fair Boat . 02 w 2 . JO l . 0
Tungsten
Boride 2900 12. 75 Poor
Tungsten
carbide w,c 267 15. 6 .18 1480 2120 Xlnt.
Tungsten
Silicide wsi 2 2165 9. 4
Tungsten
Sulfide ws, 1250 7 . 51
Tungsten
Telluride WTe 1 9 . 49
Uranium u 11)2 19. 07 . 24 1132 1582 Good Boat .02 w Films oxidize.
Uranium
carbide 2260 11 . 28 2100 Decornposes
Uranium
Fluoride :::1000 )00 Boat Mo
Uranium (IV)
oxide U0 2 2500 10 . 9 Boat w Ta causes decomposition.
Uranium
Phosphide a . 57 1200 Boat Ta Decomposes .
uranium
Sulfide u,s, 1400 Boat w Slight decomposition .
Wets Ho . EB evaporated
Vanadium v 1890 6.11 SJ 1890 1547 Xlnt. Boat .02 w J.OJ @ .58 films preferred.
Vanadium
Boride 2400 5.10
Vanadium
Carbide vc 2810 5. 77 "1800
Vanadium
Nitride 2320 6. lJ
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.0:25
Thermal Evaporation
vanadium (V)
Oxide Vz05 690 ). 36 =500
Vanadium
Silicide VSi 2 1700 4 . 42
Ytterbium l . 48 2. 2
Fluoride YbF 3 1161 8 . 19 =780 Boat Mo l . 32 14
Ytterbium
Oxide Yb2o 3 2227 9.17 =1500 Sublimes. Loses oxygen.
Yttrium
Aluminum Y3Al5012 1990 Good Films not ferroelectric.
nv ~ .....
Yttrium l . 46 2. 5
Fluoride YF 3 1152 4. 01 1.42 10
Zinc
Bromide ZnBr 2 391 4. 99 =300 Boat Mo l . 58 . 58 Decomposes~
Zinc
Fluoride ZnF 2 872 4 .95 =BOO Boat Ta
Zinc
Nitride zn3N2 6.22 Boat Mo Decomposes.
Zirconium
Bromide ZrB2 3000 6. 09 Good
Zirconium
carbide zrc 3540 6. 73 =2500
Zirconium
Silicide zrsi 2 1790 4 .88
Al.0:26 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Resistance Evaporation
Al.U) INTRODUCTION
Deposition of thin films from resistance heated evaporation sources was the first practical thin film
deposition technique [1]. Since then, despite being technologically eclipsed by electron beam heated
evaporation sources and sputtering, resistance evaporation finds wide use, offering a unique combination
of reliability and economy.
During the 1930s the art of lens blooming became the science of optical coating [2, 3]. By the late 1930s
multilayer optical coatings, all deposited with resistance heated sources, were invented [4]. The multilayer
resistance source deposited films used in most production dichroic lighting reflectors [5] are an example
of cost efficient, modern production technology.
The resistance heated evaporation system (figure A 1.1.1) consists of:
Lower shield I
reflector
To
vacuum
pumps
Transformer
Figure Al.1.1. Vacuum system for deposition from resistance heated sources.
Al.1.1 SOURCES
The resistance heated refractory metal source is the heart of the evaporation system. Selection of a suitable
.source is the critical decision in designing a resistance evaporation process. The source is the evaporant
heater and container and the most probable cause of failure. Tabulated data in appendix 1 [6] is the best
starting place in selecting a optimal source. The catalogues of source manufacturers contain a plethora
of designs and little specific performance or application data. A source is selected to meet the following
@ 1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.1:1
./1
Thermal Evaporation
process requirements: evaporant compatibility, capacity and power availability. The most difficult of
these to meet is evaporant compatibility because many important evaporants (Al, Au, Fe, Inconel and Pt)
dissolve all refractory metals to some extent.
In general, sources are made ofrefractory metal (Mo, Ta, W) of three design classes (figure Al.1.2): coils,
boats and special purpose designs.
(a) Coils support the evaporant by melting into a coil of stranded tungsten wire. Coils are isotropic
emitters of low capacity and low cost. They are commonly used for Al and less commonly for Au,
Ag, Cu and Sn. As isotropic emitters they are particularly useful in the large drum coaters used for
decorative metalizing. The central position of numerous coils in these coaters permits use of both the
upward and downward emission of evaporant from coils. Coils are often inefficient in typical 'coat up'
belljar applications because the vast majority of evaporant is lost. The evaporation of Al from coils has
been studied in detail [7] due to its extensive decorative application. Efficient coil operation requires the
Al charge be less than 10% of the coil weight. To minimize dripping of the melting aluminum, the coil
temperature must be increased immediately to the evaporation temperature of 1200-1500C. Multiturn
coils of heavy multistrand wire have the highest capacity. Such coils also require much more current and
power to operate, often exceeding the capabilities of small power supplies. The coil class of sources also
includes rod sources. This relatively modern design (figure Al.1.2) separates the heating (the central rod)
from the evaporant supporting (wire overwind) functions. In so doing, the source capacity and life are
increased significantly. Unfortunately, the required current and power are also increased. Where sufficient
power is available, a rod source is usually preferred over coils, particularly for difficult metals such as
Al, Ni and Inconel. Conical coils are also available to contain small droplets of evaporant or pieces of
subliming material.
(b) After the coil, boats are the classic resistance evaporation sources. These generally small refractory
stampings (figure Al.1.3) hold the evaporant in dimples or folds and are heated by the passage of electric
current. Boats are available in an endless range of sizes and shapes. Most boat designs differ little and
have come into the manufacturers' catalogues at the request of an influential customer. The folded box
boats are the most useful because of their high capacity and flexibility. When a tungsten boat is necessary,
those with an oval dimple have a higher capacity while those with round dimples are less effected by
shorting through the evaporant (see section Al.1.4).
Because the boat, like a coil, is the source of the heat for evaporation, the boat must operate at a temperature
considerably above that required to evaporate the material it contains. This temperature is the principal
limitation of the capability of boats. Because the boat is so hot, the mutual solubilities (reactions) of
the evaporant and boat are enhanced. The alloying inevitably thins and embrittles the boat leading to
cracking. In some cases (Pt) the melting point of the boat/evaporant alloy is so much lower than that of
the pure refractory that the boat evaporant alloy melts long before any evaporation occurs. Alloying can
be minimized by the following three techniques.
(1) Selecting the most suitable material for the boat. Field experience is the best guide to an optimum
refractory metal. Specific recommendations are contained in appendix 1 [6] and in the recipes section of
this handbook. In depth alloying information can be found in Constitution of Binary Alloys [8].
(2) Coating the boat with an inert surface, usually A}i0 3 . The protective oxide coating on a boat can be a
source of evaporant contamination and functions as a thermal barrier. As a result, these coatings require
significantly higher boat temperatures, accelerating cracking. This limits the usefulness of coatings for
high evaporation temperature materials where th~y are most needed.
(3) Making the boat massive enough, compared to the amount of evaporant, that saturating the evaporant
with refractory does not destroy the boat. This can be done by using very thick 0.5-1 mm boats. These
are, unfortunately, difficult to fabricate and expensive. An approximation of a very thick boat can be
made by mixing pieces of additional refractory metal (usually W) with an equal volume of evaporant. A
convenient source of refractory metal is used broken boats. This system will double the life of boats used
for evaporating Inconel and Ge.
Boats, because of the evaporant cavity formed into them, are usually very rigid. As a result, the effects of
thermal expansion induced distortion are very great. A 100 mm long W boat heated to 1500C will grow
A1.1 :2 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Resistance Evaporation
(a)
----~----\f\{W\(\(
(b)
( c)
c
J
( d)
( e)
( f)
Figure Al.1.2. Resistance heated evaporation sources: (a) coil, (b) dimple
boat, (c) canoe boat, (d) aluminium rod source, (e) baffled furnace and (f)
1 mm thick Ge source.
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.1:3
Thermal Evaporation
(a)
(b)
0.5 mm, buckling sufficiently in just a few cycles to be useless. This buckling can be minimized by using
flexible busbars as described in section Al.1.2, Cables and Busbars. Folded box boats (figure Al.1.3) do
not require flexible busbars because of their intrinsic flexibility. A number of materials (i.e. Cu) wet and
flow toward the cold contacts of boats. This problem can be controlled by designing boats with hot bands
in the contact legs. Folded box boats intrinsically have hot ends. In dimple boats, localized heating is
generated by notches or holes in the legs, reducing their cross section. These designs are useful but more
fragile than boats without notches.
(c) Special purpose sources. Resistance heated sources designed for specific materials are common. Rod
sources for aluminium were discussed under coils. For Cr, tungsten rods, plated with 0.5 mm of high purity
Cr work very well. They require less heat than subliming Cr from a conical basket or boat because the
metal is in intimate contact with the heated rod. Baffled furnace sources are commonly used for subliming
dielectrics such as SiO that frequently spit. These sources are often well heat shielded, reducing the power
required and providing a uniform sublimation temperature. The internal baffles filter out the particles
evolved during sublimation. Crucible heaters hold and heat inert crucibles of evaporant. Crucibles of
Pyrex, quartz, carbon and BN of up to 1 L capacity are common and can be heated by an open coil or
heat shielded refractory metal. A wide range of low to moderate temperature metals (Pd, Sn, Se, As, In)
evaporate well from crucibles (as well as from boats). Many users find crucibles to be less prone to failure
and more stable than boats. A principal advantage of crucibles is the complete isolation of the evaporant
from the heater, eliminating shorting and alloying. When selecting baffled furnaces, crucible heaters and
other expensive special purpose assembled sources, the most important criteria, after compatibility .with..
the evaporant material, is flexibility. These sources are costly and should operate for 50-100 or more
cycles before distortion and cracking force replacement. Source and busbar flexibility and the ability of
the source to tolerate distortion will determine its life.
Roll to roll Al coated plastics find tonnage applications as packaging materials and capacitors. Bar sources
of TiBi-BN intermetallic permit the evaporation of Al wire at rates up to several gm/min onto plastic web
as fast as 200 m/min. Large scale specialized coating plants [9] coat 1000 kg rolls of plastic at $0.0l/m2
These durable TiB 2-BN intermetallic bar sources are also suitable for the deposition of a wide range of
other materials. The requirement to wire feed the source material and the requirement for up to 5 kW/Bar
of drive power means these sources are less frequently used than they perhaps should be.
A 1.1 :4 Handbook of Thin Film Process Technology @ 1995 IOP Publishing Ltd
Resistance Evaporation
The placement of the feedthroughs and the busbar design (figure Al.1.4) have a very significant impact on
the success of resistance evaporation and on source life. The busbars must make reliable, low resistance,
electrical contact from the feedthrough to the source, conduct high current and be sufficiently flexible to
accommodate up to 0.5 mm thermal expansion of the source. Failing to accommodate source thermal
expansion will cause the source to buckle and fail prematurely. Figure A 1.1.4 illustrates a well designed
system layout. The feedthrough and source clamps on the busbars are preferably split type rather than screw
locking. The parallel busbars allow source thermal expansion to deflect them in the weakest direction.
Opposed busbars should be avoided because they cannot give and the thermal expansion of the busbars
adds to the source expansion. Where possible the thermal expansion problem can be solved by putting
a 'hinge' (figure Al.1.4) of 20-40 layers of 0.005 in (0.1 mm) copper foil in one busbar. This will not
cause undue power loss and may double the life of the source.
Source
Thermal
expansion
di rec ti on
Hinge detail
Figure Al.1.4. Parallel busbar positions for resistance evaporation, showing split clamps and
hinge detail.
High current feedthroughs carry the current into the vacuum system and are usually of standard commercial
manufacture. The resistance heated source is coupled to the feedthroughs and supported by them.
Applications requiring over 500 A must use water cooled feedthroughs and usually two electric cables in
parallel. This can make the installation cumbersome, requiring careful planning. There is a temptation to
use only one feedthrough and allow the chamber to carry the other secondary leg. This should be avoided
because it can lead to ground loops and large stray currents. The current requirements of a large resistance
heated source can exceed 1000 A. This is outside the normal range of experience. In most equipment
the cables are only about half the size required leading to large power losses and increased risk of fire.
Recommended cable sizes are as follows:
Table Al.1.1.
Wire Nominal conductor
diameter
USA (AWG) Europe (CSA)
250 A 1 25 mm2 7 mm
500 A 410 or 250MCM 2 x 35 or 100 15 mm
1000 A 2 x 3/0 or 500MCM 2 x 100 25 mm
These cables can be purchased through welding supply dealers. The preferred cable terminating lugs are
cast or forged copper alloy with 2 clamping bolts. Experience has shown that solder lugs and crimp lugs
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.1:5
Thermal Evaporation
are not reliable at high current. The holes in the lug tang must often be enlarged to fit standard 1 inch
(2.5 cm) feedthroughs and bent to permit installation.
The source temperature and thereby the evaporation rate is controlled by the power supply output (0-220
V AC), through a stepdown transformer. The power supply may be electronic SCR or a variable voltage
transformer. SCR or motor driven transformers are normally controlled by the 0-9 V DC output of a film
thickness and rate monitor. The response time and stability requirements of the power supply are low.
Large resistance heated sources have long time constants requiring very slow power control response to
avoid over controlling. Given the choice, SCR power controls are best operated in the constant power or
constant voltage mode. This minimizes the chance of rapid source overheating during a source failure.
Resistance heated sources require 1-20 V and up to 5 kW of power. The stepdown transformer matches
this load to power supply allowing efficient use of its output. These transformers have multiple independent
windings (for example: 2 at 2.5 V, 500 A and 2 at 10 V, 125 A) that can be connected in series or parallel
to provide a range of output voltages (2.5, 5, 10, 15 and 20) with very high (1000 A at 2x2.5 V) currents
available at the lowest voltages. To efficiently use the power supply the secondary voltage is set with
jumper busbars on the transformer to match the requirements of the source being used. An extensive table
of source power requirements is published by R D Mathis Co. [ 1O]. Proper matching of the power supply
to the load resistance is critical. The transformer secondary voltage is adjusted, using the transformer
jumper busbars to 'balance' the output. (This is done by measuring the power supply primary voltage
(0-220 V AC) and current (at 2.5 kVA, 0-12 A under load.) The power supply output amperage and
voltage when properly balanced should be approximately the same percent of maximum. For example, a
large tray boat filled with a nonconducting evaporant requiring 1500 Watts draws the following:
Table Al.1.2.
Transformer Watts Primary Primary Secondary Secondary
winding voltage(%) amperage(%) voltage amperage
2.5 v 765 220 (100%) 3.5 (35%) 2.5 305
5V 1500 160 (72%) 9.3 (77%) 3.5 430
lOV 1500 80 (36%) 18.75 (155%) 3.5 430
Maximum at 2.5 kW 220 (100%) 12.0 (100%)
In the foregoing example, using the 5 V tap results in a close balance of the primary voltage required,
72% and the resultant primary current 77% indicating efficient leading of the transformer. If one were to
attempt to run this boat on the 2.5 V tap the output power at the maximum available 220 V (maximum
primary voltage) would only be 765 W. This is insufficient for the boat. Similarly, using the 10 V tap at
1500 W output yields an acceptable 80 V primary voltage but requires about 18 A of current. This is 150%
of the maximum rated current for the 2.5 kW transformer and power supply. This example shows that
for the source power to approach the maximum available the output tap voltages of the power transformer
must match those required quite closely. As a result of the likelihood of an imperfect match between the
source requirements and the available output voltages, it is unwise to anticipate loading a resistance power
supply to more than 75% of its rated output.
Many of the materials evaporated from resistance heated sources are electric conductors. If the source
selected is a refractory metal coil or boat the high conductivity metal evaporant will partially short the
source. This shorting may double the current required to heat large boats and cause superheating of the
contacts. Shorting can be controlled by filling boats less than half (sometimes less than 1/4) full, selecting
a boat that is refractory coated or indirectly heating the evaporant in an insulating crucible.
~ A 1.1 :6 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Resistance Evaporation
The installation of the source, that is the relative arrangement of the busbars, source heat shields, evaporant
shields and substrates, is critical to efficient thin film production. A typical source installation (figure
Al.1.1) shows the arrangement of the key parts. Most of this installation (source and substrate positions,
evaporant shields, shutter and monitor) is typical of many evaporation systems. The busbar arrangement,
detailed in figure Al.1.4. minimizes thermal distortion of the source. A small heat and evaporant shield
has been added near the source. This shield can sharply reduce chamber heating and the required source
power. Each resistance heated evaporation source has a unique evaporant distribution (figure Al.0.4).
There are often large distribution variations from run to run due to differences in wetting and fill of the
source. As a result the thickness uniformity on a 25 cm flat plate 40 cm from the source may vary by
20-30%. By using a planetary substrate holder or similar well designed substrate motion system this
error can easily be reduced to less than 5%. The run to run variability of the distribution will, however,
introduce significant run to run thickness errors. These thickness errors occur because the measuring head
is at a fixed location, not benefiting from the averaging of the substrate motion system.
REFERENCES
[l] Phol R and Pringsheim P 1912 Verh. DTSCH. Phys. Ges. 14 546
[2] Bauer G 1934 Ann. Phys., Lpz. 19 434
[3] Strong J 1936 J. Opt. Soc. Am. 26 73
[4] Cartwright C and Truner A 1939 Multilayer films of high reflecting power Phys. Rev. 55 1128A
[5] Eastman Kodak Co., private communication
[6] Lebow Databook 1990, Lebow Company, Goleta, CA
[7] 1975 Research/Development Magazine 26 (11) 49-54 (November); (12) 45-49 (December); 1976 27 (2) 65-71
(February)
[8] Hansen M and Anderko K 1958 Constitution of Binary Alloys 2nd edn (New York: McGraw-Hill)
[9] Baxter I 1994 New technology for high rate thick film deposition Soc. Vac. Coaters, 37th Proc. pp 218-26
[10] RD Mathis Co, 2840 Gundry Ave., Long Beach, CA 90806
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.1:7
-,
l l
Electron Beam Evaporation
l.~
Al.2.0 INTRODUCTION
The electron beam heated evaporation source is a thermal evaporator as is a resistance heated source.
Electron beam heated sources differ from resistance heated sources in two ways: the heating energy is
su12plied.to the top of the evaporant by the kinetic-en-ergy of a high current electro1r beam,-amFlhe-evaporant
is contained in a water cooled cavity or hearth. Heating by elect~on beam allows attainment ofJi .mperatures
limited only by radiation and conduction to the hearth. Evaporants contained in a water cooled hearth do
:-- - - - - -
not significantly react with the hearth, thus providing a nearly universal evaporan container.
. . -. ' " . '
The bent beam evaporation source as is now used in thin film fabrication for electronics, optics and
research was first envisioned by L Holland [1] and the modern 270 un was develo ed in the earl 1960s
by Hugh Smith and Charles Hanks of Temescal Corporation [2]. In the late 1960s and 1970s electron
beam guns were principally used for the 4P-O..SiJi9n of ~hi.rn.in,l!.m met.a.Uization on semiconductor devices.
Today these sources have been largely supplanted for semiconductors because of the switch to alloy
metallization coupled with the development of high rate, easily automated planar magnetron sputtering
sources. Magnetron sources are much better suited for alloy deposition and more stable in automated
production applications. The use of lift-off metallization technology for gallium arsenide and other high
performance devices is causing a resurgence of electron beam deposited semiconductor metallization. The
development of the new generation of hard multilayer optical coatings has made electron beam evaporation
the technology of choice in optics.
The electron beam heated evaporation source, as used for high technology thin film fabrication (figure
Al.2.1) has 3 basic sections: the electron gun, the beam deflection magnetic lens, and the evaporant-
containing hearth. The beam is formed in the gun, passes through the magnetic lens and is focused upon
the evaporant. The specifics of the design of an electron beam source will be treated in section Al .2.3.
Application of an electron beam heated source is governed by three complex relationships.
(1) The energy balance of ~he ~~~mq the rem1jr_~m~nt for st~ble dissi ation of the beam
energy.
-
(2) The complex distribution of the evaporant vapour flux from the evaporant surface caused by the pressure
within this vapor and the resultant evaporant surface geometry.
(3) The ionizing effect of the electron beam, as it passes through the evaporant vapour cloud, impacts the
melt surface and is partially reflected from that surface.
For useful evaporation to occur, the evaporant surface must reach a temperature such that the surface
vapour pressure is greater than 10- 1 Torr. At very high evaporation rates, this pressure may reach 10
Torr. The electron beam is conventionally accelerated to 10 kV at a current of up to 1.5 A. This beam
impacts an area of 1/4-1 cm 2 with an energy of up to 60 kW/cm 2 . Stable evaporation requires that a
thermal equilibrium exist in the evaporant and the energy dissipation must be stable. Energy is dissipated
from the evaporant principally by conduction to the hearth, then by evaporant phase change heat and
least by thermal radiation. A portion of the beam, often containing considerable energy, is also reflected
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.2:1
Thermal Evaporation
Water coot in g
Electron beam
Magnetic tens
poles
Magnetic tens
pole
Electron beam
Magnet
from the evaporant surface [3] : A 10 kV electron beam upon impact with a surface gives up its energy
essentially at the surface (within a small fraction of a millimetre). For an ideal evaporant (Ag, Al, Au,
Cu), as the temperature is increased, evaporation will begin at a vapour pressure of about 10-4 Torr,
increasing in rate with increasing energy input until the evaporant pressure over the beam impact point
reaches the viscous fl.ow range (about 10- 1 Torr), and the mean free path is reduced to a fraction of a
millimetre. At this point added beam energy is principally absorbed by the ev~porant cloud, limiting the
further increase in evaporation rate. Because the temperature change in degrees K is relatively small from
the point of first evaporaiio'~ to instability, (for aluminum 1245 Kat 10-4 Torr to 1640 Kat 10- 1 Torr),
heat losses 'by. conduction. tatfie' ilearih and radiation do no mcrease rapidly. As a result, the evaporation
rate (figure Al.2.2) increases exponentially. with; i.nc:r~asing , pow~e,1 .u,ntil Ii.~ited__ by the vapour de.n.~ity over
t e evaporant surf ace.' The evaporation rate can be further increased at constant O! J ncreased power only
by increasing .the elt?.~~!:2JLQ.~~.mjrnR~!.~!;l!.!?,,~?~. !~~~?,X reducing the power density.
~ ".e !" ~-r-..~ ~... ~
Understanding the evaporant distribution of electron beam heated s~~;~~s i; necessary for the production of
films of uniform thickness. The flux of evapgHn.t b~~J:?~~r:i, qescr~ped. ~~ secti<;m ~) .0.4, equation (Al.0._14 ).
~ ;itt!><I .A ::i,.i;t. ~~ v ~ ,, , ...... '.~(. 4'1 . , ~~_,. . ._, ,.,,.._.";, .., ";;jo;
1b~.!9Je ..dependent.coefficients are ,given in section A~_ .0.4, figure Al.0.5. c ~ ~ , c,.,,
Before evaporation begins, an operating electron beam source emits a small current of energetic electrons
toward the substrates most likely reflected from the evaporant surface. In addition the electron beam source
emits a modest flux of soft X-rays. These emissions from the source, although-smaff,~'can aamage' sensiti've
-ae~-{~~s:13~~~~~ ~ sfmiia~-"o'r".ffic>"ie' l'ntense ion and electron bombardment emanates from sputtering sources
the most radiation sensitive devices must be coated using resistance heated sources. As the electron beam
passes through the evaporant cloud, impacts the evaporant and is absorbed or reflected, several processes
occur. Meas_!!.r:e.m.e.nts . of ~he ion . CJHl:Y..111.J!~.."45 cm above an operating electron beam ev~pqraiion source
~-- ~-...... - . ' .. ,, . . . . . . , .. , . ,~ .,.. "" ~ k . :, ,y .,,..... ,,,~.;.... ~ , ~
[4] show that the ,e~mpgi;ant- is- partially ion'ized by "the-electron beani. For aluminum the ionization is only
1%, but for silver and copper it may be g:iu(ih Q~gher. The path :of these ions in the evaporant during their
first 1 to 2 cm of travel is influenced by the magnetic lens and the-1.9 _kV beam accel~r~~i_ng.._.f, jc;ld of the
e~ap~~~!~~~~,~-~~~.~ :.v_,so_me- ions fol~ow curved trajectQr~ impacting th-e lilgn~-~gi- f~~g~s and
wires, s uttenn matenal and causm arc~. The material sputtered from t 'e feedtnroughs is :igmficant
A l.2:2 Handbook of Thin Film Process Technology @ 1995 IOP Publishing Ltd
/'
t
Electron Beam Evaporation
so--~~~~~~~~~--
0.8 cm 2 beam
x103
40
I
I/)
o<t
30
cu
+-
c
t...
c
0
~ 20
I/)
0
a.
cu
D
10
0 2 4 6 8 10 12 14
Beam power {kW)
Figure Al.2.2. The aluminum deposition rate (at 25 cm) for
a focused 0.03 cm2 and a diffuse 0.8 cm2 beam.
enough to destroy them over a period otye~ -~- ~------_lightly c.Qntaminate the .films.-- Th~~~ feedthroughs
and wires must be protected by ~unded metal shield. ----- ---..~- -- --
- -----------------..._.,, __,,__..~.......--.~---------'"----._.-~
The electron beam heated source is made,,-Up-ef~ po.wer -supply and evaporation .~ _ource. Great care must
be taken to assure that their design is sufficiently durable to survive continual arcing-and attack by hot
molten metal.
The component parts of electron beam heated evaporation sources have, for many years, been made by
a number of small suppliers. As a result, the gun/power supply maximum operating specifications have
been informally standardized as follows:
High voltage: 5 to 10 kV DC
Electron beam current: 0 to 1.5 A DC
Filament current: 0 to 40 A AC
Filament voltage: 0 to 6 V AC
Position magnet voltage: -2 to +2 V DC
Position magnet current: 0 to 3 A DC
Emission control voltage: 0 to -10 V DC
The electron beam heated source power supply consists of three interconnected but independent power
supplies, high voltage, filament and magnet, as shown in figure Al.2.3. Two designs .. areuse-a
to protect
the -higl'CvoTtage supply from arc damage: a regulating pass tube in the ground leg or, less commonly, a
switch on the power source or in the ground leg. Most large power supplies are pass tube regulated, using
a high dissipation triode. On sensing an arc as an increase in current or a voltage drop across the tube,
these pass tubes are shut off sufficiently to limit the current and, if necessary, reduce it to extinguish the
arc. These very expensive tubes are durable and capable of actively regulating the high voltage output.
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.2:3
Thermal Evaporation
Switching protected supplies are compact and relatively simple. In all power supplies the high voltage
is generated in a vacuum potted 3 phase transforriier havin.g...a--1~12 kV output and rectified with a full
wave bridge using low cost 20 kV avalanche diodes.
(a)
Series regulated
(b)
Switching protected
Figure Al.2.3. Electron beam source (a) series regulated and (b) switching protected power
supplies.
The filament and magnet power supplies are independent of the high voltage power supply and are
unusual only in their exceptional immunity to electromagnetic noise. The SCR filament supply is normally
controlled by the 0 to 9 V power control voltage with the output to the gun coupled through a transformer
with 20 kV of isolation. There are two feedback paths from the filament supply to the pass tube/switch,
one for beam current regulation (if used) and a second to 'fold back' the filament power during arcs to
assist arc recovery. The magnet power supply is of basic current regulated design with adjustable current
limits, putting out 2 amps and crossing through zero smoothly. This supply is often interlocked with the
filament to minimize the chance of the beam being directed outside the hearth.
The high voltage used in electron beam sources and power supplies is exceptionally dangerous. All
components must be located with grounded safety shields and a thorough safety interlock system on the
cabinet, feedthroughs and vacuum chamber must be used. In addition, the water cooling is critical to
operation of both the power supply and sources, re uirin . water flow metering interlocks and a low fl.,.ow
alai: . Operating an electron beam gun for less than one minute at only 5 kW output without water cooling
w1 1 usually destroy it. . - - ~
Electron beam heated evaporation sources are an integrated assembly (figure Al.2.2) of electron gun,
magnetic lens and water cooled, evaporant containing hearth. The electron gun, the least integrated
module of the evaporation source, consists of a pair of filament holding blocks and a coil filament (of
0.8-1 mm diameter tungsten wire) mounted on ceramic insulators and positioned about 2 mm from a
grounded anode. The filament cavity (cathode) and anode configuration are designed empirically for ease
of beam extraction and mechanical stability at high temperature. To minimize arc formation in the ion rich
vacuum surrounding the cathode it must be carefully fabricated, and molybdenum is usually the material
of choice. Because the filament blocks operate at up to 500C, the alumina mounting insulators must be
of high purity, carefully shielded and heatsunk on the grounded end.
A 1.2:4 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
n, .
[ / ii ( ('
Electron Beam Evaporation .
The magnetic lens and the evapora~t cont~ining h~~th are usually a totally integrated asseIIlbly, _though ( i'. .'
for clarity they ..~.ill."be .treated sep~~tefy~ fn this disc.ussfon. A".~gneifc' .f}el4
O 1f ~..
.!;;r .... ~-
a
...
hundred
~.,.--.-'!T-i''t.-;.......; -:;..... - -
~is"
....Y-~~
.of .fe.w gauss > t ~I i ,- :l
reqmred_tQJ?.~.!i.<L th~ _ <?.lectron bearnJ rpm .the .ele_ct(Qf} _gyn th.~~~gh ....270 and .fo~~.s.jt,u1'-,2!1.Jhe~. ~y~9r'l_nt,
The -rrm:gnetic brcuif
'to'ns'1s"fs"-of' a~pair'o.fsof(lr9n"'s1de . "wlth~-~~dj'u's_t~ble .beam formers and two pfates
magnets, an electromf!gne~ and ~. p~~IJWU:~:Q( .m~g~ei'::'~i~ parallel. The field of -thep~~~~nent magnet is
chosen such that the beam is positioned within the he~rt~. - ~~d the parallel electrgrri ~g.n.~! .i JJ..~ed to in.crease_
and decrease th~ main magnefiCfreld for longituch naf" beam position control. Like the cathode cavity the _
m~g'n'etic le.ns. is empiri~aliy design~d"~ i;c~~- the-beam to abo.u t 0;5-~ cm.. diamet~r C:tt .a riearfy vertical angle
of incidence upon the eva orant charge. The principal focusing elements of the ~~g~e.tic" lensare"'tfle..b~im
ormers . 5) ~ 1ch .create . an adjustable 's harp field concentration and gradient. This focus adjustability is
of critical importance, as discussed in section 1.2.4, for the evaporation of dielectrics. For lateral beam
positioning, a small secondary magnetic circuit is often included, particularly in large hearth (over 20 cm 3 )
sources. An electromagnet with one pole near the electron gun and the other loosely coupled to the side
plates performs this function. The required field is very low, and no detailed lens design is necessary.
All the permanent and electromagnets in a evaporation source must be embedded within th.e.. w~t~r-coQled
heart.h__,9!._!_~2f.12Qrnffi.-~.~~~~-~~-?..?.,!.~!1:~t.!?..~~-~~~-- <?.f...t!:t.<?....bigh . ambieotJemp_~JJttre . __ PmicuJat: . C.C:tr~ . mJJ&t be \
taken in winding the electromagnets or they will be damaged by voltage spikes induced into them by high
voltage arcs' ~ ~ "' .O>'w''"' "" -~ "''' .__,,,,,,,.,~, '"'~> ;~
,. -..
0
.. , ' " """N"-'- <-,
energy, a~VY i,..n1~mmtj op o.i.:. reducJion ,in flow.r..d.Yt!ng, qperation can destroy the hearth. ~ t. '"~' -~ ~' a . M ~; ., : ~,.;'i.:,~ ~~iJo>,. ;;,'1~-. -, ~~~~:r;-of'::.. .A.~.~ ~'i\f.2''t ~
An important extension of the basic single hearth source is the multihearth source. By fabricating the hearth
as a rotary turntable with four or more independent hearth cavities and a water-distributing central axis, a
great increase in versatility can be achieved. Most films today are multilayer films, and multihearth sources
are commonly used, particularly for optics. These hearths are significantly larger and more complex to
design and fabricate than single hearth sources. Of special importance is the 0-ring sealed central rotary
water joint. Unless this joint is designed with exceptional stiffness and close tolerances, it will tilt or
unduly V'!.~.aLthe.. 0-rings during rotation, and subsequently leak. For reliable service careful maintenan.ce
and lubricat(9n. of the . h~arth.-0-rings is necessary .. The magnetic fields couple easily across 'a 1''ffi~;~-g~p
betw~en thinxecf"mag~etic lens and that p~rt -of the lens incorporated into the rotating hearth.
In addition to containing the evaporant, the hearth must contain the electrons reflected from the evaporant
charge. Because the energy in the reflected electron beam may contain 30% of the beam energy, capturing
( these electrons is of significant importance. This is usually done by mounting the permanent magnet on
top of the hearth, behind a water-cooled copper shield, in the path of the reflected electrons. This creates a
strong magnetic field concentration which bends these electrons into the shield over the magnet, capturing
' - . . . . ..... "'" > . .... . .. ' .. ~.
them. . .......-~---~- ....... _. - .
Success in depositing films using electron beam heated sources is much more sensitive to application
technology than is the case for sputtering. The evaporant quantity required by an electron beam heated
source may be as little as 1 cm 3 , yet the selection and conditioning of the evaporant are of utmost
importance for successful film deposition.
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.2:5
Thermal Evaporation
The conditioning of the evaporant consists of a slow increase in power (or rate for subliming materials) over
several minutes to one and one half to two times the level desired before opening the shutter to clean and
degas the charge. Occasionally, during this process, a melting material will give off such large quantities
of gas that the system pressure will rise above the maximum acceptable (4 x 10-5 Torr for example), and
' degas very slowly, even at ht gfi power. 'Exp_~~i-etrce -trn.s sfiOwn that by repeatedly pulsing the power to the
highest practical for a few seconds, then when the pressure reaches the maximum allowable, or the charge
becomes unstable, reducing the power by 90% and allowing the syst_em . to ~~cover, significantly speeds
conditioning. Severe gas evolution m~so be a sign theevaporant is decomposing. Eacii"foc remen tal
power 1 hctea~e is a testing process in which the evaporant charge must be examined for stability and
I
spitting by using a small mirror mounted within the vacuum chamber. In some cases, extreme patience is
required as the-c:ondltioning may require over an hour. Conditioning is completed by running the gun at
maximum conditioning power, while sweeping the beam back and forth over the gun charge and watching
for instability in the charge. Semimelting materials (those melting only at the point of electron beam
impact) are particularly difficult to_150ndition:~as-rns-- ii~suaffy i"iripossible to-- melt-the entirn charg~y,0lume. ,
As-semimeltfhg'materiaTsare consumed by evaporation, new subsurface material is melted and-exposed to
the substrates, often with violent instability. This can be minimized by conditioning the charge in several
pumpdowns, reiriovi'iigifaru.riu~~~g "TC90 after each pumpdown. With subliming materials, the amoun
of conditioning tnaCocturs is necessarily fow, and slow due to low gas mobility in solids.
- -- - - - - - - - - - - . . . . - - -... -- ~------....- - - - - .. ..--v .--""""" _ _ _ _ --""" -#"-~- ,___ ....~-- ... _..,.., .. _. __ . . - ..~--0 . . . . . ~-,.,. . . .__,., _,__ .
-~ ~-:;..,._.,,:.,,..,.,, ""'_~.
The deposition rate is strongly influenced by the variable characteristics of the ~i~~tron oeam~ g~~ ., - ~~d
1
- :
-'
the evaporability of the material [7]. The influence of the electron beam gun hearth size and beam spot \\
~ size is seen in table Al.2.1. For melting materials (those which melt before evaporating, Al, Au, Cu), \
l the rate increases with increased power density (decreased spot size) in the melt, to the power dissipation
f~ limit. With semimelting materials (those which melt only in the beam, sapphire), and subliming materials
{l (chromium and silicon monoxide), which are not able to absorb the full power of the beam, the power and
\1i rate can be increased only by increasing the spot area. For these materials a fixed beam spot is similar to
;,ts !
q~ a swept one of equal area. The small spot, swept at 60 Hz, though to the eye a good approximation of a }
i,fixed large spot, is still only an approximation and is not fast enough to allow nearly as much power into
silicon monoxide as does the fixed spot at 1 cm2 area .
/
...... - - -----------"'"--- ~-~------ _..._ __ .
Table Al.2.1. The effect of electron beam gun parameters on the deposition rate of
representative materials. Column headings in cm 2 are spot sizes. Rates in Almin; power in
kW.
Maximum deposition
3
3.2 cm 3 hearth 30 cm hearth
State at
Material evaporation 1/4 cm2 1 cm2 a 1/4 cm 2 1 cm2 a 1 cm2 2 cm 2 a
The melting materials deposit at higher rates from the large hearth due to the longer thermal path to the
water cooling. Some subliming materials also exhibit in creased rates in the large hearth, due perhaps to
A l .2:((D Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
Electron Beam Evaporation
- - - - - --
.........
.J'
the larger charge permitting heating a greater volume to evaporation temperature. -~ With all dielectrics,
~~ -- $~ .. """ '..r.:~ ....
~ea~ .po~~ .. ~u~_,t"" ~~,, ! ~c~e~~~~....~lowlx C'Q.O , .~~,son 1,~ -:~~Yf.t J~:!~~t: , -~o -~e~'- heat.,..~l'. owi!}_g the el~cP:ic.al
conductivity to increase to a pomt where the surface charge on the matenal can Clisslpate ..to,. ground .
.-.......-~ ,. ;:;-.~ ., . _ .,..,-~. _ t...f" -? .,.... '"'. '""'"~ \~./~ "'l "A;,"-...; r .n ..... ~"'"f'. ~ _ ,. .. .4 l.' . i.,-;?.'<'-"l't''t..+t~ <1"';;_:<~Y:7.:~. ," ' '~);-...~' "*"f,
The charge -form affects strongly the evaporability of the ;~terials (see table Al.2.2). Materials that. ~e,
good thermal conductors and melt well below evaporation temperature, e.g. zinc, can be evaporated from \.
\ any form of raw material. All other materials are difficult or impossible to evaporate fr~IJ! ~ny!h.~pg , but
chunks or hot-pressed pieces.--Mo.st materials, when heated as pO\vders ..or granuies: form a localized hot
, spot_~~Y~~l.~?:1aB volume at too low a power to permit conditioning . or signjfj~a.!1.t.. ?Y~E2r~'!!iqn_. 1Ae large .' 1
; ~. surface are.a of pf>.,,~-~lt:rx.s.. and granuks. ,C\l~p permits the absorption of large amounts of water vapour and
~
, oth-ef'ccl'ni~;;.;r~~nts which are released
'- ...,;;;:;,.... - - ~ ~.-4..... ,
upon heating, causing localized
....,j, ; -.:::
..;,-::_t1'i _'"". , ~"-~;:
"expfodi"rig"'
- . . .. " .,.
from ttie' guh. -
~;_~q;,r,,_!>' ! ".i'. ii' - - .A ~" ~. '<..;.;r;..
"'l.*'"'"!'' .-)'.~:.:,!>.'.~
\
..
\ \ S1hcon \
\~
Sublimes
Hot pressedJ y; 1.3
Granular i : c
\.
\ 12 000 )
I
1.l monoxide , . c i
.J Chunks I
" <. f; ~ft Hot pressed \ 0.6 41 000 l
'':'i 1
.' ,
1
;
Silicon i
:semi melts Chunks .c
Lumpsb ~. O 7000
'~ Zinc :Melts i' Powder 'J.5 39 000
L !~
' Chunks
i
i\ i.5 39 000
Wire 1.5 39 000
\ ~~~~--;.~~~~~~~~~------~~~~~~~~~~
The detailed evaporation characteristics for a wide range of materials are reported in table Al.2.3 [8].
In this table, the evaporation rate reported is not the maximum but a practical rate obtainable without
significant charge instability. The state at evaporation is tabulated at the indicated deposition rate. At
significantly different rates the state may be different. For example, titanium sublimes at very low rates
and magnesium melts at extremely high rates.
'Stability' is perhaps an excessively subjective term to use, but it is also the one which best describes
the most desirable behaviour. Stability, as used in table Al.2.3, is graded from excellent for materials
that, once conditioned, can be evaporated with virtually no risk of spitting or melt explosion, to poor for
materials that, though evaporable, require intensive attention and may explode or avalanche at any moment.
A material avalanches when a slight increase in rate decreases the impedance of the electron beam gun,
resulting in an uncommanded increase in power (beam current) followed by a further increase in rate,
perhaps continuing to the point of emptying the hearth in moments at a rate of thousands of angstroms
per second.
The 'visual clues' in table Al.2.3 are what will be seen in the mirror during evaporation. Those materials
listed as 'none' evaporate or sublime below the temperature of incandescence and do not fluoresce. The
listed hazards are only the most serious and acute encountered during film deposition. All materials used
in thin films are hazardous in some context and all must be investigated for hazard before use. Particular
care must be exercised during equipment clean-up to minimize exposure to toxic materials, minimize the
risk of fire, and to assure safe disposal of waste.
The following are four widely applicable guidelines for successful deposition of thin films from an electron
beam heated evaporation source.
(1) Select a charge form with the largest possible volume to area ratio. Avoid trying to evaporate powdered
1995 IOP Publishing Ltd Handbook of Thin Film Process Technology Al.2:7
Thermal Evaporation
or granular materials.
(2) Use the largest hearth volume consistent with available evaporant charge and the desired film.
(3) Use the largest beam spot area possible but still attain the required deposition rate. I
;
/
(4:)...!~crease the spot size if increasing the beam power causes instability or film pinholing.
.. ..""
'~ ~
.. '~ . . ....." .
1
t 2
60
Al.2:8 Handbook of Thin Film Process Technology 1995 IOP Publishing Ltd
@
~
\Cl
Vl
.,,c~I TABLE Al. 2. 3 The Evaporation Characteristics of Materials from an Electron Beam Heated Source (8)
~
;;;
::r
:;
~I
Conditioning
Evaporant Symbol Charge Time State at 0
Rate Stability Similar Film Visual Hazards Notes
Fonn (min) Deposition (A/sec@ kW) To Appearance Clues
Aluminum Al 20 3 Sapphire 3 Semimeltsd 10@ .5 Good Si0 2 Clear Very Bright ... Start slowly; sapphire will explode.
Oxide
Antimony Sb Lump 2 Melts 50@ .5 Avalanches ... Gray Nonee Toxic Melt viscous. Use Mo 1 iner.
Antimony Sb 2o3 Powder 2 Sublimes 10@ .2 Good ... Clear None Toxic
Trioxide
Unstable Silver/
Arsenic As Lump None Sublimes 100@ .1 Avalanches Pb, Se Gray None Toxic Condenses on N2 trap, contaminates equipment.
Fluoresces
Barium Ba Lump 1 Melts 20@ . 1 Avalanches Pb, Se Silver Blue Toxic Getters
Fluoresces
Barium BaC1 2 AM fn Mob 1 Melts 30@ .2 Good ... Clear Blue Toxic
Chloride
g: Fluoresces
::.
:
Barium BaF2 Powder 3 Sublimes 50@ .1 Good ... Clear Blue Toxic
c Fluoride
c
..... Extremely
~ Beryllil;m Be VM 2 Melts 100@ 1.5 Excellent Al Gray Bright Toxi c Difficult to degas.
;;;l
s Extremely
~ Beryllium BeO HPc 2 Melts 40@ 1 Excellent ... Clear Transparent Toxic
"
.,, Oxide
tT.1
0
~
Bismuth Bi Lump 2 Melts 100@ .5 Excellent ... Gray Pale Red ... (")
r::;
"''"'
~ 0
Start @ 1 Ang/sec; do not exceed 10 Ang/sec.
~ Boron B Lump 13 Hr Melts 10@ 1.5 Explodes Unique Black Very Bright ... Melt shatters on cooling; remove and replace ='
'::.-
::."'
each run t:C
c (!)
0 Boron B4C Lump 3 Semimeltsd 35@ 1 Smooth Cr Black .Very Bright ... ~
""'C Carbide 3
tT.1
Cadmium Cd Shot 1 Melts 30@ .3 Avalanches ... White None Toxic Destroys mirror @ . 5 micron. Use Mo liner. ,~
"O
0
'""I
ac;
='
->
iv
'C
"O
-
>--
~
~
e
Conditioning
g: --3
:::::;'
;:i Evaporant Symbol Charge Time State at Rate Stability Similar Film Visual Hazards Notes (!)
0
: Form (min) Deposition (A/sec @ kW) To Appearance Clues "'1
Cl
..,.
Cl 3
~
e:.
~ Lump 3 Sublimes 10@ .25 Smooth ... Toxic
tTl
Cadmium CdS Yellow Fluoresces <!
;; Purple i:i:i
Sulfide "d
:::-i 0
"" "'1
Calcium Ca Shot 1 Sublimes 20@ .2 Poor Mg Gray None Flanrnable Spits occasionally throughout evaporation. ~
""
;:i
,.., <:
::s
~ Calcium CaF 2 1rrm Pieces 3-5 Semimeltsd 30 @ .05 Good ... Clear Bright Red .. Heat to red in air before using.
~
,.., Fluoride
::r-
;:i Eye
Cl
c Carbon c 5rrm Rod 1 Sublimes 30@ l Poor Cr Black Very Bright Protection Coats entire jar and tooling.
<>o
'C
Required
Chromium Cr Shot .5 Sublimes 15@ .3 Good .. Silver Bright . .. May spit, meits at hottest points .
Gray
Cobalt Co VM 3 Melts 20@ 2 Fair Ti Silver Very Bright ... Clean hearth after every run.
Copper Cu 3rrm Rod 1 Melts 50@ .2 Good Ag Copper Bright ... Getters. Use oxygen-free Cu. Use Mo l i ner .
Dysprosium Dy Lump 2 Semimeltsd 30@ .3 Excellent Sc Black Bright Flanrnable Getters, burns in air.
Gold Au Any .5 Melts 30@ 6 High Cu,Al Gold Bright ... Avoid C liners. Me l t oscillates.
Inconel Ni/Cr/Fe VM 3 Melts 50@ 3 Good Ni Gray Bright ... Very low rates best, will spit.
Matte Fluoresces
Indium In Any 1 Melts 100@ . 1 High Sn, Pb White violet before . . . Use Mo liner. Films may melt .
melting
Lanthanum La Lump 3 Melts 100@ 3.5 Good Ti White Bright Flanrnable Burns in air.
ii
0'ti
Lanthanum
Boride
LaB 6 HP 5 Semimeltsd 100 @ 1. 5 Low Zr02 Black Very Bright Toxic Power very unstable. Getters.
'ti
=
er
~
a:
=
(IQ
r
&
ii
Conditioning
Evaporant Symbol Charge Time State at Rate Stabi 1 ity Similar Film Visual Hazards Notes
0
Form (min) Deposition (A/sec @ kW) To Appearance Clues
0"tl
"tl Shiny
c Any 2 Melts 30@ .1 High Sn, Zn Black Slight Red Toxic Use Mo 1 iner. Films oxidize quickly.
er Lead Pb
~
:r Gray
5 Lithium Li Shot 1 Melts 100@ .05 High Zn Matte None Flammable Use Mo 1 i ner. Vent to Ar, not N2 .
Otl
r Melts,
s. Lithium
Fluoride LiF Hp 3 Viscous 10@ . 15 Poor BaCl Clear Red ... Use Mo 1 i ner . Spits easily; bubbles at high rate
Fluoresces
Magnesium Mg lcm Rod None Sublimes 100 @ .04 Poor ... White Green Flammable Do not use turnings or ribbon .
Magnesium Fluoresces
Fluoride MgF2 VM Lumps 1 Semimeltsd 30@ .01 Good .. Clear White
2 .
Manganese Mn 1-Srrm Pieces 3 Semimeltsd 40@ .4 Fair Mg Black Slight Red Flammable Vent to N @ room temperature.
Molybdenum Mo VM 5 Melts 40@ 4 Fair w Gray Very Bright ... Extremely hot substrates required; will melt
hard-soldered tooling.
Neodymium Nd lcm Pieces 5 Melts 20@ 1 .4 Good Co White Bright Flammable Getters
Gray/
Nickel Ni VM 5 Melts 25@ 2 Poor, Spits Ti White Bright None Spits most severely of all conmen materials.
Gray/ Eye
Niobium Nb Any 3 Melts 20@ 5 Good Ti White Very Bright Protection Very hot. Will spit late in run.
Required
ti
Chloride KCL VM in Mo 1 Semimeltsd 30 @ .05 Good BaCl Clear Red
Potassium
Fluoride KF VM in Mo 1 Sublimes 40 @ .05 Poor SiO Clear None ... Deliquescent. Beam tunnels thru charge easily.
.....
~
;;:! Praseodymium Pr 1-Srrm Pieces 5 Melts 25@ 1.5 Good .. White Bright Flammable Oxidizes easily; films react to brown.
Gassy at melt.
s Gray/ Eye
~ Rhenium Re 3mm Rod 15 Melt~ 3@ 3 Poor Ni White Bright Protection Spits very badly; VM material not available
" Required
tI'j
"ti
Cl Scandium Sc 5rrm Pieces 2 Semimeltsd 50@ .5 Good Ti White Bright Flammable Getters ~
(")
<"'l
::::;-
"'~ Selenium Se Shot None Semimeltsd Unlimited Excellent .. Black None ~ Toxic High rates contaminate chamber. 0
~ :::s
<"'l
;::.- Poor/ t;O
;:s
c Silicon Si 1-Srrm Chips 10 Melts 15@ .15 Explodes Ge Black Bright ... Clean hearth after every run. CD
~
6"
C>Q
'C a
tI'j
<
~
"O
0
'""
~
g
I
:::s
>--
......
~
~
Id
.I.Ji
-
-::>
>
~
~
N
~ Conditioning
~
;:i
:
Evaporant Symbol Charge Time State at 0
Rate Stability Similar Film Visual Hazards Notes ::r
~
c Form (min) Deposition (A/sec@ kW) To Appearance Clues '""I
c
;.,. 3
~ !::..
Silicon
~ Dioxide Si0 1-3trm Chips 2 Semimeltsd 20@ . 7 Good Cr Clear Bright ... Corrrnercia l quartz rod spits. trJ
;; 2 <
s;.i
:!l Silicon "O
3
..,, Monoxide SiO 5+mn Pieces 1 Sublimes 20@ 1 Poor Cr Clear Red ... 0
'"1
~ ~
<"I Silver Ag Shot l Melts 100@ .5 Good Cu Silver Bright ... Use Mo liner. c;
~ ::s
Sodium
~ Chloride NaCl AM in Mo 3 Semimeltsd 50@ .15 Good BaCl Clear Fluoresces ... Films stable, use U.S. P. sodium chloride.
<"I
;::,.
;:i
c Sodium Semimeltsd Fluoresces
S" Fluoride NaF AM in Mo 3 Sublimes 40@ .2 Good BaCl Clear Green Toxic
<>o
'...:
Fluoresces
Strontium Sr lcm Rod 1 Melts 50@ . 1 Poor Mg White Blue Toxic Melts under oxide, which collapses.
Gray/ Causes exceptional equipment contamination.
Sulphur s AM in Mo 3 Melts 100 @ .02 Poor Zn Yellow None Toxic Minimize vacuum exposure as films evaporate.
Eye
Tantalum Ta 5nrn Rod 3 Semi melts 100@ 5 Good Mo Gray Very Bright Protection Extremely hot
Required
Red,
Tin Sn Shot 2 Melts 10@ 2 Excellent Cu Gray Beam Flickers . . . Use Mo liner. Fogs mirror.
Titanium
Dioxide Ti0 2 HP 5 Melts 10@ l Good .. Clear Bright ... Films absorbing @ high rate;0
2
atmosphere required
Eye
Tungsten w 4nvn Rod 5 Melts 20 @ 5.5 Good Co Black Very Bright Protection Extremely hot.
Required
Tungsten
Trioxide h03 HP 5 Sublimes 30@ .1 Good SiO Yellow Red
Vanadium v VM 3 Melts 100@ 4.5 Good Co Black Very Bright ... Getters
Gray/
Ytterbium Yb Lump 3 Sublimes 20@ . l Good Mg White .. ... Numerous small flakes of Yb on substrates.
Ytterbium
Oxide Y203 HP 20 Sublimes 20@ 2 Good .. Clear Red ... Beam tunnels thru charge easily.
Zirconium Zr VM 2 Melts 20@ 5 Excellent Co White Very Bright Flammable Will burn on gun in air.
ii
Lii
0'"tl
Zirconium
Oxide Zr2o3 HP 3 Semimeltsd 20@ l Good .. Clear Very Bright
Eye
Protection Beam tunnels thru charge easily.
Requir-ed