1 /pages
61 to 64/l
994
0042-207X/94$6.00+.00
Electron beam evaporators with magnetic field and hot filament are widely used in vacuum technology. The lack of adequate models for the electron beam evaporating process makes it difficult to predict the optimum parameters for quality and low cost process control. A mathematical model for transforming the filament heating power into evaporation power is developed. A discrete model for computer control and an algorithm for the models parameter estimation are presented.
1. Introduction Electron hcam evaporators are used in vacuum coating systems, thin film production, tool coating and vacuum metallurgy. Their implcmcntation in the ion-getter pumping is also well known , The most widely used type of evaporator is shown in Figure I. The material is evaporated from anode A which is heated by electron bombardment. This three-electrode system, Figure 2, is placed in a magnetic field generated by a coil B. Film growth takes place at room temperature T,,. The parameters characterizing the fihn deposition can be varied over a wide range by changing the evaporation power P,,. The power ncccssary to heat the anode is a product of the anode current I., and the constant voltage U, applied to anode (U;, = 6 x 10 V) : P,, = I:,6 x IO , PC, depends on the thermionic emission from the filament and thus on the filament heating power P,-. No problems would be faced during the different technological processes if P,, could be controlled adequately. In this case the manipulated variable is P, and the controlled one is PC,. The offset characteristic in Figure 2 is obtained by means of a tungsten filament of diameter d and length 1. whereby I>> d
Figure I.
Figure 2. 61
N I Donkov
Electron
beam evaporation
AI,., = I,.,,C,A7:
AC,,,,,, = P,,,C,( 7,
(7)
, -~ 1 )
(8)
magnetic fit&l. yxlcc chargo
;II-L (rang~
whcrc Cis the spccitic heat oltun~stcn On the basis of the power
onic cniission translbriiiatioii and elcclron of bonibardnicnt
Al,,
= :,I t,,,AP.
\aluc\ 01 the
halancc 01 lihnicnt
At liiglicr
obscrvccl II).
by the equations
which
lend to changes
in the olfsct
charactct-lhtic
(I)
(4):
This nccezsitatcs
coethcicn!s
in equation
(8). With
ccluation
(5) in \icb.
and ncglccling
pet- unit
(1)
~iicrwsc
lenpth
on the tilament
demands intcrwls (I)
niatcrial. 01 I~ncat -I in as
the synthesis
3. oquution
I:rom controlled
ccluations mcmbcr
rcsponsc
01 the
:
(IO)
A/,,,( /j = (1
I ~implilicd rc\ult
-I i\ :
(II)
(6)
500 r 400 4 h 5 ia.: 3 200
100 t *
300 1
The
unit
step arc
i-esponscz aho~n
loi- the
lour
irango
Irom
tlic
oil*ct
chalactcristic
in Figure mcmbcr.
1. To dc\clop model
it ix ncccssar!
:I\ ;I controlled
space mhA
2 >
6y
range lour ij
) = .3.\-(X) +
%I Iigurc 3 62
I
100
.\-(I\ + I
_I(/<)
BL/(/i).
70
q (W)
C.\-(k).
Time Figure 4.
(s) Figure 6.
Diameter
(mm)
u(k) arc the controlled members scalar output corresponds to P,,,(k) and u(k) is proportional C are :
-0
0 A=00010, 0 0
0 1
0 0
01 0
0 0
0 0
0 0
0
0 0
I F
01, 8.264 4.110 2.0431.
The pole z = 0.503 is less than unity and thcrcforc the controlled member is asymptotically stable. Effects such as sputtering, oxidation etc. can influence R, and 4, which in turn can change the digital models parameters and lower the control quality. Figures 6 and 7 show the variation of the models parameters with d. Calculations arc made using equation (10) with At = constant and ubeing variable. The detcrmination of d in real time can be performed easily using the available information for I/, and I, :
c =
[l
B = [33.423
16.620
Its componentsh are determined from the transient response (Figure 5). The sampling time AT has been chosen to be 0.5 s. F can be obtained from :
Ap%,.,, -
,z, h1
in the z domain is : 1 -O.503z~-
where p<, is the specific rcsistancc and a is the temperature coefficient of resistance. The algorithm shown in Figure 8 corrects the models parametcrs with dependence on the changes of the working interval as well as with the changes of r/ and R, ; AL is the timing period for data checking. The algorithm prevents the anode from mclting due to excess maximal heating power level. For the control law we have :
In the same four ranges. F is 0.503. The transfer function of the evaporator Gp(z-
) =
33.423~~+0.08;~+0.1,-P+o.~1-_~+0.02z~5
0.6 r
*Or
Time Figure 5.
(s) Figure 7.
Diameter
(mm)
63
N I Donkov
Electron
beam evaporation
Read
ADC
Uf Jf
1 Computing 4 I Computing F,bi ,i=1+4
Read I, ADC
Figure 8.
Protection
level
Ktftrcncc~
Time
(s)
obs~r-\cr) .
ofthc
state
which cm
not dircclly
The control
rcsponae
I in Figure parameter
system
the estimation
paraniclers.
64