Monitoring 
Each installation has its own moni- 
toring problems that are peculiar to the 
process at hand. We cannot afford a 
single instance of lack of sterility in our 
product. 
sean width and penetration as it affects 
our product under processing condi- 
tions. We prefer continuous monitor- 
ing; a second choice is intermittent 
monitoring. 
Problems. 
trol of machine parameters. 
We want to measure dose, 
We had to consider con- 
With a 
linac this problem is a bit more compli- 
cated than with electrostatic acceler- 
ators. In an electrostatic accelerator 
like a Van de Graaff, controls are fairly 
simple. Beam current and scan-coil 
current can be read on ammeters. 
Accelerating voltage, which is very 
stable, can be measured on an electro- 
static voltmeter. 
After determination of 
allowable variations in any of the three 
important parameters these changes 
can be fed into a computer circuit. 
This circuit gives a signa] whenever the 
irradiation dose is reduced by a certain 
amount as a result of changes in any 
combination of parameters. The sig- 
nal can be recorded or used to actuate a 
marking device or kickoff mechanism in 
conjunction with the conveyor belt. 
Linear accelerators present a more 
difficult problem because the energy of 
the electrons can not be measured 
directly. Electron pulses are acceler- 
ated by an r-f traveling wave (2). 
Energy is affected by field strength and 
frequency of the r-f wave. Peak fre- 
quency is affected by various factors 
such as changes in line voltage, drift 
maximum 
Center 
a 
Ss 
e 
a 
a 
BS 
s 
@ 
2 
@ 
no 
fo) 
[=) 
4 8 
Scan Width (in) 
FIG. 3. SCAN CURVES show lack of 
lateral uniformity in dose (A), which 
preceded adjustments of magnet pole 
Pieces and more uniform scan that 
followed (B) 
in phase, defocusing, etc. Any change 
in energy of the r-f wave affécts pene- 
tration, scan width and dose. 
Monitoring can be achieved in a man- 
ner like that for the Van de Graaff, ex- 
cept that a frequency parameter has to 
be used instead of voltage. Scan- 
width changes can be measured by 
metal absorbers placed at each end of 
Two are spaced at each end 
~7g in. apart and wired to separate 
the scan. 
ammeters. The inner pair are always 
touched by the scanned beam. The 
outer pair normally are not. If the 
scan widens, both sets of absorbers 
record current; if it narrows, neither 
does. 
A meter on the exit window records 
the electron current that the window 
absorbs. In general the more current 
absorbed in the window, the lower the 
energy of the electron beam. 
For preventive trouble shooting we 
find it important to monitor focus-coil 
currents, operating and pulse frequen- 
cies, vacuum, r-f power, line voltage, 
power-supply voltage and accelerator- 
room radiation level. 
Conveyor speed must be accurately 
controlled at the point of irradiation. 
In normal operation our conveyor 
speed is 15-18 in./min, and it is con- 
tinuously measured. For maximum 
beam utilization our conveyor system 
is designed so that the product is irradi- 
ated continuously without spaces or 
gaps in the flow of material. 
Linac operation. Our only way of 
measuring linac energy is to absorb the 
beam completely in a stack of dosimeter 
slides. From the readings of the slides 
in proper order, useful penetration is 
calculated and converted into electron 
energy. An energy of 1 Mev implies a 
penetration of 0.130 in. in unit-density 
material. Plotting individual readings 
against thickness produces an ioniza- 
tion-distribution curve. 
Scan uniformity with the linac was at 
first unsatisfactory. Dose range within 
the 14-in. scan was as great as 1.0 to 1.2 
megarads around a nominal dose of 2.5 
megarads. After proper adjustments 
in the pole pieces of the scanning mecha- 
nism the range was reduced to 0.5-0.7 
megarads (Fig. 3). This nonuniform- 
ity across the scan is still greater than 
that of the Van de Graaff, but for a 
sterilization process we can live with it 
if the average dose is right. 
Another problem is beam-energy 
fluctuation during runs. Initially we 
had some large and many small fluctu- 
We learned to take care of the 
large ones by keeping the machine con- 
tinuously adjusted to optimum oper- 
ational settings. The many little fluc- 
tuations are still with us, but they do 
ations. 
not affect our processing greatly. Ona 
few occasions when the energy was 
gradually dropping off over several days 
in spite of the fact that the machine 
was operating at the right settings, we 
could predict impending major klystron 
trouble through 
continuous 
our dosimetry. A 
indicator on the 
linac would be a major improvement. 
energy 
With a year’s experience under pro- 
duction conditions, we can say that we 
required about 6 months to obtain 
FIG. 4. 
PRODUCTION TRAY for testing 
machine output carries row of slides 
to test scan uniformity and stack of 
slides in cutaway slot to test penetration 
efficient and consistent operation of the 
equipment. 
Process monitoring. We have been 
using ceric-cerous dosimetry (3) as a 
primary standard, and blue cellophane 
(4) and rigid vinyl film (5) as secondary 
standards. Rigid vinyl film as a do- 
simeter is suitable for the following 
measurements: (a) total and useful 
penetrations, (b) ionization distribu- 
tion, (c) absorption in nonhomogeneous 
objects, (d) scan width and uniformity. 
At the beginning of each shift on the 
linac a production tray is put through 
(Fig. 4). It contains a row of 14 rigid 
vinyl! slides to test scan width and uni- 
formity and a stack of 60 slides, which 
is thick enough to absorb the beam 
completely. Each slide measures 3 X 
1 X 0.015in. The slides are placed on 
a 34-in. plywood board to reduce back- 
scatter. The row of 14 slides are 
placed side by side across the board 
with their long dimensions in the direc- 
tion of travel. They are numbered 
consecutively so that their positions 
147 
