ELECTRON MULTIPLIER TUBES DEVELOPMENTS UTILIZATION" BY ANDRE LALLEMAND

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78-03300A001600050013-4
• Release Decision: RIPPUB
• Original Classification: C
• Document Page Count: 7
• Document Creation Date: December 22, 2016
• Document Release Date: February 8, 2012
• Sequence Number: 13
• Content Type: REPORT

Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
 Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
ELECTRON MULTIPLIER TUBES DEVELOPMENTS. UTILIZATION
By ANDRE LALLEMAND
Astronomer at the Observatory of Paris
Summary - Determination of the fluctuation in output current of an electron multiplier
tube in terms of the number of stages and the multiplication factor of each stage, under
ideal conditions. Other causes of current fluctuations are discussed along with the construc-
tion of a tube in which fluctuations are nearly eliminated. Results obtained are given as well
as some ideas on the use of electron multiplier tubes: a high voltage feed. The multiplica-
tion obtained is determined if a galvanometer or amplifier is used as receiving apparatus.
Use of multiplier tubes with a very stable balanced amplifier and use of a neon tube relaxa-
tion circuit are discussed.
Let us consider a photoelectric cell illuminated
by a luminous flux, D,,. At saturation it supplies a
current, I,, _ Kd,,,. This current, often very weak,
for example 10-1-, amp. cannot be measured by
direct methods. Amplification is necessary.
Usually one uses a vacuum tube amplifier, but
since this amplifier is sensitive to voltage and not
to current one will change the current into a
voltage drop, RI, across a load resistance, R. The
larger the R the more advantageous the change,
but with large R's certain difficulties arise: the
response becomes slow and the grid current of the
following amplifier tube interferes. We must
therefore use special tubes called electrometric
tubes. The current, I,,, is not a perfectly contin-
uous current, it undergoes fluctuations or Shottky
effect which are obviously amplified. In a funda-
mental manner these fluctuations limit the small-
est measurable luminous flux. But there exists
another source of very important disturbance
in the load resistance, R, which behaves like a
generator furnishing an e.m.f. of perfectly irregu-
lar character, i.e. that all the frequency com-
ponents are equally probable and with the same
amplitude.
If we consider the frequency interval, of, the
mean square of the disturbing voltage or Johnson
effect is:
E2 - 4KTRAf
K, Boltzmann constant: 1.380 X 10-` W's per
degree
R, the resistance in ohms
T, the absolute surrounding temperature
whereas the mean square of the current fluctu-
ations of the photocathode is:
i2 = 2e1?of
e - 1.60 'x; 10'? coulombs, charge of the electron.
It is necessary to compare the size of these two
disturbances. In the load resistor the current, i,
creates a difference in potential, iR, from which
we obtain the relationship :
i2R2 I ,,R I?R
E=2KT5X10'
e
That is, the Schottky effect is equal to the John-
son effect when y = 1, if the current furnished by
the cell produces a voltage drop of 5 X 10-2 V.
across the resistor.
Thus the smallest measurable flux is always
masked by the Johnson effect of the input
resistor.
We must therefore try to amplify the current
without using any load resistance. This can be
done by means of an electron multiplier.
If I? is the photocathode current, n the number
of multipliers and 8 the multiplication factor, the
current furnished by the multiplier tube is
I = I?8".
But this current is composed of fluctuations.
We shall try to calculate the smallest measurable
luminous flux by assuming ideal conditions where
no other sources of disturbance exist other than
those produced by the granular nature of
electricity.
Let us proceed as follows :
1) The electronic emission arising from any
multiplier follows Schottky's law
12 =2eI?of
2) The Schottky effect is multiplied by the
successive stages, like any signal (all frequencies
are equally amplified) so that if I? is the current
arising from the photocathode, its fluctuation is
given by:
i,2, = 2eI?of
At the output of the first multiplier the current
is I, = 81,,, its fluctuation
if = 2e81?of + 82(2eI?of) = 2eI,of
Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
 Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
At the n"' multiplier the output current fluctu-
ation is:
i; = 21,Af [8"+an+1+ .... 8211]
1-Sn+1
= 2I ~f8 (1 .... ~8") =2I~af8'Z 1-8
The disturbance at the input is given by :
io = 2eI(,of,
The signal at the input is given by: x,
The disturbance at the output is given by:
i', = 2eI?of8"1_Sn+1
The signal at the output is given: x8",
From which:
signal
( noise input _ 1-811+1 1
(8"(1-8)) - A.
signal )
noise output
This equation permits us to calculate the dis-
turbance brought about by the multiplier system.
Let us take several numerical examples :
8=2
n=1 wefind A=x/1.5
82
n-7 66 A- /2
8=4
n=1 44 A= /1.25
84
n=7 66 A: /1.3
The signal to noise ratio at the input is always
larger than the signal to noise ratio at the output.
A multiplier tube cannot improve the signal to
noise ratio of a photosensitive layer. The larger
is 8 and the fewer the multiplications, the less
the disturbances are. But practically speaking the
supplementary disturbances produced by this
manner of amplification are negligible as soon as
8>2 no matter how many stages there are (under
the ideal conditions which we have assumed).
The great value of the electron multiplier tube
is that it makes it possible to measure very
weak photoelectric currents without adding any
large amount of disturbances to them. But if we
are working with luminous fluxes of sufficient
intensity to be able to measure these currents
without amplification, or if the photoelectric cur-
rent produces a drop in voltage greater than 5 X
10-2 V in the load resistance, the multiplier tube
ceases to be of any value.
We shall show later, that a high multiplication
cannot compensate for the weak sensitivity of
the photocathode.
We have assumed ideal conditions. Unfortu-
nately the multiplier tube possesses other sources
of disturbance and the research which we have
undertaken had as its aim the cancellation or
reduction of these disturbances while conserving
a large enough factor S and a high total multipli-
cation.
It is essential to have a photocathode of high
efficiency, yet it is a difficult matter to produce
very sensitive photocathodes. It is hardly likely
that one will obtain optimum activation in the
same tube for the photocathode and the multi-
plier target. We have thus chosen for multiplier
targets a silver-magnesium alloy which can be
activated separately, in part, outside the cell.
When the activation is finished we can focus all
our attention on the creation of the photocathode.
We can thus obtain a sensitivity with a luminous
source at 2500?K of 80-100 ?A per lumen for the
antimony-cesium cathodes. The Ag-Mg alloy
possesses several advantageous properties also: it
is not photoelectric and its thermal emission at
ordinary temperatures is nil.
The arrangement to be used should be such that
the desired paths of the electrons should be
affected as little as possible by stray magnetic
and electrical fields. The stray electrostatic fields
which are produced in the interior of the cell are
particularly disastrous and are often produced
by dielectrics which are in the vicinity of the
targets (wall of the cell, mica separators, glass
support stems). They become charged to un-
Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-03300A001600050013-4
 Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-03300A001600050013-4
known potentials and create uncontrollable fields.
We thus decided to choose a mounting in which
all these insulators were eliminated (except of
course the wall of the tube whose action will be
minimized).
The multiplier targets are made of semi-trans-
parent screens (Fig. 1) placed one behind the
other at as small a distance as possible. Even
when strongly deviated from their course, the
electrons cannot miss the targets ; they are
focussed toward the axis by giving the targets
suitable shapes like that of a basin whose curva-
ture was determined empirically by tracing the
trajectories of the electrons through fluorescence,
in order to insure a sufficient concentration with-
out losing electrons nor creating an excess charge
in the center of the targets. Another advantage
of this design is that the mounting need not be
very precise, thereby permitting elimination of
the glass or mica separators. The multiplier elec-
trodes are mounted increasingly close together so
as to augment the extraction field of the secondary
electrodes. The anode is well protected by a guard
ring.
We were obliged to change this device for
several reasons:
1) Great difficulty in procuring suitable silver-
magnesium sheets.
2) The extraction of secondary electrons was
not good.
3) It was found to be difficult suitably to acti-
vate a sheet.
We created a mounting as shown in diagram
2. The multiplier parts are formed from parallel
and inclined plates. A loose grill of fine wires
permits focussing and extraction of the secondary
electrons. A study of the electronic trajectories
was made by the rubber membrane method.
We kept the advantages of design 1: ease of
mounting, insensitivity to magnetic disturbances
and omission of separators. In addition we have
better extraction and more easily produced acti-
vation.
We have thus been able to make cells of 7, 12
and 18 stages.
We admitted, for the ideal case, that the non-
illuminated cell supplies no current. In reality this
is not the case because in the dark the cell always
has a weak current known as the residual cur-
rent. On it depends, when the mounting has been
done correctly, the smallest measurable luminous
flux. The various causes for this current have
been well presented by Rajchmann. This current
originates from:
1) The ionic reaction. Although the vacuum
in the cell is good there are always some ions
tending to rise toward the photocathode which
they bombard, and thus provoke a multiplied
electronic emission normally contributing to the
residual current. One must prevent the ions from
reaching the photocathode. In our mounting they
are efficaciously stopped by the multipliers them-
selves.
2) The cold emission. These are the electrons
which are torn away from the different electrodes
by the electric field. In order to avoid it one must
eliminate all unevennesses or points. The arrange-
ment of the supports is such that the electrons
torn from the edges of the electrodes cannot par-
ticipate in the multiplication process. It is neces-
sary that two conductors or two electrodes
carried at two very different potentials be
separated as far as possible one from the other.
This we have found the best in our large 18 stage
tube.
3) The current of ohmic origin. In our mount-
ing it can only be produced in the foot of the cell.
Since experience has shown that it is of super-
ficial origin, we have lengthened the escape lines
to a maximum from one electrode to another, as
is done for high tension insulators. It is well to
note that a multiplier tube utilizes very high
voltages in relation to the weak intensities
brought into play, and it is not reasonable to
reduce the dimensions of the cell too much, which
necessarily leads to insufficient insulators and
gradient potentials which are too high and sources
of parasitic emission. The last source of residual
current is found in the thermic emission of the
photocathode - it is linked to the extraction work
of the layer, i.e. to its sensitivity limit toward
the infra-red, and the antimony-cesium cells
which are much less sensitive toward the red than
Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-03300A001600050013-4
 Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
the cells of the type Ag-Cs-O-Cs have a much
smaller thermic current. We hope to develop
layers of very low sensitivity to the long wave-
lengths of the spectrum, having a very weak
thermic emission. In this case we must avoid the
in the layer to a high degree.
These different studies and adjustments have
brought about the following results:
Antimony-cesium cell - Light source: tungsten
filament lamp at 2500cK (temperature of color).
Cells at room temperature (20?C.)
Sensitivity
Lumens giving a
No. of
cell
No. of
stages
of the
photocathode
Residual current
at 120 V.
Multiplication
at 120 V.
current equal to the
residual current
A 11 ------------------------------------
7
88
? A/lumen
3.7 X 10-11)
1215
3.5 X 10-~'
B 24 -----------------------------------
7
103
1.5 X 10-10
2080
6 X 10-70
A 19 -----------------------------------
12
66
128 X 10-10
100 000
3 X 10-"
BG 1 ------------------------------------
17
67
30 X 10-10
850 000
5 X 10-11
BG 3 ---- ------------------------------
19
60
44
1.3 X 10-11 3
700 000
5.8 X 1W,
1953 ------------------------------------
1.3 X 10-8
50,000
Considerations as to the Use of Multiplier Tubes.
The supply voltage of the different stages con-
stitutes an essential problem since the multiplica-
tion factor, 8, varies quickly with the voltage. If
one desires coherent results a constant feed volt-
age is necessary. This source is the more easily
realized the fewer stages the cell possesses. This
is why one must limit oneself to the number of
indispensable multiplier stages. The simplest
source consists of a battery of the voltaic pile
type. These batteries feed the multiplier directly
without need of a voltage divider. One must take
care to introduce a protective resistance of about
50 K 12 for each multiplier in order to avoid serious
annoyance in case of short circuits. If one wants
a feed taken directly from the power line one
can use rectified and filtered current to feed a
chain of neon lamps in series with a large stabiliz-
ing resistance. The desired voltages can be taken
directly from the terminals of the lamps, this
arrangement having the inconvenience that one
cannot regulate the applied voltage applied to
each multiplier except by steps of about 60V
which represents the voltage drop in each lamp.
When the luminous flux to be measured is
sufficiently intense or when the multiplier tube
possesses enough stages, one can measure the
current with the aid of a galvanometer. The mul-
tiplication factor which the cell must possess in
order to attain the maximum sensitivity limited
by the Schottky effect is interesting to calculate,
for a given galvanometer. If I? is the current in
amperes supplied by the photocathode, T, the time
constant of the circuit, usually given by the
period, 0, of the galvanometer:
T 27r
and e the charge of the electron in colombs, then
one can show that :
i"= e- I2., ,
2T
i' being the mean square of the fluctuation of
current, I.
The minimum current, I,,, will be the thermic
current of the photocathode at ordinary tempera=
ture; its order of magnitude can be taken equal
to 10-11 or 10-10 Amps. Let us take I. = 10-10 Amp,
T= 2s,
i = 0.6 x 10-1 -- Amp.
Let us take a galvanometer whose sensitivity is
10-111 Amp/mm at 1 m. So that the inevitable
fluctuation of the current gives a zero fluctuation
of the 1 mm galvanometer, the coefficient of mul-
tiplication of the cell must be:
M= 10-1?_ 1.7X10,
0.6 x 10-17 -
which is quite difficult to obtain without other
disturbances.
If one cools the cell the situation is even more
critical.
Thus one often resorts to amplifying by tubes
if one wishes to obtain all its sensitivity from the
cell. Assuming this to be the case, letting R be
the value of the load resistance, of the band width
of the amplifier. The fluctuation found at the
output of the multiplier is:
T; = 2cI?.fM'
The Johnson effect in the resistance is 4 KTR. f
and one should obtain :
2eI?AfM2R2 X 4KTRof.
M' \ 2KT Ri with 2 KT = 5 X 10- Volts
Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
 Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-03300A001600050013-4
Let us suppose a cell at ordinary temperature
to have I. = 10-1,, Amp. Let us take R = 10?s2,
which produces no difficulty due to the grid
current and the stability of the amplifier. We
thus find
M\7\10'.
We see that in this case an amplification of
10,000 is largely sufficient. Naturally the ampli-
fier itself must possess an amplification sufficient
to make its fluctuations apparent at the output
terminals of the apparatus.
If the cell can be cooled I. diminishes; here one
would have to increase the number of multipliers
of the cell.
We have tried to find what multiplication was
necessary to permit the attainment of the ulti-
mate sensitivity of the photocathode. For multipli-
cation permitting correct measurement of the
photocurrents without adding disturbances to
them, let us assume I(, the photoelectric current
of the photocathode. I(, the current at the output
of the multiplier. The signal is:
I+ =1q\M
The current of thermic origin is i ; at the
output it causes a residual current I = i X M.
At the output we will have the following fluc-
tuations:
MI2e(i+l,).f I
and the relationship
signal _ I(rM Iq,
noise - MI2e(i+I,)Af I 12e(i+Ir)of]i
M disappears as one might suppose, and we see
that this ratio increases:
1) If I(, increase, i.e. for a given luminous
flux, if the sensitivity of the photocathode in-
creases.
2) If I decreases;
3) If Af decreases.
Thanks to the use of the silver-magnesium
alloy, very sensitive photocathodes can be pro-
duced (100 ,Amps per lumen and more). More-
over, this alloy furnishes multipliers which them-
selves do not possess any thermic emission, just
as we had supposed.
i can be rendered smaller by cooling the layer.
One can also make it smaller by electronic optic
methods.
Improvement in the signal ratio can be ob-
noise
tained very efficiently by reducing of. This
possibility has not yet been investigated
thoroughly because it involves delicate mountings.
The modulation frequency must not slip during
the measurements in relation to the passing fre-
quencies of the amplifier.
When the frequencies to be amplified are low,
for example of the order of 1 cycle per second,
it is difficult to create a stable amplifier. Under
these conditions the multiplier tube permits the
use of a balanced amplifier which is much more
stable (Fig. 3). The amplifier consists of a com-
mon resistance in the cathode circuit which con-
tributes a great deal to the stability. The
application of signals of equal amplitude and
180? out of phase to the two grids is easily
produced by taking one signal from the multi-
plier anode and the other from the last diode,
the amplification, 8, of this plate being always
high enough (of the order of 5). The two cur-
rents, out of phase by 180? furnish voltages of
the same amplitudes in the load resistance R, of
the anode and R. of the diode by taking:
Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-03300A001600050013-4
 Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4
Let us note another circuit which is very
simple and sensitive (Fig. 4). The output of the
multiplier is applied to a condenser of small
capacity (100cm), well isolated and shunted by a
neon lamp. The multiplier current charges the
condenser which discharges into the lamp when
the illuminating voltage is reached. Thus, in a
given time period, we have a certain number of
flashes which are sufficient to be counted in
order to have a measure of the luminous flux.
This count can perhaps be made automatically
by amplifying without special precautions that
current which is furnished by the lamp and
thereby activating a small telephonic counter.
The sensitivity can be diminished by increasing
the capacity of the condenser.
Declassified in Part - Sanitized Copy Approved for Release 2012/02/08: CIA-RDP78-0330OA001600050013-4