Wide-band
detector
for
micro-microampere low-energy
electron
currents
by
T.
E.
EVERHART,
Ph.D.,*
and
R.
F.
M.
THORNLEY,
B.A.,
Department
of
Engineering,
University
of
Cambridge
[Paper
received
25
January,
19601
Abstract
Electrons with
a
mean
energy
of
a few
electron
volts
ewerging
from
a source
a few
nzillimetres
in
diameter
are
accelerated
on
to
a positiyely
biased
plastic
scintillator.
The
light generated
in the
scintillator is guided
by
a
Perspex
light-pipe
to
a photomultiplier.
If
the
scintillator
bias
is greater than
10
kV,
no
noise is introduced
by the
scintillator-light-pipe-photomultiplier
system.
If
the ori-
ginal electron
signal
is
modulated,
the
maximuin
modula-
tion
frequency
transmitted
by
the
system
i3
greater
than
10
Mcis. The
ininimum
detectable current
is
limited
by
the photomultiplier dark current,
and
may
be
less
than
A
Mith
a suitably chosen
tube.
Introduction
HE
properties
of
the
detector
described
in
this
paper
were largely
determined
by
the
instrument
for
which
it was designed,
the
scanning electron
microscope.
In
this instrument, a
fine
electron
probe
is
scanned
in
a
rectangular raster across
the
surface
to
be examined.
Secondary electrons
leaving
the
surface
are
collected,
amplified
and
used
to
control
the
brightness of
a cathode-ray
tube scanned
in
synchronism with
the
probe. Details
of
the
instrument
and
the
contrast
mechanism
have
been
published
by
McMullan,(') Smith
and
Oatley,(2)
and
Everhart,
Wells
and
Oatle~.(~)
For
the
full
capabilities
of
the
instrument
to
be realized, relatively noise-free amplification
of
the
modulated secondary electron
current
is
required.
As
this
current
ranges
from
to
A,
and
contains frequency
components
up
to
200
kc/s:
conventional thermionic
ampli-
fiers
cannot
comply with
the
low-noise
and
bandwidth
requirements.
Early workers
used
secondary
emission
electron
multipliers, but
these
were
not
very satisfactory,
electrically
or
mechanically.
General description
The
new
detector
(shown
in
Fig.
1)
consists
of
a cylindrical
brass
shield which
is
closed
at
the
end
facing the
specimen
by
a grid
of
copper
gauze (etched
in
nitric
acid
to
increase
the grid porosity)
and
is biased positively
in
order
to
attract
the
low-energy electrons.
Once
through
the grid,
the
electrons
are
accelerated
toward
the
hemisphere
of
plastic
scintillator, the
surface of which is covered with
a
700A
layer
of
aluminium (maintained
at
7
to
12
kV
positive).
The
intense electrostatic
field,
shaped
by
the focusing
electrode,
*
Now
at
Department
of
Electrical Engineering,
University
of
California,
U.S.A.
causes
most electrons
to
strike the hemisphere near
its
apex.
The
light
generated
in
the scintillator
is guided
by
a Perspex
light
pipe
to
a
commercial photomultiplier
tube
which then
converts
the
light back
to
an
electron
signal
and
provides
most
of
the required
amplification.
;-primary
electron
beom
1
volts)
\
\
\
copper enfry
gauze
%
Fig.
1,
Diagram
showing
collector
geometry
Theoretical performance
For
such
a system
to
be
reasonably
noise-free,
each
electron
entering the collector
must
liberate:
on
the
average,
at
least
one
electron
from
the photomultiplier
cathode. The
number
of
photoelectrons produced
by
one
zero-energy secondary
electron entering the
collector
is
given
by:
where
E,
=
energy
of
the electron striking
the
scintillator
E,
=
energy
lost
by
the
electron
in
passing
through
Cp
=
energy
conversion
efficiency
into photons
of
CO
=
efficiency
of
the optical
system,
C,
=
conversion
efficiency
of
the
photocathode,
surface
(eV),
the aluminium
film
(ev),
average energy
€,,
f(v)
=
spectral response
of
the
photocathode
to
E,
can
be
calculated
from the
Thompson-Whiddington
law:
photons
of
energy
Ep
(ev).
where
x
is
the
film thickness
in
cm,
a
is
4
x
10"
(eV)l
cni','g,
and
d
is the
density in
g'cmj.
JOURNAL
OF SCIENTIFIC
INSTRUkfESTS
246
VOL.
37,
JVLY
1960
WIDE-BAND
DETECTOR FOR MICRO-MICROAMPERE
LOW-ENERGY ELECTRON CURRENTS
In
the early
work
on
the
collector,
the
manufacturer's
figure
for
Cp
was used
and
CO
was
assumed
to
be
unity.
This
gave
n
equal
to
,2.2
for
E,
=
7
kV,
indicating
that
under
these conditions
no
noise
was
introduced.
It
was
found
in
practice
that
noise
was being
introduced
and
a
detailed investigation
of
Cp
and
CO
was
carried out.
Scintillator
efficiency
Several
worker^(^-^)
have reported that the
conversion
efficiency of
scintillators
falls
as
the
incident particle
energy
is
reduced
below
50
keV,
but
the
losses
quoted
vary
con-
siderably.
It
was
therefore
necessary
to
measure
Cp
at
the
energy levels
in
question.
Two makes of
plastic
scintillator
were investigated-
Pamelon,
made
by
Isotopes Development
Ltd.;
and Naton
11,
made
by
Nash and
Thompson
Ltd.
No
significant difference
in
performance
was detected.
A
10
kV
electron
beam
carrying
about
10-I2A
was
arranged
to
strike
the
scintillator
surface,
the potential
of
which
could
be
varied
between
-10
and -10kV.
It
was
found
that
the
response varied approximately linearly with
incident
energy
for
energies
above a threshold value
which
varied
between
2
and
10
kV.
Typical curves
are
shown
in
Fig.
2.
Further
investigation
showed
that the method
used
B
-
+I
.-
*'
:
sol
-
=,
~
,/'
c
2-
,
,e
,
,
I
,
,
,
C
i
8
I2
I6
20
:iciceiir
electron
enerqy
(ite?';'
Fig.
2.
Scintillator performance
Curve
(U),
wet
machined, French
chaik
polished: curve
(b),
dry
machined, polished
with
metal polish
to
prepare
the scintillator surface
was of
critical
importance
in
determining the performance
for
low
incident
energies.
Scintillators machined
under
water-cooled
conditions
and
polished with
French
chalk
in
water showed threshold
energies
of
2
to
2
'5
keV.
This
energy
is
that
required
io
penetrate the
700A
aluminium
film used
to
make the
scintillator surface conducting.
Dry
machining
and/or
polishing
with proprietary
brands
of
metal
polish
raised the
threshold
into
the
5
to
10
keV
range.
Any
slight
overheating
during
the
metallizing
had
a similar
deleterious effect
on
the
performance.
Measurements
with
a calibrated photomultiplier showed
that
the conversion
efficiency
of
a well
prepared scintillator
was
0.02
0.005
for
7
keV
electrons.
This figure agrees,
within
the
limits of experimental
error,
with
the
manu-
facturer's
value.
The
scintillator
did not
show
any-
saturation
effects
until
the
input current
density
reached
about
1
A,"?.
Under
normal
operating
conditions,
the
input
current
density
is
less
than
1 ,uA/cm2
so
saturation
is not
a
limiting
factor
in
the
performance
of
the
collector.
Optical
system
efficiency
The
optical
efficiency
of
the
scintillator-light-pipe
system
is
estimated
by
assuming
that
all
electrons
entering
the
collector
shield
(see
Fig.
1)
strike
the
scintillator
near the
apex of the
hemisphere.
The
critical
angle
Bc
of
the scintillator
material (measured
from
a
surface
normal
to
the ray
path)
is
40.5".
If
no
light
is absorbed
by
the
scintillator,
and
no
loss
occurs
at
surface reflexions,
then
by
using
a simple
geometrical
argument,
over
90%
of
the
generated light
is
found
to
enter
the
light-pipe.
Further, all
these light
rays
entering the light-pipe
subtend
an
angle
to
the
surface
normal
greater
than
the
critical angle
of
the
light-pipe material,
Perspex.
Thus in
the absence
of
absorption
in
the
light-pipe,
all light entering the
pipe
should reach
the
photomultiplier.
If
it is assumed
that
a
10%
loss
of
intensity occurs
at
each
surface reflexion
in
the
scintillator,
about
80%
of
the
generated
light enters the
light-pipe.
The
transmission
efficiency
of
the
light-pipes used was
measured
wjth
a special
neon
light
source
having
an
intensity
distribution
similar
to that
expected
from
a
scintillator
hemisphere.
The
light-pipes were
made
from
Q
in.
diameter
Perspex
rod cut
in
lengths
varying
from
9
to
11
cm.
The
ends were polished. Bends
of
45"
and
90"
with a radius
of
2.5
cm
were
made
on
some light-pipes
by
gentiy
warming
the
rod
in
a low bunsen
flame
and
then bending
to
the
desired
shape.
It
was
found
that
in
no
case
did
the
trans-
mission exceed
65
%,
and
if extreme
care
was
not
taken
with
the
surface finish,
the
transmission
fell below
40%.
The
bends did
not
increase
the
loss
by
more
than
10%
in
any
case. This loss, which is
rather
surprising
in
view
of
the
high
transparency
of Perspex, is believed
to
be
mainly caused
by
reflexion losses
on
entering
and
leaving
the
pipe.
A
drop
of
vacuum oil
placed
in
a joint
between
two
pipes
increased
the transmission
by
about
5
%
provided the oil did
not
spread
along
the
ferrule
supporting the joint.
If
this occurred,
the
loss was
greater
than for
a dry
joint.
Measured
performance
Putting
CO
=
0.4,
Cp
=
0.02
and
n
=
1 in the
transfer
equation
gives
E,
=
8
keV.
This
analysis
depends
on
the
accuracy
of
the
photomultiplier calibration
which
was
not
better
than
2OpL.
An
independent
check
on
the result
was
made
by
observing the
signal/noise
ratio
of
the
photo-
multiplier
output for
a constant
input
current,
with
the
scintillator
voltage varied
to
give values of
E,
between
0
and
20
keV. This gave
the curve shown
in
Fig.
3,
indicating
til
0
4
a
I2
I6
20
Wenfia;
OR
:he
scintiiiator
surfcce
(tV)
Fig.
3.
Signalpoise ratio
variation
with
scintillator potential
VOL.
37,
JCLY
1960
247
JOGR
NAL
OF SCIESTIFIC INSTRLMEhTS
WIDE-BAND
DETECTOR
FOR
MICRO-MICROAMPERE
LOW-ENERGY
ELECTRON
CURRENTS
that
the
noise is
effectively
constant
for
scintillator potentials
greater
than
10 kV.
(The
18
dB signal-to-noise
ratio
is
determined
by
the shot
noise
in
the
primary
electron beam
plus
the
noise introduced in
the
secondary emission process.)
Conclusions
A
wide-band
detector
for
micro-microampere secondary
electron
currents
has
been
constructed
and
tested. Noise-
free current
gains
ranging
from
lo5
to
IO8
have
been observed
with
a bandwidth exceeding 200 kcjs.
The
minimum
signal
which
can
be
detected is
set
by
the
dark
current
of
the
photo-
multiplier used.
For
the tube
used
in
this investigation,
an
E.M.I.
type
6094B,
this
dark
current
is
less
than
10-’jA,
and
it
could be
reduced
further
by
cooling
the photocathode.
The
maximum
bandwidth
is
set
by
the
decay time
of
the
scintillator
or
the transit
time
spread
of
the
photomultiplier,
or
both.
The
decay time
for
Pamelon
is
stated
to
be
less
than
s,
and
the transit time spread
for
thephotomultiplier
used is
about
10-8s.
Thus the
bandwidth
of
the
present
system
is
estimated
to
be greater
than
10
Mc/s.
As
used
in
the scanning
electron microscope,
this detector
offers
important
advantages.
The
mechanical flexibility is
much
better
than
that
of
the old
secondary electron detector.
A
movement
of
the
detector
input
requires
only a
new
piece
of
light-pipe bent
to
the proper shape
with
the
new
detector,
while previously
an
entire
new
brass
bottom
plate
for
the
specimen chamber
had
to
be constructed.
Previously,
the
input
of
the
electron multiplier
was
necessarily
near
ground potential, requiring the
output
to
be
approximately
6 kV positive. This
in
turn
required that
the
initial
stapes
of
valve amplification float
at
6 kV positive.
With
the
new
detector,
the
output
of
the
photomultiplier
can
be
at
ground
potential:
making a
direct-coupled
video-
amplifier possible.
Acknowledgements
We
wish
to
express
our
thanks
to
our
supervisor,
Mr.
C.
W.
Oatley,
for
much useful advice
and
encouragement.
One
of
us
(T.
E.
E.)
was
supported
by
a Marshall
Scholar-
ship, while
the
other (R.
F.
M.
T.)
was
in
receipt
of
a
D.S.I.R. Research Studentship.
References
(1)
MCIMULLAK,
D.
Proc.
bstn
Elect.
Engrs,
100,
Part
11,
p.
245 (1953).
(2)
SMITH,
K.
C.
A.,
and
OATLEY, C.
W.
Brif.
J.
Appl.
Phys.,
6,
p.
391
(1955).
(3)
EVERHART,
T.
E.,
WELLS,
0.
C.,
and
OATLEY,
C.
W.
J.
Elect.
&
ControI,
7,
p. 97
(1959).
(4)
TAYLOR,
C.
J.,
JENTSCHKE,
W.
K., REMLEY:
M.
E.,
EBY,
F.
S.,
and KRUGER,
P.
G.
Phys. Rev.,
84,
p.
1034
(1951).
(5)
BIRKS,
J.
B.,
and
KING,
J.
W.
Phys. Rev.,
86,
p.
568
(1952).
(6)
BELCHER,
E.
H.
Brit.
J.
Radiol.,
30,
p.
103
(1957).
LABORATORY
AND
WORKSHOP
NOTES
Modified
cylindrical
cell
for
the microelectrophoresis
of
suspended
particles
by
E.
J. AMBROSE,
M.A.,
and
J.
RYDER,
B.Sc., Chester
Beatty Research
Institute, Institute
of
Cancer Research:
Royal
Cancer
Hospital,
London,
S.W.3
[Note
received
6
January,
19601
Satisfactory chambers
for the
measurement
of
the
electrical
mobility
of very
small particles have been
known
for
some
time.(’)
An
apparatus
suitable
for
the
measurement
of
sus-
pended bacteria with
a rectangular chamber
(0.5
x
25
mm
cross-section) has been described
by
Loveday
and
James.(2)
In
these cases
the
particles
remain
continuously
in
suspension
and
they
can
always
be
observed
at
the
stationary
level,
where
the
true
electrophoretic mobility
can
be measured.
The
stationary
level lies
at
a distance
of
0.21
times
the total depth
of
chamber from the
chamber
wall
and
parallel
to
the
wall.
Larger particles sediment under gravity
and
their mobility
can
only be measured accurately
if they
are
allowed
to
fall
freely,
in the plane
of
the stationary
level, while
the
electric
field
is
applied
at
right angles
to
this direction.
The
rect-
angular chamber must
therefore be arranged vertically
so
that
the
particles can fall
through
the
whole
depth
of
the
cham-
ber.c3)
This
arrangement
is
reasonably satisfactory
but
is
most sensitive
to
small thermal changes
owing
to the
vertical
depth
(25
within which convection
currents can
arise.
The
location
of
the
large particles
(up to
20
p
in
diameter)
at
the stationary
level
in
the small
total depth
(500
p)
of
the
chamber
also presents
difficulties.
Bangham, Flemans,
Heard
and
Seaman(’)
described
an
apparatus
using
a
cylindrical
chamber
which has several
advantages.
The
height
of
the chamber
is only
equal
to
the
diameter
of
the
tube
(2.33
mm)
and
convection
currents
leading
to
drift
are
therefore
not
troublesome.
The depth
of
the
chamber
(2330
p)
is relztively much larger in
relation
to
the
diameter of
the
particles
(20
p).
The apparatus
is
satisfactory
for
moderately large particles, such
as
red
blood
cells,
but
larger particles sediment extremely rapidly
in
the
small
depth
of
the
tube
and
it is difficult
to take
more
than
ten
readings
after
filling
the
chamber, even when working
at
great
speed.
To
enable more
readings
to
be
taken,
and
with
greater accuracy,
a method
of
resuspending
the
cells
in
the
chamber has
been devised.
A
small piece
of
Mumetal
(approximately
3 mm
long)
sealed
into
thin-walled capillary
tubing
(Fig.
1)
is
placed
in
Fig.
1.
Enlarged
view
of
stirrer
Y,
Mumetal;
q,
thin
glass
the
electrophoresis
chamber.
It
is
drawn along through
the
chamber
by
a magnet (Fig.
2).
Its
passage stirs
up
the
sus-
pension
and
resuspends
the
cells
without disturbing the
thermal
equilibrium.
The
magnet is moved
along a metal
JOURNAL
OF
SCIENTIFIC
INSTRUMENTS
248
VOL.
31,
JULY
1960