of 4
IEEE
Transactions
on
Nuclear
Science,
Vol.
33,
No.
1,
February
1986
LOW
ENERGY
X-RAY
SPECTRA
MEASURED
WITH
A
MERCURIC
IODIDE
ENERGY
DISPERSIVE
SPECTROMETER
IN
A
SCANNING
ELECTRON
MICROSCOPE
J.S.
Iwanczyk,
A.J.
Dabrowski,
G.C.
Huth
University
of
Southern
California
Institute
of
Physics
4676
Admiralty
Way,
Suite
932
Marina
del
Rey,
CA
90292
J.G.
Bradley,
J.M.
Conley
Jet
Propulsion
Laboratory
California
Institute
of
Technology
4800
Oak
Grove
Drive
Pasadena,
CA
91109
A.L.
Albee
California
Institute
of
Technology
1201
East
California
Blvd.
Pasadena,
CA
91125
ABSTRACT
A
mercuric
iodide
energy
dispersive
x-ray
spectrometer,
with
Peltier
cooling
provided
for
the
detector
and
input
field
effect
transistor,
has
been
developed
and
tested
in
a
scanning
electron
microscope.
X-ray
spectra
were
obtained
with
the
15
keV
electron
beam.
An
energy
resolution
of
225 eV
(FWHM)
for
Mn-Kac
at
5.9
keV
and
195
eV
(FWHM)
for
Mg-K
line
at
1.25
keV
has
been
measured.
Overall
system
noise
level
was
175
eV
(FWHM).
The
detector
system
characterization
with
a
carbon
target
demonstrated
good
energy
sensitivity
at
low
energies
and
lack
of
significant
spectral
artifacts
at
higher
energies.
INTRODUCTION
Mercuric
iodide
x-ray
spectrometery
has
been
advanced
considerably
in
the
last
several
years
(1-10).
The
exceptionally
low
room
temperature
leakage
current
which
is
possible
with
HgI2,
combined
with
the
good
transport
properties
of
electrons,
have
formed
a
good
basis
for
high
energy
resolution
x-ray
detectors
capable
of
operation
at
room
temperature
or
slightly
below.
Because
there
is
no
need
for
the
cryogenic
coolant
and
its
associated
vacuum
cryostat,
the
design
of
compact
and
lightweight
detection
system
is
possible.
This
simplicity
and
considerable
size
advantage
can
be
important
for
terrestial
applications
as
well
as
for
space
exploration.
Mercuric
iodide
detectors
have
previously
demonstrated
300
eV
(FWHM)
energy
resolution
for
the
5.9
keV
Mn-Ka
line
from
an
Fe-55
source
and
245
eV
(FWHM)
for
the
1.25
keV
Mg-K
line
with
both
the
detector
and
preamplifier
operated
at
room
temperature
(1-3).
The
sensitivity
of
HgI2
detector
have
also
been
shown
for
ultra
soft
x-ray
below
1
keV.
The
characteristic
x-ray
peak
of
oxygen
at
523
eV
was
clearly
resolved
from
noise
(9).
With
the
input
field
effect
transistor
(FET)
cooled
to
its
optimum
temperature
of
approximately
1400
K
and
the
HgI2
detector
at
room
temperature,
a
resolution
of
175
eV
(FWHM)
for
1.5
keV
(Al-K)
has
been
demonstrated
(7).
These
latter
resolution
figures
are
comparable
to
those
achieved
with
commercial
Si(Li)
detector
systems.
Currently
work
is
underway
to
achieve
comparable
HgI2
resolution
values
with
the
preamplifier
input
FET
cooled
using
a
thermoelectric
(Peltier)
cooler.
Work
on
laboratory
research
prototype
version
of
the
Scanning
Electron
Microscope
and
Particle
Analyser
(SEMPA)
instrument
at
The
Jet
Propulsion
Laboratory
(JPL)
provided
motivation
for
development
of
a
minature,
lightweight
and
low
power
HgI2
x-ray
spectrometer
for
scanning
electron
microscope
(SEM)
(11,12).
The
SEMPA
instrument
is
proposed
as
one
of
the
instruments
for
a
Mariner
Mark
II
mission
to
rendezvous
with
a
comet
in
the
1990's
(13).
As
proposed,
the
spacecraft
would
travel
with
the
comet
for
several
years
while
periodically
performing
analyses
to
provide
information
on
the
composition
and
character
of
the
comet
dust.
The
SEMPA
instrument
would
collect
solid
comet
dust,
image
individual
grains
and
collect
x-ray
spectra
for
quantitative
determination
of Na
and
key
elements
with
higher
atomic
number,
expected
to
be
in
the
minerals
and
rocks.
The
total
actual
analytical
time
for
imaging
and
x-ray
analysis
is
expected
to
be
about
1000
hours.
The
requirements
of
such
an
interplanetary
mission
place
severe
constraints
on
the
selection
of
analytical
instruments,
including
the
choice
of
an
x-ray
detector
for
SEMPA.
The
use
of
a
Si(Li)
detector
would
require
the
use
of
a
costly
radiative
cooler
to
achieve
the
required
near
liquid
nitrogen
temperatures.
There
is
evidence
that
a
Si(Li)
detector
system
can
be
operated
at
degraded
performance
level
with
thermoelectrical
cooling,
however
high
electrical
power
and
heat
dissipation
capacity
would
be
needed
(14,15).
Therefore
the
use
of
a
HgI2
detector
was
identified
as
a
good
choice
to
minimize
power
consumption
and
weight
of
the
SEMPA
instrument
since
the
preamplifier
input
FET
and
detector
would
clearly
need
to
be
cooled
using
only
small
thermoelectric
coolers.
A
series
of
experiments
have
been
conducted
using
a
mercuric
iodide
energy
dispersive
x-ray
spectrometer
installed
in
the
JPL
SEMPA
research
prototype
instrument.
These
are
continuing
experiments
designed
to
study
and
improve
such
factors
as
obtainable
x-ray
energy
resolution,
effects
of
detector
positioning
and
proximity
to
the
target,
and
optimization
of
thin
filters
in
front
of
the
detector
to
eliminate
unwanted
backscattered
electrons.
EXPERIMENTAL
SET
UP
An
outline
drawing
of
the
SEMPA
target
chamber
is
presented
as
Figure
1.
The
detector
and
first
stage
of
field
effect
transistor
preamplification
are
inside
of
the
flanged
housing
(to
the
left
in
the
views)
and
thus
extend
into
the
microscope
vacuum.
The
vacuum
flange
and
electrical
feedthroughs
lead
to
the
subsequent
stages
of
amplification
which
are
housed
externally.
0018-9499/86/0200-0355$01.00(1986
IEEE
355
I)3
m
Thile
liter
iH
Abso
rber
iAIOlaeasiess
Is
cies)
Figure
1.
Cross
sectional
view
of
SEMPA
instrument
target
chamber
Tne
mercuric
iodide
detector
and
associated
pulsed
light
feedback
preamplifier
are
outlined.
The
detectors
in
these
experiments
had
2
to
5
mm2
active
area
and
were
mounted
on
a
single
stage
Marlow
MI
1021
electrical
Peltier
element
so
the
detector
could
be
slightly
cooled
to
the
temperature
of
about
00
C.
The
detector
angle
relative
to
the
electron
beam
was
400
and
the
distance
between
the
target
and
the
HgI2
detector
plane
was
about
10
mm.
The
solid
angle
of
collection
was
0.02
to
0.05
steradians.
The
input
FET
was
cooled
with
a
three
stage
Peltier
element
Marlow
MI
3026
to
approximately
-
400
C.
This
is
still
not
cold
enough
for
optimum
noise
reduction
-for
this
type
of
silicon
transistor
(2N4416),
that
temperature
being
about
-120
C.
The
total
power
supplied
to
the
detector-preamplifier
system
including
both
Peltier
coolers
was
about
3
Watts.
A
metal
shield
formed
from
thin
stainless
steel
covered
the
front
end
assembly
containing
the
detector
and
first
stage
FET
within
the
target
chamber.
A
thin
section
of
beryllium
was
placed
between
target
and
detector
to
absorb
unwanted
backscattered
electrons.
The
construction
allowed
for
easy
replacement
of
absorbers.
Experiments
were
performed
with
two
different
thicknesses
of
beryllium
sections
of
12
pm
and
8
pm.
The
x-ray
spectra
were
obtained
with
15
keV
electron
beam
provided
by
SEMPA
instrument.
The
current
of
the
beam
was
kept
below
few
nanoampers.
The
metal
targets
used
were
high
purity
polished
standards.
The
carbon
target
was
a
commercial,
unpolished,
SEM
sample
mounting
stub.
Electrical
signals
from
detector-preamplifier
system
were
fed
to
a
Canberra
2020
amplifier
utilizing
a
12
us
shaping
time
constant.
No
pileup
rejection
or
reset
blanking
was
used.
The
pulse
height
analyzer
was
a
Tracor
TN
1242
in
the
combination
with
TN
4000
system
in
the
standard
configuration
as
for
Si(Li)
detectors.
The
standard
Tracor
programs
were
used
for
peak
identification
and
energy
resolution
calculations.
X-RAY
SPECTROSCOPY
Figure
2
shows
the
spectrum
obtained
from
a
copper
target.
Three
lines
Cu
L
(0.93
keV),
Cu
Ka
(8.05
keV)
and
Cu
Ka
(8.90
keV)
are
clearly
visible.
The
L
line
of
copper
is
seen
with
it's
intensity
diminished
by
an
8
um
beryllium
absorber
and
due
to
the
detector
carbon
entrance
electrode.
The
energy
resolution
of
the
Ka
line
is
230
eV
(FWHM).
Figure
3
shows
the
Ka
and
K1
lines
manganese
at
5.90
keV
and
6.49
keV,
respectively.
In
this
case
energy
resolution
is
225
eV
(FWHM)
for
the
Mn
Ka
peak.
Figure
4
shows
the
K
line
of
the
spectrum
obtained
from
magnesium
target.
The
195
eV
(FWHM)
energy
resolution
is
the
first
obtained
below
200
eV
using
Peltier
coolers.
Also
shown
in
Figure
4
is
a
pulser
peak.
A
measurement
of
the
electronic
noise
level
of the
mercuric
iodide
spectrometer
made
by
the
pulser
method
indicated
a
value
of
175
eV
(FWHM).
All
measured
x-ray
peaks
show
excellent
symmetry.
The
shape
and
intensity
of
background
counts
is
typical
for
electron
excitation.
Within
acquired
counting
statistics,
I
or
Hg
escape
peaks
(AE-3.9
keV
and
2.2
keV,
respectively)
are
not
visible
in
the
spectra.
This
is
expected
from
the
fact
that
the
escape
peaks
would
be
from
iodine
L
and
mercury
M
levels
which
have
very
small
fluorescence
yields
(16).
The
energy
resolution
of
the
x-ray
peaks
are
significantly
improved
over
values
obtained
in
the
previous
initial
experiments
in
the
scannings
electron
microscope
(8).
L1,
1WIL
LEAit
L
KLL
WI
n.
LIJJJ
_LtA
III,,I,
Hgl2
SPECTROMETER
COPPER
TARGET
Cu
Ka
8.05
keV
2_
C)k
FWHA
230
e
Cu
L
0.93
keV
IV
0
1
2
3
4
5
6
7
8
9
10
ENERGY
(keV)
Figure
2.
X-ray
spectrum
obtained
from
copper
target
356
357
ii
.
..
..
I,
.,.I.I,
1
2
3
4
5
6
7
8
9
ENIERGY
(keV)
Figure
3.
X-ray
target
0
10
spectrum
obtained
from
manganese
0
1
2
3
4
5
6
7
8
9
10
ENERGY
(keV)
Figure
4.
X-ray
spectrum
obtained
from
magnesium
target.
(Electronic
pulser
indicates
system
noise)
The
Fano
factor
calculated
using
energy
resolution
230
eV
(FWHM)
for
8.05
keV
(Cu-Ka
line)
and
pulser
width
175
eV
(FWHM)
gives
a
value
of
0.12.
This
value
is
similar
to
the
lowest
previously
reported
for
HgI2
(7)
and
also
close
to
the
experimental
value
of
Fano
factor
obtained
for
silicon.
A
lower
measured
value
for
the
Fano
factor
reflects
a
lower
contribution
from
the
trapping
phenomena
in
the
x-ray
spectra
and
indicates
that
electronic
noise
is
limiting
factor
in
the
obtainable
energy
resolution.
DETECTOR
CHARACTERIZATION
WITH
A
CARBON
TARGET
Every
energy
dispersive
detector
system
creates
artifacts
in
an
x-ray
spectrum
that
are
characteristic
of
the
detector
rather
than
the
excited
target.
The
most
obvious
of
these
are
limited
efficiency
for
very
high
and
low
energies
because
of
the
presence
of
absorbing
windows
and
contacts,
and
finite
detector
thickness.
More
subtle,
but
always
present
to
some
degree
are
escape
peaks,
absorption
edges
and
secondary
fluorescence
peaks
due
to
the
crystal
properties
and
materials.
One
convenient
way
to
partially
characterize
an
SEM
detector
system
for
energy
response
and
artifacts
is
to
observe
the
spectrum
generated
by
a
carbon
target.
The
carbon
characteristic
K
line
is
at
0.282
keV
so
the
spectrum
above
that
energy
is
entirely
Bremstrahlung
continuum
if
no
heavier
elements
are
present.
The
shape
of
the
energy
spectrum
of
the
continuum
leaving
the
sample
is
calculable,
so
any
observed
differences
are
due
to
the
detector.
Figure
5
shows
the
spectrum
from
a
carbon
target
measured
for
one
of
the
HgI2
detectors
used
in
experiments.
The
spectrum
contains
1.6
X
106
counts.
The
shape
of
the
spectrum
above
2
keV
is
the
smooth
shape
expected
from
100%
efficiency
detection
of
continuum
x-rays.
Below
2
keV
the
detected
intensities
are
greatly
reduced
by
absorption
in
the
12
pm
thick
Be
window
and
detector
front
carbon
contact.
Calculation
using
tabulate
a
mass
absorption
coefficients
indicates
that
50%
of
the
detection
efficiency
decrease
was
due
to
the
Be
window
and
50%
due
to
the
carbon
entrance
electrode.
The
detector's
net
efficiency
at
1
keV
is
about
10%.
The
only
other
spectral
artifact
detectable
above
1
keV
in
this
spectrum
is
the
"peak"
at
1.75
keV.
This
may
have
been
due
to
secondary
fluorescence
of
Si
in
the
silicone
rubber
near
the
detector
active
area,
or
due
to
Si
in
the
carbon
target.
There
are
no
Hg
or
I
lines
or
absorption
edges
that
could
produce
a
peak
at
this
location.
The
largest
absorption
edge
jump
should
be
due
to
Hg-M
at
2.4
keV,
but
no
such
feature
is
observable
in
this
detector.
This
indicates
a
very
thin
HgI2
deadlayer.
An
approximate
computer
model
of
the
continuum
and
the
effect
of
absorbers
and
deadlayer
has
been
created.
The
model
confirmed
the
discrete
energy
hand
calculation
of
the
absorber
effect,
and
VI-
0
2-
l
LI
A
I
ALlIA
1
,.1.1
11.
i
1.1
LI
J
-L
All
LII.
LI
_ULll
Hgl2
SPECTROMETER
CARBON
CONTINUUM
r-FT
T-
-TJ
-r-FT-{F-rrT--Tl--T
T
0
1
2
3
4
5
6
ENERGY
(keV)
7
8
9
10
Figure
5.
X-ray
spectrum
obtained
from
carbon
target
(/)
z
0
C.
HgI
2
SPECTROMETER
Mn
Ka
5.90
keV
Mn
TARGET
FWHM
225
eV
Mn
K,B
6.49
keV
It
f
1-
-1
T
l
I
0
0
C.)
1
L
.