Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~bradw/cv/papers/NeKLL.ps.gz Äàòà èçìåíåíèÿ: Thu Jan 18 00:56:02 2001 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 03:11:03 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http astrokuban.info astrokuban

Dielectronic satellite contributions to Ne VIII and Ne IX K­shell spectra
B. J. Wargelin* and S. M. Kahn +
Department of Physics and Space Sciences Laboratory, University of California, Berkeley, California 94720
P. Beiersdorfer
Lawrence Livermore National Laboratory, 7000 East Avenue, Livermore, California 94550
#Received 8 August 2000; published 17 January 2001#
K# spectra of heliumlike neon and associated lithiumlike, berylliumlike, and boronlike satellite line emis­
sion have been observed with a high­resolution crystal spectrometer on the Lawrence Livermore Electron
Beam Ion Trap. The KLL dielectronic recombination satellites were resolved from their He­like parent lines in
electron energy space, and their wavelengths and resonance strengths measured. The wavelength measure­
ments achieved a typical accuracy of a few må , with two measurements accurate to better than one part in
10 000. By normalizing to the He­like resonance line, w, we measure Li­like satellite resonance strengths that
are 10% to 46% lower than predicted by theoretical models. The wavelengths and relative strengths of Be­like
KLL satellites were also measured, and absolute strengths were obtained by normalizing to the collisionally
excited Li­like qr satellite blend.
DOI: 10.1103/PhysRevA.63.022710 PACS number#s#: 32.80.Hd, 32.30.Rj, 32.70.Fw, 34.80.Kw
I. INTRODUCTION
In addition to its role in determining a plasma's charge
balance, dielectronic recombination #DR# is important in
x­ray emitting astrophysical and laboratory plasmas because
of the resulting satellite emission lines. These satellites often
blend with their parent emission lines or lines from other
ions, thus affecting those lines' apparent intensities. Like­
wise, the closely related process of resonant excitation can
lead to enhanced emission of the primary emission lines
themselves. These resonance phenomena are particularly im­
portant for He­like ions, whose K# lines have several diag­
nostic uses, including the determination of electron density,
temperature, and element abundance. The relative intensities
of the satellites can also, when resolvable, provide informa­
tion on the electron temperature, or deviations from a Max­
wellian energy distribution.
From an astrophysical point of view, the He­like neon
spectrum is particularly rewarding to study because of its
electron density diagnostic, which is most sensitive at
roughly 10 12 cm #3 . This is just above the typical density of
stellar coronae (#10 10 cm #3 ) , and a good match for the
densities in stellar flares, which can range up to several times
10 13 cm #3 . The He­like neon spectrum is, however, chal­
lenging to study because it overlaps with the ubiquitous iron
L­shell emission. Even in the high­resolution spectra ob­
tained with the Chandra and XMM­Newton x­ray observato­
ries, great care must be used to determine the contributions
of individual lines. This requires accurate knowledge of line
wavelengths and the relative intensities of primary emission
lines and their satellites.
Much of the confusion associated with line blending can
be alleviated by using an electron beam ion trap #EBIT#,
which allows study of collisional emission spectra as a func­
tion of wavelength and electron collision energy. Measure­
ments are therefore level­specific, permitting the determina­
tion of the wavelength, excitation energy, and strength of
individual resonances. The degree of completeness and detail
available in such measurements often exceeds that provided
in theoretical studies, which may not list all the relevant
autoionization and radiative rates needed for proper compari­
son. Indeed, published results usually integrate over all en­
ergies and sum individual satellites together.
This paper presents results from an EBIT study of the
lowest­energy DR resonances in He­like, Li­like, and Be­like
neon, and is part of a larger program to experimentally char­
acterize K# line emission from heliumlike neon as com­
pletely as possible. Measurements of the radiative lifetime of
the metastable 1s2s 3 S 1 level, the basis of the important den­
sity diagnostic, have already been reported #1,2#. Future pa­
pers will present results on the measured cross sections for
collisional transfer from the metastable level, resonant exci­
tation and near­threshold DR, inner­shell ionization of Li­
like Ne #which populates the He­like metastable level#, and
direct electron impact excitation.
II. THEORY
A. Overview of He­like K# lines and their satellites
As shown in Fig. 1, there are six levels in the He­like
1s2l configuration, four of which have single­photon radia­
tive decays of the form
1s2l#1s 2 #h# , #1#
giving rise to the so­called resonance line w (1s2p 1 P 1
#1s 2 1 S 0 ) , intercombination lines x and y (1s2p 3 P 2,1
#1s 2 1 S 0 ) , and forbidden line z (1s2s 3 S 1 #1s 2 1 S 0 ) ,
where we have used the w , x , y , z notation of Gabriel #3#.
*Present address: Harvard­Smithsonian Center for Astrophysics,
60 Garden Street, Cambridge, MA 02138.
+ Present address: Department of Physics, Columbia University,
538 West 120th Street, New York, NY 10027.
PHYSICAL REVIEW A, VOLUME 63, 022710
1050­2947/2001/63#2#/022710#11#/$15.00 ©2001 The American Physical Society
63 022710­1

Satellite lines are produced when at least one extra spec­
tator electron is present, in transitions of the form
1s2ln#l##1s 2 n#l##h##. #2#
The spectator n#l# electron partially shields the nuclear
charge, so that satellite lines are slightly shifted toward
longer wavelengths relative to their parent lines. The shift is
largest for n##2 spectator electrons and approaches zero as
n### . Satellite intensity decreases with increasing n#; this
paper addresses only the strongest, 2 l # satellites.
There are 16 levels in the Li­like 1s2l2l# configuration,
with 22 dipole­allowed 2#1 transitions to the three Li­like
1s 2 2l ``ground'' levels. Gabriel #3# labels these satellite
lines with the letters a through v . Similarly, there are 30
levels of the type 1s2l(2l#) 2 in Be­like ions and 35
1s2l(2l#) 3 levels in B­like, which can, respectively, decay
to 10 and 15 ground levels, producing a total of 102 and 217
dipole­allowed #n#1 satellite transitions. Only a fraction of
these lines are strong enough to be observed, and the main
satellites in Li­like, Be­like, and B­like neon are grouped
around three or four wavelengths for each ion.
Multiply excited states such as those described above can
either #i# emit a satellite line via a radiatively stabilizing
transition to a bound #nonautoionizing# state, or #ii# autoion­
ize #undergo Auger emission#, in which one of the excited
electrons falls to a lower­energy subshell while another ex­
cited electron carries away the energy of that transition by
being ejected into the continuum. The net emission cross
section for a satellite line is therefore equal to the cross sec­
tion for excitation of the state that emits the line times the
fluorescence yield, W, which is the probability that the state
decays radiatively and emits the line. For excitation of a
satellite line s from an initial state #i # which at low densities
is virtually always the ground state, to an intermediate au­
toionizing state #s# and then to a final state # f # by radiative
decay, this net cross section can be written as
# s # E ### is # E #W s f , #3#
where E is the energy of the colliding electron.
Several transitions may be possible from an excited state.
The fluorescence yield for a particular satellite line is given
by
W s #
A s f
rad
# j
A s j
auto
##
k
A sk
rad
, #4#
where A s f
rad is the rate #in sec #1 ) for radiative decay from
state #s# to state # f # , the second summed term in the denomi­
nator represents the total rate of radiative decay from state
#s# , and the first sum extends over all levels which are popu­
lated by autoionization of level #s# #for instance, a 1s2lnl#
state autoionizes to 1s 2 , while a 1s3lnl# state can autoionize
to 1s2l or 1s 2 ) . Note that for singly excited #nonautoioniz­
ing# states, the fluorescence yield simply reduces to a radia­
tive branching ratio.
B. Dielectronic recombination
There are three ways to excite an autoionizing state via
electron collisions: inner­shell ionization, inner­shell excita­
tion, and dielectronic recombination #DR#. Each of these
processes tends to favor the production of a different set of
autoionizing states, although individual states can often be
excited in more than one way. Inner­shell ionization of a
Be­like ion creates an autoionizing Li­like ion (1s2s 2 ) , and
ionization of a B­like ion leaves a Be­like ion (1s2s 2 2p) .
The fluorescence yields from those excited configurations,
however, are very low and satellite emission is negligible.
Inner­shell ionization of Li­like ions, in contrast, often leads
to the emission of the He­like forbidden line, z.
Likewise, innershell excitation of a Li­like ion to a
1s2sn#l# level, particularly in the 1s2s2p configuration,
produces significant satellite line emission, primarily from
the q and r blend #see Table I#, which has a combined exci­
tation cross section nearly as large as that of the He­like
parent line, w. This direct collisional excitation occurs only
above a threshold energy, which is slightly below that nec­
essary for excitation of the parent He­like line due to the
shielding effect of the n##2 electron. Satellites can also be
produced when the colliding electron has an energy below
the direct excitation threshold via dielectronic recombina­
tion.
The first step in the DR process is dielectronic capture,
which is like radiative recombination except that instead of
carrying away the recombination energy via a photon, one of
the initially bound electrons is excited to a higher­energy
level. Since these energy levels are quantized, this is a reso­
nant process, i.e., it can only occur at discrete energies. DR
resonances are usually labeled using the notation for the in­
verse autoionization process, e.g., KLM denotes the DR
resonance in which a K­shell electron is excited to the L shell
by the capture of a continuum electron into the M shell. After
dielectronic capture, the ion is left in a multiply excited state
which will, as explained earlier, either decay radiatively to a
FIG. 1. Energy­level diagram showing ground state and first
excited states (1s2l) of a He­like ion. Line w is the resonance line,
x and y are the intercombination lines, and z is the forbidden line.
Energy differences are not to scale; all excited states are much
closer to each other than to the ground state. Satellite lines have
essentially the same transitions but with one or more spectator elec­
trons.
B. J. WARGELIN, S. M. KAHN AND P. BEIERSDORFER PHYSICAL REVIEW A 63 022710
022710­2

TABLE
I.
Theoretical
atomic
data
for
KLL
dielectronic
satellite
transitions
1s2l2l##1s
2
2l#
in
Li­like
Ne
7#
.
Transitions
are
labeled
as
by
Gabriel
#3#
using
LS
coupling
notation,
and
with
j
marked
for
the
p
electrons
of
the
upper
level,
as
given
by
Nilsen
#19#.
E
res
is
the
resonance
energy,
using
the
average
of
the
values
of
Chen
#12#
and
Karim
and
Bhalla
#13#,
which
agree
to
within
1.3
eV
for
all
lines.
P
is
the
line
polarization
and
G
s
/G
w
is
the
spectrometer
response
function
for
line
s
relative
to
that
for
w.
S
is
the
transition
strength,
in
units
of
10
#20
cm
2
eV;
negative
numbers
in
brackets
indicate
powers
of
10.
Line
Transition
E
res
P
a
G
s
/G
w
#
b
(å)
#
c
(å)
#
d
(å)
#
e
(å)
S
b
S
c
S
d
S
e
a
1s2p
3/2
2
2
P
3/2#1s
2
2p
2
P
3/2
683.70
#0.75
0.387
13.675
13.6755
13.6995
13.6702
0.652
0.384
0.452
0.583
b
1s2p
3/2
2
2
P
3/2#1s
2
2p
2
P
1/2
683.70
#0.60
0.939
13.672
13.6724
13.6967
13.6667
0.110
0.065
0.076
0.144
c
1s2p
1/2
2p
3/2
2
P
1/2#1s
2
2p
2
P
3/2
683.50
0
0.632
13.678
13.6785
13.7026
13.6732
7.5##4#
2.5##3#
9.2##4#
2.8##4#
d
1s2p
1/2
2p
3/2
2
P
1/2#1s
2
2p
2
P
1/2
683.50
0
0.632
13.675
13.6755
13.6998
13.6697
1.6##3#
5.2##3#
1.9##3#
5.7##4#
e
1s2p
3/2
2
4
P
5/2#1s
2
2p
2
P
3/2
673.88
0.50
0.867
13.819
13.8332
13.8557
13.8187
3.6##3#
2.3##3#
2.8##3#
2.1##3#
f
1s2p
1/2
2p
3/2
4
P
3/2#1s
2
2p
2
P
3/2
673.52
#0.75
0.379
13.820
13.8347
13.8573
13.8211
1.4##3#
2.0##4#
1.0##3#
2.6##4#
g
1s2p
1/2
2p
3/2
4
P
3/2#1s
2
2p
2
P
1/2
673.52
#0.60
0.928
13.8316
13.8175
5.6##6#
4.5##4#
h
1s2p
1/2
2
4
P
1/2#1s
2
2p
2
P
3/2
673.43
0
0.623
13.8363
13.8226
2.0##8#
9.4##8#
i
1s2p
1/2
2
4
P
1/2#1s
2
2p
2
P
1/2
673.43
0
0.623
13.819
13.8332
13.8559
13.8190
1.6##4#
5.1##7#
1.7##4#
2.1##6#
j
1s2p
1/2
2p
3/2
2
D
5/2#1s
2
2p
2
P
3/2
681.95
#0.50
0.875
13.711
13.7104
13.7344
13.6969
8.783
7.295
8.674
7.643
k
1s2p
1/2
2p
3/2
2
D
3/2#1s
2
2p
2
P
1/2
681.97
#0.60
0.936
13.707
13.7062
13.7305
13.6936
5.299
4.391
5.226
3.782
l
1s2p
1/2
2p
3/2
2
D
3/2#1s
2
2p
2
P
3/2
681.97
#0.75
0.385
13.710
13.7093
13.7333
13.6971
0.610
0.494
0.584
1.366
m
1s2p
3/2
2
2
S
1/2#1s
2
2p
2
P
3/2
693.84
0
0.639
13.539
13.5293
13.5500
13.5225
1.584
1.395
1.538
1.298
n
1s2p
3/2
2
2
S
1/2#1s
2
2p
2
P
1/2
693.84
0
0.639
13.536
13.5263
13.5473
13.5190
0.699
0.621
0.689
0.579
o
1s2s
2
2
S
1/2#1s
2
2p
2
P
3/2
651.40
0
0.595
14.163
14.1842
14.1956
14.1692
0.189
0.168
0.234
0.174
p
1s2s
2
2
S
1/2#1s
2
2p
2
P
1/2
651.40
0
0.596
14.160
14.1809
14.1926
14.1654
0.098
0.087
0.121
0.090
q
1s(2s2p
3/2
3
P)
2
P
3/2#1s
2
2s
2
S
1/2
667.99
#0.60
0.936
13.652
13.6681
13.6759
13.6551
4.073
5.887
5.591
4.576
r
1s(2s2p
3/2
3
P)
2
P
1/2#1s
2
2s
2
S
1/2
667.84
0
0.633
13.654
13.6705
13.6779
13.6571
2.455
3.074
3.070
2.470
s
1s(2s2p
3/2
1
P)
2
P
3/2#1s
2
2s
2
S
1/2
675.62
#0.60
0.942
13.564
13.5655
13.5796
13.5452
1.025
1.351
1.095
1.205
t
1s(2s2p
3/2
1
P)
2
P
1/2#1s
2
2s
2
S
1/2
675.59
0
0.638
13.565
13.5655
13.5800
13.5461
0.637
0.773
0.663
0.692
u
1s2s2p
1/2
4
P
3/2#1s
2
2s
2
S
1/2
655.99
#0.60
0.922
13.835
13.8537
13.8692
13.8356
9.0##4#
3.7##4#
7.2##4#
5.4##4#
v
1s2s2p
1/2
4
P
1/2#1s
2
2s
2
S
1/2
655.92
0
0.622
13.836
13.8544
13.8699
13.8371
2.4##4#
2.4##4#
8.0##5#
1.1##4#
a
Inal
and
Dubau
#9#.
b
Vainshtein
and
Safronova
#8#.
c
Chen
#12#.
d
Nilsen
#19#.
e
Karim
and
Bhalla
#13#.
DIELECTRONIC SATELLITE CONTRIBUTIONS TO Ne VIII . . . PHYSICAL REVIEW A 63 022710
022710­3

singly excited state, resulting in dielectronic recombination
and the retention of the additional electron, or autoionize,
resulting in either resonant excitation or resonant elastic scat­
tering.
Cross sections for a resonant process such as DR are of
course extremely energy­dependent, with Lorentzian shapes
and natural widths of less than 1 eV. For satellite lines ex­
cited by DR, it is therefore more useful to speak of an inte­
grated cross section, or resonance strength. If we define the
satellite resonance strength S s as the integral over all ener­
gies of cross section # s (E) for emission of line s via the
process #i ###s### f # , then Eq. #3# becomes
S s ##
0
#
# s # E #dE# # 0
#
# is # E #W s dE#S is W s . #5#
This simply states that the satellite resonance strength is
equal to the dielectronic capture resonance strength, S is ,
times the fluorescence yield.
Because dielectronic capture is the inverse of autoioniza­
tion, S is is proportional to A si
auto by the principle of detailed
balance. The full equation is
S is #
# 2
k i
2
g s
g i
A si
auto #
2# 2 a 0
3
R 2
E res
# m e
2R y
g s
g i
A si
auto , #6#
where k i is the electron wave number corresponding to the
resonance energy, g s is the statistical weight of the autoion­
izing level #equal to 2J s #1) , g i is the weight of the initial
level #equal to 1 for the He­like ground state, 2 for Li­like#,
a 0 is the Bohr radius, R is the Rydberg energy, and m e is the
electron mass. It is common to refer to a satellite intensity
factor, defined as
Q s #g s A si
auto W s #
g s A si
auto A s f
rad
# j
A s j
auto
##
k
A sk
rad
sec #1 , #7#
so that Eq. #5# may be written as
S s ##2.475#10 #30 #
Q s
g i E res
cm 2 eV, #8#
where E res is in eV.
Note that DR satellites are strongest when their upper
levels have equal and large radiative and autoionization
rates. Because most autoionization rates are all roughly the
same (10 12 -- 10 14 sec #1 ) , S s is primarily determined by the
radiative rate. Radiative rates are not much affected by spec­
tator electrons, so lines emitted from states with short life­
times #large A rad ) , such as the electric dipole lines w, y, and
q, have strong satellites, while line z, a magnetic dipole de­
cay from the long­lived 3 S 1 state, and line x, a magnetic
quadrupole transition #blended with y) , have essentially no
satellites.
III. EXPERIMENT
A. EBIT, spectrometer, and data acquisition system
Neon K# spectra were collected on the Livermore EBIT
#4# with a low­energy flat­crystal spectrometer #5#. The de­
tector is a position­sensitive proportional counter and the
spectrometer has a net resolving power of #500 when using
a thallium acid phthalate #TAP# crystal. In this experiment,
neutral neon atoms are continuously injected into the central
trap region of EBIT where they are ionized and then electro­
statically confined #radially# by a 70­#m­diam electron
beam. A set of three electrodes provides longitudinal con­
finement. The nearly monoenergetic beam ##75 eV full
width at half maximum #FWHM## can be quickly raised and
lowered, permitting study of the plasma while ionizing, re­
combining, or in equilibrium. For our experiment, the beam
energy was ramped between 600 and 1200 eV with a 2.5­
msec­period triangular wave form, reaching from the lowest
DR resonances up to the He­like ionization threshold at 1196
eV. For every detected photon, the time and electron beam
energy were recorded using the EBIT event­mode data ac­
quisition system.
The period of the beam­energy sweep is comparable to
the time scale for ionization of Li­like ions to the He­like
state (#0.5 msec with an average electron beam density of
4#10 12 cm #3 ) , but the ionization cross section #with thresh­
olds of 239 eV for direct 2s ionization, and #900 eV for the
various inner­shell excitation­autoionization channels# is
constant to #8% over this energy range. Charge exchange
with neutral neon atoms is the dominant recombination
mechanism, with a time scale on the order of 10 msec. Ra­
diative recombination of He­like ions occurs on a time scale
of roughly 1 sec. KLL DR is much faster, but only occurs
during the small fraction of time when the beam energy is on
a resonance. Simulations show that the energy sweep is fast
enough to keep the charge balance constant to better than
1%. Note that charge exchange recombination does not lead
to any kind of 2#1 emission here, since the recombined
ions #Li­like and lower charge# are singly excited and cannot
decay any lower than the n#2 level.
Once ionization equilibrium has been attained, following
a periodic trap dump to remove higher­Z contaminants
which build up over time, spectra are collected. Cross sec­
tions for excitation of the neon K# lines are of order
10 #20 cm 2 near threshold, corresponding to time scales of
approximately 10 msec. The typical time scale for radiative
decay of excited ions is on the order of picoseconds
(10 #12 sec) , so the chance of an excited ion undergoing fur­
ther collisions before decaying is negligible. An exception is
the highly forbidden transition from the 1s2s 3 S 1 state that
decays to produce line z; the radiative lifetime of that level is
91# sec #1,2#, which is slow enough compared to the elec­
tron energy slew rate to create the ``tail'' extending directly
below z's excitation threshold in Fig. 2, as well as permit
collisional transfer to the 1s2p 3 P 2,1 states that give rise to x
and y, respectively. This, however, has no effect on our sat­
ellite line measurements.
Figure 2 is a plot of the spectral data, with wavelength
along the horizontal axis and electron energy along the ver­
tical. The He­like K# lines are prominent, with direct exci­
B. J. WARGELIN, S. M. KAHN AND P. BEIERSDORFER PHYSICAL REVIEW A 63 022710
022710­4

tation thresholds for w, xy , and z of 922, 916, and 905 eV,
respectively. The blend of q and r #the strongest Li­like sat­
ellites# is visible above its inner­shell collisional excitation
threshold of 908 eV, and a fainter feature appears to the right
of z. With roughly one­third the intensity of qr , this feature
is the blend of Li­like u and v and the Be­like satellite #
(1s2s 2 2p 1 P 1 #1s 2 2s 2 1 S 0 ) . Further to the right are faint
B­like collisional satellites.
Below threshold are the DR satellites, the most prominent
of which appear as curved tails on w and qr . #Recall that the
straight tail below z is a radiative lifetime effect.# The KLM
band (n#3 spectator electron# at the bottom of the tails
#about 820 eV# is partially resolved while the KLL band
(n#2) around 680 eV is quite distinct. The relatively strong
KLL satellites below about 13.8 å are Li­like satellites,
while satellites with ##13.8 å are Be­like. Some weak
B­like KLL satellites can be seen beyond about 14 å .
B. Intensity normalization method
Because the electron beam energy spread in EBIT
(#75 eV FWHM# is so much wider than the intrinsic DR
resonance widths, the intensity of a Li­like DR satellite line s
emitted from EBIT is #ignoring presumably constant geomet­
ric factors such as the sizes and overlap of the electron beam
and ion cloud#
I s # E av ##v e n e n He S s f # E res #E av #, #9#
where E av is the average energy of the beam electrons and
the function f (E#E av ) describes the energy distribution of
the electrons, which is approximately Gaussian. In our ex­
periment, n He is constant, and the electron beam current is
held fixed so that v e n e also can be assumed to be constant. If
the energy distribution function is normalized so that
# ##
#
f # E res #E av #dE#1, #10#
then integrating both sides of Eq. #9# over all energies #ex­
perimentally, by sweeping the beam energy up and down
across a resonance# yields
I s #v e n e n He S s . #11#
Because the intensity of Li­like satellites is proportional
to the He­like ion density, their strengths must be normalized
to some spectral feature whose intensity is also proportional
to the He­like ion abundance. The best choice, since its the­
oretical cross section as a function of energy should be quite
accurate, would be radiative recombination of He­like to Li­
like ions, but the continuum photons produced by this pro­
cess are too few to be observed. The next best choice is line
w, the brightest line in the spectrum. Its emission processes
are the best understood, and it has a smoothly varying cross
section with only minor contributions from cascades and
resonant excitation. Its intensity is given by
I w # E av ##v e n e n He # w # E av #. #12#
Integrating over a range of energies, in this case from 920 to
1200 eV, then yields
I w #v e n e n He
# 920
1200
# w # E #dE . #13#
When Eqs. #11# and #13# are divided, v e n e n He cancels out,
giving
S s #
I s
I w
# 920
1200
# w # E #dE . #14#
The resonance strengths of Be­like satellites are given by the
same formula, but by normalizing to the collisionally excited
Li­like qr blend instead of He­like w.
C. Polarization and spectrometer efficiency
The intensity of a line as observed by our spectrometer is
equal to I l
obs #G l I l , where G l , the spectrometer response
function for line l, includes terms for spectrometer collecting
area and efficiency, as well as the effects of polarization.
Ion­electron collisions in EBIT are unidirectional, leading to
unequal populations of magnetic sublevels, and so emission
will in general be polarized and nonisotropic. Equation #14#
then becomes
S s #
I s
obs
G s I w
obs
# 920
1200
G w # E ## w # E #dE . #15#
For unresolved satellites, G s is an average for the blended
lines, weighted by their theoretical resonance strengths.
To determine G for any line, we start with the equation
for polarization, which is defined as
P#
I # #I #
I # #I #
, #16#
FIG. 2. Neon spectra, plotted versus wavelength and electron
energy. Curved tails seen on several lines below the direct excita­
tion threshold are from high­n DR satellites. The KLL spectrum
contained within the angled box is shown in Fig. 3. Intensities of
lines w and qr were summed from 920 to 1200 eV for normaliza­
tion of Li­like and Be­like satellites.
DIELECTRONIC SATELLITE CONTRIBUTIONS TO Ne VIII . . . PHYSICAL REVIEW A 63 022710
022710­5

where I # is the intensity with polarization parallel to the elec­
tron beam direction and I # is for perpendicular polarization.
The total intensity emitted toward the spectrometer, at 90°
relative to the electron beam, is I(90°)#I # #I # . The net
observed intensity, after diffraction by a TAP crystal with
reflectivities R # and R # , is then
I obs #A## I # R # #I # R # #, #17#
where A is a geometrical term which includes the solid angle
acceptance of the spectrometer and energy­independent
variations in reflectivity across the crystal surface, and # is
the quantum efficiency of the proportional counter detector,
including the transmission of a thin window separating EBIT
from the spectrometer. This can be rewritten, using the two
preceding relations for P and I(90°) , as
I obs #A#I#90° # R #
# 1
2
# # # 1#P ###1#P #
R #
R #
# . #18#
In order to relate the total line intensity integrated over all
angles to that emitted toward the spectrometer, we use the
equation #6#
I#
3#P
3 I#90° #, #19#
where P and I(90°) are, respectively, the polarization and
intensity of x rays emitted 90° to the electron beam. This
relation applies for electric dipole transitions---which in­
cludes w, q, r, and all the other satellite lines under study---
from ions excited by unidirectional electron collisions. Equa­
tion #18# can then be written as
I obs #
3A#R #
2#3#P #
# # 1#P ###1#P #
R #
R #
# I#GI . #20#
Apart from the polarization dependence, G is roughly the
same for all the lines being studied. A was measured by
scanning line w across the TAP crystal, and has a small
linear rise at positions corresponding to longer wavelengths
#5.3% higher at q than at w) . R # is constant with energy to
within a fraction of 1%, but R # /R # gradually falls from
0.2444 for w to 0.2286 for q according to cos 1.792 # B , as
derived from theoretical calculations by Gullikson #7#. # also
decreases slightly at longer wavelengths, by about 5.5%
from w to q.
IV. ANALYSIS
Evaluation of resonance strengths using Eq. #15# requires
a mix of experimental measurement (I s
obs and I w
obs ) and the­
oretical modeling (G s , G w , and # w #. I s
obs was measured for
each DR satellite by collapsing the spectra contained in the
angled extraction region shown in Fig. 2. The resulting spec­
trum is shown in Fig. 3. The integrated observed intensities
of w and qr were found by summing all spectra above 920
eV and fitting each line or blend to determine the number of
counts. Lorentzian profiles gave excellent fits to the above­
threshold lines, and all linewidths were linked to the value
obtained for w. The nm and st peaks, which blend together,
were fit by setting their wavelength separation to 0.025 å ,
the value predicted by Vainshtein and Safronova #8#.
The instrumental response factor G was evaluated for
each DR satellite according to Eq. #20#, with polarizations
taken from Inal and Dubau #9# for the Li­like satellites and
from Shlyaptseva et al. #10# for the Be­like, but modified to
account for the fact that electron collisions in EBIT are not
perfectly unidirectional because of a thermally induced trans­
verse electron velocity component. As explained by Gu,
Savin, and Beiersdorfer #11#, the net #reduced# polarization
for electric dipole transitions is given by
P#P
2#3#
2##P , #21#
where # is the ratio of the transverse energy component (E # )
and the total electron energy, and P is the polarization with
zero transverse energy. We use the average of Chen's #12#
and Karim and Bhalla's #13# predictions for the total #reso­
nance# energies for each Li­like satellite, and Chen's #14#
values for the Be­like lines. Estimates of E # on EBIT range
FIG. 3. Comparison of experimental and theoretical KLL spec­
tra. Theoretical spectrum uses results from Chen #12,14,20#, ad­
justed for spectrometer response.
B. J. WARGELIN, S. M. KAHN AND P. BEIERSDORFER PHYSICAL REVIEW A 63 022710
022710­6

from 100 to 250 eV. Here we assume E # #150 eV, but as
explained later, our results depend very little on the exact
value.
To evaluate #G w (E av )# w (E av )dE av , we rely on theoreti­
cal calculations by Reed #15# of the collisional excitation
cross sections for the 1s2p 1 P 1 state and the 1s3s 1 S 0 and
1s3d 1 D 2 states which feed it by cascades with essentially
100% branching ratios. Cascades from corresponding higher­
n levels were also included, with the assumptions that cross
sections scale as n #3.6 #by extension of Reed's results for
n#2 and 3# and that branching ratios to 1s2p 1 P 1 are still
near 100%. Excitation to F, G, and higher terms is negli­
gible, as is cascade feeding of the 1s2p 1 P 1 state from triplet
levels #16#. We find that ## w (E)dE from 920 to 1200 eV is
5.50#10 #18 cm 2 eV, where we have extrapolated w's direct
excitation cross section slightly below its 922­eV threshold
to include the contributions of high­n DR, with cascades
contributing just under 5% of the total. For comparison, the
integrated cross section using the results of Sampson et al.
#17# is 5.57#10 #18 cm 2 eV, including cascades.
Reed #15# computed cross sections for each magnetic sub­
level (m J ) to permit calculation of P w , the polarization of w,
following the formalism of Alder and Steffen #18#, which is
summarized by Gu, Savin, and Beiersdorfer #11# for appli­
cation to EBIT. The resulting theoretical polarization of line
w when excited by unidirectional electron collisions and
viewed at 90° to the collision axis is 0.613 at threshold, with
very little change up to 1200 eV. After including the effects
of transverse electron velocity, the net polarization is re­
duced to about 0.49, depending only slightly on energy. The
cross section and polarization results were then used to com­
pute G w (E av ) .
For the Be­like satellites, we normalize to #G qr # qr dE ,
using
G qr # qr #G q # q #G r # r #G q # iq W q #G r # ir W r , #22#
where # iq and # ir are the cross sections for excitation of the
states that decay to produce q and r #see Eq. 3#. Lines q and
r arise from identical configurations, with J values of 3/2 and
1/2, respectively. Since they are excited from the same Li­
like ground state, on statistical grounds one would expect the
ratio # ir /# iq to be 1/2. Using the results of Reed #15#, which
include excitation cross sections #by magnetic sublevel# for
all 16 autoionizing Li­like Ne levels of the type 1s2l2l#, we
indeed find that # ir /# iq is very near 1/2 #0.493# for all en­
ergies. We also used those results to compute the polariza­
tion of q, which is 0.341 at threshold and #0.27 after depo­
larization. As will be discussed in Sec. V B, the actual values
of W q and W r are not well known, but their ratio is, allowing
us to accurately determine G qr (E) .
To help understand the uncertainties involved in Eq. #15#,
one can rewrite it as
S s #
I s
obs
I w
obs
G s
G w
# 920
1200
# w # E #dE , #23#
where G w is the cross­section­weighted average of G w (E)
#i.e., #G w # w (E)dE/## w (E)dE#. As seen from Eq. #20#,
when taking the ratio G s /G w , the A#R # factors essentially
cancel out, leaving only polarization­dependent terms. The
intrinsic Li­like satellite polarizations are well known, as is
the energy­dependent polarization of w, so the primary
source of polarization uncertainty lies in the degree of depo­
larization, which is simply determined by E # , according to
Eq. #21#. The G s /G w values listed in Table I were computed
assuming that E # #150 eV, but using a different value has
little effect. G s /G w varies by only about #3% for a 50­eV
difference in E # , except for the negatively polarized a and l
lines, which are quite weak. The net uncertainty in G s /G w
for blended lines is less than 1%.
Some of the Be­like satellites we observe are not listed by
Shlyaptseva et al. #10#, but we know the polarization of a
few of them. The Be0 transition at 13.7806 å #theoretical
wavelength# involves an upper level with L#0 , so its polar­
ization is zero. The Be2 transition at 13.9566 å has the same
LSJ values as the Be1 transition at 13.8926, and so has the
same polarization of #0.439. The six Be1 3 P# 3 P transi­
tions are between levels that are a mixture of S and D states,
with the degree of polarization determined by how much of
each component contributes to the levels involved. The
strength of these lines is negligible, however, and we have
simply assumed that P#0 in our calculations, as marked in
Table II. We have also assumed P#0 for the Be1 line at
13.8887 å . If the true polarization of that line were
#0.50 (#0.50) , then G s /G qr would be 40% larger
#smaller#, and the net measured strength of the Be1 blend
would be 3% smaller #larger#.
V. RESULTS AND COMPARISON WITH THEORY
Many theoretical papers on dielectronic recombination
have been published, usually providing total recombination
rates and summing over all n and l configurations. Although
such results can be compared with laboratory measurements,
most of the information about individual transitions and their
rates is lost, and no line­ or energy­specific comparisons with
laboratory measurements can be made. A few authors, how­
ever, have published data with which we can make direct
comparisons.
Vainshtein and Safronova #VS# #8# employed a perturba­
tion technique on a basis of Coulomb functions using an
expansion in powers of 1/Z to calculate radiative and Auger
decay rates for all of the Li­like Ne 1s2s2l# satellites except
g and h. It should be noted that they list q
” , which is mislead­
ingly called a satellite intensity factor, but which uses only
A s f
rad instead of # k A sk
rad in the denominator of Eq. #7#. For
those pairs of satellite lines which share the same upper
state--- ab , cd , hi , kl , mn , op ---we therefore determined
the relevant radiative rates from their tables and computed
the value of Q s ourselves. Radiative decays other than those
producing n#2#1 satellites were not considered by VS.
This is generally not important for the relatively strong lines
we observed, but means that the strengths of some of the
weaker satellites listed in Table I are overestimated.
Chen #12# carried out calculations for 1s2ln#l# levels
with n##2 and 3 #corresponding to KLL and KLM satel­
DIELECTRONIC SATELLITE CONTRIBUTIONS TO Ne VIII . . . PHYSICAL REVIEW A 63 022710
022710­7

lites# with configuration interaction using the multiconfigu­
rational Dirac­Fock method #MCDF#. Results for the m and
n satellites, which were not listed in that paper, were ob­
tained by private communication, along with values of E res
for each line.
Nilsen #19# used relativistic MCDF bound states and
distorted­wave Dirac continuum states in his calculations for
1s2ln#l##1s 2 n#l# transitions, with n#,n##2,3,4 . Karim
and Bhalla #KB# #13# also studied KLL , KLM , and KLN
DR, using intermediate coupling with a multiconfiguration
Hartree­Fock­Slater atomic model, but only listed the radia­
tive and Auger rates for selected transitions, omitting a, b, q,
r, and several others. Wavelengths and rate data for all the
lines were obtained by private communication #13#.
The only published reference for Be­like DR rates is by
Chen and Crasemann #20#. Although they list what they call
``Q'' values as well as wavelengths, those Q's were calcu­
lated using A si
auto rather than # j A a j
auto in the denominator of
Eq. #7#. While the two are equivalent for Li­like 1s2l2l#
levels, which can only autoionize to the single ground state
of the He­like ion, Be­like 1s2l2l#2l# levels autoionize to
create a Li­like ion which has three possible ground levels,
namely 1s 2 2s 2 S 1/2 , 1s 2 2p 2 P 1/2 , and 1s 2 2p 2 P 3/2 . For the
purposes of the present comparison, Chen #20# provided the
necessary Auger rates to let us compute the resonance
strengths listed in Table II.
A. Wavelengths
Satellite wavelengths were determined by using w, xy ,
and z as reference lines with wavelengths of 13.4473,
13.5530, and 13.6990 å , respectively, taken from Drake
#21#, where for # xy we have used the intensity­weighted av­
erage of # x and # y with I y /I x #26, as computed by Oster­
held #16#. Drake's predictions for the corresponding He­like
Al lines have been experimentally confirmed to approxi­
mately one part in 40 000 #22#, so the above wavelengths are
considered accurate to #0.0003 å . Results for the blended
DR satellites are shown in Table III for Li­like lines and
Table IV for Be­like lines. The wavelength of qr was also
measured from the summed 920--1200 eV spectrum to be
##13.6522#0.0005 å , but spectra from other EBIT mea­
surements #1# with more counts indicate a wavelength of
13.6532#0.0004 å . The weighted average is 13.6528
#0.0004 å .
Although other collisionally excited lines are apparent in
Fig. 2, their weakness, distance from lines of known wave­
length, and blending make the determination of wavelengths
rather problematic. Making generous allowance for uncer­
tainties in extrapolating our wavelength scale, we find that
the blend of Li­like u and v , along with some contribution
from the Be­like # line, is at ##13.839#0.004 å , and a
B­like collisional satellite feature is at ##14.041#0.005 å .
The below­threshold features labeled as B1 and B2 are
B­like KLL DR satellites, with wavelengths of 14.057
#0.005 and 14.103#0.007 å , respectively.
As can be seen, the wavelength predictions of VS #8# are
clearly superior to the others, with excellent self­consistency
and agreement with our measurements. #We include separate
results for nm and st , but those features are not well resolved
from each other and the listed uncertainties, which are based
purely on counting statistics, are too small. A more reliable
result, for both wavelength and strength, is obtained from the
TABLE II. Theoretical atomic data for principal KLL dielectronic satellite transitions 1s2l 3
#1s 2 2l 2 in
Be­like Ne 6# . Polarizations are taken, when available, from Shlyaptseva et al. #10#. Theoretical values of #
and S are derived from results of Chen and Crasemann #20#, with E res taken from Chen #14#. Transitions are
grouped to correspond with observed spectral features. Transition strengths are in units of 10 #20 cm 2 eV.
Blend Transition E res #eV# P G s /G qr # (å) S
Be0 # 1s2s( 3 S)2p 2 3 S 1 #1s 2 2s2p 3 P 0 708.5 0 0.820 13.7806 0.18
1s2s( 3 S)2p 2 3 S 1 #1s 2 2s2p 3 P 1 708.5 0 0.820 13.7813 0.55
1s2s( 3 S)2p 2 3 S 1 #1s 2 2s2p 3 P 2 708.5 0 0.820 13.7831 0.96
Be1
¦ 1s2p 3 3 P 1 #1s 2 2p 2 3 P 0 727.4 0 a 0.815 a 13.8391 0.02
1s2p 3 3 P 0 #1s 2 2p 2 3 P 1 727.4 0 a 0.815 a 13.8393 0.02
1s2p 3 3 P 2 #1s 2 2p 2 3 P 1 727.4 0 a 0.815 a 13.8400 0.02
1s2p 3 3 P 1 #1s 2 2p 2 3 P 1 727.4 0 a 0.815 a 13.8400 0.01
1s2p 3 3 P 2 #1s 2 2p 2 3 P 2 727.4 0 a 0.815 a 13.8415 0.06
1s2p 3 3 P 1 #1s 2 2p 2 3 P 2 727.4 0 a 0.815 a 13.8415 0.02
1s2s( 1 S)2p 2 1 S 0 #1s 2 2s 2 1 P 1 718.0 0 0.814 13.8593 0.24
1s2s( 3 S)2p 2 3 D 1 #1s 2 2s2p 3 P 0 701.5 0 a 0.811 a 13.8887 0.80
1s2s( 3 S)2p 2 3 D 1 #1s 2 2s2p 3 P 1 701.5 #0.429 0.607 13.8895 0.87
1s2s( 3 S)2p 2 3 D 2 #1s 2 2s2p 3 P 1 701.4 #0.333 1.014 13.8899 1.71
1s2s( 3 S)2p 2 3 D 2 #1s 2 2s2p 3 P 2 701.4 #0.429 0.607 13.8917 1.28
1s2s( 3 S)2p 2 3 D 3 #1s 2 2s2p 3 P 2 701.4 0.439 1.089 13.8926 4.14
Be2 # 1s2p 3 3 D 3 #1s 2 2p 2 3 P 2 704.0 #0.439 1.083 13.9566 1.00
1s2s( 1 S)2p 2 1 D 2 #1s 2 2s2p 1 P 1 710.8 #0.60 1.211 13.9704 1.75
a Polarizations unknown. G factors computed assuming P#0 .
B. J. WARGELIN, S. M. KAHN AND P. BEIERSDORFER PHYSICAL REVIEW A 63 022710
022710­8

composite feature, nm#st .# VS also provide the best agree­
ment with the uv# blend at 13.839 å; although, as noted
above, this feature is difficult to interpret, we believe from
examining higher­energy spectra that # lies slightly long­
ward of uv , which is consistent with VS's prediction of
# uv #13.835 å . See Nilsen and Safronova #26# for a detailed
comparison of wavelengths and radiative and autoionization
rates predicted by Nilsen #19# and VS #8# and measured in
other EBIT experiments.
B. Intensities
Theoretical resonance strengths for KLL lines were com­
puted according to Eq. #8# using satellite intensity factors
(Q s 's# from the theoretical papers listed above. Measured
strengths were computed according to Eq. #15#, where I s
obs is
the number of counts in the Li­like #Be­like# satellite and
I w
obs (I qr
obs ) is the number of counts in w (qr) in the 920--
1200­eV summed spectrum. The general evaluation of
#G w # w dE and #G qr # qr dE was described earlier; we now
discuss the details of how G qr # qr was determined.
We use Reed's #15# predictions for # iq and # ir #see Eq.
#22##. The integral of # iq from 920 to 1200 eV is 3.04
#10 #18 cm 2 eV, and 0.493 times that for r, but we also need
to know the fluorescence yields for those two lines. VS,
Chen, and Nilsen all agree that W r /W q #0.875 to within
one­half percent, while KB predict a slightly higher value
of 0.909. We have adopted 0.875 for our calculations. There
was much greater disagreement on the value of W q , how­
ever, with predictions of 0.351 #Chen#, 0.519 #Nilsen#, 0.524
#KB#, and 0.637 #VS#. From a measurement of the ratio
# zISI /# qr #23#, where # zISI is the cross section for production
of line z via inner­shell ionization of Li­like Ne #which, for
direct ionization, leaves the He­like ion in a 1s2s 3 S 1 state
3/4 of the time#, and using Reed's #15# q and r excitation
cross sections and Younger's #24# direct inner­shell ioniza­
tion cross section, we derive W q #0.68, a value higher than
any of the theoretical predictions, although the VS value is
not too different. Using Lotz's #25# formula for the ioniza­
tion cross section yields essentially the same value (W q
#0.65) , and we have used 0.68 in our calculations.
The resulting theoretical resonance strengths are pre­
sented in Tables III and IV. Individual results for nm and st
are somewhat suspect because of blending, but their compos­
ite strength should be reliable. For comparison with the ob­
served spectrum, we used Chen's predictions #12,14# to pre­
dict what the spectrum would look like after accounting for
polarization and instrumental effects #see Fig. 3#, assuming
an ion abundance ratio of n Li /n He #0.10.
Reasonably good agreement is found between theory and
experiment for the Li­like blends, although total theoretical
resonance strengths range from 16% to 32% #average 23%#
higher than measured values. Variations are larger for the
TABLE III. Comparison of measurement and theory for blended Li­like satellite lines. G s /G w is the net spectrometer response function
for each blend #primarily dependent upon line polarization#, normalized to the value for w. Experimental wavelengths were measured with
respect to the He­like lines w, xy , and z using wavelengths from Drake #21#. Numbers in parentheses denote the uncertainty in the last
digit#s#. Wavelength differences are listed as #### theory ## expt , where # theory is the average for the blended lines, weighted by their
theoretical as­observed intensities. Experimental resonance strengths were measured with respect to the integrated intensity of line w from
920 to 1200 eV, equal to 5.50#10 #18 cm 2 eV according to theoretical cross sections calculated by Reed #15#. Listed uncertainties are
statistical. R#S th /S expt is the ratio of the theoretical and experimental resonance strengths for the blend. ``KLL sum'' refers to the sum of
all observed KLL blends: nm , st , qrba , and kl j . The wavelength of the qr blend, measured from data taken above its direct excitation
threshold, is 13.6528#0.0004 å .
Blend # expt (å) ## a (å) ## b (å) ## c (å) ## d (å) G s /G w S expt R a R b R c R d
nm 13.533#4# 0.005 #0.005 0.016 #0.012 0.639 1.08#22# 2.11 1.86 2.06 1.74
st 13.561#3# 0.003 0.005 0.019 #0.016 0.829 2.30#28# 0.72 0.92 0.76 0.83
nm#st 13.552#2# #0.001 #0.002 0.013 #0.017 0.728 3.38#35# 1.17 1.22 1.18 1.12
qrba 13.655#2# #0.001 0.014 0.022 0.001 0.802 6.45#51# 1.13 1.46 1.43 1.21
kl j 13.7083#12# 0.001 0.001 0.025 #0.012 0.868 11.08#65# 1.33 1.10 1.31 1.15
KLL sum 20.91#96# 1.24 1.23 1.32 1.16
a Vainshtein and Safronova #8#.
b Chen #12#.
c Nilsen #19#.
d Karim and Bhalla #13#.
TABLE IV. Comparison of measurement and theory for
blended Be­like satellite lines. G s /G qr is the net spectrometer re­
sponse function for each blend, normalized to the value for the
Li­like collisionally excited satellite blend qr . Experimental reso­
nance strengths were measured with respect to the integrated inten­
sity of qr , using theoretical excitation cross sections calculated by
Reed #15# and integrated from 920 to 1200 eV #which yields ## iq
#3.04#10 #18 cm 2 eV) and fluorescence yields of W q #0.680 and
W r #0.595. Listed uncertainties in experimental wavelengths and
strengths are statistical.
Key G s /G qr # th (å) # expt (å) S th S expt
Be0 0.820 13.782 1.69 #1
Be1 0.926 13.890 13.879#0.003 9.19 9.50#0.72
Be2 1.064 13.966 13.941#0.004 2.75 2.29#0.35
DIELECTRONIC SATELLITE CONTRIBUTIONS TO Ne VIII . . . PHYSICAL REVIEW A 63 022710
022710­9

three individual blends, and range from 10% to 46% higher
than measured values. In particular, the predictions of Chen
and Nilsen seem quite high for q and r, and Nilsen and VS
have notably greater strengths for j and k. Statistical uncer­
tainties (1#) for the measured strengths range from 10% #for
nm#st) to 5% #for the total of all Li­like KLL lines#.
In addition to statistical uncertainties in the observed
spectra, there are several possible sources of systematic ex­
perimental error. Uncertainties related to polarization and
relative spectrometer efficiency are at the 1% level, and so
the primary sources of error are likely to be in the measured
and theoretical integrated intensity of w, and in our assump­
tion of constant v e n e n ion #see Eqs. #11# and #12##.
The estimated uncertainty in the electron beam energy is
##10 eV. When summing spectra over a range of 280 eV,
from 920 to 1200 eV, this corresponds to #5% uncertainty
in the number of integrated counts. Uncertainty in the theo­
retical cross section for w is difficult to estimate, but believed
to be of order 10%. Our measured satellite resonance
strengths are directly proportional to ## w , so if a value more
reliable than 5.50#10 #18 cm 2 eV becomes available, our re­
sults can be easily scaled.
Variations in the value of v e n e n ion arise from changes in
trap conditions, particularly the He­like/Li­like ion ratio and
the beam­ion overlap, as the electron beam sweeps up and
down over a factor of 2 in energy. As mentioned before,
simulations of time­dependent charge balance show that the
energy sweep was fast enough to keep the charge balance
constant at the 1% level. Since beam current was held fixed
for this measurement, electron density increased as electrons
moved more slowly at lower energies. This would tend to
increase the beam­ion overlap as the electron space charge
attracted ions more strongly. The result of this extra attrac­
tion, however, would be more trapped ions at low energies,
thus increasing the apparent satellite intensities and the dis­
agreement between theory and measurement if this effect
were taken into account. Beiersdorfer et al. #27# estimated
the increase at between 5% and 20% in their measurement of
resonance strengths for He­like Fe, but at the very low ener­
gies required to study neon, beam instabilities are expected
to counteract some or all of this increase, and we make no
net adjustment for these effects in our analysis.
We do, however, conservatively increase our net system­
atic error to #15% , apart from uncertainties in # w . Given
that statistical and line­fitting uncertainties are between 5%
and 10%, and that theoretical predictions of Li­like satellite
resonance strengths are between 10% and 46% higher than
measured values #and on average 23% higher#, our results
suggest that existing models overpredict KLL DR rates.
Our measurements agree best #a 16% overall difference#
with the predictions of KB. In the only other published mea­
surement of He­like neon DR, which used what we assume is
essentially the same atomic modeling code as KB, Ali et al.
#28# reported results from an electron beam ion source
#which counts ions, rather than x rays# and compared the
predicted and measured total #non­line­specific# KLL #and
KLM#) DR cross sections onto He­like neon. They ob­
tained good agreement, although systematic errors were not
well understood, and all relevant recombination channels
may not have been included in the model.
For the Be­like KLL blends, experimental and theoretical
strengths agree to within roughly the statistical uncertainty of
the measurement. Systematic uncertainties are similar to
those for the Li­like lines. The potentially largest error lies in
our normalization to the theoretical value of # q , which de­
pends on the excitation cross section for the upper level and
the fluorescence yield. As noted earlier, theoretical values of
W q range from 0.35 to 0.64, and we believe the actual value
is close to 0.68. Given such large disparities, it is rather
surprising that the theoretical Li­like satellite resonance
strengths agree with each other as well as they do.
VI. CONCLUSIONS
We have described an experimental measurement of the
strengths and wavelengths of the KLL dielectronic recombi­
nation satellites in Li­like and Be­like neon. We find that
existing theoretical calculations for Li­like satellites predict
strengths that are marginally too large, and that the wave­
length predictions of Vainshtein and Safronova #8# are by far
the most accurate, agreeing with our measurements to within
the experimental uncertainty, which was as good as one part
in 30 000. The single set of theoretical calculations for Be­
like satellites agrees well with our measured resonance
strengths, although the intensity normalization method high­
lights significant uncertainty over the fluorescence yield for
the Li­like satellite line q; existing predictions for that yield
vary by nearly a factor of 2, and only one is close to what we
measure.
ACKNOWLEDGMENTS
We would like to thank M. Chen, K. Reed, and A. Oster­
held for their atomic modeling calculations, K. Karim for
providing an extended version of previously published cal­
culations, R. Ali for a reprint of his EBIS measurements, and
D. Nelson and E. Magee for their technical support. This
work was performed under the auspices of the U.S. Depart­
ment of Energy by the Univeristy of California, Lawrence
Livermore National Laboratory under Contract No. W­7405­
ENG­48 and was supported by the NASA Space Astrophys­
ics Research and Analysis Program.
#1# B.J. Wargelin, P. Beiersdorfer, and S.M. Kahn, Phys. Rev.
Lett. 71, 2196 #1993#.
#2# E. Tra ˜ bert, P. Beiersdorfer, G.V. Brown, A.J. Smith, S.B. Ut­
ter, M.F. Gu, and D.W. Savin, Phys. Rev. A 60, 2034 #1999#.
#3# A.H. Gabriel, Mon. Not. R. Astron. Soc. 160, 99 #1972#.
#4# M.A. Levine, R.E. Marrs, J.R. Henderson, D.A. Knapp, and
M.B. Schneider, Phys. Scr. T22, 157 #1988#.
#5# P. Beiersdorfer and B.J. Wargelin, Rev. Sci. Instrum. 65, 13
B. J. WARGELIN, S. M. KAHN AND P. BEIERSDORFER PHYSICAL REVIEW A 63 022710
022710­10

#1994#.
#6# I.C. Percival and M.J. Seaton, Philos. Trans. R. Soc. London,
Ser. A 251, 113 #1958#.
#7# E. Gullikson #private communication#.
#8# L.A. Vainshtein and U.I. Safronova, At. Data Nucl. Data
Tables 21, 49 #1978#.
#9# M.K. Inal and J. Dubau, J. Phys. B 22, 3329 #1989#.
#10# A.S. Shlyaptseva, R.C. Mancini, P. Neill, P. Beiersdorfer, J.R.
Crespo Lo ’ pez­Urrutia, and K. Widmann, Phys. Rev. A 57, 888
#1998#.
#11# M.F. Gu, D.W. Savin, and P. Beiersdorfer, J. Phys. B 32, 5371
#1999#.
#12# M.H. Chen, At. Data Nucl. Data Tables 34, 301 #1986#; rates
for m and n and resonance energies for all lines #private com­
munication#.
#13# K.R. Karim and C.P. Bhalla, J. Quant. Spectrosc. Radiat.
Transf. 51, 557 #1994#; K. R. Karim---wavelengths, resonance
energies, and rates for individual lines #private communica­
tion#.
#14# M.H. Chen, Phys. Rev. A 31, 1449 #1985#; additional reso­
nance energies #private communication#.
#15# K. Reed #private communication#.
#16# A.L. Osterheld #private communication#.
#17# D.H. Sampson, S.J. Goett, and R.E.H. Clark, At. Data Nucl.
Data Tables 29, 467 #1983#.
#18# K. Alder and R.M. Steffen, in The Electromagnetic Interaction
in Nuclear Spectroscopy, edited by W.D. Hamilton #American
Elsevier, New York, 1975#, Chap. 1, pp. 1--38; R. M. Steffen
and K. Alder, ibid., Chap. 12, pp. 505--582.
#19# J. Nilsen, At. Data Nucl. Data Tables 38, 339 #1988#.
#20# M.H. Chen and B. Crasemann, At. Data Nucl. Data Tables 37,
420 #1987#; M.H. Chen---additional Auger rates #private com­
munication#.
#21# G.W.F. Drake, Can. J. Phys. 66, 586 #1988#.
#22# B.J. Wargelin, P. Beiersdorfer, D.A. Liedahl, S.M. Kahn, and
S. von Goeler, Astrophys. J. 496, 1031 #1998#.
#23# B.J. Wargelin, Ph.D. thesis, University of California, Berkeley,
1993.
#24# S.M. Younger, J. Quant. Spectrosc. Radiat. Transf. 26, 329
#1981#.
#25# W. Lotz, Z. Phys. 202, 205 #1967#.
#26# J. Nilsen and U.I. Safronova, J. Quant. Spectrosc. Radiat.
Transf. 49, 371 #1993#.
#27# P. Beiersdorfer et al., Phys. Rev. A 46, 3812 #1992#.
#28# R. Ali, C.P. Bhalla, C.L. Cocke, M. Schulz, and M. Sto ˜ ckli, in
Recombination of Atomic Ions, edited by W.G. Graham et al.
#Plenum Press, New York, 1992#, p. 193.
DIELECTRONIC SATELLITE CONTRIBUTIONS TO Ne VIII . . . PHYSICAL REVIEW A 63 022710
022710­11