Документ взят из кэша поисковой машины. Адрес оригинального документа : http://star.arm.ac.uk/preprints/383.ps Дата изменения: Thu Sep 5 17:27:27 2002 Дата индексирования: Mon Oct 1 21:34:03 2012 Кодировка: Поисковые слова: http www.stsci.edu science goods

The O iv and S iv intercombination lines in the ultraviolet
spectra of astrophysical sources
F.P. Keenan 1 , S. Ahmed 1 , T. Brage 2 , J.G. Doyle 3 , B.R. Espey 4 , K.M.
Exter 5 , A. Hibbert 6 , M.T.C. Keenan 7 , M.S. Madjarska 3 , M.
Mathioudakis 1 and D.L. Pollacco 1
1 Department of Pure and Applied Physics, Queen's University, Belfast BT7 1NN
2 Kurslab LU, Department of Physics, University of Lund, S{221 00, Lund, Sweden
3 Armagh Observatory, Armagh BT61 9DG
4 Physics Department, Trinity College Dublin, Dublin 2, Ireland
5 School of Cosmic Physics, Dublin Institute for Advanced Studies, 5 Merrion
Square, Dublin 2, Ireland
6 Department of Applied Mathematics and Theoretical Physics, Queen's University,
Belfast BT7 1NN
7 Hazelwood College, 70 Whitewell Road, Belfast BT36 7ES
Abstract.
New electron density diagnostic line ratios are presented for the O iv 2s 2 2p 2 P{
2s2p 2 4 P and S iv 3s 2 3p 2 P{3s3p 2 4 P intercombination lines around 1400  A. A
comparison of these with observational data for the symbiotic star RR Telescopii
(RR Tel), obtained with the Space Telescope Imaging Spectrograph (stis), reveals
generally very good agreement between theory and observation. However the S iv
2 P 3=2 { 4 P 1=2 transition at 1423.824  A is found to be blended with an unknown
feature at 1423.774  A. The line width for the latter indicates that the feature arises
from a species with a large ionization potential. In addition, the S iv 2 P 1=2 { 4 P 3=2
transition at 1398.044  A is identied for the rst time (to our knowledge) in an
astrophysical source other than the Sun, and an improved wavelength of 1397.166  A
is measured for the O iv 2 P 1=2 { 4 P 3=2 line. The O iv and S iv line ratios in a sunspot
plume spectrum, obtained with the Solar Ultraviolet Measurements of the Emitted
Radiation (sumer) instrument on the Solar and Heliospheric Observatory, are found
to be consistent, and remove discrepancies noted in previous comparisons of these
two ions.
Keywords: atomic data | Sun: UV radiation | binaries: symbiotic
1. INTRODUCTION
The O iv 2s 2 2p 2 P{2s2p 2 4 P and S iv 3s 2 3p 2 P{3s3p 2 4 P intercombi-
nation multiplets around 1400  A show prominent emission lines in the
spectra of the Sun and other astrophysical sources (see, for example,
Feldman et al. 1997; Keenan et al. 1993). It has long been known
that intensity ratios involving these transitions provide very useful
electron density (N e ) diagnostics for the emitting plasma (Flower &
Nussbaumer 1975; Bhatia, Doschek & Feldman 1980). However more
recently several authors have noted discrepancies between theoretical
c
2002 Kluwer Academic Publishers. Printed in the Netherlands.
s4.tex; 5/09/2002; 9:46; p.1

2 Keenan et al.
line ratios and observational data. For example, Cook et al. (1995)
found that the O iv ratios imply densities which dier by up to an order
of magnitude, while some observed S iv line ratios lay outside the range
of values allowed by theory. Although some of these discrepancies were
subsequently resolved by improved atomic data for O iv (Brage, Judge
& Brekke 1996), several still remain, such as a disagreement between
theory and observation for the I(1397.2  A)/I(1404.7  A) intensity ratio
in O iv (Harper et al. 1999).
In this paper we use the most up-to-date atomic physics calculations
for O iv and S iv to derive intercombination line ratios applicable
to a wide range of astrophysical phenomena. These calculations are
subsequently compared with high spectral resolution and signal-to-
noise observational data, in particular a spectrum of the symbiotic
star RR Telescopii (RR Tel) recently obtained with the Hubble Space
Telescope. The aim of this work is to investigate the accuracy of the
O iv and S iv diagnostics, and to assess the importance of possible
blending in the observations.
2. ADOPTED ATOMIC DATA AND THEORETICAL
LINE RATIOS
The model ion for O iv consisted of the 15 energetically lowest ne-
structure levels (2s 2 2p 2 P 1=2;3=2 ; 2s2p 2 4 P 1=2;3=2;5=2 , 2 D 3=2;5=2 , 2 S, 2 P 1=2;3=2 ;
2p 3 4 S, 2 D 3=2;5=2 , 2 P 1=2;3=2 ), while that for S iv comprised the lowest
5 levels (3s 2 3p 2 P 1=2;3=2 ; 3s3p 2 4 P 1=2;3=2;5=2 ). Energies for these levels
were obtained from Safronova, Johnson & Safronova (1996) and Martin,
Zalubas & Musgrove (1990) for O iv and S iv, respectively. For both
ions, we note that test calculations including higher-lying levels had a
negligible eect on the theoretical line ratios considered in this paper.
Electron impact excitation rates for transitions in O iv and S iv
were taken from Zhang, Graziani & Pradhan (1994) and Tayal (2000),
respectively. Einstein A-coeфcients for O iv were obtained from Brage
et al. (1996) for 2s 2 2p{2s2p 2 transitions, Dankwort & Tretz (1978)
for 2s 2 2p{2p 3 and Nussbaumer & Storey (1982) for 2s 2 2p 2 P 1=2 {2s 2 2p
2 P 3=2 . Data for the 3s 2 3p 2 P{3s3p 2 4 P intercombination lines in S iv
were taken from Hibbert, Brage & Fleming (2002), while for all other
radiative rates in this ion the calculations of Johnson, Kingston &
Dufton (1986) and Bhatia et al. (1980) were adopted. Proton impact
excitation rates for O iv were obtained from Foster, Keenan & Reid
(1996) and Foster, Reid & Keenan (1997), and those for S iv from
Bhatia et al. (1980).
s4.tex; 5/09/2002; 9:46; p.2

The O iv and S iv intercombination lines 3
Using the atomic data discussed above in conjunction with the sta-
tistical equilibrium code of Dufton (1977), relative O iv and S iv level
populations and hence emission line strengths were calculated for a
range of electron temperatures (T e ) and densities (N e ). Details of the
procedures involved and approximations made may be found in Dufton
(1977) and Dufton et al. (1978). Given typical uncertainties in the
adopted atomic data of 10 per cent (see references above), we would
expect our line ratio calculations to be in error by at most 15 per
cent.
In Figs 1{4 we present O iv and S iv theoretical line intensity ratios
generated under two sets of plasma conditions. The rst set of results
(Figs 1 and 3) are appropriate to solar and stellar transition region ob-
servations. These have been calculated at the electron temperatures of
maximum O iv and S iv fractional abundances in ionization equilibrium
(log T max (O iv) = 5.2; log T max (S iv) = 5.0; Mazzotta et al. 1998),
plus log T max {0.2 and log T max +0.2 for O iv and S iv, respectively, and
for electron densities in the range N e = 10 8 {10 13 cm 3 . The second set
of theoretical line ratios (Figs 2 and 4) are applicable to observations of
gaseous nebulae, and have been calculated at T e = 10 000 and 20 000 K,
and electron densities between 10 3 {10 10 cm 3 .
The O iv line intensity ratios in Figs 1 and 2 are
R 1 = I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 1=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 5=2 ) =
I(1407.3  A)/I(1401.1  A) and
R 2 = I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 1=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 3=2 ) =
I(1407.3  A)/I(1404.7  A) while we note that the ratio
R 3 = I(2s 2 2p 2 P 1=2 {2s2p 2 4 P 1=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 5=2 ) =
I(1399.7  A)/I(1401.1  A) has the same T e and N e dependence as R 1 due
to common upper levels, but with R 3 = 1.03  R 1 . Similarly, the ratio
R 4 = I(2s 2 2p 2 P 1=2 {2s2p 2 4 P 3=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 3=2 ) =
I(1397.2  A)/I(1404.7  A) is independent of T e and N e due to common
upper levels, and has the constant value R 4 = 0.131. For S iv, the ratios
plotted in Figs 3 and 4 are
R 1 = I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 3=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 ) =
I(1416.9  A)/I(1406.0  A) R 2 = I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 1=2 )/I(3s 2 3p 2 P 3=2 {
3s3p 2 4 P 3=2 ) = I(1423.8  A)/I(1416.9  A) and
R 3 = I(3s 2 3p 2 P 1=2 {3s3p 2 4 P 1=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 ) =
I(1404.7  A)/I(1406.0  A). The ratio
R 4 = I(3s 2 3p 2 P 1=2 {3s3p 2 4 P 3=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 ) =
I(1398.0  A)/I(1406.0  A) has the same T e and N e dependence as R 1
due to common upper levels, but with R 4 = 0.0492  R 1 , while for the
same reason the ratios
R 5 = I(3s 2 3p 2 P 1=2 {3s3p 2 4 P 1=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 1=2 ) =
s4.tex; 5/09/2002; 9:46; p.3

4 Keenan et al.
Figure 1. The theoretical O iv line intensity ratios R1 = I(2s 2 2p 2 P 3=2
{2s2p 2
4 P 1=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 5=2 ) = I(1407.3  A)/I(1401.1  A) and R2 = I(2s 2 2p
2 P 3=2 {2s2p 2 4 P 1=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 3=2 ) = I(1407.3  A)/I(1404.7  A), where I
is in energy units, plotted as a function of logarithmic electron density (Ne in cm 3 )
at logarithmic electron temperatures (Te in K) of log Te = 5.0 (solid line) and 5.2
(dashed line).
I(1404.7  A)/I(1423.8  A) and
R 6 = I(3s 2 3p 2 P 1=2 {3s3p 2 4 P 3=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 3=2 ) =
I(1398.0  A)/I(1416.9  A) are independent of T e and N e and have the
constant values R 5 = 1.37 and R 6 = 0.0492.
We note that the O iv line ratios in Figs 1{4 are very similar (within
a few per cent) of those calculated by Harper et al. (1999). However the
S iv results dier by up to 30 per cent from those presented by Dufton
et al. (1982), due primarily to the adoption of the improved A-value
calculations of Hibbert et al. (2002) in the present analysis.
s4.tex; 5/09/2002; 9:46; p.4

The O iv and S iv intercombination lines 5
Figure 2. The theoretical O iv line intensity ratios R1 = I(2s 2 2p 2 P 3=2
{2s2p 2
4 P 1=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 5=2 ) = I(1407.3  A)/I(1401.1  A) and R2 = I(2s 2 2p
2 P 3=2 {2s2p 2 4 P 1=2 )/I(2s 2 2p 2 P 3=2 {2s2p 2 4 P 3=2 ) = I(1407.3  A)/I(1404.7  A), where I
is in energy units, plotted as a function of logarithmic electron density (Ne in cm 3 )
at electron temperatures of Te = 10 000 K (solid line) and 20 000 K (dashed line).
3. OBSERVATIONAL DATA
3.1. HST observations
HST data of RR Tel were obtained with the Space Telescope Imaging
Spectrograph (stis) as part of program 8098 (PI: Keenan). A 2408 s
exposure covering the O iv/S iv region (rootname O5EH01010) was
taken with the stis FUV/MAMA and the medium resolution E140M
grating on 8 October 2000. The standard 0.2 arc sec square echelle
aperture was used, yielding a point source resolution of 1.3 pixels, or
0.021  A (4.5 km s 1 ).
Data were extracted from the archive and re-calibrated using the
best available calibration sets available in June 2001. Processing in-
cluded a correction for scattered light within the echelle instrument, as
discussed in the stis Instrument Handbook Version 5.1. The signal-to-
noise per pixel directly measured from the continuum in the processed
data for each spectral order near the emission lines of interest is typ-
s4.tex; 5/09/2002; 9:46; p.5

6 Keenan et al.
Figure 3. The theoretical S iv line intensity ratios R1 = I(3s 2 3p 2 P 3=2
{3s3p 2
4 P 3=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 ) = I(1416.9  A)/I(1406.0  A), R2 = I(3s 2 3p
2 P 3=2 {3s3p 2 4 P 1=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 3=2 ) = I(1423.8  A)/I(1416.9  A) and R3 =
I(3s 2 3p 2 P 1=2 {3s3p 2 4 P 1=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 ) = I(1404.7  A)/I(1406.0  A),
where I is in energy units, plotted as a function of logarithmic electron density (Ne
in cm 3 ) at logarithmic electron temperatures (Te in K) of log Te = 5.0 (solid line)
and 5.2 (dashed line).
ically 6{7 per 0.016  A pixel, and much higher in the emission lines
themselves.
In this paper we are interested in the accuracy of the stis calibration
in terms of both ux and wavelength calibration. The stis Instrument
Handbook lists the expected ux accuracy as typically 5 per cent (1
) for the echelle. However it will be signicantly better than this over
the limited wavelength range that we are considering (three spectral
orders), and also because we are concerned with the relative (rather
than the absolute) ux accuracy.
The accuracy of the echelle wavelength calibration has been checked
by one of us (BRE) by extracting a long exposure engineering wave-
cal from the HST archive, taken with the same grating setup. Data
obtained for program 8430 as part of the Cycle 8 stis calibration
program in August 2000 (rootname O5J25LXCQ) was used. By modi-
fying the header of these observations, it is possible to calibrate them
s4.tex; 5/09/2002; 9:46; p.6

The O iv and S iv intercombination lines 7
Figure 4. The theoretical S iv line intensity ratios R1 = I(3s 2 3p 2 P 3=2
{3s3p 2
4 P 3=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 ) = I(1416.9  A)/I(1406.0  A), R2 = I(3s 2 3p
2 P 3=2 {3s3p 2 4 P 1=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 3=2 ) = I(1423.8  A)/I(1416.9  A) and R3 =
I(3s 2 3p 2 P 1=2 {3s3p 2 4 P 1=2 )/I(3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 ) = I(1404.7  A)/I(1406.0  A),
where I is in energy units, plotted as a function of logarithmic electron density (Ne in
cm 3 ) at electron temperatures of Te = 10 000 K (solid line) and 20 000 K (dashed
line).
using CALSTIS, and generate wavelength calibrated spectra. From the
measured wavelengths of a sample of known Pt Ne arc lines, the oset
between the true and assigned wavelengths can be determined. Details
of this approach can be found in stis Instrument Science Report 98{
12, available on the stis website (http:==www.stsci.edu=hst=stis). For
our calibration dataset, we nd an rms scatter of 0.002  A, and the
relative wavelength scale (Section 4.1) should therefore be accurate to
0.003  A.
In Figs 5 and 6 we show the stis spectrum of RR Tel between
1390{1425  A, along with identications of emission features in this
wavelength range (see Section 4.1).
s4.tex; 5/09/2002; 9:46; p.7

8 Keenan et al.
Figure 5. stis spectrum of RR Telescopii, obtained on 8 October 2000, showing the
1390{1425  A wavelength range. The ux is in units of 10 11 erg cm 2 s 1  A 1 .
3.2. sumer observations
The Solar Ultraviolet Measurements of the Emitted Radiation (sumer)
instrument on board the Solar and Heliospheric Observatory (SoHO)
is a high resolution, stigmatic, normal incidence spectrometer covering
the wavelength range from 660{1610  A and 465{805  A in rst and
second order, respectively (Wilhelm et al. 1997; Lemaire et al. 1997),
with an angular pixel size in the direction along the slit of 1 arc sec
and a spectral pixel size between 0.042{0.045  A. The dataset analysed
in this paper was obtained on 18 March 1999, exposing a band of 120
spatial  1024 spectral pixels from detector B with a 0.3  120 arc sec
slit, positioned through the central part of a sunspot umbra.
In the quiet Sun, all the O iv and S iv lines are weak. However,
as shown from Skylab data (for example, Doyle et al. 1985), the mid-
transition region lines are enhanced over a sunspot umbra, particularly
in the plume region. Due to the tilt of the slit, the extraction of spectral
lines taken over an extended wavelength region requires proper align-
ment. For the data considered here, this translated to a 2 arc sec
shift between the location of the O iv 1401  A line compared to the S iv
s4.tex; 5/09/2002; 9:46; p.8

The O iv and S iv intercombination lines 9
Figure 6. Expanded version of Figure 5, to show weak emission lines.
1423  A line on detector B. The sunspot plume spectrum analysed here
is shown in Figs 7 and 8.
As in the RR Tel spectrum, the sumer data are aected by blends, in
this instance from both rst and second order lines. In the sumer solar
spectrum the O iv 1397.22  A line is blended by Ni ii 1397.48  A (note
that the wavelengths given here are those derived from the calibration
of Curdt et al. 2001, which dier slightly from the more accurate stis
values). The O iv 1399.77  A feature is blended in the red wing by Fe ii
1399.97  A, with the contribution of the latter being 40 per cent for
the quiet Sun and 20 per cent in a sunspot spectrum. Similarly, O iv
1401.16  A is aected by a blend with the chromospheric S i 1401.51  A
line. Details about how the degree of blending depends on the solar
region observed may be found in Teriaca et al. (2001), but for a sunspot
spectrum the S i contribution is only 0.025 per cent. The spectral
feature at 1404  A consists of 3 lines, namely second order O iii 702.32  A
at 1404.64  A, S iv 1404.79  A and O iv 1404.81  A. In addition, O iv
1407.39  A is heavily blended by the second order O iii 703.845 and
703.85  A lines, which appear as one feature at 1407.7  A and will be
referred to as O iii 703  A. This blend can be resolved using a double
Gaussian t and the intensities of both lines can be determined. Since
s4.tex; 5/09/2002; 9:46; p.9

10 Keenan et al.
Figure 7. The sumer sunspot plume spectrum obtained on 18 March 1999 in the
1397{1409  A wavelength range.
O iii 703  A has the dominant contribution in a quiet Sun spectrum,
its parameters are much better determined than those of the O iv
line. However, in a sunspot spectrum O iv is stronger and hence its
parameters are well determined. Further discussions of O iv blending
problems in sumer spectra may be found in Teriaca et al. and Judge
et al. (1998).
The S iv 1406.04  A line is blended with Fe ii 1405.61  A and second
order O iii 702.805 and 702.9  A. Similarly, S iv 1416.93  A is heavily
blended in the blue wing by Fe ii 1416.73  A. The S iv 1423.86  A line has
a strong contribution from Fe ii 1424.07  A, and is suфciently intense
to be selected as a spectral feature only in sunspot observations.
s4.tex; 5/09/2002; 9:46; p.10

The O iv and S iv intercombination lines 11
Figure 8. The sumer sunspot plume spectrum obtained on 18 March 1999 in the
1414{1425  A wavelength range.
4. RESULTS AND DISCUSSION
4.1. Line identifications, wavelengths and energy levels
In Table I we list wavelengths and suggested identications from the
NIST Database at
http:==physics.nist.gov=PhysRefData
and the Atomic Line List of Peter van Hoof at
http:==www.pa.uky.edu=peter=atomic
for the emission features in the RR Tel spectrum between 1392{
1425  A. The rest wavelengths have been derived by shifting the mea-
sured values so that the Si iv lines at 1393.755 and 1402.770  A appear
at their rest wavelengths, which have been established to an accuracy
of better than 0.001  A (Martin & Zalubas 1983). As noted in Section
3.1, the relative wavelength scale for the RR Tel spectrum should be
accurate to 0.003  A, while the error in the wavelength measurements
s4.tex; 5/09/2002; 9:46; p.11

12 Keenan et al.
Figure 9. stis spectrum of RR Telescopii showing the 1423.790  A feature and the
Fe ii 1424.139  A line, where the ux is in units of 10 13 erg cm 2 s 1  A 1 . Also
plotted is the two component model for the 1423.790  A line prole, the parameters
of which are summarised in Table III.
from the line tting procedure is less than 0.002  A. Hence the wave-
length separation of the features in Table I should be accurate to better
than 0.004  A. We discuss our analysis of the O iv and S iv features
in this wavelength region separately below.
4.1.1. O iv
The wavelength separation of the O iv 1399.731 and 1407.333  A lines
implies a 2s 2 2p 2 P 1=2 {2s 2 2p 2 P 3=2 energy dierence of 385.910.29 cm 1 .
This is identical to the measurement of 385.910.29 cm 1 obtained
from the 1397.166 and 1404.740  A lines, and is in very good agreement
with the value of 385.9 cm 1 given by Moore (1983). We note that our
wavelength of 1397.166  A for the 2s 2 2p 2 P 1=2 {2s2p 2 4 P 3=2 line repre-
sents a signicant improvement over the measurement of 1397.20  A by
Bromander (1969), which could not be determined accurately due to
the presence of a blend. In addition, the good agreement found for the
2 P 1=2 { 2 P 3=2 energy separation using the two sets of O iv lines indicates
that the measured wavelength of the 2s 2 2p 2 P 3=2 {2s2p 2 4 P 3=2 transition
s4.tex; 5/09/2002; 9:46; p.12

The O iv and S iv intercombination lines 13
Table I. Line identications and uxes for the RR Tel spectrum.
Wavelength FWHM Identication Flux  3 error
(  A) (  A) (erg cm 2 s 1 )
1392.112 0.160 Fe ii 3d 7 b 2 F 5=2 {3d 6 4f 2 [2] 3=2 (2.470.91)10 14
1393.755 0.192 Si iv 3s 2 S{3p 2 P 3=2 (3.630.07)10 12
1397.166 0.153 O iv 2s 2 2p 2 P 1=2 {2s2p 2 4 P 3=2 (5.590.47)10 13
1397.788 0.069 Fe ii 3d 6 4s 2 P 3=2 {3d 5 4s4p 2 P 3=2 (2.290.66)10 14
1398.044 0.069 S iv 3s 2 3p 2 P 1=2 {3s3p 2 4 P 3=2 (6.744.41)10 15
1399.731 0.155 O iv 2s 2 2p 2 P 1=2 {2s2p 2 4 P 1=2 (1.200.08)10 12
1401.115 0.161 O iv 2s 2 2p 2 P 3=2 {2s2p 2 4 P 5=2 (7.270.41)10 12
1401.947 0.160 Fe ii 3d 6 4s 4 D 7=2 {3d 5 4s4p 6 P 5=2 (3.790.83)10 14
1402.770 0.196 Si iv 3s 2 S{3p 2 P 1=2
(2.090.05)10 12
1404.740 0.161 O iv 2s 2 2p 2 P 3=2
{2s2p 2 4 P 3=2
+ (4.100.27)10 12
S iv 3s 2 3p 2 P 1=2
{3s3p 2 4 P 1=2
O iv 2s 2 2p 2 P 3=2 {2s2p 2 4 P 3=2 (3.910.31)10 12 a
1406.004 0.135 S iv 3s 2 3p 2 P 3=2 {3s3p 2 4 P 5=2 (4.050.27)10 13
1407.333 0.155 O iv 2s 2 2p 2 P 3=2 {2s2p 2 4 P 1=2 (1.150.08)10 12
1411.899 0.160 N i 2s 2 2p 3 2 P{2s 2 2p 2 3s 2 D (4.560.91)10 14
1413.667 0.128 Fe ii 3d 6 4s 4 H 11=2 {3d 5 4s4p 4 H 11=2 (1.250.17)10 13
1416.872 0.126 S iv 3s 2 3p 2 P 3=2 {3s3p 2 4 P 3=2 (2.610.41)10 13
1423.790 0.213 S iv 3s 2 3p 2 P 3=2 {3s3p 2 4 P 1=2 (1.380.24)10 13
1424.139 0.063 Fe ii 3d 6 4s 4 P 5=2 {3d 6 5p 4 D 3=2 (2.190.94)10 14
a Corrected for presence of S iv line (see Section 4.2 for details).
at 1404.740  A is not aected by the presence of the S iv 3s 2 3p 2 P 1=2 {
3s3p 2 4 P 1=2 line, as also implied by the line intensity ratio analysis (see
Section 4.2).
The wavelength separation of the 1397.166 and 1399.731  A line pair,
and 1404.740 plus 1407.333  A transitions, imply a 2s2p 2 4 P 1=2 {2s2p 2
4 P 3=2 energy dierence of 131.160.29 and 131.160.29 cm 1 , respec-
tively. These are in excellent agreement, although 0.86 cm 1 greater
than the value of 130.3 cm 1 given by Moore (1983). Similarly, the
wavelengths of the 1401.115 and 1404.740  A features indicate a 2s2p 2
4 P 3=2 {2s2p 2 4 P 5=2 energy separation of 184.180.29 cm 1 , compared
with the Moore estimate of 185.4 cm 1 . Given the internal consistency
of our results, we believe that our energy levels are to be preferred over
those of Moore. The values for the 2s2p 2 4 P energy levels clearly require
further investigation in the laboratory, as also suggested by Harper et
al. (1999).
s4.tex; 5/09/2002; 9:46; p.13

14 Keenan et al.
Table II. O iv and S iv line ratios in the RR Tel spectrum.
Species Ratio designation Observed value Theory a
O iv R1 0.160.01 0.17
O iv R2 0.290.03 0.28
O iv R3 0.170.01 0.18
O iv R4 0.140.02 0.13
S iv R1 0.640.11 0.72
S iv R2 0.530.12 0.24
S iv R2 0.200.08 b 0.24
S iv R4 0.0170.011 0.035
S iv R6 0.0260.017 0.049
a Calculated at Te = 20 000 K and Ne = 10 6 cm 3 .
b Estimated using revised S iv 3s 2 3p 2 P 3=2 {3s3p 2 4 P 1=2 line intensity
from Table III (see Section 4.2 for details).
Table III. Parameters of the two component model for the 1423.790  A line prole
in RR Tel.
Wavelength FWHM Identication Flux  3 error
(  A) (  A) (erg cm 2 s 1 )
1423.774 0.333 S vi 4d 2 D 3=2 {5p 2 P 1=2 ? (1.120.39)10 13
1423.824 0.126 S iv 3s 2 3p 2 P 3=2 {3s3p 2 4 P 1=2 (5.291.88)10 14
Table IV. O iv and S iv line intensities in the
sumer sunspot spectrum.
Transition FWHM Flux  3 error
(  A) (W m 2 sr 1 )
O iv 1401.16  A 0.140 2.190.16
S iv 1406.04  A 0.137 0.1830.052
O iv 1407.39  A 0.151 0.4530.074
S iv 1416.93  A 0.126 0.1340.024
S iv 1423.86  A 0.055 0.0330.025
4.1.2. S iv
The wavelength separation of the 1398.044 and 1416.872  A lines implies
a 3s 2 3p 2 P 1=2 {3s 2 3p 2 P 3=2 interval of 950.500.29 cm 1 , in reasonable
agreement with the Kaufman & Martin (1993) value of 951.43 cm 1 .
s4.tex; 5/09/2002; 9:46; p.14

The O iv and S iv intercombination lines 15
Table V. O iv and S iv line ratios in the sumer sunspot spectrum.
Species Ratio designation Observed value Derived density or
theoretical value
O iv R1 0.210.04 10.20.4 a
S iv R1 0.730.25 0.64 b
S iv R2 0.250.19 0.24 b
a For O iv we list the derived value of log Ne (Ne in cm 3 ), using the
calculations in Figure 1 at log Tmax = 5.2 (Mazzotta et al. 1998).
b For S iv we list the theoretical ratios from Figure 3, using the
calculations for log Tmax = 5.0 (Mazzotta et al. 1998) and log Ne = 10.2.
This provides support for the identication of the line at 1398.044  A
being due to the S iv 3s 2 3p 2 P 1=2 {3s3p 2 4 P 3=2 transition, which is also
indicated from a line intensity ratio analysis (see Section 4.2). This
conrms the tentative detection of this line in the solar spectrum by
Curdt et al. (2001), and is the rst time (to our knowledge) that this
feature has been observed in an astrophysical source other than the Sun.
In addition, our wavelength determination represents an improvement
over the value of 1398.06  A measured by Curdt et al. from sumer data.
The 1416.872 and 1423.790  A transitions imply a 3s3p 2 4 P 1=2 {3s3p 2
4 P 3=2 interval of 342.930.29 cm 1 , in poor agreement with the Kauf-
man & Martin (1993) result of 344.6 cm 1 . However we note that a
consideration of line ratios involving the 1423.790  A feature indicates
possible blending, and hence the wavelength of this line may be ill-
determined (see Section 4.2). >From the 1406.004 and 1416.872  A tran-
sitions, we nd a 3s3p 2 4 P 3=2 {3s3p 2 4 P 5=2 separation of 545.550.29 cm 1 ,
in good agreement with the Kaufman & Martin value of 545.69 cm 1 .
The measured wavelength of the 1423.790  A line and the 3s 2 3p
2 P 1=2 {3s 2 3p 2 P 3=2 interval of 950.50 cm 1 implies that the 3s 2 3p 2 P 1=2 {
3s3p 2 4 P 1=2 feature should lie at 1404.779  A. This is in better agreement
with the Kelly (1987) wavelength of 1404.77  A than the Kaufman &
Martin (1993) value of 1404.808  A, and would appear to indicate that
this line is more closely blended with the O iv 1404.740  A transition
than previously thought. However as noted above the 1423.790  A wave-
length measurement is not secure, and hence little weight should be
attached to any prediction of the S iv 3s 2 3p 2 P 1=2 {3s3p 2 4 P 1=2 line
wavelength. In any event, this feature contributes at most a few per
cent to the O iv/S iv 1404.7  A line blend (see Section 4.2).
s4.tex; 5/09/2002; 9:46; p.15

16 Keenan et al.
4.2. O iv and S iv line intensity ratios in RR Tel
The full-width half maximum (FWHM) line widths and intensities of
the emission features in the RR Tel spectrum, measured using the
spectrum synthesis package dipso (Howarth, Murray & Mills 1994),
are summarised in Table I, along with the 3 errors in the latter.
Resultant values of the ratios R 1 through R 4 (O iv) and R 1 through
R 6 (S iv) are listed in Table II. The theoretical S iv ratio R 5 =
I(1404.7  A)/I(1423.8  A) = 1.37 (Section 2) has been used in conjunction
with the measured 1423.8  A line ux to determine the contribution of
the S iv component to the O iv/S iv 1404.7  A blend. We nd that
S iv contributes less than 5 per cent to the total 1404.7  A line ux,
and hence has a negligible eect on the derived R 2 and R 4 line ratios
in O iv. Indeed, further analysis indicates that the S iv contribution
may be even smaller (see below).
Diagnostic line ratios in other species in the RR Tel spectrum,
ranging from Al ii to O v, indicate an electron temperature close to
20 000 K and electron densities N e ' 10 5 {10 8 cm 3 (Keenan et al.
1994, 1999, 2002; McKenna et al. 1997; Hayes & Nussbaumer 1986).
Over this temperature and density interval, the predicted O iv and S iv
line ratios are eectively constant (see Figs 3 and 4), and in Table 2
we therefore list their theoretical values at T e = 20 000 K and N e =
10 6 cm 3 . However we note that varying the temperature by a factor
of two and the density by several orders of magnitude leads to a < 10
per cent change in the theoretical line ratios, and hence would not alter
the discussion presented below.
An inspection of Table II reveals excellent agreement between theory
and observation for all of the O iv line ratios. In the case of S iv there
is generally good agreement for R 1 , R 4 and R 6 (taking into account
the 15 per cent uncertainty in the theoretical results as well as the
observational errors), with the latter providing additional support for
the identication of the 1398.044  A feature being due to the S iv 3s 2 3p
2 P 1=2 {3s3p 2 4 P 3=2 transition. However the observed value of R 2 is about
a factor of two larger than theory predicts, probably due to blending
in the 1423.790  A line. Additional support for blending of this feature
comes from the line width, which we can see from Table I is somewhat
larger than for the other S iv features. A closer inspection of the line
prole, shown in Fig. 9, also suggests a possible asymmetry.
We have therefore tted the 1423.790  A line prole using a two
component model, shown in Fig. 9 and summarised in Table III. For
one component, we xed the wavelength to that predicted for the S iv
3s 2 3p 2 P 3=2 {3s3p 2 4 P 1=2 transition (1423.824  A), obtained from the
measured 3s 2 3p 2 P 3=2 {3s3p 2 4 P 3=2 value (1416.872  A) in conjunction
s4.tex; 5/09/2002; 9:46; p.16

The O iv and S iv intercombination lines 17
with the Kaufman & Martin (1993) 3s3p 2 4 P 1=2 {3s3p 2 4 P 3=2 separation
of 344.6 cm 1 . In addition, we set the FWHM for the 1423.824  A
component to that measured for the S iv 1416.872  A line, i.e. 0.126  A.
The resultant measured intensity of the 1423.824  A line implies a
revised R 2 = 0.200.08, in good agreement with the theoretical value
of 0.24. In addition, this new intensity measurement indicates that the
S iv 3s 2 3p 2 P 1=2 {3s3p 2 4 P 1=2 transition makes a smaller contribution
to the O iv/S iv 1404.7  A blend than previously thought (see above),
of less than 2 per cent.
The second component of the 1423.790  A feature has a rest wave-
length of 1423.774  A and a FWHM of 0.333  A. Crawford (2000) has
noted a correlation between the FWHM of an emission line in the
RR Tel spectrum and the ionization potential (IP) of the relevant
species. The FWHM { IP relationship derived by this author implies
that the IP of the ion responsible for the 1423.774  A line is 80{120 eV.
An inspection of the NIST Database and the Atomic Line List reveals
the only likely candidate to be a S vi line at 1423.846  A, although we
note that this is predicted to lie on the long wavelength side of the S iv
feature, and not shortward as indicated by the present analysis. Clearly
a further investigation of this wavelength region is highly desirable,
both to improve on wavelength measurements and identify blending
species.
We note in passing that Keenan et al. (1993) have measured O iv
line ratios in an RR Tel spectrum obtained with the International
Ultraviolet Explorer (IUE) satellite. However the IUE observations are
not of as high a quality as the HST data presented here, although they
are in good agreement with the line ratio calculations in Fig. 2, with
discrepancies between theory and experiment of less than 30 per cent.
4.3. O iv and S iv lines in the sumer sunspot spectrum
Cook et al. (1995) found very poor agreement between theory and
observation for the S iv line ratios from Skylab S082B and High Res-
olution Telescope and Spectrograph (HRTS) solar spectra, and also for
HST/Goddard High Resolution Spectrograph (GHRS) observations of
the binary star Capella. In addition, the electron densities derived
from the S iv diagnostics diered by large factors (> 10) from those
estimated using O iv line ratios. These disrepancies were probably due
to blending in the observations, especially for the solar spectra, which
were recorded on photographic emulsions and hence of relatively poor
signal-to-noise.
Brage et al. (1996) suggested that an analysis of high quality sumer
data would allow blending problems in the O iv/S iv spectral region
s4.tex; 5/09/2002; 9:46; p.17

18 Keenan et al.
to be claried. However both the O iv and S iv lines in sumer spectra
are aected by blends, from rst and/or second order lines, as detailed
in Section 3.2.
Where possible, we have measured O iv and S iv line intensities in
the sumer sunspot spectrum using the cfit procedure from the solar
software, employing prole tting techniques to remove the eects of
blends by other rst and/or second order lines. These intensities are
listed in Table IV, along with the 3 errors and line widths, while
the resultant O iv and S iv line ratios are summarised in Table V.
An inspection of this table reveals excellent agreement between the
observed S iv line ratios and the theoretical values calculated at an
electron density appropriate to the O iv emitting region of the sunspot
plasma, namely N e = 10 10:2 cm 3 . This resolves discrepancies previ-
ously found between O iv and S iv line ratios by Cook et al. (1995), as
discussed above. Clearly however, better quality solar spectra covering
the O iv/S iv wavelength range would be highly desirable, to reliably
measure and assess all lines.
ACKNOWLEDGEMENTS
KME and MSM are grateful to PPARC for nancial support, while
SA acknowledges the award of a postgraduate studentship from the De-
partment of Employment and Learning for Northern Ireland. Research
at Armagh Observatory is grant-aided by the Department of Culture,
Arts and Leisure for Northern Ireland. The RR Tel observations were
made with the NASA/ESA Hubble Space Telescope at STScI which is
operated by the Association of Universities for Research in Astronomy,
Inc. under NASA contract NAS 5{26555. The sumer project is nan-
cially supported by DLR, CNES, NASA, and PRODEX. We thank W.
Curdt for helpful discussions, and P. van Hoof for use of his Atomic
Line List.
REFERENCES
Bhatia A. K., Doschek G. A., Feldman U., 1980, A&A, 86, 32
Brage T., Judge P. G., Brekke P., 1996, ApJ, 464, 1030
Bromander J.,1969, Ark. Fys., 40, 257
Cook J. W., Keenan F. P., Dufton P. L., Kingston A. E., Pradhan
A. K., Zhang H. L., Doyle J. G., Hayes M. A., 1995, ApJ, 444, 936
Crawford F. L., 2000, PhD Thesis, Queen's University of Belfast
Curdt W., Brekke P., Feldman U., Wilhelm K., Dwivedi B. N.,
Schuhle U., Lemaire P., 2001, A&A, 375, 591
Dankwort W., Tretz E., 1978, A&A, 65, 93
s4.tex; 5/09/2002; 9:46; p.18

The O iv and S iv intercombination lines 19
Doyle J.G., Raymond J.C., Noyes R.W., Kingston A.E., 1985, ApJ
297, 816
Dufton P. L., 1977, Comput. Phys. Commun., 13, 25
Dufton P.L., Berrington K.A., Burke P.G., Kingston A.E., 1978,
A&A, 62, 111
Dufton P.L., Hibbert A., Kingston A.E., Doschek G.A., 1982, ApJ,
257, 338
Feldman U., Behring W. E., Curdt W., Schuhle U., Wilhelm K.,
Lemaire P., Moran T. M., 1997, ApJS, 113, 195
Flower D.R., Nussbaumer H., 1975, A&A, 45, 145
Foster V. J., Keenan F. P., Reid R. H. G., 1996, A&A, 308, 1009
Foster V. J., Reid R. H. G., Keenan F. P., 1997, MNRAS, 288, 973
Harper G. M., Jordan C., Judge P. G., Robinson R. D., Carpenter
K. G., Brage T., 1999, MNRAS, 303, L41
Hayes M. A., Nussbaumer H., 1986, A&A, 161, 287
Hibbert A., Brage T., Fleming J., 2002, MNRAS, 333, 885
Howarth I. D., Murray J., Mills D., 1994, Starlink User Note No.
50.15
Johnson C. T., Kingston A. E., Dufton P. L., 1986, MNRAS, 220,
155
Judge P. G., Hansteen V., Wikstl ., Wilhelm K., Schuhle U.,
Moran T., 1998, ApJ, 502, 981
Kaufman V., Martin W. C., 1993, J. Phys. Chem. Ref. Data, 22,
279
Keenan F. P., Aller L. H., Espey B. R., Exter K. M., Hyung S.,
Keenan M. T. C., Pollacco D. L., Ryans R. S. I., 2002, Proc. Natl.
Acad. Sci. USA, 99, 4152
Keenan F. P., Conlon E. S., Bowden D. A., Feibelman W. A.,
Pradhan A. K., 1993, ApJS, 88, 169
Keenan F. P., Dufton P. L., Feibelman W. A., Bell K. L., Hibbert
A., Staord R. P., 1994, ApJ, 423, 882
Keenan F. P., Espey B. R., Mathioudakis M., Aggarwal K. M.,
Crawford F. L., Feibelman W. A., McKenna F. C., 1999, MNRAS,
309, 195
Kelly R. L., 1987, J. Phys. Chem. Ref. Data, 16, Suppl. 1
Lemaire P., Wilhelm K., Curdt W., Schuhle U., Marsch E., Poland
A. I., et al., 1997, Solar Phys, 170, 105
McKenna F. C., Keenan F. P., Aller L. H., Hyung S., Feibelman W.
A., Berrington K. A., Fleming J., Hibbert A., 1997, ApJ, 486, 571
Martin W. C., Zalubas R., 1983, J. Phys. Chem. Ref. Data, 12, 323
Martin W. C., Zalubas R., Musgrove A., 1990, J. Phys. Chem. Ref.
Data, 19, 821
s4.tex; 5/09/2002; 9:46; p.19

20 Keenan et al.
Mazzotta P., Mazzitelli G., Colafrancesco S., Vittorio N., 1998, A&AS,
133, 403
Moore C.E., 1983, Natl. Stand. Ref. Data Ser., 3, 10
Nussbaumer H., Storey P. J., 1982, A&A, 115, 205
Safronova M. S., Johnson W. R., Safronova U. I., 1996, Phys. Rev.
A, 54, 2850
Tayal S. S., 2000, ApJ, 530, 1091
Teriaca L., Madjarska M. S., Doyle J.G., 2001, Solar Phys., 200, 91
Wilhelm K., Lemaire P., Curdt W., Schuhle E., Marsch E., Poland
A. I., et al., 1997, Solar Phys., 170, 75
Zhang H. L., Graziani M., Pradhan A. K., 1994, A&A, 283, 319
s4.tex; 5/09/2002; 9:46; p.20