The goal of the CHIANTI database is to reproduce the emission spectrum of astrophysical sources at wavelengths greater than about 1 Å. This initial version was developed with the expectation that it would be essentially complete for reproducing the emission line spectra at wavelengths above 50 Å. This was done in order to meet the immediate needs for analyses of astrophysical spectra obtained by experiments such as EUVE, Hubble, and SOHO. Consequently, ions in the helium isoelectronic sequence have not been included and He II is the only ion in the hydrogen isoelectronic sequence to be included in the current version. We expect to include these ions in the near future. Many transitions at wavelengths shorter than 50 Å\ are included in the database but it is not complete at the shorter wavelengths.
For He II the 25 fine structure levels of the 1s,
2l, 3l,
4l, and 5l configurations have been
included. Observed
energies are from Kelly (1987).
Radiative constants (gf and A values)
were taken from Wiese et al. (1966)
for dipole transitions. For the 1s
magnetic dipole
and two photon
electric dipole transitions, the A values of
Parpia & Johnson (1972)
are used.
Aggarwal et al. (1991b) have performed R-matrix calculations of collisions strengths for transitions among the n=1-5 levels of He II (i.e. 1s, 2s, 2p, 3s, 3p, 3d, 4s, 4p, 4d, 4f, 5s, 5p, 5d, 5f, and 5g). Aggarwal et al. (1992) noted that the 1991 calculations were performed at energies up to 7 Ryd and that the collision strengths in the intermediate energy region above those energies where the threshold resonances occur are overestimates of the actual collision strength. Consequently, they combined the close-coupling collision strengths of Unnikrishnan et al. (1991) in the 4.4 to 14.71 Ryd energy range with the collision strengths of Aggarwal et al. (1991b) in the threshold region to arrive at better estimates of the collision strengths to the 2s, 2p, 3s, 3p, and 3d levels. We have used these values. For the n=4 and n=5 levels, we have used the original calculations of Aggarwal et al. (1991b). To distribute the collision strengths between 2 LS states among the fine structure levels, we have scaled the collision strengths following the rules for the distribution of gf values under LS coupling.
The atomic structure of the lithium-like ions is
relatively simple.
Since there are no intersystem transitions, the
calculation of
collision rates and radiative data is relatively
straightforward. The
consequent lack of metastable levels means that
density sensitive line
ratios are not available but the interpretation
of lithium sequence
line intensities in terms of emission measures is
simplified. The
transitions are quite strong and
provide good diagnostics over a
range of temperatures, from C IV in the
transition regions, through Mg
X and Si XII in the quiet corona and to Fe XXIV
in flares. The
electron excitation data available for the
lithium-like ions seems to be
very accurate (McWhirter 1994).
The lithium-like ions provide an opportunity for a
precise comparison between laboratory
measurements of electron
excitation rate coefficients () and
theoretical work.
The recent laboratory measurements for C IV, N V
and O VI agree
well with theory, to within the typical
uncertainties (20%) quoted.
However, there remains a long-standing
discrepancy for Be II.
Also, they provide an opportunity for determining
the
electron temperature in the solar atmosphere
using the intensity ratios of the
relative to
transitions (McWhirter
1976). In particular, the O VI
ratio 1032 Å/173 Å is a potentially useful
diagnostic for SOHO
(Bely-Dubau 1994).
For the lithium isoelectronic sequence ions C IV
and N V, the
configurations 2s22s, 2s22p, 2s23s,
2s23p, and 2s23d
were included. Values for the observed energy
levels were taken from
Kelly (1987). Oscillator strengths
and spontaneous transition
probabilities (A values) were taken from
Wiese et al. (1966) and
Martin
et al. (1993). Gau & Henry
(1977) provided analytical fits to their
calculations of the necessary collision strengths
for several ions in
the lithium isoelectronic sequence and these have
been used here.
Their functional form (
) provides the
correct high energy behavior (
) but
it has been necessary
to modify their parameter c2 to ensure an
exact match to the
Coulomb-Bethe high energy limit.
Laboratory measurements of the absolute electron
excitation rate
coefficients for the transition in
C IV have been carried out
by Savin et al. (1995). These
measurements
agree well with the 9 state
CC calculations of Burke (1992).
Figure 1: Collision strengths (Zhang et al.
1990) for Mg X 1s22s
2S1/2 - 1s23p 2P3/2. Upper
left, the original
collision strengths, upper right, the scaled
collision strengths with
the high energy limit from the Bethe
approximation, lower left, the
values of 
, the collision strengths
convolved with a
Maxwellian over the energy range from threshold
to infinity, and, lower
right, the scaled values of following
the Burgess & Tully
(1992) scaling for an allowed transition
Observed energies for the 24 fine structure levels of 2l, 3 l, 4l, 5l configurations for O VI, Ne VIII and Ar XVI are obtained from Kelly (1987) and for Mg X, Al XI, Si XII, S XIV, Ca XVIII, Fe XXIV, and Ni XXVI are obtained from the NIST database.
Zhang et al. (1990) have calculated relativistic distorted wave collision strengths for the the lithium isoelectronic sequence from O VI through Ni XXVI. In addition, they provide oscillator strengths from which the radiative data has been calculated. Some additional oscillator strengths come from Martin et al. (1993).
As an example of the Burgess and Tully collision
strength scaling, the
data for the Mg X 1s22s 2S1/2 to
1s23p 2P3/2
dipole transition is shown in Fig. 1 (click here).
The spline fits to the scaled
's show an average deviation of about
0.2%.
The beryllium-like ions contain metastable levels (2s2p 3P0,1,2 in the first excited configuration. Since the early work with Skylab, these ions have provided useful electron density diagnostics for the solar transition region (cf. Gabriel & Jordan 1972; Dupree et al. 1976; Dere & Mason 1981). In particular the diagnostic potential of the C III lines has been extensively explored (Doschek 1997).
A comparison of the laboratory and theoretical transition probabilities for the resonance and intercombination lines in the Be-like sequence has recently been published by Ralchenko & Vainshtein (1996). Although on the whole the agreement is good, there still remain some discrepancies to resolve.
The available electron excitation rates for the beryllium-like ions was reviewed by Berrington (1994). For the low ion stages, it is important to include the channel coupling and resonance effects of the n = 3 states on the n=2 to n=2 transitions. R-matrix calculations are therefore recommended. For the highly ionized systems the CC and DW results are found to be in good agreement.
The adopted atomic model for C III includes 6 configurations: 2s2, 2s2p, 2p2, 2s3s, 2s3p and 2s3d, corresponding to 20 fine structure levels. Experimental energies from the NIST database are available for all levels, and they have been used to calculate transition wavelengths. Radiative transition probabilities for all possible transitions between the 20 levels have been taken from Bhatia & Kastner (1992). No radiative transition probability for the level is 2s2p 3P0 is available in the literature. The decay would involve a forbidden J=0 to J=0 transition to the ground level. Collisional data for transitions between 2s2, 2s2p and 2p2 levels have been taken from the R-Matrix calculations of Berrington et al. (1985). Maxwellian averaged collision strengths are calculated for 9 values of electron temperature in the range 104-106 K. Adopted collisional data for transitions from n=2 to n=3 and between n=3 levels are reported in Berrington et al. (1989): they provide R-Matrix effective collision strengths for 10 values of electron temperature. Both R-Matrix calculations adopt the LS coupling scheme for all levels. In order to transform to intermediate coupling their thermal averaged collision strengths have been scaled using the statistical weights of the levels.
The adopted atomic model for N IV includes 6
different configurations:
2s2, 2s2p, 2p2, 2s3s, 2s3p and 2s3d,
corresponding to 20
fine-structure energy levels. Experimental energy
levels come from the
NIST database and have been used to calculate all
transition
wavelengths. Theoretical energy levels come from
Ramsbottom
et al. (1994). A values and oscillator
strengths come from the
unpublished calculations of Bhatia and are
available for all possible
transitions between the 20 energy levels. Again,
no radiative
transition probability is available for 2s2p
3P0 level.
Maxwellian averaged collision strengths are taken
from the R-Matrix
calculations of Ramsbottom et al. (1994).
values are
provided for 31 values of electron temperature
for all transitions
between the 20 fine structure levels in the range
103-106 K.
For some transitions from Ramsbottom et al. the
behavior of the
effective collision strengths as a function of
electron temperature is
quite complex and the 5-point spline
interpolation technique is not
able to reproduce with the required accuracy
their dependence on 
.
For this reason the values of the effective
collision strengths for
temperatures lower than 103.6 K have been
omitted from the CHIANTI
database, so that the interpolation technique
could reproduce
adequately the behavior of the curve. This
omission of points should
not have any consequence for studies of
transition region and coronal
plasmas where the temperature is much higher than
103.6 K and the
temperature of maximum abundance for N IV is
K.
For O V, the configurations 2s2, 2s2p, 2p2,
2s3s, 2s3p and 2s3d
are included. Energy levels from the NIST
database are available for
all of the 20 fine structure levels. Zhang
& Sampson (1992) provide
electric dipole oscillator strengths and A
values for all allowed
transitions. Additional radiative transition
probabilities are taken
from Muhlethaler & Nussbaumer
(1976) for forbidden and
intercombination transitions. For transitions
from the n=3 levels, the
A values of Hibbert (1980) have
been used.
Zhang & Sampson (1992) also provide relativistic DW collision strengths among the n=2 levels. Collision strengths to the n=3 levels are provided by Kato et al. (1990) who use the R-matrix calculations of Berrington & Kingston (1990). The values of Kato et al. have been recently updated (Kato 1996).
For these members of the beryllium-like sequence, the configurations 2s2, 2s2p, 2p2, 2s3s, 2s3p, 2s3d, 2p3s, 2p3p and 2p3d are included. Energy levels from the NIST database are available for many but not all of the 46 fine structure levels. The number of unidentified levels increases with Z. Zhang & Sampson (1992) provide electric dipole oscillator strengths and A values for all allowed transitions. Additional radiative transition probabilities are taken from Muhlethaler & Nussbaumer (1976) for forbidden and intercombination transitions. Zhang & Sampson (1992) also provide relativistic DW collision strengths among the n=2 levels. The n=2 to n=3 transition probabilities and Coulomb-Born-exchange collision strengths are taken from Sampson et al. (1984).
We note that R-matrix electron impact excitation rates for Ne VII transitions up to the n=3 configurations have recently been published by Ramsbottom et al. (1994, 1995). These will be assessed in the next release of CHIANTI.
The configurations 2s2, 2s2p and 2p2, are
included. Energy
levels from the NIST database are available for
all of the 10 fine
structure levels. Zhang & Sampson
(1992) provide electric dipole
oscillator strengths and A values for all
allowed
transitions.
Additional radiative transition probabilities are
taken from
Muhlethaler & Nussbaumer (1976)
for forbidden and intercombination
transitions which have been interpolated for Al X
and Ar XV. Zhang &
Sampson (1992) also provide relativistic DW
collision strengths among
the n=2 levels. For S XIII and higher,
transitions from the n=3 levels
occur at wavelengths shorter than 50 Å. The
n=3 levels for these
ions have not been included in the present
version of the CHIANTI
database since the additional levels incur a
penalty in computational
speed when solving for level populations.
Nevertheless, they will be
included in future releases of CHIANTI.
Berrington (1994) found very good agreement between the DW results for Ca XVII (Bhatia & Mason 1983) and the R-matrix calculations by Dufton et al. (1983c), even for quite low temperatures. This gives us confidence in the adoption of relativistic DW calculations for the highly ionized Be-like ions.
The adopted atomic model for Fe XXIII includes 9 configurations, 2s2, 2s2p, 2p2, 2s3l and 2s4l corresponding to 30 fine structure energy levels. Experimental energies come from the NIST database and have been used to calculate the transition wavelengths. When no experimental energy level was available, the theoretical value has been used. Theoretical energies for configurations 2s2, 2s2p and 2p2 come from Zhang & Sampson (1992). Theoretical values for the remaining configurations, oscillator strengths and A values come from the unpublished calculations of Bhatia. Radiative transition probabilities are available for all transitions between the levels of the adopted atomic model.
Adopted collision strengths for transitions between levels belonging to 2s2, 2s2p and 2p2 configurations come from Zhang & Sampson (1992) (see previous subsection). Collision strengths to the 3l and 4l come from the unpublished distorted wave calculations of Bhatia (1996). Collision strengths are provided for 3 values of the incident electron energy at 115, 345 and 695 Rydberg for all transitions between the 30 levels.
The 's from Zhang & Sampson have been
compared with those
determined from the earlier work by Bhatia
& Mason (1981, 1986). On the
whole the agreement is found to be excellent at
around 107 K.
However, for a few (non-dipole) transitions, the
high temperature
behaviour differs. This could be due to lack of
convergence for the
high partial waves in Bhatia and Mason's
calculations.
Berrington
(1994) found good agreement between the
Fe XXIII distorted wave
calculations (Bhatia & Mason 1986)
and the sophisticated, fully
relativistic (Dirac) R-matrix calculations
(Norrington & Grant 1987;
Keenan et al. 1993). The
calculations of Bhatia and Mason do not contain
the
2p3l and
2p4l configurations. This places a severe
limitation on the accuracy
of the collision strengths for some of the n=2
to n=3, 4 transitions. This
problem is discussed by Bhatia & Mason
(1981).
The boron isoelectronic sequence has proven a
rich source for
diagnostic ratios in the solar atmosphere (cf.
Vernazza & Mason
1978). For the low ion stages,
transitions from the metastable 
levels in the excited configuration 
,
which fall at around
1400 Å, have been used as a primary diagnostic
for measuring the
electron pressure in the transition region (cf.
Dere et al. 1982). For
the coronal ions (Mg VIII, Si X, A XIV, Ca XVI),
the relative change in
population in the ground levels - 
is reflected in the
intensities of the UV transitions from the
excited configuration
. The intensity ratios of the UV
spectral lines have been used
to determine electron densities in solar flares
(cf. Dere
et al. 1979) and other solar features
(see references in
Dwivedi
1994 and Keenan 1996).
X-ray and XUV lines from the ion Fe XXII arise
from the transitions between the excited
configurations ,
and 
and the ground
configuration (Mason & Storey
1980). These have been recorded in
spectra of solar flares and more
recently in astrophysical sources (Dupree
et al. 1993;
Monsignori
et al. 1994b). These spectral lines can
be used for electron
density determination if the electron density is
high (> 1012
).
Electron excitation data for the B-like ions was
reviewed by
Sampson
et al. (1994). Extensive datasets are
available for this sequence
and include those of
Sampson et
al. (1986) who used their own
non-standard
CB exchange method, with relativistic
corrections. These are generally
in very good agreement with the IC DW
calculations using the UCL code.
These data have now been superseded by more
recent R-matrix and fully
relativistic DW calculations.
The observed energy levels for the 2s22p, 2s2p2, 2p3, 2s23s, and 2s23p configurations are from Kelly (1987) and are complete. Oscillator strengths and A values are taken from Dankwort & Trefftz (1978), Nussbaumer & Storey (1981), Lennon et al. (1985), and Wiese & Fuhr (1995). Radiative transition probabilities for C II have been measured in the laboratory by Fang et al. (1993). R-matrix values of the collision strengths among these levels have been provided by Blum & Pradhan (1992). For the allowed transitions, the collision strengths generally tend to extrapolate to the proper high energy limit derived from the oscillator strength.
Observed energies for the 20 fine structure levels of the 2s22p, 2s2p2, 2p3, 2s23s, 2s23p and 2s23d configurations are from Moore (1993) and Kelly (1987). Oscillator strengths and A values have been provided for all of the same levels by Stafford et al. (1993) except for the ground level fine structure transition where the A value is from Nussbaumer & Storey (1979). Stafford et al. (1994) report R-matrix calculations of Maxwellian-averaged collision strengths among all these configurations.
Figure 2: Collision strengths (Zhang et al.
1994) for Si X 2s22p
2P1/2 - 2s2p2 2D3/2. Left,
the original Maxwellian
averaged collision strengths , right,
the scaled collision
strengths with the high energy limit from the
Bethe approximation. The
scaling follows that of Burgess & Tully (1992)
The 125 fine structure levels of the 2s22p,
2s2p2, 2p3 and
2l2l'3l'' configurations have been
included for
these ions. The energy levels are from the NIST
database
(Martin
et al. 1995) except for the 2s2p2
4P levels which are from
Edlen
(1981). Widing (1996) has
pointed out that the Edlen values seem to
identify several lines of Ar XVI arising from the
2s2p2 4P
levels. For the n=2 configurations, the
observed energies are known.
For the 2l2l'3l'' configurations, most
of the
levels are known for O IV but as Z increases,
the number of levels for
which there are observed energies become fewer
and for Fe XXII only a
minority of these levels have been identified.
Oscillator strengths
and A values for the n=2 levels were obtained
from the same sources as
for C II above with additional data from
Bhatia et al. (1986). For the
n=3 levels, the oscillator strengths from the
unpublished calculations
of Zhang & Sampson (1995) were
used.
Zhang et al. (1994) have provided R-matrix calculations of Maxwellian averaged collision strengths between the 15 fine structure levels in the 2s22p, 2s2p2 and 2p3 configurations. The behavior of the collision strengths at the lowest temperatures (100 - 500 K) can often vary so rapidly that they can not be well represented with a 5 point spline. Often these values have been ignored so that the accuracy at the lowest temperature is about only a factor of 2. Collision strengths to the n=3 levels (110 fine structure levels) have been calculated by Zhang & Sampson (1995) using the Coulomb-Born method.
One complication for the boron sequence is a
level crossing by the
2s2p2 2S
and 2P
levels somewhere
between calcium (Z=20) and iron (Z=26).
Zhang (1995) has confirmed
that their published collision strengths
involving these two levels
must be exchanged to reflect this.
Dankwort & Trefftz (1978) appear
to have simply exchanged labels so that of the
two, the 2P levels always has the higher energy.
Figure 2 (click here) shows an example of the use of
the Zhang et al. R-matrix
collision strengths for the case of Si X. It has
been necessary to
truncate the calculated collision strengths to
temperatures above 3
104 K. Since Si X is formed near 106 K in
collisional
equilibrium, there should be no loss of accuracy
for collisionally
dominated plasmas. The high temperature limit is
determined from the
oscillator strength. In general, it would be
useful to have the
oscillator strength calculated for the same
atomic model used for the
collisional calculations to compare the
values to. The average deviation of
the spline fit to the scaled collision strengths
is about 0.5%.
Ait-Tahar et al. (1996) have
recently carried out a Dirac R-matrix
calculation for transitions between levels in the
n=2 configurations
of Fe XXII. For the high energies, they find 10%
agreement with RDW (Zhang & Sampson
1994) and R-matrix (BP + TCC)
(Zhang & Pradhan 1994, 1995). For
low energies they find some
discrepancies in the resonance structures which
need further
investigation. They have published sample plots
of 
's and
their 's are not yet available.
The atomic model included the same levels as the
O IV - Fe XXII ions
above. The values for the observed energy levels
were taken from NIST
(Martin et al. 1995) and
Edlen (1981). For the n=2
levels, the
oscillator strengths and A values were
generally from
Dankwort &
Trefftz (1978) except for the 2s2p2
2S and
2P levels which were from
Sampson et al. (1986). The
ground term fine structure A value was obtained
by extrapolating the
values of Bhatia et al. (1986).
Oscillator strenghts and A values
involving the n=3 levels are from Zhang
& Sampson (1995). For the
n=2 levels of Ni XXIV, the collision strengths
were taken from the
Coulomb-Born-Exchange calculations of
Sampson et al. (1986). For the
n=3 levels, the collision strengths are those
of Zhang & Sampson
(1995).
Lines from ions in the carbon isoelectronic
sequence have been studied
in spectra from the transition region, corona and
flares. The ground
configuration 
comprises 
, 
and 
levels.
The relative populations of these levels can be
sensitive functions of
density. References are given in Mason &
Monsignori Fossi (1994),
Dwivedi (1994) and Keenan
(1996).
Monsignori Fossi & Landini (1994c)
assessed the electron excitation data for the
C-like ions.
Burgess
et al. 1991 have critiqued the R-matrix
calculations of
Aggarwal (1985b) for
Mg VII. They find his treatment for the allowed
transitions
inadequately accounts for the long range
potential. Errors on the
order of 15% could arise and they point out that
there may be similar
problems with allowed transitions of other ions
in this isoelectronic
sequence calculated in a similar manner by
Aggarwal.
In addition to the density diagnostic ratios for the coronal lines, the solar flare lines from Fe XXI are of particular interest. The UV line at 1354.1 Å has been studied from Skylab and SMM observations and more recently stellar spectra have been obtained with the Goddard High Resolution Spectrograph on the Hubble Space Telescope (Maran et al. 1994). Highly ionized iron lines have been recorded for several stars with EUVE, and electron density values have been deduced for the stellar atmospheres from the Fe XXI line ratios (Dupree et al. 1993).
The configurations 2s22p2, 2s2p3, 2s22p3s, 2s22p3p, corresponding to 23 fine-structure levels have been included. The experimental energy levels of the NIST database were adopted. The oscillator strengths and A values of Bell et al. (1995) have been adopted. Since Bell's work did not provide radiative transition probabilities for several transitions, a new calculation has been performed using the SSTRUCT package of the University College of London. The new calculation included 17 configurations: 2s22p2, 2s2p3, 2p4, 2s22p3s, 2s22p3p, 2s22p3d, 2s2p23s, 2s2p23p, 2s2p23d, 2s22p4s, 2s22p4p, 2s22p4d, 2s22p4f, 2s2p24s, 2s2p24p, 2s2p24d and 2s2p24f. The resulting radiative transition probabilities have been corrected for the differences between experimental and theoretical values of the transition's wavelengths. The corrected values have been inserted where Bell's values were unavailable.
The effective collision strengths for
all transitions are
taken from the R-matrix calculations of
Stafford et al. (1994).
Effective collision strengths are calculated in
the electron
temperature range 5 103 to 1.25 105 K.
Lennon & Burke (1994)
have also recently published collision data for
the C-sequence,
including N II. Some of their values differ from
Stafford et al. by
more than 50%. The problem is that the position
of a large resonance
near threshold dramatically effects the low
temperature
values. The differences between the two sets of
calculations must be
taken as a measure of the uncertainty in the
calculations. The exact
position of such resonances is extremely
difficult to determine.
The adopted atomic model for O III includes 46
fine structure levels
of the 2s22p2, 2s2p3, 2p4,
2s22p3s, 2s22p3p, and
2s22p3d configurations. Experimental energies
for all levels come
from the NIST database and have been used to
calculate transition
wavelengths. Oscillator strengths and A values
for all possible
transitions between the 46 energy levels are
taken from Bhatia &
Kastner (1993a). For transitions
between the levels of the ground
configuration (first 5 levels) the R-Matrix
Maxwellian averaged
collision strengths of Lennon & Burke
(1994) were adopted. They report
values in the electron temperature range
103-105 K. Effective
collision strengths for optically allowed
transitions between the
levels belonging to 2s22p2 and 2s2p3,
2p4 configurations
have been calculated by Aggarwal
(1985a) for temperatures between 5
103 and 5 105 K. Collision strengths for
the remaining
forbidden and intercombination transitions among
these levels are
provided by Aggarwal (1983) for
temperatures between 2.5 103 and 6
105 K. Both calculations have been performed
using the R-Matrix
method. A comparison between the Maxwellian
averaged collision
strengths from Lennon & Burke
(1994) and
Aggarwal (1983) for transitions
between the ground configuration levels shows no
significant
differences. Collision
strengths for transitions to
the n=3 levels are taken from the distorted
wave calculations of
Bhatia
& Kastner (1993a). Collision strengths
are reported for 5 values of
incident electron energy at 4, 6, 8, 10 and 12
Rydberg.
The adopted atomic model for Ne V includes 6 configurations, 2s22p2, 2s2p3, 2s22p3s, 2p4, 2s22p3p, and 2s22p3d, corresponding to 46 fine structure energy levels. Experimental energy levels are taken from the NIST database for all the levels, with the exceptions of levels 2p4 1D2 and 1S0, which come from Edlen (1985). A values and oscillator strengths come from Bhatia & Doschek (1993b), and are provided for all the possible transitions between the 46 energy levels.
Maxwellian averaged collision strengths for transitions between the ground levels (configuration 2s22p2 and 2s2p3 5S2) are taken from Lennon & Burke (1994). Collision strengths for all other the transitions come from the distorted wave calculations of Bhatia & Doschek (1993b). Collision strengths are provided for 3 values of incident electron energies at 10, 15 and 20 Rydberg. Collision strengths have also been calculated using the R-Matrix by Aggarwal (1985b, 1986) for allowed transitions and by Aggarwal (1984) for forbidden and intercombination transitions.
The configurations 2s22p2, 2s2p3, 2s22p3s, 2p4, 2s22p3p, 2s22p3d, corresponding to 46 fine-structure levels, were included in the atomic models for these 2 ions. Values for the observed energy levels were taken from the NIST database. Transition wavelengths were calculated using observed energy levels, but where these values were not available theoretical energy levels were used. Theoretical energy levels, oscillator strengths, collision strengths and spontaneous transition probabilities for these ions are taken from the calculations of Bhatia & Doschek (1993c, 1995b) for Si IX and Mg VII respectively. Collision strengths were calculated using the distorted wave approximation for three values of the incident electron energy in the range 12-36 Rydberg (Mg VII) and 20-60 Rydberg (Si IX). For these ions, the incident energies are fairly high above threshold and closely spaced in energy. This presents a problem in extrapolating the collision strengths over the complete energy range from threshold to infinite energy.
R-matrix calculations of allowed transitions in Mg VII and Si IX have been performed by Aggarwal (1985b, 1986) but may be somewhat inaccurate as discussed by Burgess et al. (1991).
The adopted atomic model for S XI includes the configurations 2s22p2, 2s2p3, 2p4, 2s22p3s, 2s22p3p and 2s22p3d, corresponding to 46 fine structure energy levels. Experimental energies were taken from the NIST database while the theoretical energies for all the levels come from Bhatia & Kastner (1987). Wavelengths have been calculated with the experimental energy levels but where these values were not available the theoretical energies have been used. Radiative transition probabilities and oscillator strengths are reported by Bhatia & Kastner (1987) for all possible transitions between the levels of the adopted atomic model.
Collision strengths for transitions between all
the 46 levels are
available from Bhatia & Kastner
(1987). In this paper collision
strengths
are calculated for only one value of the incident
electron energy at 25
Rydberg. The calculation of the
Maxwellian-averaged collision
strengths is then very arbitrary. For this
reason collision strengths
for all transitions among the ground levels
(2s22p2
configuration) and between the ground
configuration and the 2s2p3
configuration have been taken from Mason &
Bhatia (1978). In this
paper collision strengths are reported for 3
values of incident
electron energy, and the calculation of the
Maxwellian-averaged
collision strengths is better defined. For the
remaining transitions,
the effective collision strengths have been
calculated using the single
energy collision strengths. For these transitions
we have artificially
introduced the 
values
at threshold and for
so that they are similar to energy
dependence of the
same transition for other members of the
isoelectronic sequence. A
comparison between the effective collision
strengths of Mason &
Bhatia (1978) and those of Bhatia
& Kastner (1987) show differences of
up to 30%.
The adopted atomic model for Ar XIII includes the configurations 2s22p2 and 2s2p3, corresponding to 15 fine structure energy levels. Experimental energies come from the NIST database and have been used to calculate transition wavelengths. Theoretical energy levels, A values, oscillator strengths and collision strengths for all the possible transitions have been taken from Dere et al. (1979). Collision strengths are calculated using the distorted wave approximation, including the configurations 2s22p2, 2s2p3 and 2p4, for 3 values of the incident electron energy at 15, 30 and 45 Rydberg.
The adopted atomic model for Ca XV includes the configurations 2s22p2 and 2s2p3, corresponding to 15 fine structure energy levels. Experimental energies come from the NIST database and have been used to calculate transition wavelengths. Oscillator strengths and A values are from the calculations of Froese-Fischer & Saha (1985).
For the Ca XV 2s22p2 and 2s2p3
configurations, we use the
R-matrix collision strengths of Aggarwal et
al. (1991a). We have also
considered distorted wave calculations for this
ion (Bhatia & Doschek
1993a; Dere et al. 1979).
For the transitions within the n=2
configurations, it was not possible to obtain
satisfactory fits to the
Bhatia & Doschek (1993a)
collision strengths because the incident
energies are too closely bunched and too high
above threshold. In
these cases, extrapolation of the collision
strength values to
threshold and the high energy limit (nondipole)
is very uncertain.
Dere et al. (1979) published
collision strengths for Ca XV at lower
energies. Although most of their values matched
those of Bhatia &
Doschek at 45 Ryd, a few transitions differed.
The probable reason is
that Dere et al. did not carry out the DW
calculations to partial wave
values as high as Bhatia and Doschek. The Dere et
al. values at 45 Ryd
may not have fully converged. Consequently, it
was not possible to
construct a consistent set of distorted wave
collision strengths that
spanned a sufficient range in energy. In
addition, the resonance
contribution is significant for the forbidden
transitions within the
ground configuration. Indeed, for some of these
transitions, the
's at 3 106 derived from
Bhatia & Doschek (1993a)
differ by more than a factor 2 from Aggarwal's
values.
For Fe XXI, 36 fine-structure levels in the configurations 2s22p2, 2s2p3, 2s22p3s, 2s22p3d, 2p4 were included. The observed energies are mostly from the NIST database and supplemented by a few from Mason et al. (1979). Oscillator strengths and A values have been computed using the SSTRUCT package. The adopted atomic model includes 17 configurations: 2s22p2, 2s2p3, 2p4, 2s22p3s, 2s22p3p, 2s22p3d, 2s2p23s, 2s2p23p, 2s2p23d, 2s22p4s, 2s22p4p, 2s22p4d, 2s22p4f, 2s2p24s, 2s2p24p, 2s2p24d and 2s2p24f. The calculated radiative transition probabilities have been corrected for the differences between experimental and theoretical values of each transition's wavelength.
Collision strengths from Aggarwal (1991) have been adopted for transitions involving the 2s22p2, 2s2p3 2p4 configurations. They have been calculated using the close-coupling method for 20 values of electron energy in the range 20-300 Rydberg. Collision strengths for transitions from the ground configuration to 2s22p3s and 2s22p3d configuration come from Mason et al. (1979). They are calculated with the distorted wave method for 3 values of electron energy in the range 20-100 Rydberg. The energies specified for the incident electron cover a sufficiently large range, at least for the n=2 excitations, that the Maxwellian integral is well determined. For several transitions, a comparison with the R-matrix calculations of Aggarwal (1991) was made and generally the agreement was found to be quite good. It is noted that electron collision rates to the n=4 levels have recently been published by Phillips et al. (1996). These will be included in future releases of CHIANTI.
The nitrogen isoelectronic sequence provides some interesting diagnostics for coronal ions, cf. Bhatia & Mason (1980a) and recent work by Dwivedi, Mohan and colleagues referenced in Dwivedi (1994). Kato (1994) has reviewed the existing data for the nitrogen isoelectronic sequence. She found that for several important ions of this sequence very few data are available (Ne IV to Mn XIX), while O II and Fe XX have been studied quite extensively. Most of the existing calculations have been carried out using the distorted wave approximation.
The adopted atomic model for O II includes the configurations 2s22p3, 2s2p4 and 2p5, corresponding to 15 fine-structure energy levels. The experimental energies are taken from the NIST database. These energies were used to calculate the transition wavelengths. Since no experimental energy values were available for the configuration 2p5, wavelengths for transitions originated from the 2p5 levels were computed using theoretical energies.
Maxwellian averaged collision strengths are taken from McLaughlin & Bell (1994). Effective collision strengths are provided for all transitions between 11 LS levels (corresponding to configurations 2s22p3, 2s2p4, and 2s22p23s). The collisional data were calculated using the R-Matrix method and fitted with a Chebyshev polynomial expansion and tabulated. Effective collision strengths among the fine structure levels have been obtained from the reported LS coupling values by scaling according to the statistical weights of the fine-structure levels involved in each transition. Levels in the 2s22p23s configuration have not been included since radiative data for the 2s22p23s levels were not available. These levels will be included in future releases of CHIANTI.
For transitions involving the 2p5 configurations and for transitions between the levels of each multiplet, collision strengths have been taken from the unpublished distorted wave calculations Bhatia (1996), provided for 3 values of incident electron energy at 4, 8 and 12 Rydberg. Since collision strengths for all other configurations are provided, a comparison has been carried out between these data and those of McLaughlin & Bell (1994). The two sets of data show significant differences both in the allowed and in the forbidden transitions.
The adopted atomic model for Ne IV includes 3 configurations: 2s22p3, 2s2p4 and 2p5, corresponding to 15 fine-structure energy levels. The experimental energies are taken from the NIST database and are used to calculate the wavelengths.
The radiative data and collision strengths have been taken from unpublished distorted wave calculations of Bhatia (1996). Collision strengths are calculated using the distorted wave method for 3 values of electron energy at 5, 10 and 15 Rydberg.
The atomic model includes two configurations: 2s22p3 and 2s2p4, corresponding to 13 fine-structure levels. Energy levels and transition wavelengths have been calculated using the experimental energy levels of the NIST database.
Radiative transition probabilities and
collision
strengths for all
transitions within 2s22p3 and between the
2s22p3 and
2s2p4 configurations are included and are from
the calculations of
Bhatia & Mason (1980a). The
scattering calculations have been
carried out using the configurations
2s22p3, 2s2p4 and 2p5,
with the distorted wave method, and collision
strengths are computed
for 3 values of the electron energy between 10
and 45 Rydberg.
The configurations included in the atomic model are 2s22p3 and 2s2p4, corresponding to 13 fine-structure energy levels. The relevant atomic data for Fe XX are taken from Bhatia & Mason (1980b). Both theoretical and experimental energy levels are provided by the authors. Transition wavelengths have been calculated using the experimental values. Collision strengths have been calculated using the distorted wave method for 3 values of electron energy: 20, 50 and 100 Rydberg.
Spectroscopic diagnostics for the oxygen-like
transition region and
coronal ions were explored by Raju &
Dwivedi (1978).
The solar flare lines from Fe XIX have been
studied in XUV
(Loulergue
et al. 1985) and X-ray
(Bhatia et al. 1989)
spectra.
The oxygen isoelectronic sequence has not been
treated extensively in
the literature. The only close-coupling
calculations are those of
Butler & Zeippen (1994) who have
calculated thermal-averaged
collision strengths for transitions between the
levels within the
ground configuration of ions from F II to Ar XI,
for electron
temperatures ranging from 103 to 105 K.
Distorted wave
calculations have been performed for all the most
abundant ions,
including transitions between the ground
configuration and excited
ones. A comparison between distorted wave and
close-coupling collision
strengths for the lighter ions shows a generally
acceptable agreement.
Only S IX shows some differences between the two
different methods,
where the close coupling data sometimes show
greater thermal-averaged
collision strengths at low temperatures. Since
the maximum abundance
temperature for oxygen-like ions with Z 
16 is 106 K
we have adopted distorted wave calculations also
for the transitions in
the ground configuration for Si VII and above.
The case of Ne III and Mg V is different. The electron temperature for the maximum abundance for these ions falls in the range 105-105.5 K and it is important to include the close-coupling effective collision strengths since the effects of resonances are important.
The main deficiency of the adopted distorted wave collision strengths is that they have been calculated only for three values of the incident electron energy near threshold. Consequently, the behavior of the non-dipole collisions strengths at high energy is not well determined.
For Ne III and Mg V, the 10 fine-structure levels belonging to the 2s22p4, 2s2p5 and 2p6 configurations have been included. Observed energy levels are taken from the NIST database and they are used to calculate the transition wavelengths. The radiative data are from the unpublished calculations of Bhatia (1996) which also provide collision strengths for all the transitions between these 10 levels. Distorted wave collision strengths for Ne III are computed for three values of the incident electron energy in the range 5-15 Rydberg, and for Mg V, 4 values of electron energy are available, in the range 10-30 Rydberg.
Butler & Zeippen (1994) report close-coupling effective collision strengths for the ground configuration transitions. These effective collision strengths are provided for 11 values of the electron temperature ranging from 103 to 105 K. This temperature range is insufficient for an accurate extrapolation to high energy. The collision strengths provided by Bhatia do allow a good determination of the effective collision strengths in the temperature range 106 - 5 106 K. Thus the two different sets of data have been merged and the resulting effective collision strengths have been inserted in the CHIANTI database. No significant inconsistencies between the two different sets of data were found.
Bhatia's collision strengths have been adopted for all other transitions.
The atomic model includes the 2s22p4,
2s2p5 and
2p6 configurations, corresponding to 10 fine
structure energy
levels.
Experimental energy levels from the NIST database have been adopted. These values have been used to calculate all the transition wavelengths. The experimental energy of the 2p6 1S0 level of Ar XI is unknown and all the wavelengths of the transitions involving this level have been computed using theoretical energies. Theoretical energy levels, radiative transition probabilities and collision strengths for Si VII, S IX and Ar XI have been provided by Bhatia et al. (1979).
The scattering problem has been solved using the distorted wave approximation. Collision strengths have been calculated for 3 values of electron energy: 10, 15, 20 Rydberg for Si VII and S IX, and 15, 20, 25 Rydberg for Ar XI. Again, this is a fairly restricted range in energy for determining the energy dependence of the collision strengths.
The atomic model of Ca XIII includes 4 configurations: 2p4, 2s2p5, 2p6 and 2p33s, corresponding to 20 fine-structure energy levels. The experimental energies of the NIST database have been adopted, but the values for the levels 2p33s 5S2 and 3P0,1,2 are unknown. Transition wavelengths have been calculated using the experimental energy levels, but if these values were not available, the theoretical ones have been used. The theoretical energy levels, radiative transition probabilities and collision strengths are taken from the unpublished calculations of Bhatia (1996).
The collision strengths were computed using the distorted wave approximation. They have been calculated for 5 values of the electron energy in the range between 40 and 60 Rydberg. The collisional data are available for all transition between and within the listed configurations. These data agree well with Mason (1975) for the n=2 transitions.
Baliyan & Bhatia (1994) published
R-matrix collision
data for Ca XIII with a more extensive set of
target configurations.
Unfortunately, their results are in LS coupling
and they do not
tabulate the 's, so we have not used
them in the CHIANTI
database. However, they have carried out a very
careful
comparison of the R-matrix and distorted wave
results, giving
particular attention to the partial wave
convergence.
In general, they found good agreement
(better than 30%) with the Bhatia distorted wave
results, although
they suggest that the target wavefunctions could
be improved.
They also found the resonance structure to be
significant for low energies.
The adopted atomic model includes 2s22p4, 2s2p5 and 2p6 configurations, corresponding to 10 fine-structure energy levels. Theoretical energy levels, radiative transition probabilities and collision strengths come from Loulergue et al. (1985), who also provide the experimental energy levels. The latter energies are used to calculate the transition wavelengths.
The authors show that for calculating the radiative transition probabilities, there is no significant improvement in extending this basis to higher configurations than those listed above. The collision strengths are calculated using the distorted wave method for 3 values of incident electron energy, 22.5, 45 and 90 Rydberg.
For the fluorine sequence ions Ne II through Ca XII, only the two lowest configurations having a total of 3 fine structure levels have been included: 2s22p5 2P3/2,1/2 and 2s2p6 2S1/2. The experimental energy levels are from Kelly (1987) for Ne II and from the NIST database (Martin et al. 1995) for the others. A values between the ground configuration levels are from Martin et al. (1995) and from Blackford & Hibbert (1994) for the others. Collision strengths between the 2 ground configuration levels are from the calculations of Saraph & Tully (1994). Sources for collision strength to the 2s2p6 level are: Ne II (Bhatia 1996, unpublished), Mg IV (Mohan et al. 1988), Si VI (Mohan & Le Dourneuf 1990), S VIII (Mohan et al. 1987) and Ca XII (Mason 1975).
The 113 fine structure levels of the
2s22p5, 2s2p6,
2s22p43l, 2s2p53l and
2p63l
configurations are included. Observed energy
levels have been taken
from the NIST database (Martin et al.
1995). The primary source of
oscillator strengths and A values were the
calculations of
Sampson
et al. (1991). Additional radiative
rates were taken from
Blackford
& Hibbert (1994) and Cornille et
al. (1992).
Sampson et al. (1991)
have calculated relativistic distorted-wave
collision strengths for
ions with Z 22 and have been incorporated
into the CHIANTI
database. These calculations are generally in
good agreement with the
distorted wave calculations of Cornille et
al. (1992) for Fe XVIII.
The adopted atomic model includes 4 configurations: 2s22p6, 2s22p53s, 2s22p53p and 2s22p53d, corresponding to 27 fine-structure levels. The experimental energy levels are taken from the NIST database and are used to calculate all the transition wavelengths. Radiative transition probabilities and collision strengths come from the calculations of Bhatia et al. (1985). Radiative data are available for all transition between the listed configurations.
Collision strengths were calculated using the distorted wave method for only one value of the incident electron energy, 15 Rydberg. For dipole transitions, a second energy point is available in the Bethe approximation high energy limit so that the integration over a Maxwellian velocity distribution can be performed. For the nondipole transitions, the variation of the collision strength with energy has been estimated from the same transition in Fe XVII, either by setting the collision strength to be constant with energy or to go to zero at high energies.
The adopted atomic model for S VII includes four configurations: 2s22p6, 2s22p53s, 2s22p53p and 2s22p53d corresponding to 27 fine-structure levels. The adopted experimental energy levels are taken from the NIST database and they are used to calculate all the transition wavelengths. The theoretical energy levels, A values and oscillator strengths are taken from Hibbert et al. (1993). Radiative data are calculated for transitions to the ground configuration, between 2s22p53s and 2s22p53p, between 2s22p53p and 2s22p53d. A values for the forbidden transitions 2s22p6 1S0 - 2s22p53s 3P2 and 2s22p53s 3P1 - 3P0 have been extrapolated from the corresponding values of the other ions of the neon isoelectronic sequence.
Thermal-averaged collision strengths are taken from Mohan et al. (1990). The collision strengths are calculated using the close coupling method for all transitions from the ground level to 2s22p53s, 2s22p53p and 2s22p53d configurations. The data are tabulated for 20 values of electron temperature ranging from 5 104 to 106 K. These data are calculated in LS coupling. Since the ground level consists of a single level, fine structure thermal-averaged collision strengths for transitions involving multiplets have been distributed according to their statistical weights.
The included configurations are 2s22p5nl and 2s2p6nl with n=3, 4 and l=0, 1, 2, 3. This atomic model includes 89 fine structure levels. The experimental energy levels are taken from the NIST database. These values have been used to calculate transition wavelengths, but where experimental energy levels were lacking (Ca XI and Ni XIX), theoretical values of Zhang et al. (1987) have been used.
Zhang et al. (1987) also calculated radiative transition probabilities for electric dipole transitions between the ground level and the excited levels. These rates provide decay rates for only a few (14) of the upper levels. Additional oscillator strengths and A values have been taken from Hibbert et al. (1993). Nevertheless for the majority of energy levels belonging to the higher energy configurations 2s22p54l and 2s2p6nl, radiative data are not available. Consequently, it is not possible to construct a model of the statistical level populations that include these levels and so they have been omitted from the adopted atomic model. Since these levels have very high energy and they are not metastable we do not expect any significant change in the level populations and line intensities in the synthetic spectra of these ions. The adopted atomic model includes all levels of configurations 2s22p6, 2s22p53s, 2s22p53p and 2s22p53d, and the levels 2s2p63p 3P1, 1P1, 2s22p54s 3P1 and 1P1, 2s22p54d 3P1, 3D1, 1P1, 2s2p64p 3P1 and 1P1. The total number of energy levels is 36.
Collisional data are taken from Zhang et al. (1987) who employed a Coulomb-Born-exchange method. The collision strengths are calculated for 9 values of the incident electron energy in threshold units in the range 1-15, allowing a straightforward determination of the collision strength values from threshold to the high energy limit. Collisional data are available for all transitions connecting the ground level and the 88 excited levels. Collision strengths for transitions involving the omitted levels have not been inserted in the database, though they are available in CHIANTI format.
The adopted atomic model includes seven configurations: 2s22p6, 2s22p53s, 2s22p53p, 2s22p53d, 2s2p63s, 2s2p63p, 2s2p63d corresponding to 37 fine structure energy levels. The experimental energy levels for the lowest 26 levels are from Bhatia & Doschek (1992) and the rest are from the NIST database. It should be noted that Bhatia & Doschek preserve the term designations of the lower Z members of the sequence. The experimental energies have been used to calculate the transition wavelengths. Where no experimental energy values were available, theoretical ones have been used. Theoretical energy values, A values, oscillator strengths and collision strengths are provided by Bhatia & Doschek (1992) for all transitions between the 37 levels of the adopted atomic model. Distorted wave collision strengths are calculated at five incident electron energies in the range 77-254 Rydberg. These collision strengths can be compared those from a similar calculations by Cornille et al. (1994).
The 21 fine structure levels of the 3
,
4, and 5 configurations have been included. The
observed energy levels are
from the NIST database (Martin et al.
1995). Oscillator strengths and
A values are from Sampson et al.
(1990) except for the 
transitions where the hydrogenic values of
Wiese et al. (1966) are
used.
Relativistic distorted wave collision strengths
have been provided by
Sampson et al. (1990) over a wide
energy range and are used in the
CHIANTI database.
Cornille et
al. (1997) have calculated distorted
wave results for Fe XVI (n=3, 4 and 5). They
discuss the importance of
high partial wave contributions, even for the
electric quadrupole
transitions. Their values agree well with those
of
Sampson et al.
(1990). Tayal (1994)
published R-matrix values for the n=3, 4
configurations. He included the resonance
contribution, which is
particularly important at lower temperatures (<
106 K). New R-matrix
calculations are in progress by Eissner et
al. (1996) as part of the
Iron Project. They find that some collision rates
published by Tayal
(1994) are very sensitive to the position
of large resonances near
threshold.
The adopted atomic model for Si III includes the configurations 3s2, 3s3p, 3p2, 3s3d, 3s4s and 3s4p, corresponding to 20 fine structure energy levels. The experimental energy levels are taken from the NIST database and have been used to calculate the transition wavelengths. The theoretical energy levels come from Baluja & Hibbert (1980). Oscillator strengths and A values come from Dufton et al. (1983a). Since only a few transitions are treated in that paper, we have been performed new calculations for all the possible transition in the adopted atomic model. The calculation has been carried out with SSTRUCT, including the configurations 3s2, 3s3p, 3p2, 3s3d, 3s4s, 3s4p, 3s4d, 3s4f, 3p3d, 3d2, 3p4s, 3p4p, 3p4d and 3p4f. The resulting A values and oscillator strengths have been adopted for all the transition for which values were not provided by Dufton et al. (1983a). A comparison between SSTRUCT results and Dufton et al. (1983a) radiative transition probabilities shows no significant differences.
Adopted collision data come from Dufton & Kingston (1989). They provide coefficients for least-squares polynomial fits for thermal averaged collision strengths between the 20 lowest states (corresponding to the 20 fine-structure levels adopted by CHIANTI). Thermal averaged collision strengths have been calculated with the R-matrix program for the temperature interval between 103.8 and 105.2 K, for which the Si III ionization fraction is significant. Fine structure transitions between triplets were reported, while transitions between singlets and triplets are presented in LS coupling. The authors show that the values for fine structure transitions between singlets and triplets may be obtained by scaling the collisional data by the appropriate degeneracies which we have done here.
The adopted atomic model for S V, Ar VII, Ca IX and Ni XVII includes the configurations 3s2, 3s3p, 3p2, 3s3d and 3s4s, corresponding to 16 fine-structure levels. The experimental energy levels are taken the NIST database. Experimental energy levels have been used to calculate the transition wavelengths. Where no experimental values were available, the theoretical ones have been used. Theoretical energy levels, A values, oscillator strengths and collisional data are taken from Christensen et al. (1986). Radiative transition probabilities are available for electric dipole transitions between 3s2 and 3s3p configurations, and between 3s3p and 3p2, 3s3d and 3s4s configurations. Collision strengths are calculated with the distorted wave method for three incident electron energies for all possible transition between the 16 fine-structure levels of the adopted atomic model.
The atomic model for Fe XV includes configurations 3s2, 3s3p, 3p2, 3s3d, 3p3d, 3s4s, 3s4p, 3s4d, 3p4s, 3p4d, 3p4f. The radiative data and electron scattering data are taken from Bhatia et al. (1997). The collision strengths, calculated in the distorted wave approximation, are given for 3 energies, 25, 50 and 75 Rydbergs. Fits were made from the levels in the 3s2 and 3s3p configurations up to all the other configurations. One limitation of these calculations was the omission of the 3d2 configuration. New calculations for the n=3 configurations have just been completed (Bhatia & Mason 1997). These are in good agreement with the earlier distorted wave results by Christensen et al. (1985). Dufton et al. (1990) compared their R-matrix results with Christensen et al. and also found good agreement (better than 20%). They did not publish their collision data and it is not available. Eissner et al. (1996) are carrying out new R-matrix calculations for Fe XV as part of the Iron Project. They find that, contrary to Dufton et al. 's conclusions, the resonance structure is important for some of the Fe XV transitions and could make a significant difference to the excitation rates.
The atomic model for Si II includes five configurations, 3s23p, 3s3p2, 3s23d, 3s24s and 3s24p, corresponding to 15 fine structure levels. The experimental energy levels are taken from the NIST database and are used for calculating the transition wavelengths. Experimental energies are available for all levels. Electric dipole allowed transitions are taken from Lanzafame (1994). Radiative transition probabilities for forbidden and intercombination transitions are obtained with SSTRUCT. The following configurations have been included in the calculation: 3s23p, 3s3p2, 3s23d, 3s24s, 3s24p, 3s24d, 3s24f, 3p3, 3s3p3d, 3s3d2, 3p3d2, 3d3, 3p23d, 3s3p4s, 3s3p4p, 3s3p4d and 3s3p4f. These radiative transition probabilities have been calculated using the theoretical energy levels but have been corrected to take into account the experimental energies.
Thermal averaged collision strengths are taken from the R-matrix calculations of Dufton & Kingston (1991a). Thermal averaged collision strengths are tabulated for 6 values of the electron temperature ranging from 103.6 to 104.6 K.
The adopted atomic model for S IV includes the 3
configurations
3s23p, 3s3p2 and 3s23d, corresponding to
12 fine-structure
energy levels. Experimental energies were
available from the NIST
database for all the included levels, and have
been used to calculate
transition wavelengths. Oscillator strengths
and A values are mainly
taken from Bhatia et al. (1980),
who provide values for transitions
between all the levels of the adopted atomic
model. Radiative decay
rates for the intercombination lines between
levels 3s23p 2P and
3s3p2 4P have been taken from
Dufton et al. (1982).
Collisional data come from 3 sources. For
transitions between
2P and 3s3p2 4P, collision
strengths
have been taken from the R-matrix calculations of
Dufton et al. (1982). They present
Maxwellian-averaged collision strengths in
the electron temperature range 104 to
105.6 K. A comparison
between these results and the close-coupling
calculations of
Bhadra &
Henry (1980) and the distorted wave
values of
Bhatia et al. (1980) has
shown that the Dufton values are larger than the
others at low
temperatures, due to scattering resonances. The
main difference between
Dufton's calculations and Bhadra's and Bhatia's
is that the former
explicitly delineates the resonances in the
collision strengths.
Collision strengths for transitions from 3s23p
2P and
3s3p2 4P to the remaining levels of the
3s3p2 configuration
come from Bhadra & Henry (1980). They provide
fine structure
collision strengths for five values of the
incident electron energy in
the range 1.3 to 6 Rydberg, calculated with the
close-coupling method.
For transitions involving the 3s23d
configuration, the Bhatia
et al. (1980) collision strengths have been
adopted. They are
calculated with the distorted wave method for
three values of incident
electron energy in the range 2 to 6 Rydberg.
The model for Fe XIV includes 12 fine structure
levels of the
, 
and 
configurations. The observed energy
levels are from the NIST database. Transition
probabilities are taken
from Froese Fischer & Liu (1986).
The R-matrix calculations of Dufton & Kingston (1991b) give upsilons for all transitions from the ground 3s23p configuration to the 3s3p2 and 3s23d configurations and are used here. No significant differences were found between this work, the earlier work by Mason (1975) and the more recent work of Bhatia & Kastner (1993b). Storey et al. (1996) give upsilons for the ground 2P1/2 - 2P3/2 transition calculated using the R-matrix codes. Only the upsilons above 104 K were fitted as the simple 5 point spline could not be applied accurately to the complete set of upsilons.
Although the Storey et al. upsilons were around 50% greater than the Dufton & Kingston upsilons, this had little effect on the level balance as the dominant process populating the 2P3/2 level is cascades from upper levels.
The model of the Fe XIII ion includes 27 fine structure levels belonging to the 3s23p2, 3s3p3 and 3s23p3d configurations. The observed energy levels are from the NIST database. A 24 configuration model of Fe XIII was used in SSTRUCT to generate transition probabilities. Most importantly, transition probabilities were derived to de-populate the metastable 3s23p3d 3F4 level, which accounts for some 20% of the level balance at densities of around 1012-1013 cm-3. In the work of Brickhouse et al. (1995), this level was omitted as there existed no data in the literature to de-populate the level, and explains the discrepancies between the CHIANTI results and theirs.
The distorted wave calculations of Fawcett & Mason (1989) were used in favor of the more recent R-matrix calculations of Tayal (1995), principally because the R-matrix results differed for some transitions (e.g., 3-13, 4-14) by a factor of 2, the reasons for which were not adequately explained. Also, upsilons were not published by Tayal for transitions up to the 3s23p3d configuration or for the 3s3p3 5S2 level. It is to be noted that Fawcett & Mason (1989) provide only collision strengths for transitions out of the ground configuration and so the effects of electron collisional excitation out of the metastable 3F4 level have not been included.
As with Fe XIII, the model of the Ni XV ion includes 27 fine structure levels belonging to the 3s23p2, 3s3p3 and 3s23p3d configurations. Experimental energies were available for only 17 of the 27 levels and so theoretical values were used for the remaining levels.
Accurate transition probabilities were calculated using SSTRUCT with a 35 configuration model. Again, we find that magnetic quadrupole transitions are significant in helping to depopulate the metastable 3s3p23d 3F4 level. We also note that there are no collision strengths available that might allow this level to be de-populated by electron collisional processes to configurations other than 3s23p2, which may affect the level population significantly.
The distorted wave calculations of Mason
(1996b) are used to
provide collision strengths at one energy (10
Ryd) for all of the
3s-3p and 3p-3d transitions. Four
configurations were included in
these calculations--3s23p2, 3s3p3,
3s3p23d and
3p4--but collision strengths were not
calculated for transitions to
or from 3p4. For dipole
transitions, the single collision strength
and the value of the high energy limit derived
from the oscillator
strength define the energy dependence of the
collision strength. For
other transitions, the collision strength is
assumed to be constant
with energy.
Forbidden lines of S II such as 
6717 and
6731 are
observed in planetary nebulae and their intensity
ratio provides a good
indicator of electron density for values near
103 cm-3
(Osterbrock 1989). Fe XII line
ratios are sensitive diagnostics of electron
density for coronal conditions (Dere
1982).
For S II, 28 fine structure levels of the 3s23p3, 3s3p4, 3s23p23d, and 3s23p24s configurations have been included. Energy levels based on observed spectra are from the NIST database (Martin et al. 1995). Sources for oscillator strengths and A values for S II include Mendoza & Zeippen (1982), Ho & Henry (1983), Huang (1984), Fawcett (1986). There are large differences among these various sources for the values of the radiative constants. In addition, none provide a radiative decay rate for the 3s23p23d 4F9/2 level. Consequently, we have used a single set of allowed and forbidden radiative rates for S II calculated with the SSTRUCT program (Binello 1996). The effective collision rates have been calculated by Cai & Pradhan (1993) using the R-matrix program.
We note that new R-matrix calculations have recently been carried out for S II by Ramsbottom et al. (1996), including 43 fine-structure levels, up to the n=4 configurations. These will be assessed in the next release of CHIANTI.
The model of the Fe XII ion includes 41 fine structure levels belong to the 3s23p3, 3s3p4 and 3s23p23d configurations. Energy levels are from the NIST database and Jupen et al. (1994). Transition probabilities were taken from a recent SSTRUCT run performed by Binello (1996) using a 24 configuration model for Fe XII.
No advances over the distorted wave calculations of Flower (1977) have been made for the important EUV transitions of Fe XII. Tayal et al. (1987) presented R-matrix upsilons for the transitions within the ground 3s2 3p3 configuration, which we use to supplement the Flower data. Tayal & Henry (1988) published R-matrix collision data between the 3s23p2 and 3s3p3 configurations. Their collision strengths agree with Flower's values to within 30%. Some strange effects in the Tayal et al. and Tayal & Henry data arise through the use of pseudo-resonances (see Mason 1994). New calculations of electron excitation data for Fe XII as part of the Iron Project (Binello et al. 1996). We await these results for incorporation into the CHIANTI database.
We note that the most important requirements for this ion are collision strengths to the 3s23p23d configuration and out of the metastable levels in that configuration, which can have substantial populations even at densities of 108 - 1010 cm-3.
The model for Fe XI includes 47 fine structure levels of the 3s23p4, 3s3p5, 3s23p33d and 3p6 configurations. Most of the energy levels were taken from the NIST database and a few from the work of Jupen et al. (1994). Observed energies for many levels do not exist. A 13 configuration model of Fe XI was used to derive transition probabilities with SSTRUCT. It was found that adding the extra configurations significantly altered the mixing coefficients of the 3s23p33d 3S1, 3P1 and 1P1 levels, thus affecting their oscillator strengths.
New distorted wave calculations by Bhatia & Doschek (1996) have finally superseded the work of Mason (1975). Collision strengths at 3 energies were calculated. In view of the changes in mixing coefficients just noted and because there were significant differences between SSTRUCT and Bhatia & Doschek for other transitions, we decided to rescale all the 3s23p4 - 3s23p33d collision strengths by the ratio of SSTRUCT oscillator strengths to the Bhatia & Doschek oscillator strengths. One consequence of this was to increase the strength of the 3P2 - 1P1 transition, and so we identify this as the 188.30 Å line, rather than the 3P2 - 3S1 identification of Jupén et al. (1994).
The model for Fe X includes 54 fine structure
levels in the
3s23p5, 3s3p6, 3s23p43d and
3s3p53d configurations.
Energy levels are from the NIST database and from
Jupen et al. (1994).
Experimental energies of many of the 3s3p53d
levels are still
unknown. Oscillator strengths together with all
transition
probabilities were derived using a 12
configuration model in SSTRUCT.
The configurations used were 3s23p5,
3s3p6, 3s23p43d,
3s3p53d, 3s23p33d2, 3s23p44s,
3s23p44p,
3s23p
, 3s23p
,
3s3p43d2 and 
.
The transition probabilities were found to be
consistent with those of Fawcett
(1991) to within 5%.
Bhatia & Doschek (1995a)
presented new distorted wave calculations for
Fe X that were an improvement over the results of
Mason (1975);
however, for the 3s23p5 2P - 3s3p6
2S transitions, the
collision strengths were severely under-estimated
and so, for these
transitions only, the accurate SSTRUCT oscillator
strengths were used
to scale upwards the collision strengths. As
part of the Iron
Project, Pelan & Berrington (1995)
calculated upsilons for the ground
2P3/2 - 2P1/2 transition using
the R-matrix program.
The inclusion of resonance structure was found to
increase the
excitation rate by a factor of 10 for low
temperatures, having
significant consequences on the level balance of
the ion, and so we use
the Pelan & Berrington upsilons for this
transition. It is to be
noted that R-matrix calculations have been
performed for Fe X by
Mohan
et al. (1994); however question marks
have been raised over this work
by Pelan & Berrington (1995) and
Foster et al. (1996) and so we
retain the distorted wave calculations for the
present.
For Fe IX, the 13 levels of the 3p6 and 3p53d configurations were included. Energy levels from the NIST database were used. A 10 configuration model of Fe IX was used in SSTRUCT to generate transition probabilities. The distorted wave calculations of Fawcett & Mason (1991) were used to fit and scale the collision strengths among these levels. Although collisional data was also given for the 3p54s configuration, we note that this would lead to two extra metastable levels (3P0 and 3P2). However, it is evident that the 3s3p53d and 3s3p43d2 configurations are energetically lower than 3p54s, and so the two levels would actually decay via these configurations. The apparent metastable nature of the two levels is thus purely a consequence of the neglect of the 3s3p53d and 3s3p43d2 configurations in the level balance, and so we ignore the 3p54s configuration in this work. Our gf value for the 1S - 1P transition (which gives rise to the strong line at 171.07 Å) was found to be 25% lower than that given by Fawcett & Mason and so their collision strengths, for this transition only, were scaled proportionately.
We note that Burgess et al. (1996) have published and fitted non-exchange DW electron excitation data for Ca II using OMEUPS. These will be included in the next release of CHIANTI.
The present model of Fe VIII includes the 3p63d, 3p64p, 3p64f, 3p65f, 3p66f, 3p67f configurations. Observed energies for the 12 fine structure levels are from the NIST database. Oscillator strengths and A values are from Czyzak & Krueger (1966), Fawcett (1989) and NIST. Czyzak & Krueger (1966) have calculated a number of collision strengths for Fe VIII using the Coulomb-Born approximation. Pindzola et al. (1989) have calculated both close-coupling and distorted wave cross sections for the 3d-4s and 3d-4p and 4s-4p. Their close-coupling and distorted wave cross sections tend to show good agreement. The values of Pindzola et al. collision strengths for the 3d-4p are about a factor of 3-4 lower than the values calculated by Czyzak & Krueger. The collision rates of Pindzola for the 3d-4p transition have been used and the data of Czyzak & Krueger for the others. Calculations of excitation rates to the 3p63d2 configuration are not available and this is a major weakness of the model. In fact, the present values of the collision strengths appear to be of limited accuracy.
The 9 fine structure levels of the ground term 3p63d2 have been included. Energy levels are taken from the NIST database (Martin et al. 1995) and the oscillator strengths and A values are those of Nussbaumer & Storey (1982). All of the lines involve forbidden transitions and the shortest wavelength is 1490 Å. Collision strengths among the 9 levels of the ground term 3p63d2 have been calculated by Norrington & Grant (1987) and their averages over a Maxwellian velocity distribution tabulated by Keenan & Norrington (1987). Their values are in good agreement with the distorted wave values of Nussbaumer & Storey (1982) but diverge at higher energies. The values of Keenan & Norrington (1987) have been included in the CHIANTI data base. Lines from this ion provide potentially useful density diagnostics for electron densities between 105 and 108 cm-3.
Fe II is a complex ion and produces a large number of lines that are observed at infrared, visible and ultraviolet wavelengths in a variety of astrophysical sources. It plays a major role in radiative losses from solar and stellar chromospheres (Anderson & Athay 1989).
The model of Fe II uses 142 levels of the quartet and sextet terms of the 3d64s, 3d7 and 3d64p configurations. New calculations of dipole transition probabilities have been made by Nahar (1995). These calculations used the energy levels of Fe II from Johansson (1978) as well as unpublished values by Johansson to arrive at a fairly complete description of the energy levels of Fe II. A values for forbidden transitions are from the NIST compilation (Martin et al. 1995). Recently, Zhang & Pradhan (1995) have calculated collisions strengths among the 142 fine structure levels levels and are all included in the CHIANTI database. At densities characteristic of stellar chromospheres, the lowest 23 levels can have significant populations.