The atomic energy levels needed for the
calculation were taken from Sugar &
Musgrove (1991). Oscillator strengths have been calculated using
the method of Bates & Damgaard (1949). Our investigation of
the influence of the departure from the LS coupling scheme on the Xe II
Stark broadening parameters (Popovic & Dimitrijevic
1996a) has demonstrated that the differences between results
obtained by using the LS and jK coupling schemes are relatively small
in comparison with the accuracy of the modified semiempirical method (
in average). The differences for the homologous Kr II radiator should
not be larger, so that the use of the LS coupling approximation should
not influence the accuracy significantly, especially for transitions
between lower levels. In Table 1 (accessible only in electronic form),
we present Stark widths for 37 Kr II lines
obtained for temperatures from 5000 K up to 50000 K and at an electron
density of 
. Since we have found that the ion
broadening contribution to the line widths is several percent, only the
electron broadening contribution is given.
Our results for Stark widths have been compared in Table 2
(accessible only in electronic form) and Figs. 1 (click here)-3 (click here) with
available experimental data (Brandt et al. 1981; Vitel &
Skowronek 1987; Uzelac & Konjevic 1989; Lesage
et al. 1989; Bertuccelli & Di Rocco 1990).
The ratio of measured and calculated data varies between 0.73 and 1.97 (or
1.60
if we exclude the results of Bertuccelli & Di Rocco 1991).
This is similar agreement as for the homologous 
Xe II
transitions, where this ratio varies between 0.7 and 1.4 (Popovic &
Dimitrijevic 1996a). One can see from Table 2 that the
largest disagreement exists
between our theoretical Stark widths and the experimental
data given by Bertuccelli & Di Rocco (1990) (
vary
from 0.73 up to 1.97). One should notice however, that they have not
determined the electron density directly. They estimated the electron
density by comparing the experimental widths of five selected lines with
those obtained in Lesage et al. (1989) and Vitel & Skowronek
(1987). The measured and calculated width ratios in the case of
Uzelac & Konjevic (1989), Brandt et al. (1981)
and Vitel & Skowronek (1987) are well within the error bars
of the modified semiempirical method (
). This is the case for
Lesage et al. (1989) as well, with the exception of the 461.9
nm line. Since all experimental data are within a narrow temperature range
between 11000 K and 17400 K, where the Stark widths decrease quicker than
for higher temperatures, it is significant to provide new reliable
experimental data for higher temperatures in order to check the theoretical
temperature dependence.
Figure 1: Stark full-width (FWHM) for Kr II 
spectral line as a function of temperature, at an electron density of
Figure 2: Same as in Fig. 1 (click here), but for line 
Figure 3: Same as in Fig. 1 (click here), but for line 
In Table 1, data
are presented for each particular line within a multiplet. One should notice
that
the atomic energy level differences within the considered multiplets
are comparable to the distances to the nearest perturbing level. Consequently,
the line widths within a multiplet differ (Dimitrijevic
1982). For example, for the lines
(
) and
(
), the lower
(
) level is the same. On the other hand, the upper level
of the 476.57 line, i.e. 
, is much closer to the
perturbing levels and especially to the 
and
perturbing levels than the upper level
(
) of the 429.29 nm line. Consequently, the 476.57 nm
line width should be larger than the 429.29 nm line width. Indeed, the
ratio of the corresponding widths is w476.57/w429.29=1.34 from our
calculations, w476.57/w429.29=1.7 (Lesage et al. 1989)
and w476.57/w429.29=1.16 (Bertuccelli & Di Rocco 1991)
from experiment.
The averaged value of the experimental and the theoretical data ratio is
, where the indicated error is an
average quadratic error calculated in the same way as in Popovic &
Dimitrijevic (1996a). If there are several values at
different temperatures from the same reference for the same line,
these values have been averaged before making
an average ratio for the line.
As one can see, the calculated Stark widths give a satisfactory
agreement with experimental values on average.
The theoretical Stark widths for several lines are given in Bertuccelli &
Di Rocco (1993). They have calculated Stark widths (only for T=10
000 and 
) by using Griem's
semiempirical formula (Griem 1968) and by using the approach
based on the Born approximation with and without the empirical modification
for the collision strength suggested by Robb (see
Bertuccelli & Di Rocco 1993). In Table 3 (accessible only in
electronic form), we have compared our theoretical
data with the theoretical data calculated by Bertucceli & Di Rocco (1993) and
with the experimental data (Brandt et al. 1981;
Vitel
& Skowronek 1987; Uzelac & Konjevic 1989;
Lesage et al. 1989; Bertuccelli & Di Rocco
1991) for several lines. Data presented for a particular line are
averaged if several values exist. As one
can see, the ratios of calculations performed by Bertuccelli & Di Rocco
(1993) and our calculations are 1.95, 1.76 and 1.56 for the
calculations by using the approach based on the Born approximation with and
without the empirical modification for the collision strength suggested by
Robb and by using the Griem's semiempirical approach, respectively.
Also, their calculations are significantly larger in comparison with
experimental data.
There are Stark shift experimental data for three Kr II lines (Vitel & Skowronek 1987). The comparison between our calculations and experimental Stark shifts is shown in Table 4 (accessible only in electronic form). One should notice that the theoretical shifts are generally of lower accuracy than widths (see e.g. Dimitrijevic et al. 1981; Popovic et al. 1993). Namely, the contributions from different perturbing levels to the shift have different signs. If contributions with both signs are similar, shift accuracy is much lower. In view of these facts, the agreement between calculated and experimental shifts is satisfactory with the exception of the 435.55 nm line. The calculated data for Stark shifts can be obtained on request from the authors.
We hope that the presented data set on Kr II Stark widths will be of help for the analysis of the trace element spectral lines and abundances (especially by using space-borne telescopes and instruments such as HST/GHRS), as well as for the investigation of regularities within homologous atom/ion sequences and their use for the interpolation of new data.
Acknowledgements
This work has been supported by the Ministry of Science and Technology of Serbia through the project "Astrometrical, Astrodynamical and Astrophysical Researches".