1: \begin{abstract}

2: 

3: 

4: We have carried-out two intermediate coupling frame transformation (ICFT) 

5: $R$-matrix  calculations for the electron-impact excitation of $\mathrm{C}$-like 

6: $\mathrm{Fe}^{20+}$, both of which use the same expansions for their

7: configuration interaction (CI) and close-coupling (CC) representations.

8: The first expansion arises from the configurations  

9: $\mathrm{2s^2\,2p^2, \,2s\,2p^3, \,2p^4}$, 

10: $\mathrm{\{2s^2\,2p,\,2s\,2p^2,\,2p^3\}}\,nl$, 

11: with $n=3,4$, for $l=0 - 3$, which give rise to 564 CI/CC levels.

12: The second adds configurations $\mathrm{2s^2\,2p\,5l}$, for $l=0-2$,

13: which give rise to 590 CI/CC levels in total. 

14: 

15: Comparison of oscillator

16: strengths and effective collision strengths from these two calculations

17: demonstrates the lack of convergence in data for $n=4$ from the smaller one.

18: Comparison of results for the 564 CI/CC level calculation with an

19: earlier ICFT $R$-matrix calculation which used the exact same CI

20: expansion but truncated the CC expansion to only 200 levels

21: demonstrates the lack of convergence of the earlier data, particularly

22: for $n=3$ levels. Also, we find that the results of our 590 CC $R$-matrix calculation are

23: significantly and systematically larger than those of an earlier comparable 

24: Distorted Wave-plus-resonances calculation.

25: 

26: Thus, it is important still to take note of the (lack of) convergence in

27: both atomic structural and collisional data, even in such a highly-charged

28: ion as Fe$^{20+}$, and to treat resonances non-perturbatively. 

29: This is of particular importance for Fe ions given

30: their importance in the spectroscopic diagnostic modelling of 

31: astrophysical plasmas.

32: 

33: 

34: \end{abstract}

35: