1. Ionization and Recombination with Electrons: Laboratory Measurements and Observational Consequences Daniel Wolf Savin Columbia Astrophysics Laboratory

2. Collaborators Warit Mitthumsiri, Michael Schnell – Columbia University Mark Bannister – Oak Ridge National Lab (ORNL) Martin Laming, Enrico Landi – Naval Research Laboratory Andreas Wolf – Max Planck Institute for Nuclear Physics Alfred Müller, Stefan Schippers – University of Giessen

3. Outline I.Motivation II.Types of Cosmic Plasmas III.Electron Impact Ionization (EII) IV.Dielectronic Recombination (DR) V.Future Needs

4. Spectra observations can be used to infer properties of the cosmos. The aim of laboratory astrophysics is to reduce atomic physics uncertainties so that discrepancies between spectral observations and models tells us something about the properties of the observed sources and cannot be attributed to errors in the atomic data used in the models.

5. For example, to infer relative abundances, we note that Rewriting this gives Clearly, accurate ionization and recombination data are needed for reliable ionization balance calculations to get reliable relative abundances. Ionization balance calculations are used to infer the properties of cosmic objects.

6. Outline I.Motivation II.Types of Cosmic Plasmas III.Electron Impact Ionization (EII) IV.Dielectronic Recombination (DR) V.Future Needs

7. Cosmic plasmas can be divided into two broad classes: Collisionally-ionized (stars, galaxies,...) Ionization due to electrons. In equilibrium an ion forms at T e ~ I p /2. High T e DR dominant recombination process. Photoionized (PNe, IGM, XRBs, AGN,…) Ionization due to photons and resulting electrons. In equilibrium an ion forms at T e ~ I p /20. Low T e DR dominant recombination process.

8. Outline I.Motivation II.Types of Cosmic Plasmas III.Electron Impact Ionization (EII) IV.Dielectronic Recombination (DR) V.Future Needs

9. Electron impact ionization (EII) e - + O 7+ → e - + e - + O 8+ EII requires E k > E b.

10. Published recommended EII rate coefficients have yet to converge.

11. In collisional ionization equilibrium (CIE) we have Rewriting gives Errors in either the ionization or recombination data will affect predicted or interpreted line ratios involving ions q and q+1. Errors in EII data translate directly into errors in predicted line ratios.

12. We are carrying out a series of new EII measurements at ORNL. Ionization data can be collected for collision energies 3-2000 eV. (Bannister 1996, Phys. Rev. A 54, 1435)

13. We have carried out preliminary measure- ments for EII of Be-like C 2+ → C 3+ Ground-state (2s 2 1 S 0 ) IP = 47.89 eV Metastable (2s2p 3 P) IP = 41.39 eV Lifetime = 9.7 ms (J=1) ≥ 200s (J=0,2)

14. Initial C 2+ EII measurements are discrepant with theory. Arrows indicate threshold for metastable and ground-state C 2+. Metastable fraction inferred by comparing electron impact excitation data (using same ion source) to theory. Curve shows configuration- average distorted-wave theory for our mixed state ion beam.

15. Extracted ground state cross section is a factor of 2 smaller than published theory. Lotz formula used for energy dependence of EII cross sections σ G and σ M. Fit to lab data gives σ G and σ M (solid curves). Also shown are distorted wave theory (dashed curve, Younger, 1981) and the recommended data (dash-dot curve, Bell et al., 1983)

16. Outline I.Motivation II.Types of Cosmic Plasmas III.Electron Impact Ionization (EII) IV.Dielectronic Recombination (DR) V.Future Needs

17. Energy conservation requires ΔE = E k + E b. Both ΔE and E b quantized  E k quantized. Low temperature DR occurs for E k << ΔE. High temperature DR occurs for E k ~ ΔE. Dielectronic Recombination (DR) e - + Fe 23+ ↔ (Fe 22+ )** → (Fe 22+ )* + h

18. DR theory for L- and M-shell ions are theore- tically and computationally challenging. Until recently modelers have had few modern calculations to use. Comparisons show these data to have factor of 2 or more uncertainties. (Savin et al. 2002, ApJ, 576, 1098)

19. In photoionized gas DR uncertainties affect predicted temperature and gas stability. (Savin et al. 1999, ApJS, 123, 687) Using XSTAR and varying the low T e DR data for Fe 17+ to Fe 23+ by a factor of 2. Line emission seen from ions predicted to form in region of thermal instability. Temperature Phase diagram

20. In electron-ionized plasmas DR errors affect predicted relative abundances. Using older DR data inferred relative abundances in the solar corona can be a factor of 5 smaller or 1.6 times larger. (Savin & Laming 2002, ApJ, 566, 1166) Line Ratio Variation MinimumMaximum Mg VI/Ne VI0.601.11 Mg VII/Ne VII0.671.22 Mg IX/S IX0.331.29 Mg IX/S X0.511.64 Si IX/S IX0.201.01 Si IX/S X0.361.14 Si X/S X0.431.60

21. We are carrying out a series of DR measure- ments using the Test Storage Ring (TSR).

22. Schematic of the electron cooler

23. Measurements can be carried out for low and high temperature DR. (Savin et al. 1999, ApJS, 123, 687; 2001, ApJ, 576, 1098) DR of O-like Fe XIX forming F-like Fe XVIII

24. We can use these data to produce Maxwellian rate coefficients for plasma modeling. Pre-experimentPost-experiment Measurements are used to benchmark modern DR theory which is then used to calculate DR for other ions in the tested isoelectronic sequence.

25. Even with benchmarking modern DR theory has still not converged for all L-shell ions

26. Current AGN spectral models over-predict the ionization stages of M-shell iron ions. (Netzer et al. 2003, ApJ, 599, 933) Models that match spectral features from abundant 2nd row elements, over-predict the average Fe charge stage. This is believed to be due to the absence of low T e DR data for M- shell Fe (Kraemer et al. 2004; Netzer 2004).

27. Published laboratory work supports that poor Fe M-shell DR data is the cause. Published DR data were for tokamaks, stars, etc. and did not attempt to treat properly the low energy DR resonances. This is an example of how better communica- tion between atomic physics and astro- physics could have predicted this problem (Müller 1999, Int. J. Mass Spectrom. 192, 9) DR of Fe XVI forming Fe XV

28. We are carrying out further M-shell Fe DR measurements to address this issue. DR of Fe XV forming Fe XIV

29. Conclusions Significant errors exist in EII data base. Much experimental and theoretical EII work needs to be done. L-shell DR data has improved recently but room remains for theoretical improvement. More L-shell benchmark DR measurements needed. Lots of experimental and theoretical work is needed to improve the M-shell DR data. More accurate structure calculations are needed for low-lying autoionization levels.

30. We have added a beam attenuation cell to determine directly the metastable fraction. If the electron capture cross section for metastable and ground state ions differ significantly, then one state will be lost first as the target gas density increases. Plot of the log of ion current vs. target gas density is bi- linear (slopes proportional to capture cross sections) and can be used to infer relative populations of ground-state and metastable ions. (Zuo et al. 1995, ApJ, 440, 421)

31. These DR uncertaintes also affect predicted line emission. (Savin et al. 2000, AIP CP547, 267) XSTAR spectra for gas at log(ζ)=2.1 erg cm s -1

32. In CIE we have Rewriting gives Errors in either the ionization or recombination data will affect predicted or interpreted line ratios involving ions q and q+1. In CIE, errors in DR data translate directly into errors in predicted line ratios.

33. Theory has also has a problem with high charge states for L-shell ions.

34. Recent AGN Observations have indicated the importance of Fe M-shell DR. (Sako et al. 2001, A&A, 365, L168) A new AGN spectral feature at λ ≈ 16-17 Å has been identified as being due to absorp- tion in M-shell iron ions.