HETGS Line Profiles

Comparison of new and old grating RMFs

We compare a set of new RMFs based on the new LSFPARM products with the canned grating RMF (version N0004) in CALDB. In this example we also use a 43 ksec Calibration/GTO observation of Capella (ObsID 1103) taken with Chandra HETG/ACIS-S.

Once again we use isis to perform a simple line profile fitting, though at this time using a Gaussian function with a fixed narrow width (1e-10 Angstrom). Likewise we also add a 1st order polynomial function to adjust the baseline.

Once again, the plots themselves speak a thousand word. Please look at fitting results shown in the following figures.

Line Profile Fitting

HETG/ACIS-S HEG +1st order spectra:

Fe XVII $\bf{\lambda{15.015}}$

Figure 11: Results of line profile fitting at Fe XVII with different RMF/fitting kernels. The red line shows the best-fit model. The difference $\delta\chi$ is also displayed for your entertainment: (top) New RMF based on the new LSFPARMS (goodness of fit: ${\chi}^{2}$ =0.24); (bottom) Old RMF N0004 (+1st order HEG) in CALDB (goodness of fit: ${\chi}^{2}$ =0.39).

Mg XII $\bf{\lambda{8.421}}$

Figure 12: The same as for Figure 11. (Top) New (goodness of fit: ${\chi}^{2}$ =0.298); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.292).

Si XIII Triplet $\bf{\lambda\lambda{6.647, 6.685, 6.740}}$

Figure 13: The same as for Figure 11. (Top) New (goodness of fit: ${\chi}^{2}$ =0.484); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.382). (NB: For some technical binning reason we could not center one of Si XIII lines (6.647AA) accurately with the new RMF. Upon changing the rebinning scale, it results in a better fit than the given goodness value here.)

Fe XVII + Ne X Blend $\bf{\lambda\lambda{12.125, 12.136}}$

HEG: Fe XVII + Ne X: Old Canned RMF (N0004)

Figure 14: The same as for Figure 11. (Top) New (goodness of fit: ${\chi}^{2}$ =0.211); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.274).

HETG/ACIS-S HEG -1st order spectra:

Fe XVII $\bf{\lambda{15.015}}$

Figure 15: Results of line profile fitting at Fe XVII with different RMF/fitting kernels. The red line shows the best-fit model. The difference $\delta\chi$ is also displayed for your entertainment: (top) RMF based on the new LSFPARMS (goodness of fit: ${\chi}^{2}$ =0.400); (bottom) RMF N0004 (+1st order HEG) in CALDB (goodness of fit: ${\chi}^{2}$ =0.405).

Mg XII $\bf{\lambda{8.421}}$

Figure 16: The same as for Figure 15. (Top) New (goodness of fit: ${\chi}^{2}$ =0.318); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.315).

Si XIII Triplet $\bf{\lambda\lambda{6.647, 6.685, 6.740}}$

Figure 17: The same as for Figure 15. (Top) New (goodness of fit: ${\chi}^{2}$ =0.256); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.244).

Fe XVII + Ne X Blend $\bf{\lambda\lambda{12.125, 12.136}}$

MEG: Fe XVII + Ne X: Old Canned RMF (N0004)

Figure 18: The same as for Figure 15. (Top) New (goodness of fit: ${\chi}^{2}$ =0.365); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.407).

HETG/ACIS-S MEG +1st order spectra:

Fe XVII $\bf{\lambda{15.015}}$

Figure 19: Results of line profile fitting at Fe XVII with different RMF/fitting kernels. The red line shows the best-fit model. The difference $\delta\chi$ is also displayed for your entertainment: (top) RMF based on the new LSFPARMS (goodness of fit: ${\chi}^{2}$ =1.70); (bottom) RMF N0004 (+1st order MEG) in CALDB (goodness of fit: ${\chi}^{2}$ =1.73).

Mg XII $\bf{\lambda{8.421}}$

Figure 20: The same as for Figure 19. (Top) New (goodness of fit: ${\chi}^{2}$ =0.436); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.494).

Si XIII Triplet $\bf{\lambda\lambda{6.647, 6.685, 6.740}}$

Figure 21: The same as for Figure 19. (Top) New (goodness of fit: ${\chi}^{2}$ =0.843); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.876).

Fe XVII + Ne X Blend $\bf{\lambda\lambda{12.125, 12.136}}$

Figure 22: The same as for Figure 19. (Top) New (goodness of fit: ${\chi}^{2}$ =0.665); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.743).

HETG/ACIS-S MEG -1st order spectra:

Fe XVII $\bf{\lambda{15.015}}$

Figure 23: Results of line profile fitting at Fe XVII with different RMF/fitting kernels. The red line shows the best-fit model. The difference $\delta\chi$ is also displayed for your entertainment: (top) RMF based on the new LSFPARMS (goodness of fit: ${\chi}^{2}$ =4.914); (bottom) RMF N0004 (+1st order MEG since there is no minus 1st order MEG) in CALDB (goodness of fit: ${\chi}^{2}$ =4.324). (N.B: Large ${\chi}^{2}$ values because of its neighboring line not being modeled in the fitting properly.)

Mg XII $\bf{\lambda{8.421}}$

Figure 24: The same as for Figure 23. (Top) New (goodness of fit: ${\chi}^{2}$ =0.724); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.941).

Si XIII Triplet $\bf{\lambda\lambda{6.647, 6.685, 6.740}}$

Figure 25: The same as for Figure 23. (Top) New (goodness of fit: ${\chi}^{2}$ =1.06); (bottom) Old (goodness of fit: ${\chi}^{2}$ =0.643). (N.B: see caption in Figure 13.)

Fe XVII + Ne X Blend $\bf{\lambda\lambda{12.125, 12.136}}$

Figure 26: The same as for Figure 23. (Top) New (goodness of fit: ${\chi}^{2}$ =0.908); (bottom) Old (goodness of fit: ${\chi}^{2}$ =1.10).

Note that the derived goodness of fit for each case is more or less the same, i.e., the old RMFs are still O.K. for line profile analysis of a low signal-to-noise dataset. However, note that the old RMF assumes intrumental line profile to be symmetric, whearas the \it{true} instrumental line profile is very slightly skewed (esp. in HEG +/- orders).

Line Centroid and Flux Estimates

We have estimated line centroid and flux for each line (per order per grating) and the results are tabulated in Table 3.

Table 3: Validation of HEG +1st Order Centroid and Flux Estimates with different RMF kernels

HEG +1st
Atomic	Wavelength(AA)		Flux (photon/cm2/s)		Flux Ratio
Elements	New	Old	New	Old	(Old/New)
Fe XVII	15.0125	15.0125	3.91e-3	3.47e-3	89%
Mg XII	8.420	8.420	2.23e-4	2.20e-4	99%
Si XIII	6.6475	6.6475	2.02e-4	2.09e-4	103%
Si XIII	6.6851	6.6851	3.90e-5	3.83e-5	98%
Si XIII	6.7375	6.7381	1.87e-4	1.83e-4	98%
Ne X	12.135	12.1339	8.89e-4	8.68e-4	98%
Fe XVII	12.1225	12.1228	5.00e-4	4.70e-4	94%
HEG -1st
Atomic	Wavelength(AA)		Flux (photon/cm2/s)		Flux Ratio
Elements	New	Old	New	Old	(Old/New)
Fe XVII	15.015	15.015	3.64e-3	3.42e-3	94%
Mg XII	8.4225	8.4225	2.34e-4	2.29e-4	98%
Si XIII	6.6497	6.6500	2.04e-4	2.00e-4	98%
Si XIII	6.6899	6.6899	5.39e-5	5.22e-5	97%
Si XIII	6.7400	6.7425	1.79e-4	1.68e-4	94%
Ne X	12.1375	12.1375	8.16e-4	8.92e-4	109%
Fe XVII	12.1250	12.1250	5.99e-4	6.25e-4	104%
MEG +1st
Atomic	Wavelength(AA)		Flux (photon/cm2/s)		Flux Ratio
Elements	New	Old	New	Old	(Old/New)
Fe XVII	15.0061	15.0050	3.48e-3	3.29e-3	95%
Mg XII	8.415	8.415	2.41e-4	2.29e-4	95%
Si XIII	6.645	6.645	2.33e-4	2.23e-4	96%
Si XIII	6.685	6.685	6.52e-5	5.96e-5	91%
Si XIII	6.735	6.735	1.43e-4	1.42e-4	99%
Ne X	12.130	12.130	4.26e-4	5.02e-4	118%
Fe XVII	12.120	12.125	6.49e-4	6.71e-4	103%
MEG -1st
Atomic	Wavelength(AA)		Flux (photon/cm2/s)		Flux Ratio
Elements	New	Old	New	Old	(Old/New)
Fe XVII	15.015	15.015	3.88e-3	3.67e-3	95%
Mg XII	8.4200	8.4202	2.02e-4	1.89e-4	94%
Si XIII	6.6500	6.6500	2.07e-4	2.02e-4	98%
Si XIII	6.6850	6.6850	5.05e-5	4.68e-5	93%
Si XIII	6.7400	6.7400	1.52e-4	1.50e-4	99%
Ne X	12.1355	12.1400	7.99e-4	8.52e-4	107%
Fe XVII	12.1250	12.1273	4.06e-4	3.08e-4	76%

The results show that the derived centroid wavelengths are nearly identical at any given wavelength, while the new fluxes are generally higher than the old ones. The systematic difference in flux is due to the corrected normalization factors (or a.k.a. encircled energy fraction) provided with the new LSFPARM. Note, however, that counting statistics in this case dominates over the systemic error as the derived fluxes vary in a quite large scale between orders and gratings.

Validity of the old RMF N0004 (CALDB)

In brief, we have demonstrated that the old RMF N0004 products in CALDB is likely suitable for simple spectral analysis. However, if users wish to perform much more careful analysis on line emission/absorption profile of Chandra X-ray spectra, then we strongly recommend users to make their own RMFs based on the new LSFPARM.

Previous: Verification: Capella dataset Next: MARX vs. ChaRT

This page was last updated Apr 11, 2003 by Bish K. Ishibashi. Accessibility To comment on it or the material presented here, send email to bish@space.mit.edu.