Examining the Uniformity C-K edge Across the ACIS Detector

There were two observations of 3C 273 taken less than 2 days apart using the LETG with ACIS where the SIM was shifted by 0 mm (obs ID 2471) or -8 mm (obs ID 2464). Previously, these observations were fit independently to spectral models with 5 source-specific parameters (N
H and 2 PL components with a normalization and photon index, Gamma, each) and several instrument-specific parameter characterizing the absorption due to the contaminant. The previous results gave consistent opacities at the C-K edge but the uncertainties were rather large due to the presence of many degrees of freedom.

For this analysis, all source spectral shape parameters were held fixed due to the lack of significant differences between the two observations. The parameters are Gamma1 = 2.55, Gamma2 = 1.35, N_H = 1.68e20, and A1 = A2 = A. This commonality is not too surprising for 3C 273 -- most variations are small over 2 days. Furthermore, the absorption model was fixed except for a factor t that gives the optical depth of the contaminant relative to the value predicted by the model of the time dependence for the median date of about 2001.45. The figures show the fits.

Fig. 1: Fit to the LETG ACIS spectrum of 3C 273 taken in June 2001 at a SIM Z offset of zero. The NH is fixed and other spectral parameters are fixed, as described above. The contamination factor gives the optical depth of the contaminant relative to the value expected from the time dependence model. The jump near 0.5 keV is due to a systematic error in BI QEs relative to FI QEs that has not yet been determined at the 10% level.

Fig. 2: Same as Fig. 1 but for the observation obtained at a SIM Z position of -8 mm. The optical depth due to the contaminant is slightly larger than in the center of the detector (SIM Z = 0 mm, Fig. 1).

Adjusting the uncertainties for the resultant chisq values (1.45 and 1.36), t increases from 0.928 +/- 0.027 to 1.063 +/- 0.024 by shifting -8 mm. The chisq values are very close to those obtained from full 5 parameter models, so there appears to be very little intrinsic spectral shape change, as required for this test. The ratio of the values of t is 1.145 +/- 0.042, a 3.5 sigma difference from 1.00, indicating a significant increase in absorption from the row 512 to row 180.


Herman Marshall
Last updated July 7, 2003