About Chandra Archive Proposer Instruments & Calibration Newsletters Data Analysis HelpDesk Calibration Database NASA Archives & Centers Chandra Science Links

Analysis of ACIS Data Affected by the Low Energy QE Degradation

This memo is an update to the report of April 5, 2002 concerning the reduction in the ACIS low energy QE. Below we summarize the results of our investigation into this problem and provide observers with advice and new tools to use in their spectral analysis of ACIS data. These provide a significant improvement in the spectral analysis below 1 keV. The results presented here represent the combined efforts of many groups, including, the CXC, the IPI teams at PSU and MIT, and Project Science at MSFC.



An Examination of archived astronomical observations and data acquired from the on-board ACIS calibration source (55Fe) shows that there has been a continuous degradation in the ACIS QE since launch. Our best interpretation is that this is due to molecular contamination building up on the cold optical blocking filter, and/or the CCD chips. This degradation is the most severe at low energies. Above 1 keV, the degradation is less than 10% to date. Analysis of the on-board calibration source shows that the L-complex (about half Mn-L and half Fe-L lines) to Mn-K alpha line ratio has decreased at a steady rate since launch corresponding to a decrease in the QE at 670 eV of about 10% per year. Using periodic calibration observations of PKS2155-304 with the LETG, we find that the contamination build-up on ACIS amounts to an increase of 0.41 optical depths per year at the C K-edge (284 eV). In the most recent PKS2155-304 observation (taken June 11, 2002), excess absorption at the O K-edge (543 eV) is clearly detected, and there is some suggestion of excess absorption at the N-K edge (410 eV). All indications are that the contamination is uniform across both ACIS-I and ACIS-S.

With the contamination rate well constrained, and an assumed hydrocarbon model for the contaminant, an XSPEC absorption model (ACISABS) has been developed that can be adjusted to a specific observing date based on the measured rate of QE degradation at 670 eV. This model can be added as an extra component when fitting ACIS spectra with an uncorrected effective area file (.arf). An additional tool (corrarf) has been developed that directly applies the ACISABS absorption profile to the effective area file generated by the CIAO tool 'mkarf'. Using 'corrarf' eliminates the need for observers to include an extra model in their spectral analysis. Below we present several examples that demonstrate the improvement in the spectral analysis of ACIS imaging data using the 'corrarf' tool. Directions on how to obtain and execute these tools are given below . Since the calibration uncertainties are more pronounced in gratings data, we recommend that the observer include the ACISABS model as an additional component in their spectral analysis, with an uncorrected effective area file. This procedure provides more flexibility in analyzing gratings data.

Caveat - While these tools provide a significant improvement in ACIS low energy spectral analysis, the user must be aware that this is an on-going investigation and we have not yet determined the exact chemical composition of the contaminant, which is necessary to provide a more definitive absorption model. Residuals of 10% will still be noticeable in ACIS imaging observations with high photon statistics using the "corrarf" tool. Thus, any residuals of order 10% in the spectral analysis of imaging data below 1 keV should be treated as consistent with the present calibration uncertainties, and not new scientific discoveries.

Spectral Analysis of ACIS Data in Imaging Mode

When ACIS is in the stowed position, i.e., the HRC-S is at the focal point, it is illuminated by an 55Fe source. Since the discovery that the ACIS CTI significantly increased during the first few weeks of the mission, observations of the 55Fe source have been taken before and after each perigee. Figure 1 shows that the L-complex to Mn-K alpha line ratio has continuously decreased since launch. Since the Mn-K alpha line at 5.9 keV is not affected by absorption, the decrease in the L-complex to Mn-K alpha line ratio must be due to increased opacity at the L-complex at 670 eV. Figure 1 indicates that the opacity at 670 eV is increasing at about 10% per year. To demonstrate that nothing unusual is happening to the 55Fe source, Figure 2 shows that the Mn-K alpha line intensity, produced in the 55Fe decay process, is decreasing at a rate consistent with the half-life of 55Fe.

Figure 1: L-complex to Mn-K alpha line ratio during the Chandra mission.

Figure 2: The Mn-K alpha line intensity during the Chandra mission.

Abell 1795 was observed with ACIS-S in April 2000 and again in June 2002. A comparison of the spectra extracted from the central 1' minute region in these two observations is shown in Figure 3. Notice that there is essentially no change in the QE above 1 keV. Figure 4 shows the ratio of the two A1795 spectra and the prediction for the increased absorption between the two observations from the ACISABS absorption model. Notice that all of the residuals, are less than about 10% except near the C K-edge. This figure also provides a fiducial reference for the decrease in QE as a function of energy over a span of 26 months. Figure 5 demonstrates that applying this absorption model to the effective area files produces ACIS derived cluster temperatures in good agreement with those derived from ASCA data.

Figure 3: ACIS-S spectra of A1795 from observations in April 2000 (black) and June 2002 (red).

Figure 4: Ratio of the June 2002 to the April 2000 spectrum of A1795. The solid green line shows the prediction for the increased absorption between these two observations from the ACISABS model. This curve also gives the decrease in QE as a function of energy over a 26 month period.

Figure 5: A comparison of cluster temperatures using ACIS effective area files corrected for excess absorption with ASCA derived temperatures.

Spectral Analysis of ACIS Data with the Gratings

The first indication that there were problems with the low energy ACIS QE was a May 2000 observation of PKS2155-304 with the LETG that showed excess absorption near the C-K edge. A similar feature was also observed in a LETG/ACIS-S observation of 3C273. Since an observation of 3C273 with the LETG/HRC-S did not show this feature, it was concluded that the problem resided with ACIS. The first attempts at reconciling the excess absorption near the C-K edge with a contamination model were unsuccessful (see this memo for more detailed information). To account for this absorption feature, an optional ACIS QE calibration file (qeN0004) was created for gratings analysis. The standard ACIS QE file non-gratings observations is qeN0003.

A recent LETG/ACIS-S calibration observation of PKS2155-304 was carried-out in June 2002 to constrain the composition of the molecular contaminant. Figures 6 and 7 show the data along with the best fit broken power-law plus edges at C, N,O, and F. Even with the depths of the four K-edges treated as free parameters there are still significant residuals due to uncertainties in the ISM absorption model and incomplete modeling of the atomic physics near the absorption edges.

By performing a similar analysis on previous PKS2155-304 observations, the optical depth at the C-K edge as a function of time can be determined and is shown in Figure 8. The dashed line shows a steady increase of 0.41 optical depths per year at the C K-edge. The solid line shows the required Carbon build-up to produce 100% of the increased opacity at 670eV. This curves demonstrates the need for additional contaminants, in particular, oxygen.

At the present time, users may include the ACISABS absorption model as an added component in their gratings spectral analysis, along with an uncorrected effective file. We are working on an improved absorption model for use with gratings data. For consistency, the user must use the qeN0003 file when generating gratings effective area files with the CIAO tools 'mkgarf' or 'fullgarf'. Directions for changing the version of the QE table used when generating gratings effective area files can be found here.

If ACISABS is used, then the user should treat the normalization and chemical composition as free parameters (i.e., parameters 3 through 7 in the ASICABS absorption model) when fitting gratings spectra. Table 1 shows the chemical composition of the hydrocarbon used in the ACISABS absorption model (N_ACISABS) along with the range of acceptable values (from N_Low to N_High) obtained when this model is fit to the recent LETG/ACIS-S observation of PKS2155-304. The user must realize that the edge energies in the ACISABS model are based on the atomic form of these elements and not the edge energies appropriate for the molecular form of the contaminant on ACIS. Edge shifts of a few eV and additional structure near the edges will be present in high signal-to-noise gratings spectra.

Figure 6: Fit to the LETG/ACIS-S spectrum of PKS2155-304 taken in June 2002. The 68% uncertainties in the optical depths are 0.015, 0.007, 0.008, and 0.1 for the C,N,O, and F K-edges, respectively.

Figure 7: Same as Figure 7 except for the 0.2 to 1.0keV energy band.

Figure 8: The optical depth at the C K-edge for all PKS2155-304 observations. The dashed line shows an increase of 0.41 optical depths per year. If the contaminant were pure C, then a much greater deposition rate (shown by the solid line) would be required to produce the measured QE decrease at 670eV.

Table 1 - Range of acceptable chemical compositions

Element N_ACISABS N_Low N_High
Hydrogen 20 0 50
Carbon 10 9 11
Nitrogen 1 0 2
Oxygen 2 1 3

Obtaining the corrarf and ACISABS tools

The effect of the contamination can be corrected by using CIAO tools and some of the contributed software described below.

A Fortran code has been written that corrects the effective area file generated by the CIAO tool 'mkarf' using the ACISABS absorption model. The source code, an executable version of the code, and a description of the code can be downloaded from corrarf.tar.gz . Once installed, the user simply needs to execute the following command

corrarf -gtifile gti.fits -arf arf.fits -o new_arf.fits -qe_cont qe_cor.H20.C10.O2.N1

and the original "arf.fits" file generated by mkarf will be corrected and stored in the file "new_arf.fits". The "gti.fits" file is used to determine the date of the observation.

The XSPEC ACISABS absorption model and installation instructions can be downloaded from the CXC contributed software page or from HEASARC .

The CIAO team provides an S-Lang version of the ACISABS model that is usable from Sherpa: see the "Using the ACISABS model in Sherpa" thread for more details. A script has been developed to apply this model directly to an ARF, as described in the "Apply the ACISABS model to an ARF" thread.

Last modified: 11/15/10

The Chandra X-Ray Center (CXC) is operated for NASA by the Smithsonian Astrophysical Observatory.
60 Garden Street, Cambridge, MA 02138 USA.    Email: cxcweb@head.cfa.harvard.edu
Smithsonian Institution, Copyright © 1998-2004. All rights reserved.