Previous  Next  Contents

Instruments: ACIS

Calibration: Response Matrix Products and CTI Correction

A great deal of work was done in calendar year 2002 on the development of response matrix products for use with the CXC CTI (charge transfer inefficiency) corrector. This software, an option in acis_process_events, corrects some of the effects of the radiation damage the front-illuminated (FI) ACIS devices experienced in the first two months of the Chandra mission. This damage produced a large CTI, which means that both the mean charge collected and the energy resolution of ACIS is reduced, in an approximately linear function of the number of rows in the imaging array through which a given packet must be clocked in order to reach the readout register.

The CTI corrector is able to very nearly restore the mean of the charge distribution (the "gain" of the detector), and to restore approximately half of the resolution degradation. There is a stochastic component to the degradation that cannot be completely corrected after the fact.

This new, better, performance of the ACIS imaging chips (the I array and S2) then needed to be calibrated. We have developed a new FEF (FITS Encoded Function) file, released in early November 2002 (with a small correction to be released in March 2003). We have run of order 108 photons through the MIT CCD simulator, and degraded the pulse heights of the simulated events in a manner analogous to the hardware. The undegraded events at each energy were fit to a function consisting of a handful of Gaussian parameters (two for the main peak, one each for Si K escape and fluorescence, and one for low-pulse-height noise), plus some broad functions to model the continuum "tail" produced by partial charge collection. The CTI degradation was modeled for each energy in a "CTI scatter matrix" consisting of two Gaussian parameters. This was then convolved analytically with the Gaussian portions of the undegraded response function. The resulting coefficients are stored in the FEF file. A small upgrade to the FEF file function set was needed for this work; it was incorporated into the CIAO mkrmf tool.

The resulting FEF files were tested against data taken in spring, 2000, from the on board external calibration source, which has strong lines at 1.49, 4.51, and 5.90 keV. Energy scales were adjusted such that a match was found within 0.3% for all chip regions for all three energies. The resulting FEF files were also used to fit the SMC supernova remnant 1E0102-72.3 ("E0102"), which has a particularly simple spectrum (He- and H-like lines from oxygen and neon, with little to no iron) and a high surface brightness. Good agreement (of order 0.5%) was found in the energy scales at these energies (down to about 0.55 keV). Further testing against various grating data taken in non-standard modes (with the source off axis, and falling on the ACIS-I array) is in progress to assess the precision of the low-energy portion of the response.

We are also beginning to test the FEF files to look for time dependent effects as the CTI increases at expected rates over the life of the mission. The current mean I-array parallel CTI is about 1.31 x 10-4, while that of S3 is 1.7 x 10-5. These numbers are increasing at a rate consistent with our expectations: 3.6 x 10-6 / yr for the mean I-array and 1.2 x 10-6 / yr for S3.

Calibration: Degradation of the Low-Energy (E < 1 keV or so) QE of ACIS

Another effect discovered this year is that the quantum efficiency (QE) of ACIS is decreasing with time for energies less than about 1 keV. The effect is consistent with slow buildup of a contaminant(s) on the ACIS Optical Blocking Filter (OBF), an aluminized polyimide film designed to reject optical light. Thanks to the HETG and LETG, we have obtained high signal-to-noise spectra of the contaminant (or mixture of contaminants) in absorption against a variety of astrophysical continuum sources. Work is proceeding to obtain the composition of the substance(s). We observe a strong carbon K edge, a clear oxygen K edge, and a weak fluorine K edge, leading to the tentative conclusion that it may be related to lubricants used in the translation stage of the Chandra science instrument module. The (model) spectrum of the degradation is given in Figure 8, for four epochs (as a function of wavelength for the MEG first order): 2000.0, 2001.5, 2003.0, and a projection for 2004.5, the middle of the AO-5 observing period.

FIGURE 8: A model of the filter transmission at various epochs

If the particular substance identified proves to be sufficiently volatile, we may attempt to evaporate it, by temporarily raising the temperature of ACIS and/or the OBF. The prime contractor for Chandra (Northrup-Grumman, formerly TRW) is carrying out studies to assess the temperature and duration of such a bakeout, and to predict how effectively material might be removed from the vicinity of the ACIS. A risk assessment for such an option is in progress as of this writing (January 2003). In the meantime, users can correct for the effect using one of a variety of tools. One, called corrarf (G. Chartas), designed to correct the Ancillary Response Function (ARF file), accepts as inputs a standard ARF file and a time since launch. Another called contamarf (D. Huenemoerder) is high resolution, suitable for gratings as well as images. Two time dependency functions are under study (linear and exponentially decaying buildup), and either can be selected. These tools are available from :

Another tool is an xspec- or sherpa-usable model component which the user can multiply by an astrophysical model. This allows for some flexibility in the composition of the contaminant and the rate of buildup (via the time parameter). These models are available here:
(for the xspec model)
(for the sherpa version)

Calibration Workshop, November 2002

Elsewhere in this newsletter, Hank Donnelly describes the first annual Calibration Workshop, held in Cambridge in November 2002. Many talks were given detailing various aspects of the calibration of the observatory, and most of these presentations are available on the web.

This excellent resource should be a standard place to go when calibration questions arise in the course of data analysis.


In 2002 a long (53 ks) observation was taken in a novel mode. The translation table was moved to put ACIS between the on-board external calibration source and the optical bench, so that ACIS could see neither the cal source nor the sky. This put the focal point of the telescope on the HRC-I. A target was selected and an exposure of the neutron star RXJ 1856.5-3754, useful for calibration of the point spread function far off-axis (27.33 arcmin) was obtained using a small window (to minimize bandwidth) on the HRC-I.

The ACIS data are useful as backgrounds, since they include a charged particle environment very similar to that at the focal point of Chandra, and no sky-looking x-ray photons. The resulting spectra are described at (web page, and the event list files are available (see links in the above memo). The spectra are quite comparable to backgrounds obtained in less desirable modes (dark moon observations, or histogram mode ACIS data taken when HRC-I is at the focal point).


ACIS continues to operate smoothly. We continue to obtain a great deal of on-board calibration source data when the observatory is in the near-earth radiation belts. These data are used to monitor the performance of the ACIS devices. Users may find it useful to obtain on board calibration source data taken near in time to their GO data, for comparison purposes. The observatory was shut down from time to time due to radiation events associated with solar Coronal Mass Ejections, as the sun comes down from its period maximum activity.

Richard Edgar, on behalf of the extended ACIS Instrument, Operations, and Calibration Teams.

Previous  Next  Contents