Quantum Efficiency of ACIS CCD w134c4r

Notes for pre-release quantum detection efficiency curve for ACIS back-illuminated detector S3 = w134c4r. Note: these data represents work in progress; substantial uncertainties pertain. See "intended use" and "notes/bugs" below.

M. Bautz
14 September




ASCII table suitable for use with qdp. There are two columns:

  1. Energy (keV)
  2. Quantum Efficiency (for ASCA grades 0,2,3,4,6)


This file provides the best currently available estimate of the quantum efficiency of the ACIS S3 CCD detector (MIT Lincoln Laboratory model ccid17, serial number w134c4r, a back-illuminated device.) The data are for ASCA grades 0,2,3,4,6, with a split-event threshold of 15 electrons (13 adu; adu= analog-to-digital converter units) and an event threshold of 45 electrons (20 adu).

These estimates were derived from MIT CSR subassembly calibration measurements and from XRCF Phase I (flat-field) measurements and reflect the spatially averaged detection efficiency of the device.

A simple three-component model of the device response was fit to relative quantum efficiency measurements. The model includes uniform dead layers of silicon and silicon dioxide; the total thickness of the photosensitive region is the third model parameter.

Two data sets were used. Relative quantum efficiency (QE) data were obtained at CSR with respect to a standard detector (w203c2, a front-illuminated device for calibrated at PTB/Bessy) To allow joint fits with XRCF Phase I data, the MIT relative QE data were converted to relative directly efficiencies with respect to the S2 detector, w182c4r, using the relative QE of w182c4r and reference detector w103c4r. The BESSY calibration of w203c2 was not used in this conversion.

XRCF (ACIS telemetry) data from Phase I provided additional measurements of the QE of w134c4r relative to S2 (w182c4r). At four of the six energies sampled by both the XRCF and CSR data (viz., 1.74, 2.1, 4.5 and 8.0 keV), there is agreement in the efficiency with respect to S2 within 6%. At lower energies (0.53 and 0.7 keV) there are disagreements in the the two data sets that are as large as 25%. The source of this discrepancy is unknown. Somewhat arbitrarily, we have chosen to ignore the XRCF data at 0.525 and 0.705 keV, although we include XRCF phase I data at 0.277 keV, in the process of modeeling the QE of w134c4r.

The relative QE of w134c4r (S3) with respect to w182c4r(S2) was converted to an absolute detection efficiency using the best-fit model for the absolute detection efficiency of w182c4r contained in w182c4r_eff_997.qdp. Data used in the fit are listed below:

Absolute QE of w134c4r obtained from relative QE measurements at MIT follows:


The absolute QE of w134c4r obtained from relative QE measurements with respect to S2 at XRCF follows:


* excluded from w134c4r model fit

The best-fit model parameters for w134c4r used to generate this curve follow:

Deadlayer Silicon Thickness 1x10-4 ±m
Deadlayer Si02 Thickness: 2.3x10-2 +7.3x10-3 -2x10-3 µm*
Photo-sensitive Thickness 39.7 ±4 µm*

* 90% 1-parm conf. int.

Intended use:

This curve is intended for use by ASC to illustrate the performance of ACIS back-illuminated detectors. Given the (yet-to-be-resolved) inconsistencies in data analyzed to date, it is difficult to assign an accurate errors to this QE estimate. See notes for an attempt. In any event, THIS CURVE SHOULD NOT BE USED FOR PURPOSES REQUIRING ERRORS LESS THAN 25% BELOW 1.5 keV.

  1. Given as-yet unexplained inconsistencies between MIT and XRCF data below 1.5 keV (see above), it is not possible to form reliable error estimates for this qe curve in this spectral region. Taking the magnitude of the inconsistencies as a measure of the likely uncertainties suggests errors no greater than 10% at energies exceeding 1.5 kev. Errors at lower energies are undoutedly larger, and may be as large as 25% in the immediate vicinity of the oxygen edge.
  2. The relative QE data from MIT CSR should be reanalyzed and related directly to BESSY-calibration of w203c2. Additional data (comparing w134c4r to w147c3, a BESSY-calibrated, back-illuminated device) should be used to constrain the quantum efficiency model.
  3. The quantum efficiency model effectively assumes branching ratios which are independent of energy for both front- and back-illuminated devices. This assumption is not strictly correct.
  4. The quantum efficiency w134c4r is similar to that of the other back-illuminated device in the ACIS focal plane, w140c4r(S1) at energies low energies (less than 4 keV), but S3 is considerably more efficient than S1 at higher energies (25% more efficient at 8 keV).
  5. Spatial variations in detection efficiency across w134c4r can be as large as 15-20%, RMS.
  6. The model qe data in this file (w134c4r_eff_pre_997.qdp) are replica of /benz/h4/mwb/asc/obs_guide/qe/newbi/w134c4r_model_data_rev.qdp of 0637 Sep 15

Mark W. Bautz
Thu Dec 4 18:33:45 EST 1997 Last modified: Thu Dec 4 19:21:54 EST