Low-energy Energy Scale of ACIS S3 Chip

        Richard J. Edgar (mailto:edgar@head-cfa.harvard.edu)
        November 17, 2000
        revised July 17, 2001
        disk location: /data/asc8/iCXC/htdocs/cal/ACIS-S-LETG

In September 2000, a series of observations were conducted of a bright continuum source (the quasar PKS 2155-304) using the LETG and ACIS-S, with large y-offsets (i.e. in the dispersion direction). The point of these observations was to cause the low energy portion of the spectrum to fall at various places on the ACIS-S array, in particular the S3 chip. Five different offsets were used (6, 8, 10, 12, and 14 arcmin), and the observations were taken at two focal plane temperatures: -110 and -120 C.

Data Used for these analyses:

        obsid   y-off   FP_temp
        -----   arcmin  degrees_C
        1790    6.0     -110
        1791    8.0     -110
        1792    10.0    -110
        1793    12.0    -110
        1794    14.0    -110
        1795    6.0     -120
        1796    8.0     -120
        1797    10.0    -120
        1798    12.0    -120
        1799    14.0    -120

We present here the analysis of the -120 C data. For each of the four quadrants of the chip, we construct a new column in the level 1.5 FITS event file (acis*evt1a.fits) called TG_ENERGY, which is the energy derived from the transmission grating geometry (i.e. hc/lambda). We then merge the event lists from the five observations. This is then plotted against the pulse-height-derived energy from the standard reduction. For 10 eV slices in TG Energy, the difference is taken, and we plot the mean and standard deviation (as error bars) of both PI Energy and the residual as functions of TG Energy. These plots represent the (unnormalized) response matrix of the ACIS: for photons of a given true energy (a vertical slice on these plots), the distribution of pulse heights is plotted.

The conversion from raw pulse heights to PI and energy was done by the acis_process_events tool, using the gain file:

CALDB/data/chandra/acis/bcf/gain/acisD2000-01-29gainN0001.fits

Figures:

scatterplots            gain plots              gain residuals
------------            ----------              --------------
http:S3c0_scatter.ps    http:gain_0.ps          http:resid_0.ps
http:S3c1_scatter.ps    http:gain_1.ps          http:resid_1.ps
http:S3c2_scatter.ps    http:gain_2.ps          http:resid_2.ps
http:S3c3_scatter.ps    http:gain_3.ps          http:resid_3.ps

NOTE that the errorbars in the gain and residual plots represent the one sigma points of the distribution of points, not the errors on the mean.

All of these experiments were done at the same CHIPY positions, so this represents the response of the chip for 150 <= CHIPY <= 190 pixels. Errors in the TG Energy due to the off-axis point response function have been ignored; they are small compared to the intrinsic resolution of the ACIS pulse height response. It can be seen that we have differing numbers of events in different energy ranges; in particular we have few events above about 600 eV for most of the quadrants of the chip.

We see that the energy scale is remarkably linear down to an energy of about 400 eV. There is a systematic offset of about 20 eV in the sense that the true energy is higher than the mean pulse-height derived energy. Below this, the event threshold begins to cut off a significant portion of the pulse height distribution, which skews the mean of the distribution to higher values.