Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0012-1-0.ps.gz Äàòà èçìåíåíèÿ: Thu Jan 11 19:29:40 2001 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:16:17 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http astrokuban.info astrokuban

In­orbit calibration activities of the XMM­Newton EPIC
Cameras
D.H. Lumb a , Ph. Gondoin a , M.J.L. Turner b , A.F. Abbey b , P.J. Bennie b , S. Sembay b ,
G. Griffiths b , P. Ferrando c , J­L. Sauvageot c , E.Belsole c , C.Pigot c , U.G. Briel d , K. Dennerl d ,
F. Haberl d , G. Hartner d , E. Kendziorra f , M. Kirsch f , M. Kuster f , S. Molendi g , G. Villa g ,
A. Tiengo g,e , A. Lagostina g,e , N La Palombara g E. Serpell e
a European Space Research and Technology Center, 2200 AG Noordwijk zh, the Netherlands
b X­Ray Astronomy Group, University of Leicester, Leicester LE1 7RH, U.K.
c Commissariat a l'Energie Atomique, 91191 Gif­sur­Yvette cedex, France
d Max­Planck­Institut fur Extraterrestrische Physik, 85748 Garching, Germany
e XMM­Newton Science Operations Centre, Villafranca Satellite Tracking Station, 28080 Madrid, Spain
f Institut fur Ast. & Astrophysik, University Tubingen, D72076 Tubingen, Germany
g IFCTR del CNR, via Bassini 15, 20133 Milano, Italy
ABSTRACT
The combined effective area of the three EPIC cameras of the XMM­Newton Observatory, offers the greatest col­
lecting power ever deployed in an X­ray imaging system. The resulting potential for high sensitivity, broad­band
spectroscopic investigations demands an accurate calibration. This work summarises the initial in­orbit calibration
activities that address these requirements. We highlight the first steps towards effective area determination, which
includes the maintenance of gain and CTI calibration to allow accurate energy determination.We discuss observations
concerning the timing and count­rate capabilities of the detectors. Finally we note some performance implications
of the optical blocking filters.
Keywords: XMM\GammaNewton, x­ray astronomy, CCDs, calibration
1. INTRODUCTION
On December 10 1999 the XMM­Newton space observatory was placed in a 48 hours period orbit by the first
commercial Ariane V launcher. The High Throughput X­ray Spectroscopy Mission XMM­Newton 1 is a ''Cornerstone''
project in the ESA long­term programme for space science. Its primary objective is to perform high throughput
spectroscopy of cosmic x­ray sources over a broad band of energies ranging from 0.1 keV to 10 keV down to a limiting
flux of 10 \Gamma15 ergs s \Gamma1 cm \Gamma2 . The XMM­Newton payload includes:
ffl Three grazing incidence telescopes 2 which provide an effective area higher than 4000 cm 2 at 2 keV and 1600
cm 2 at 8 keV and an image quality better than 15 arcsec Half Energy Width (HEW),
ffl Three CCD imaging cameras (EPIC) 3;4 , one at the prime focus of each telescope, which provide imaging in a
30 arcmin field of view and broadband spectroscopy with a resolving power of between 5 and 60 in the energy
band 0.1 to 15 keV,
ffl Two reflection grating spectrometers (RGS) 5 which provide spectroscopy between 0.2 and 2 keV with a resolving
power of over 250 at 0.5 keV,
ffl A Uv/optical monitor (OM) 6 which permits simultaneous monitoring of x­ray sources in the UV and optical
range included between 160 and 600nm.

The EPIC (European Photon Imaging Camera) consortium is a pan­European collaboration of Universities and
research facilities from the U.K., France, Germany and Italy. Behind two mirror systems equipped with Reflection
Grating Arrays are mounted the EPIC MOS cameras, and behind the unobscured mirror system an EPIC PN
camera. The details of these two detector systems are provided in references 3 and 4. The first in­flight images and
performance highlights were presented in Turner et al 7 , and Briel et al 8 . Briefly, the MOS cameras each comprise
7 conventional CCDs, closely butted to form a ¸65mm (30 arcminutes) diameter field of view. The CCDs are
nonetheless specially optimised to enhance the low­ and high­energy X­ray responses. They offer 1 arcsecond pixel
sizes, a normal readout time resolution of 2.6 seconds and energy resolution at 1keV of 70eV FWHM. The PN
camera is comprised of a wafer scale monolithic array of 12 CCDs, 60 mm square. The pixel size of 150¯m subtends
4 arcsec on the sky. A fast multi­parallel readout allows 73 ms frame times in normal full­field coverage. The energy
resolution is comparable with the MOS camera, while its combination of thick detection layer and thin dead layer
provides excellent efficiency over the whole XMM energy band.
The spatial domain performance of EPIC is intimately related to that of the XMM­Newton mirrors, and the ac­
companying paper 9 should be consulted for such aspects. This paper concentrates on the spectroscopic performance.
Table 1 summarises some of the more important celestial observations made during the initial in­flight calibration
programme for EPIC.
Target Name Revolution Goal
NGC2516 58,60 & 92 Determine boresight and CCD alignments
LMC X­3 66 & 92 PSF at different count rates
G21.5­09 60,61, 62 Effective area and Vignetting
A496 48 Vignetting
PKS0537­28 51 Effective Area
MS0737+7441 63 Effective area
MS1229+6430 82 Effective area
Mkn205 75 Effective area
PKS0312­770 57 On axis PSF of MOS in low pile­up flux rate,
comparison with CHANDRA.
1ES0102­72 65 Low energy emission line gain and
and N132D 76 & 83 provide cross check with CHANDRA
Crab Nebula 54 & 56 On­axis effective area in diffuse source.
Off­axis stray light measurements
in X­ray baffle performance
CAL 83 68 Filter X­ray transmission (O band)
HZ43 89 Filter X­ray transmission (C band)
Alpha Pic 79 Optical filter transmission
EXO0748­676 50 Mode Timing calibration
PSR0540­693 85 Mode Timing calibration
Table 1. Summary of key celestial calibration targets for EPIC
2. CTI AND GAIN
Before launch there was a concern that the proton damage experienced by the CHANDRA ACIS front­illuminated
CCDs 10 could affect in a similar way the EPIC MOS cameras (the two satellites are in comparable, highly eccentric
orbits, and the CCD technologies are very similar). The effects are reduced in EPIC by using a closed filter position,
whereby 1mm of aluminium shielding can be placed in the path of soft protons scattered from the mirrors. This filter
mechanism was designed with high reliability, to be employed at every perigee passage. Following the CHANDRA
experience it was also decided to close the filter when the on­board radiation monitor detected particle flares. This
practice, using conservative monitor thresholds, would prevent any significant degradation of the charge transfer
efficiency. The major damage could then result from the total dose of high energy protons from major solar flares or
encountered at perigee belt passage.

Figure 1. Internal calibration source spectrum, as observed with EPIC PN
Figure 2. Evaluation of EPIC MOS CTI with time. Mn Kff emision line (¸1790ADU peak position)
To monitor the charge transfer loss as a function of pixel position, an internal calibration source is deployed
regularly. This source comprises a primary Fe 55 source, with secondary Al K fluorescence, which illuminates quasi­
uniformly the whole CCD array (Figure 1). By measuring the centroid of the detected emission line energies as a
function of position, the charge loss per pixel transfer can be calculated.

Figure 2 shows the measurement of CTI as a function of time. Close examination shows that an on­axis source
located at pixel (300,300) in the MOS camera would have experienced a change in charge loss of ¸1 ADU since
launch. On a signal of 1790ADU for the calibration source, this is less than 0.1% of the mean value. Extrapolating
for an extended mission lifetime of 10 years, we therefore predict that loss of energy resolution will not affect scientific
goals, providing that the CTI is monitored continually and corrected appropriately. Potential damage (e.g. due to
major solar flares) could be accommodated in the MOS cameras by a change in operation temperature to ¸20 degrees
colder than currently used. For the PN cameras we see no measureable degradation since launch.
The internal calibration source is also used to measure the gain of the cameras, i.e. the energy to Analogue­Digital
conversion units. Celestial measurements using supernova remnants with bright emission lines have in addition been
used, because the slight non­linearities in the gain function are not well­sampled by the 3 bright emission line
energies of the calibration source. The analysis of these data is more complex, as a result of both the uncertainties
in the parameters of the astrophysical plasma models, and the intimate connection between gain and the energy
redistribution response functions for spectral fitting. Figure 3 shows some typical results from spectral fitting
obtained during the verification of the gain with celestial targets. We note there are minor inconsistencies still to be
resolved, but in general energies are calibrated to a few eV.
Figure 3. Spectral fitting and energy resolution in EPIC MOS ­ 1ES0102­72 supernova remnant data (crosses),
thermal spectrum+power law model(histogram)
The correctness of the energy calibration is also indicated by observations of the iron line energy in cluster A496.
Figure 4 displays the spectrum of this source measured with the PN camera. The centroid of the iron line feature
matches exactly that which would be predicted for the known red­shift of the cluster.
3. EFFECTIVE AREA
First measurements of the effective area confirm the predictions pre­launch (see accompanying papers 8;12;13 ). With
an effective area larger than previously deployed, any systematic effects of incorrectly modelled effective area become
larger than statistical variations for fainter fluxes, or for shorter observations than heretofore. Determining the
effective area and spectral response distributions are thus the most important goals for XMM­Newton calibration.
The flight model detector of the PN camera was replaced with the Flight Spare, rather late in the hardware
programme. As a result the ground calibration of energy response was not as detailed as originally planned. Using
the limited set of ground calibration data, and the in­flight calibration sources, a response model for this flight spare
detector has been developed. Fortunately the quantum efficiency of the PN CCDs is uniformly high, well­modelled

XMM­Newton EPIC­PN Observation of A496
(Garanteed Time Observation of J. Bleeker)
kod ­ XMM EPIC pn, spectra of selected areas : pnspc1/kod pnspc1.ps (0,1.3), 26­Mar­00 14:30:31, P 1
MPE Offline Analysis
­3 ­2 ­1 0 1 2
cm
­3
­2
­1
0
1
2
cm
­15' ­10' ­5' 0' 5' 10'
­10'
­5'
0'
5'
10'
0 191
128
1 127
64
2 63
0
Q
0
0
199
0
0 63 1
64 127 2
128 191
Q
1
0
199
0
191 128 1
127 64 2
63 0
Q
2
0
199
0 0
63
1 64
127
2 128
191
Q
3
0
199
1 10
10 ­4
10 ­3
10 ­2
10 ­1
1
E [keV]
singles
s
­1
keV
­1
arcmin
­2 Al­K Cu­K
C­K Ti­L O­K Fe­L Cu­L Si­K Ti­K Fe­K
1 10
10 ­6
10 ­5
10 ­4
10 ­3
E [keV]
photons
cm
­2
s
­1
keV
­1
arcmin
­2
Al­K Cu­K
C­K Ti­L O­K Fe­L Cu­L Si­K Ti­K Fe­K
(net) exp.: 23.9 ks
filter : thin
s ph = 0.70 (ass.)
area: s04 b04
pixels 3129 52156
cm 0.70 11.74
arcmin 14.8 246.9
ph. flux [10 ­3 cm ­2 s ­1 arcmin ­2
0.3 ­ 0.5 0.06894 +/­ 0.00057
0.5 ­ 0.7 0.04753 +/­ 0.00036
0.7 ­ 0.9 0.04092 +/­ 0.00032
0.9 ­ 1.3 0.07429 +/­ 0.00041
1.3 ­ 2.0 0.06032 +/­ 0.00037
2.0 ­ 4.0 0.05048 +/­ 0.00042
4.0 ­ 7.0 0.01622 +/­ 0.00024
7.0 ­ 12. 0.00327 +/­ 0.00027
0.3 ­ 12. 0.36197 +/­ 0.00108
XMM / EPIC pn pn048111.001
s04 b04 s04 ­ b04
Figure 4. Spectral fitting of the EPIC PN observation A496. Note the redshifted iron line feature at ¸6keV. The
large effective area of the EPIC PN allows measurements to –10keV, and hence enables much improved measure­
ments of hot plasmas and leverage on measuring power law slopes.Data courtesy of XMM­Newton Guaranteed Time
Observations of J Bleeker, SRON, Netherlands
and confirmed in a number of flight­like detectors in white­light synchrotron beams. Likewise, for the MOS CCDs,
some individual CCD chips were replaced after the nominal calibration programme, in order to furnish the flight
cameras with a complement of high performance devices. The spectral response distributions of the MOS cameras
were extensively calibrated with monochromatised radiation at a synchrotron facility. However the quantum efficiency
could not be accurately measured at some energies. The efficiency itself proved complex to model, due to the many
different surface and electrode features of the CCD design.
To verify these different effective area features, the in­flight calibration programme has used a number of targets
possessing simple power­law spectra, such as plerionic supernova remnants and AGN. Anomalous residual features
in the fitted spectra are used to highlight discrepant features in the effective area (mainly confined to absorption
edges) and response functions (mainly low energy residuals in absorbed sources). This work is still in progress, but
already the calibration goals for allowing the first release of data to guest observers have been met, and exciting
science is already enabled. We use the initial response matrices as a baseline for testing aspects of mode performance
and blocking filter transmission as described in the rest of the paper.

4. MODE PERFORMANCE
One characteristic of CCD detectors that is in marked contrast with prompt (i.e. gas and M.C.P.) counters, is the
finite readout time, which makes CCDs prone to pile­up effects: The energy resolving capability is compromised
when two or more photons are detected in a pixel, or adjacent pixels, in one readout frame. Such degradation is
minimised in EPIC by an appropriate choice of flexible readout modes. A reduced size window can be selected for
the CCD at the focal point of each camera, reducing the frame time for readout, and the fraction of piled­up events
in proportion.
Mode Name Field of View Comment Frame Time Pt. Source Flux
(arc min) (mCrab)
PN (ms)
Full Frame 27.5 x 26.4 7% out of time events 73 1
Full Frame Extended 27.5 x 26.4 Reduced out of time events 200 0.3
Small Window 4.4 x 4.4 One CCD only 6 15
Large Window 13.8 x 26.4 A frame­store mode 45 2
Timing 4.4 x 13.8 1­d imaging only 0.03 150
Burst 4.4 x 1.4 Low duty cycle 0.007 6000
For Timing and Burst modes Frame
time = source integration time
MOS (s)
Full Frame 33 x 33 2.6 0.5
Small Window 1.8 x 1.8 For inner CCD 0.3 15
Outer CCDs full coverage 2.7
Large Window 5.5 x 5.5 For inner CCD 0.9 4
Outer CCDs full coverage 2.7
Timing 1.8 x 10 1­d imaging 0.001 150
Outer CCDs full coverage 2.7
Table 2. Summary of EPIC readout modes used in flight
A trade­off has to be made in selecting such modes. A very small window does not encompass all the energy
enclosed in the Point Spread Function (PSF), and flux loss outside the window must be accounted for, in determining
the effective area. In slow readout modes, pile­up in the core of the PSF also leads to flux loss. In both cases the
flux loss is energy dependent. Pile­up itself leads to the combining of multiple low energy photon signals into single
events of higher energies. In other words piled­up spectra show spectral hardening and flux loss. These effects are
discussed in reference 11. Guest Observers applying to use EPIC were advised that a count rate not exceeding 1
photon/CCD readout ensures minimal spectral distortion due to pile­up. Following initial commissioning tests, the
performance of readout modes of the two camera types have been consolidated. Table 2 indicates their summary
parameters, including flux rate capability based on the above criterion.
The MS0737+7441 AGN exhibits a simple absorbed power law spectrum, and was observed in two different
modes in EPIC MOS. The extracted spectral fit parameters are shown in Table 3. This table shows that not only are
measurements in different modes consistent, but also that the spectral distortion at a few counts/second is negligible.
An extreme example of count rate capability is provided by the observations of black­hole binary, LMC X­3.
This was observed to provide a high S:N measurement in the PSF wings. The piled­up core was ignored (Figure
5), and spectra from outer regions were extracted in two exposures with different modes. Again the spectral fits
are compared in the two modes (Table 4), and except for the normalisation factors (as expected due to pile­up flux
losses), are very similar. This demonstrates that reliable spectral fitting can still be obtained by appropriate data
selections on targets far brighter than nominal limits in Table 2.

MOS Mode Counts/Frame Hydrogen Column Power Law Normalisation
Full Frame 3.6 4.16 \Sigma0.10 2.39\Sigma0.05 2.16\Sigma0.05
(x 10 20 cm \Gamma2 ) (x 10 \Gamma3 )
Large Window 1.2 4.28\Sigma0.13 2.41\Sigma0.03 2.24\Sigma0.04
Table 3. Comparison of spectral fitting results in different EPIC readout modes
Figure 5. Zoomed and log­scaled image of LMC X­3 PSF (EPIC MOS). Dashed circle indicates limit of region of
piled­up core to be excised (¸13 arcsecs radius)
MOS Mode Cts/sec Cts/Frame H Column Disc Temp Norm. Power Law Norm.
(Measured) (Inferred) (x 10 20 cm \Gamma2 ) (keV) (x 10 \Gamma2 )
Full Frame 10.8 126 9.1 \Sigma0.6 0.8\Sigma0.03 5.45\Sigma0.46 2.60\Sigma0.07 1.07\Sigma0.07
Large Window 11.3 44 9.0\Sigma0.43 0.75\Sigma0.02 7.34\Sigma0.47 2.57\Sigma0.04 1.15\Sigma0.05
Table 4. Results of spectral fitting achieved on LMCX­3 data, after excision of the pile­up core

A final note on the use of readout modes is that both EPIC cameras provide a timing capability through continual
readout modes. In the PN case this is demonstrated with observations of X­ray pulsars, where the rapid readout
allows Ümillisecond resolution or the ability to measure bright targets without imaging pile­up. Figure 6 shows the
folded pulse profile of PSR0540­69, re­binned at 1msec samples.
XMM­Newton EPIC­PN Observation of PSR 0540­69 in Timing Mode
AIT
6
IAAT
PSR 0540­69 (Timing Mode)
6
AIT
7
IAAT
PSR 0540­69 Pulse Profile
Pulse Profile PSR0540­69 (Timing­Mode)
Phase (assuming P=0.05051462s)
0.0
0.2
0.4
0.6
0.8
1.0
Intensity
[arbitrary]
0.0 0.5 1.0 0.5 1.0
Pulse Profile PSR0540­69 (Timing­Mode)
Phase (assuming P=0.05051462s)
0
2
4
6
8
10
Intensity
[arbitrary]
0.0 0.5 1.0 0.5 1.0
7
Figure 6. EPIC PN observation of PSR0540­69 in timing mode. Data rebinned and folded to 1msec time resolution
5. OPTICAL BLOCKING FILTERS
Visible photons produce a signal in the X­ray cameras indistinguishable from that of X­ray photons. This additional
background signal creates an offset in the X­ray energy scale and an additional shot noise component. This noise
degrades the energy resolution and contributes to a change in response distribution function. The filter wheel in EPIC
carries different optical blocking filters optimised for particular investigations. The thick filter comprising 200nm Al
on a 350nm polypropylene support, reduces optical flux below 1 electron per pixel per frame for stars as bright as
0 magnitude. Medium and thin filters produced with 80 or 40 nm of Al on 160nm polyimide films have a higher
transmission for soft X­rays, but with magnitude limits for minimal noise degradation of ¸8 and 14 respectively.
Their performance is verified with observations of the CAL83 super­soft source. In Figure 7 we show the results
of spectral fitting of a black body spectrum with interstellar photoelectric absorption and an broad absorption edge
representing the NLTE absorption models. In this case the observation was made with the THIN filter. In Figure
8 this fixed model is compared with response in the MEDIUM and THICK filters. This illustrates the significant
improvement in soft X­ray response that can be achieved by an appropriate choice of filter. If EPIC had been
deployed with a fixed filter, with an aluminium thickness necessary to attenuate optical light from most stellar
targets, a significant factor ¸3 reduction in soft X­ray collection area would have been suffered, compared with the
thinnest of filters now employed. With its set of filters, EPIC can now provide an energy response over two decades,
from 0.15 to 15keV.

Figure 7. Data and model fit for CAL83 source in EPIC MOS THIN filter
Figure 8. Comparison of THIN filter response (upper histogram) to CAL83 super soft source, with data points
measured with MEDIUM (middle) and THICK (lower) filters
6. CONCLUSIONS
The photon energy determinations have not been significantly degraded since launch, and the effective area of the
EPIC cameras seems to meet all expectations. The calibration knowledge, following modifications deduced as a result
of a comprehensive in­flight calibration programme, is already acceptable to start performing scientific investigations
of the Guest Observer programme. The final refinements for energy response will be challenging to achieve, and
need to consider details of mode pile­up performance and optical light leakage. Nevertheless we demonstrate that
the basic expectations of count rate capability and filter transmissions are amply confirmed

ACKNOWLEDGMENTS
We are grateful to our colleagues from the XMM\GammaNewton Science Operation Center, at Villafranca Satellite Track­
ing Station, Madrid, Spain. Their support in implementing the in­orbit calibration programme in many diverse
operational aspects was invaluable. Many thanks also go to the development team of the XMM\GammaNewton Science
Analysis Software, which was used for much of this work. The data shown in Figure 4 were provided from Guranteed
Time Observations of Prof. J. Bleeker, of the Space Research Organisation Netherlands. We gratefully acknowledge
the use of this data to help to demonstrate the capabilities of the EPIC cameras.
7. REFERENCES
1 F. Jansen, ''XMM: Advancing Science with the High­Throughput X­ray Spectroscopic Mission'', ESA bulletin 100,
pp.15­20, 1999.
2 Ph. Gondoin, D. de Chambure, K. van Katwijk, Ph. Kletzkine, D. Stramacicioni, B. Aschenbach, O. Citterio, R.
Willingale, ''The XMM Telescope'', SPIE Proc.2279, pp.86­100, 1994.
3 A.D.Holland, M.J.L.Turner, A.F.Abbey, P.J. Pool, ''MOS CCDs for the EPIC on XMM'', SPIE Proc.2808, pp.
414­420, 1996.
4 N. Meidinger, H.W. Brauninger, R. Hartmann, G. Hartner, N. Krause, E. Pfeffermann, C. Reppin, G. Schwaab,
L. Struder, J. Trumper, P. Holl, J. Kemmer, S. Krisch, H. Soltau, C. van Zanthier, D. Hauf, R. Richter, ''PN­CCD
detector for the European Photon Imaging Camera on XMM'', SPIE Proc.2808, pp. 492­503, 1996.
5 A.C.Brinkman, H.J.M.Aaarts, A.J.F.den Boggende, L.Dubbeldam, J.W.den Herder, J.S.Kaastra, P.A.J.de Korte,
R.Mewe, C.J.Hailey, S.M.Kahn, F.B.S.Paerels, G.Branduardi­Raymont, J.V.Bixler, K.Thomsen, A.Zehnder, ''Re­
flection Grating Spectrometer on­board XMM'', SPIE Proc.2808, pp. 463­480, 1996.
6 R. Much, D. Lumb, M.S. Cropper, R. Hunt, K.O. Mason, F.A. Cordova, T. Sasseen, C. Ho, W. Priedhorsky,
C.Jamar, E. Antonello, ``The Optival/UV Monitor on the X­ray Multi Mirror Mission'',Proceedings of the Confer­
ence '' Ultraviolet Astrophysics, Beyond the IUE Final Archive'', Sevilla, Spain, ESA SP­413, p.815, 1998.
7 M.J.L. Turner et al, Presented at SPIE 4012
8 U.G.Briel , Proceedings SPIE 4012, in press 2000
9 Ph. Gondoin et al, these Proceedings
10 L.K. Townsley, P. Broos, G. Garmire, J. A. Nousek, ``Mitigating Charge Transfer Inefficiency in the Chandra
X­Ray Observatory Advanced CCD Imaging Spectrometer'', Astrophysical Journal 534, L139­L142, 2000
11 J. Ballet, ``Pile­up on X­ray CCD instruments'', Astronomy and Astrophysics Supplement 135, p.371­381, 1999
12 P. Marty, et al, These Proceedings
13 C. Pigot C., J.L. Sauvageot, P. Ferrando and E.Belsole, these proceedings