Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0030.pdf Äàòà èçìåíåíèÿ: Thu Jul 16 15:59:16 2015 Äàòà èíäåêñèðîâàíèÿ: Sun Apr 10 00:07:17 2016 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: dark nebula

XMM-Newton Calibration Technical Note
XMM-SOC-CAL-TN-0030 issue 7.4 Status of the RGS Calibration
Compiled by R. Gonz´lez-Riestra, on behalf of the RGS consortium a July 16, 2015

Contents
1 Intro duction 2 The RGS Calibration 2.1 Effective area . . . . . . . . . . . . 2.1.1 Empirical effective-area corr 2.1.2 Contamination correction . 2.1.3 RGS1 dispersion-dependent 2.1.4 Sensitivity of RGS2 CCD2 . 2.1.5 Instrumental absorptions . 2.1.6 Higher-order corrections . . 2.1.7 The final CCF model of the 2.2 Line spread function . . . . . . . . 2.3 Wavelength scale . . . . . . . . . . 2.4 Cross-dispersion distribution . . . 2.5 RGS temporal resolution . . . . . . 3 Ca 3.1 3.2 3.3 3.4 3.5 libration-related asp ec RGS Background . . . Pile-up . . . . . . . . . CCD detector defects . Pixel offset values . . . Fixed-pattern noise . . ts .. .. .. .. .. of . . . . . RGS .... .... .... .... .... 2 4 6 7 7 8 8 8 9 10 11 14 14 15 16 16 17 18 18 18 19

........ ections . . . . ........ correction . . ........ ........ ........ RGS effective ........ ........ ........ ........ data ... ... ... ... ... a . . . . . naly ... ... ... ... ...

. . . . . . . ar . . . .

. . . . . . . e . . . . . . . . .

. . . . . . . a . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

sis .. .. .. .. ..

References

1


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

2

1

Intro duction

The two high-resolution RGS instruments aboard the XMM-Newton Observatory have been operating since first light a few weeks after launch in 1999. The set of CCF calibration files that describes the instruments has undergone significant development since then to include in particular time-variable empirical effective area corrections for RGS1 and RGS2 over the 6-38 ° bandwidth. A As described in more detail below, these corrections were derived independently for the RGS based on reasonable assumptions concerning the spectral form of a few well-known X-ray sources, and have had the effect of greatly improving the agreement between RGS and the EPIC instruments, the differences between which now conservatively stand at a few percent. In addition to the slowly-changing, wavelength-dependent changes in the sensitivity of the RGS, one CCD assembly of the nine in each RGS failed early in the mission because of electronics problems. These were RGS1 CCD7 and RGS2 CCD4, roughly covering the wavelength ranges 11-14 ° and A °, respectively, in first order. Coverage of the wavelengths affected has been maintained by 20-24 A the built-in redundancy between RGS1 and RGS2 or between first and second order or both. Concerning its high-resolution capabilities, measurements of wavelengths are accurate to better A than about 6 m°. The dramatic reduction of the effects of radiation damage after the operating temperatures of RGS1 and RGS2 were lowered from -80o C to -110o C in November 2002 has been maintained. Otherwise the instruments continue to operate as well as before. The purpose of this document is to alert users to the quality of the instrumental calibrations developed so far in order to provide a proper context for the scientific interpretation of RGS data. Documents of varying levels of complexity are available to people interested in the details of the instrument's construction and behaviour: a summary of important instrumental parameters is shown below, a more thorough description of the instrument and the flight performance is given in [1] and [2], and detailed individual calibration documents, written mainly by instrument experts at SRON Utrecht, Columbia University and the XMM-Newton SOC, are available as shown by the list of references at the end of this document.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

3

The line and may

RGS delivers high spectral resolution for an effective area of typically 100 cm2 . The instrumental width is weakly wavelength dependent with mean FWHM of about 70 m° and 50 m° in first A A second order, respectively. Pending more detailed discussion below, the overall performance be summarised thus: RGS calibrated bandwidth ° 6 (A) 38 0.33 E (keV) 2.07

Maximum effective area Spectral Resolution Wavelength accuracy RGS1+RGS2



1st order 125 cm2 at 15 ° A 250 at 15 ° A 6 m° A

2nd order 57 cm2 at 10 ° A 300 at 10 ° A 4 m° A

Typical full-detector background rates 6-8 ° A 5 cts ks-1 °-1 A 8-27 ° A 2 cts ks-1 °-1 A A 27-34 ° A 7 cts ks-1 °-1 34-38 ° A 2 cts ks-1 °-1 A

CCD integration time RGS1 0.6 s RGS2 1.2 s Table 1: Summary of the main characteristics of the XMM-Newton Reflection Grating Spectrometers in standard spectroscopy mode.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

4

2

The RGS Calibration

The instrument calibration is based on a physical model of the various instrument components: · the mirror response (not part of this document) · the grating response: ­ reflectivity ­ figure errors ­ scattering ­ alignment ­ vignetting · the CCD detector response: ­ quantum efficiency (QE) ­ monochromatic pulse-height redistribution function ­ transmission of the optical blocking filter ­ charge transfer inefficiency (CTI) ­ gain During ground calibrations, these various components were calibrated and a physical model for each constructed and later verified in flight. Subsequent adjustments in the form of time-dependent and wavelength-dependent empirical corrections have been based on in-flight observations of specific sources. · empirical effective-area corrections: ­ assumed power-law blazar spectra ­ detector contamination The resulting knowledge with some important rele only nominal values and model is given in [1] and of the instrument is encapsulated in the RGS CCF files shown in Table 2 vant general-purpose components. Two other redundant CCF files contain are not used in any analysis. An extensive description of the instrument details may also be found in the XMM-Newton Users Handbook [3].

During routine operations, many instrumental parameters are monitored (detector contamination, detector gain, CTI evolution due to radiation damage, etc) and this leads to periodic modifications of the relevant RGS CCF components. It should be emphasised that new CCF components may be released at any time, not necessarily coordinated with new versions of the SAS or new issues of this document. Each new CCF component made publicly available is accompanied by a release note available at the XMM-Newton CCFs Release Notes Page with a name of the form XMM-CCFREL-vvvv describing its relevance and applicability.


XMM-Newton Technical Note

Purpose CCD event selection criteria CCD bad pixels CCD readout parameters CCD cool pixels Cross-dispersion PSF parameters CCD CTI correction parameters CCD dark response CCD effective area correction factors CCD quantum efficiency correction factors HK parameter selection criteria RGS system geometry Dispersion PSF model component parameters CCD readout parameters CCD quantum efficiency physical model CCD energy response parameters Wavelength-scale correction parameters Observed RGS background templates Henke X-ray absorption coefficient tables Includes RGS Euler angles Includes assorted RGS data

CCF component RGS[12] ADUCONV RGS[12] BADPIX RGS[12] CLOCKPATTERNS RGS[12] COOLPIX RGS[12] CROSSPSF RGS[12] CTI RGS[12] DARKFRAME RGS[12] EFFAREACORR RGS[12] EXAFS RGS[12] HKPARMINT RGS[12] LINCOORD RGS[12] LINESPREADFUNC RGS[12] MODEPARAM RGS[12] QUANTUMEF RGS[12] REDIST RGS[12] SAACORR RGS[12] TEMPLATEBCKGND XMM ABSCOEFS XMM BORESIGHT XMM MISCDATA

RGS1 RGS2 0025 0032 0033 0032 0001 0001 0001 0001 0004 0004 0012 0013 0005 0006 0009 0009 0005 0005 0014 0013 0008 0008 0005 0005 0005 0005 0015 0016 0004 0004 0001 0001 0006 0008 0004 0024 0022

Release Note 319/327 311 015 218 142 289 173 314 212 307 081 275 015 215 067 297 261 077 315 134

date July 2014/May 2015 February 2014 September 2000 June 2006 December2002 September 2012 July 2004 April 2014 June 2006 October 2013 June 2001 December 2011 September 2000 June 2006 March 2001 April 2013 April 2010 May 2001 July 2014 February 2007

XMM-SOC-CAL-TN-0030 issue 7.4 Page: 5

Table 2: RGS components of the XMM-Newton CCF in June 2015. RGS[12] shows that there are separate files for RGS1 and RGS2. RGS-specific version numbers and release notes refer to the latest release, while the general-purpose files refer to the latest versions relevant to the RGS. Please be aware that CCF releases generally occur more often than new editions of this document.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

6

Figure 1: Left: Model of the effective area derived from ground calibrations with a number of measured data points [1]. Right: the CCF in-flight effective area of a typical observation for RGS1 in black and RGS2 in red [3]. Due to scattering the effective area has no sharp features.

2.1

Effective area

Knowledge of the effective area of the two RGS instruments comes from a combination of ground measurements and observations in flight. It has developed significantly since launch with the introduction of two particularly important additions to the calibration that was originally based on an end-to-end physical model of the whole system. Empirical corrections were first introduced in 2006, based on the assumed power-law form of blazar spectra [5], followed a year later by the recognition of wavelength-dependent sensitivity changes which are consistent with a build-up of hydrocarbon contamination on the cold CCD detector surfaces (see 2.1.2). The absolute accuracy of the effective area is estimated at 10%. The overall characteristics of the effective area are illustrated in Fig. 1. Obvious features include the two failing CCD readout chains and the effect of a number of hot columns and pixels shown by narrow drops in the effective area. Instrumental Oxygen absorption is clearly visible near 23 °. A Less obvious instrumental features revealed on closer inspection include a shallow Al edge around ° ° 8.3 A from the blocking filter on top of the CCDs and of Mg and F edges around 9.5 A and ° 17.9 A respectively from the MgF2 isolation layer between the CCD and the blocking filter. As a result of measurements in flight, the effective-area model incorporates these and a number of other corrections which apply to both RGS1 and RGS2 unless stated otherwise: · empirical corrections based on power-law blazar spectra · absorption in a contamination layer · dispersion-dependent corrections (RGS1 only) · sensitivity adjustment of RGS2 CCD2 A · instrumental Oxygen and MgF2 features near 23 and 17.9 ° · empirical higher-order corrections.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

7

2.1.1

Empirical effective-area corrections

The long-wavelength part of the RGS has been difficult to calibrate due to the scarcity of both suitable ground-based facilities and celestial X-ray standards. It has become clear that blazars have smooth spectra well characterised by power-laws sub ject to interstellar absorption in the RGS waveband. This was empirically supported by the fact that the RGS spectra calculated with earlier versions of the calibration could be corrected for interstellar absorption and power-law slope to reveal a universal form reflecting the shape of the effective area of the RGS instruments [5]. Empirical effective area corrections have been derived using measurements of Mkn 421 at different epochs during the mission and fitting a power-law models to the stable, well-calibrated part of the RGS waveband between 10 and 25 °. Using the Crab as a reference for both power-law slope and normalisation [6], A the extrapolation of the resulting models to long and short wavelengths has been used in comparison with the observed spectra to define correction factors. These correction factors are not thought to be time dependent but when combined with the contamination correction discussed immediately below, give rise to the time and wavelength-dependent tabulations in the EFFAREACORR CCFs for the calculation of RGS response matrices. At the shortest breaks down at ground and so, theoretical shor 2.1.2 wavelengths, the underlying gratings model based on electromagnetic scalar theory the small angles involved. This could not be compensated by measurements on the at these shortest wavelengths, the empirical correction serves to compensate for the tcomings.

Contamination correction

Figure 2: Thickness of spectra of the neutron of which are effectively of the RGS effective ar

the layer of carbon contamination implied by differences in the RGS fluxed star RX J1856.6-3754 (black) and the Vela Pulsar Wind Nebula (red), both constant. An exponential fit accounts for the time-variable part of the model ea.

The RGS sensitivity has been decreasing at long wavelengths due almost certainly to the appearance of a layer of contamination on the CCDs. The layer was initially modelled as linearly increasing


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

8

in thickness [7] but has been increasing more slowly in recent years as shown in Fig. 2 [8]. The exponential law now in place is used to extrapolate into the future. 2.1.3 RGS1 disp ersion-dep endent correction

Data of calibration sources obtained early in the mission showed unexpected systematic differences between RGS1 and RGS2, with RGS1 20% less efficient at larger dispersion angle, , than RGS2. Although not entirely satisfactory in physical terms, this could be explained by additional blocking of the beam halfway between the gratings and the detectors - although the required mismatch of the optical blocking shield by 5 cm in one of the two instruments would seem unrealistic. In any case, an empirical -dependent correction ranging from 1.00 at = 0.038 rad (6.5 ° for first order) A °) has been derived and is applied during SAS data analysis [9]. to 1.25 at = 0.075 rad (37.7 A 2.1.4 Sensitivity of RGS2 CCD2

For smooth continuum sources, comparison of the count rate of RGS2 CCD2 with its neighbouring CCDs 1 and 3, and with RGS1 CCD2 shows its sensitivity is about 10% lower than expected, likely due to an extra absorbing layer deposited during the manufacturing process. In order to ensure a smooth wavelength-dependent effective area, of various possibilities for an extra passive absorption layer to account for the lower sensitivity of this CCD, a 40 nm layer of SiO2 gives the best description [10] . 2.1.5 Instrumental absorptions

Figure 3: The effect of an additional Oxygen layer on the detectors as determined by comparison of sources of varying interstellar column densities.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

9

Oxygen There is a significant reduction of about 25 % in the effective area around the O-edge, roughly equivalent to a 30 nm layer of Al2 O3 . Details of this edge, including its fine structure, have been calibrated in flight using a number of sources with low and high column density to separate the detector response from interstellar absorption [11]. Fig. 3 gives both the global structure, that follows the Henke absorption coefficients, and the fine structure. The magnitude of the Oxygen layer on the detector has been confirmed by ground measurements on a flight-spare CCD. Nevertheless, the value of the oxide layer is rather large compared with the two native SiO2 and Al2 O3 oxide layers, each of assumed thickness 5 nm, so it is plausible that a layer of ice formed on the detector when they were cooled after launch. Fluorine and Magnesium The 25 nm MgF2 layer on the CCDs that isolates them from the optical Al filter gives rise to Mg and F absorption edges near 17.9 °, see [12]. A 2.1.6 Higher-order corrections

Figure 4: Corrections for second order (above) and third order (below) for the RGS instruments, shown with the solid-line approximations used in the CCF. RGS1 before and after the loss of data from CCD7 is in red and green; RGS2 in blue. See [13] for details. Between the first and higher spectral orders, wavelength­dependent corrections have been introduced to enforce agreement. Identical correction factors for RGS1 and RGS2 were derived from the relative intensities of Mkn421's first and higher orders as shown in Fig. 4. The largest differences are at low wavelengths, where the effective area is small and a steep function of wavelength. It should be emphasised that such efficiency adjustments do not affect the better resolution of the higher orders which can be useful for the identification of lines or the resolution of blends.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

10

Figure 5: Evolution of the RGS effective area model since launch until May 2015 at different labelled wavelengths. The points plotted were extracted from the RGS1 and RGS2 first and second order response matrices calculated with SAS v13.5. The effective area is dependent on a combination of conditions and shows, for example, the build up of contamination particularly at long wavelengths and the significantly fewer bad pixels, in particular in RGS2, after cooling the CCDs in November 2002 (dotted line). 2.1.7 The final CCF mo del of the RGS effective area

The RGS effective area in an observation is a result of the combination of all the considerations discussed above and thus is a complicated function of time, wavelength and source geometry. Fig. 5 shows how the area at the wavelengths of some prominent emission lines has varied since launch according to the CCF model. The quality of the RGS calibration is best judged by the success of the resulting models in two types of comparison: reproducing the known spectra of calibration standards, and the level of agreement with EPIC on calibration standards and a variety of other ob jects whose spectra are not known a priori. Detailed comparisons of this type using a large sample of observations throughout the mission are described in the XMM-Newton Cross-Calibration Status document [14].


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

11

2.2

Line spread function

Figure 6: The three main components of the line response, known as monochromatic line at 15 °: the pro jected mirror response is shown in A the grating response in red; and after application of the energy selecti blue. The small discontinuities are due to the approximation used in the small-angle scattering.

the LSF, of the RGS to a green; after broadening by ons to reduce the wings in numerical treatment of the

The monochromatic response, known as the line spread function (LSF), results from successive convolutions of the mirror response pro jected in the dispersion direction; the grating response; and the detector response. The mirror response is approximated by a Lorentzian profile and was calibrated against the flight data of PKS 0312-770 in orbit 057 [15]. Different components contribute to the grating response: · accuracy of the variable line density · figure errors · misalignments between individual grating elements · X-ray scattering by surface roughness Two scattering components have been identified corresponding to different scales of the surface roughness: a small-angle Gaussian component and a large-angle Lorentzian component. This scattering is mainly a function of , the outgoing angle from the gratings in the dispersion direction, but has also a component in , the cross-dispersion angle. For each order, is related to wavelength through the diffraction equation: m = d (cos -cos ) where is the incident angle on the gratings and d is the line spacing. The LSF components in the dispersion direction are illustrated in Fig. 6 where small-angle scattering causes the additional broadening at a few percent of the peak intensity and large-angle scattering dominates at larger


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

12

Figure 7: The O VIII Lyman line of the active star AB Dor in RGS2 first order spectra, showing stacked data in black and narrow-line model in blue from 34 observations spread through the mission. The observed width is of instrumental origin. The vertical dotted line shows the laboratory wavelength. C shows the C-statistic of the fit of the stacked model to the stacked data with the number of degrees-of-freedom in brackets. distances from the core. Finally, the wings in the line-spread function were reduced by applying an energy selection in the CCD response. The effect of this selection is asymmetric as can be seen in Fig. 6. The LSF was compared early in the mission with lines observed in a number of stellar coronnae with agreement between model and data after modification of the instrument model to include a scaling factor applied to the pre-flight grating misalignment distribution of RGS1. This forms the basis for the RGS response matrices calculated by the SAS and for the comparison of observed and predicted line profiles. Changes made in 2011 to the LSF models improved the description of the sharp core seen in strong lines [16]. Fig. 7 shows an example fit to a coronal O VIII 18.967,18.973 ° line. A Although several components of differing angular scales contribute to the complex shape of the LSF, the FWHM gives a rough measure of the instrument's performance. The empirically-determined width of strong, relatively isolated emission lines observed in coronal spectra is a slowly-varying function of wavelength in both RGS1 and RGS2 with mean FWHM of about 70 m° in first order A and 50 m° in second order giving a spectral resolution that increases with wavelength with typical A values of a few hundred (Fig. 9). It is estimated that an observed line broadening of more 10% of the FWHM can be considered to be significant for strong lines. To give a concrete example, Fig. 8 shows data around a coronal NeIX He-like triplet, showing the ability of the RGS to resolve these lines. At longer wavelengths, the He-like triplets of O VII and N VI are comfortably resolved. At shorter wavelengths, Mg XI and Si XIII are more difficult to separate.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

13

Figure 8: Small parts of the RGS spectra errors from RGS2 first order in black and in red and green, respectively. The solid continuum. The excesses near 13.5 ° and A model.

of AB Dor near the Ne IX fir triplet showing data and the higher-resolution second orders of RGS1 and RGS2 lines show a simple model of a narrow-lined triplet and 13.8 ° are due to Fe XIX, which was not included in the A

600

400 Resolution 200
RGS1, RGS2, RGS1, RGS2, m=-1 m=-1 m=-2 m=-2

0 5

10

15

20 25 Wavelength (å)

30

35

Figure 9: Comparison of measured (points) and modelled (lines) RGS1 and RGS2 spectral resolution for first and second order. The data are measurements of the width of a number of strong and relatively isolated emission lines of stellar coronnae [1].


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

14

2.3

Wavelength scale

Figure 10: Shift between observed and laboratory wavelengths of a variety of narrow lines from measurements throughout the mission of several stellar coronnae in first and second order for RGS1 and RGS2. The wavelength scale is determined by the geometry of the various instrument components. While the relative positions of the nine individual CCDs in each RGS instrument are known to high accuracy from measurements made during instrument integration, the overall instrument orientation and position of the camera reference points were sub ject to verification in flight using measurements of strong emission lines [17]. Subsequent work using many more lines in more observations has confirmed these measurements, allowed an accurate assessment of the misalignment between RGS1 and RGS2 [18, 19] and enabled a correction to the wavelength scale dependent on the angle between Sun and spacecraft [20]. After these adjustments, the wavelength scale shows rms residuals of ±6 m° A ° for first and second order, respectively, as shown in Fig. 10. and ±5 mA As a rule of thumb, every 1 arcsecond error in source coordinates causes a systematic error of 2.3 m° A in the wavelength scale, emphasising the importance of using as accurate positions as possible in data analysis.

2.4

Cross-disp ersion distribution

The distribution in the orthogonal cross-dispersion direction is somewhat different from the optimum mirror PSF as the detector follows the Rowland circle in the dispersion direction, which is not ideal for the mirror response, so an empirical determination is made instead. The cross-dispersion distribution of the smooth continuum of the X-ray bright blazar Mkn 421 has been parametrised [21] as a function of the dispersion angle . This distribution is important as it provides the basis for the SAS to determine selection regions during calculation of point-source spectra.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

15

2.5

RGS temp oral resolution
by the CCD readout procedure. s, although this sometimes can thus the integration time of an temporal resolution.

The instrument time resolution in spectroscopy mode is determined In double-node readout for RGS one CCD is read out every 0.5741 reach 0.61 s; the full set of 8 CCDs then takes about 4.8 s, which is individual CCD and the figure that best describes the instrument's

In August 2007, the way in which RGS2 is operated was changed. For several years, RGS2 had been sub ject to occasional electronics problems with a current limiter during activation at the beginning of some orbits that caused a delay in switching the instrument to its normal spectroscopy mode. Because of the increasing frequency of this condition the decision was made to change the CCD read-out method in RGS2 from double-node, in which data from the two halves of the chips are retrieved separately, to single-node, in which data from the whole chip are read out in through a single amplifier. As the single-node readout takes twice as long, RGS2 frame times are correspondingly twice as long as those from RGS1, giving a CCD integration time of 9.6 s. The consequences for pile-up of bright sources is discussed below in section 3.2.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

16

3
3.1

Calibration-related asp ects of RGS data analysis
RGS Background

Figure 11: Count rate for the data selections {first order, 90% in CCD pulse height, all crossdispersion angles} for a quiet part of an observation of the Lockman Hole In the instrumental background various components have been identified due to: · minimum ionising particles · low-energy electrons · fluorescence lines from the housing · soft protons entering through the mirrors · calibration sources · read-out noise The net effect of this background can be illustrated using data from the Lockman Hole field which is blank for the RGS. The data shown in Fig. 11 were integrated over the ful l cross-dispersion direction of the detector for periods of a low soft-proton background. In general for point sources, the background can be reliably estimated from parts of the detector in the cross-dispersion direction outside the source-selection region, taking care to exclude the on-board calibration sources which illuminate parts of the CCDs. The locations of these calibration sources were chosen to interfere minimally with astronomical data. For extended sources, which can cover an arbitrarily large fraction of the detector's 5 arcmin cross-dispersion width, background region may be selected by inspection or independently by using blank-field RGS background template files [22, 23]. It is important to point out that the background is sometimes highly variable. It tends to be strong near the beginning and end of the spacecraft orbit due to the radiation belts but also to suffer at other times from a highly variable and even overwhelming contribution from soft protons due to


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

17

solar events. It is recommended to use events in CCD9, which is closest to the optical axis, as a monitor to select for processing Good Time Intervals of low particle background. Readers are referred to [24], which discusses the evolution of the RGS background along the mission, and is regularly updated.

3.2

Pile-up

Figure 12: Short wavelength part of spectra of Capella taken after the introduction of RGS2 singlenode readout in August 2007. First order spectra of RGS1 and RGS2 are in black and the corresponding second orders in red. The second order of RGS2 has clearly been contaminated by pile-up of the strong lines of Fe XVII 15.015,17.053,17.096 ° from the first order spectrum, that show up A as spurious features in the second order spectrum at exactly half the wavelength (marked with the arrows). The effect, though present, is much less severe in RGS1 due to the shorter readout time. Pile-up is a familiar effect when observing bright X-ray sources with the EPIC instruments, which are non-dispersive and concentrate events from a point source within the small area of the focal plane covered by the PSF. Pile-up occurs when two or more photons are detected in the same pixel during the same integration, in which case they become combined into a multiple event. This can also happen in the RGS for very bright point sources when, for example, two first-order events arrive in the same pixel close enough in time to be combined into a single apparent second-order photon. The introduction of RGS2 single-node operations discussed above in 2.5 with its longer read-out time has worsened the effect in RGS2. As each RGS pixel suffers a level of pile-up dependent on its individual illumination, it is wise to consider separately line-rich and smooth-continuum sources. Fig. 12 shows the appearance of the strongest lines in the spectrum of Capella in the second order of RGS2 at half their wavelength. Capella is one of the brightest X-ray sources in the sky with a high line-to-continuum ratio so this is more a worst case than a typical example. In the context of the new RGS small-window mode, specifically designed for bright sources, detailed comparison of first and second order spectra of Capella showed that pile-up losses amount to a few percent in the brightest lines [25].


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

18

For smooth-continuum sources, the pile-up fraction reaches about 2% for CCD count rates cts s-1 in RGS1 and 6 cts s-1 in RGS2. In the rare cases when sources are bright enough significant level of pile-up to have occurred, it is not an easy problem to treat, although in the cases of which we are aware in the brightest X-ray binaries the pile-up fraction reached no than 8%.

of 12 for a worse more

3.3

CCD detector defects

Removal of hot pixels and hot columns is a vital part of data analysis, as even a cursory glance at an unfiltered RGS event list will show. While this routine part of the SAS data analysis is usually quite successful, warm or cool - as opposed to hot - pixels that flicker occasionally are sometimes more difficult to detect Especially for low-intensity sources this may result in distortions of the spectra that, nevertheless, can easily be identified. They are much narrower than the LSF and do not occur in both spectrometers at the same wavelength. In order to enhance the ability to distinguish sometimes subtle absorption or emission spectral features in cosmic sources from detector defects, the RGS offers the Multi-Pointing Mode [3] in which a series of spacecraft repointings are executed automatically in order to move detector blemishes to different parts of the spectrum.

3.4

Pixel offset values

In common with other CCD detectors, offset values are needed to calculate the energy of each detected RGS event. Early in the mission, a single offset value was used for each CCD node. Since SASv6 in 2004, an improved method uses dynamic values calculated for each individual pixel from diagnostic data averaged over three XMM-Newton revolutions [26]. These data are supplied with the ODF and used in RGS data reduction by the SAS with the default switch withdiagoffset=yes.

3.5

Fixed-pattern noise

At long wavelengths high-frequency noise can be visible for data with good statistics due to socalled fixed pattern noise in the detector that gives an unmistakable spatial modulation This can be reduced by either applying somewhat different event reconstruction thresholds or by an improved subtraction of the detector spatial background. Since the introduction of single-node operations, fixed pattern noise in RGS2 has decreased.


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

19

References
[1] Calibration and in-orbit Performance of the Reflection Grating Spectrometer on board XMMNewton, A&A, 573,128, de Vries, C. et al., 2015. [2] The Reflection Grating Spectrometer on board XMM-Newton, A&A, 365,L7, den Herder, J.W. et al., 2001. [3] XMM-Newton User's Handbook [4] http://xmm2.esac.esa.int/external/xmm sw cal/calib/ccf.shtml [5] RGS blazar spectra and prospects for an Effective Area correction, XMM-SOC-TN-0063, A. Pollock, 2004 [6] Effective area calibration of the Reflection Grating Spectrometers of XMM-Newton. I. X-ray spectroscopy of the Crab nebula, A&A, 497, 291, Kaastra, J., de Vries, C., Costantini, E.and den Herder, J. W., 2009 [7] An improved model the RGS effective area based on the build-up of carbon contamination, XMM-CCF-REL-238, A.Pollock, 2007 [8] The RGS effective area incorporating exponential contamination and a mechanism for rectification, XMM-CCF-REL-262, A. Pollock, 2010 [9] Correcting RGS Quantum Efficiency calibration defects, XMM-CCF-REL-108, C. Gabriel, 2002 [10] RGS2 CCD2 Thickness of SiO2 Layer, XMM-CCF-REL-71, C. Erd, 2001 [11] The interstellar Oxygen-K absorption edge as observed by XMM-Newton. Separation of instrumental and interstellar components, A&A, 404, 959, de Vries, C.,den Herder, J.W., Kaastra, J., Paerels, F., den Boggende, A., Rasmussen, A., 2003 [12] Tuning RGS Instrumental MgF2 Absorption with Mkn421, XMM-CCF-REL-212, A. Pollock, 2006 [13] RGS individual CCD sensitivities XMM-CCF-REL-215, A. Pollock, 2006 and 2nd and 3rd order grating efficiencies,

[14] Status of the XMM-Newton Instrument Cross-Calibration, XMM-SOC-TN-0052, M. Stuhlinger et al., 2010 [15] XRT PSF Parameterization for RGS, XMM-CCF-REL-63, C. Erd, 2001 [16] Modification of the RGS line-spread function, XMM-CCF-REL-275, A. Pollock, 2012 [17] The RGS wavelength scale, XMM-SOC-TN-0041, R. Lorente, A. Pollock and C. Gabriel, 2003 [18] The RGS Wavelength Scale, XMM-SOC-TN-0079, D. Coia and A. Pollock, 2007 [19] Alignment of the RGS1 and RGS2 Wavelength Axes, XMM-SOC-TN-0080, D. Coia and A. Pollock, 2008 [20] Sun-angle correction to the RGS wavelength scale, XMM-CCF-REL-297, R. Gonz´lez-Riestra, a 2013 [21] RGS Cross-Dispersion PSF, XMM-CCF-REL-114, C. Gabriel, 2002


XMM-Newton Technical Note

XMM-SOC-CAL-TN-0030 issue 7.4

Page:

20

[22] RGS Background Spectra Templates, XMM-CCF-REL-229, R. Gonz´lez-Riestra, 2007 a [23] Templates for the RGS Background, XMM-SOC-TN-0058, R. Gonz´lez-Riestra, 2004 a a [24] The Behaviour of the XMM-Newton Background, XMM-GEN-TN-0014, Gonz´lez-Riestra, R. and Rodr´ uez-Pascual, P., 2015 ig [25] RGS small window mode, XMM-SOC-TN-0086, A. Pollock and M. D´ z-Trigo, 2010 ia [26] System offsets using diagnostic images, XMM-SOC-TN-0046, C. de Vries, 2003