Xray instrumentation and calibration mostly XMM'

1
Lecture 18

• X-ray instrumentation and calibration (mostly
  XMM).
• XMM users guide
• http//xmm.esac.esa.int/external/xmm_user_
• support/documentation/uhb/index.html

2
XMM-Newton

• 3 x-ray EPIC telescopes
• European Photon Imaging Cameras.
• 2 of these have
• MOS detectors
• Reflection Grating Arrays (RGAs)
• The 3rd has
• a pn detector
• No RGA.
• Mirrors all the same
• nested Wolter
• f 7 m.

Schematic of the satellite
3
Attitude

• The orientation of the spacecraft in the sky is
  called its attitude.
• It isnt just the direction it points to,
  attitude specifies the roll angle as well.
• One way to define this is via a pointing vector
  and a parallactic angle.
• However, as with any trigonometric system, this
  has problems at poles.
• Better is to define a Cartesian coordinate frame
  of 3 orthogonal vectors.
• Each vector defined by direction cosines in the
  sky frame.
• Attitude is then an attitude matrix A.
• This is isotropic (no trouble at poles).
• Easy to convert between different reference
  frames.
• Whatever you do, avoid messing about with Euler
  angles. Ugh.

4
Sky frame or basis.
z at dec90
Unless you have good reason not to, always
construct a Cartesian basis according to the
Right-Hand Rule
x?y screws toward z y?z screws toward x z?x
screws toward y.
y at RA6 hr
x at RA0
Direction cosines really just means Cartesian
coordinates.
5
Boresights

• Ideally, each instrument on a satellite would be
  perfectly aligned with the spacecraft coordinate
  axes.
• In real life, there is always some misalignment.
  This is called the boresight of the instrument (I
  think it is an old artillery term).
• Define a set of Cartesian axes for each
  instrument.
• Components of these axes in the spacecraft
  reference system (basis) form the boresight
  matrix for that instrument.

6
Boresights

• With these matrices, conversion between
  coordinate systems is easy. Suppose we have an
  x-ray detection which has a position vector v in
  the instrument frame. If we want to find where in
  the sky that x-ray came from, we first have to
  express this vector in the sky frame Cartesian
  system (new vector u). This is simple
• Then all we need to do is calculate
• QED.

7
Attitude and boresights

• NOTE that attitude varies with time as the
  spacecraft slews from target to target but
  there is ALSO attitude jitter within an
  observation.
• XMM has a star tracker to measure the attitude.
• Attitude samples are available at 10 second
  intervals.
• So to build up a sky picture from x-ray positions
  in the instrument frame, one has to change to a
  new attitude matrix whenever the deviation grows
  too large.
• Boresights can also change with time, due to
  flexion of the structure, but this is slow.
• Calibration teams measure them from time to time.

8
Other coordinate systems

• The fundamental spatial coord system is the chip
  coordinate system in CCD pixels.
• Note that for time and energy as well as in the
  spatial coordinates, coordinate values are
  ultimately pixellized or discrete.
• This defines the uncertainty with which they are
  known (to half the pixel width).
• Rebinning can give rise to Moiré effects
  (somewhat similar to aliasing in the Fourier
  world).
• Dithering can avoid this.

9
Moiré example
Original binned data bin widths are 2 units.
Re-binned data bin widths are 3.5 units. Moiré
effect causes a dip every 4th bin (since 4 is the
smallest integer n such that nx3.5 is
exactly divisible by 2).
10
EPIC telescope schematic(not to scale)
Reflection Grating Spectrometer
MOS
Optic axis
Reflection Grating Array
Mask
Mirror assemblies
Filter wheel
Optic axis
CCDs
pn
11
Mirror effects PSF

• No mirror system of finite aperture can produce a
  perfectly sharp image.
• Rather, each point source is smeared out
  (convolved) by a Point Spread Function (PSF).
• More usual, high F-number optics produce a PSF
  which is reasonably independent of off-axis
  angle
• This isnt true for x-ray grazing-incidence
  optics.
• For both XMM and Chandra, the PSF varies markedly
  with off-axis angle.

12
Mirror effects PSF
13
Mirror effects PSF

• XMM PSF is complicated.
• Asymmetrical core.
• Inner star.
• Outer wings with shadows from the mirror
  spider.
• RGA streak.
• The average radial profile is best described by a
  King function
• r0, a depend on energy.

14
Mirror effects vignetting.

• The mirror assemblies have a small acceptance
  angle transmitted flux drops by a factor of 2
  to 3 (its energy dependent!) from optic axis to
  outside of field of view (FOV).
• The ratio of transmittance at any position on the
  detector plane to that at the optic axis is
  called the vignetting function.

15
X-ray interaction with matter

• Can break it into continuum and resonant.
• Both sorts generate ions.
• Continuum absorption scales with
• Density
• 1/E.
• Resonant absorption
• electron is kicked out from an inner orbital.

X-ray
e-
Atom


M
L
K
16
Resonant absorption continued

• Because it is an inner orbital, doesnt much
  matter if atom is in a gas or a solid. The inner
  orbitals are pretty well insulated from the
  outside world.
• X-ray must have energy gt the amount needed to
  just ionize the electron.
• Hence absorption edges located at energies
  characteristic of that orbital (labelled eg K or
  L) and that element.

Absorption
X-ray energy
17
EPIC cameras

• MOS
• Front-illuminated means that the charge
  detection and movement electronics are on the
  illuminated surface (same as the retina).
• This means that
• pixels can be smaller (1.1)
• the MOS cameras are not very sensitive to soft
  x-rays (because these are absorbed in the
  electronics before reaching the detection
  substrate)
• theyre not very sensitive to hard x-rays either
  (because the substrate is too thin to absorb
  many).
• 7 chips (each 600x600 pixels square) in a
  hexagonal array, staggered in height to (very
  roughly) follow the curved focal surface.
• This causes slight shadowing of the edges of the
  central chip by the others.
• Readout time is 2 seconds (full window imaging
  mode).

18
EPIC cameras

• pn
• Back-illuminated the charge detection and
  movement electronics are on the rear. X-rays
  strike the detection substrate first.
• This means that
• pixels have to be larger (4.1)
• the pn camera is sensitive to x-rays over a much
  wider bandwidth than MOS.
• 9 chips, 200x64 rectangles, but all on the same
  rigid squarish block of silicon.
• Readout time (in normal imaging mode) is 0.07
  seconds (much faster than MOS).

19
X-rays to events.

• It isnt as simple as 1 CCD pixel per incident
  x-ray.
• Each x-ray creates a charge cloud of electrons,
  with a certain radius.
• The charge cloud can overlap more than 1 pixel.
• Thus patterns of excited pixels which correspond
  to a single x-ray have to be identified
• then all charge from that set of pixels must be
  added up ? total energy of the x-ray.
• Each recognized pattern is called an event.
• What XMM calls patterns, Chandra calls grades.

20
Example MOS patterns
21
X-rays to events.

• Complications
• X-rays are not the only things which can cause
  ionization in the chips can also have cosmic
  rays.
• However, these tend, on average, to produce
  elongated electron clouds.
• These patterns are easy to filter out.
• What cant be avoided however is a slight loss of
  detection capability where a cosmic ray has
  struck, an x-ray cant also be detected (for that
  frame). See later discussion of exposure.
• Dead or hot CCD pixels.
• Chip edges.

22
Out Of Time Events (OOTEs)

• As said last lecture, CCDs (at least in imaging
  mode) are operated in a cyclic fashion.
• Each cycle (called a frame) is composed of an
  integration interval followed by a readout
  interval.
• But! X-ray cameras dont have shutters! So even
  during the readout part of the frame, as the rows
  are being shunted towards the base of the CCD,
  x-rays are being absorbed.
• This results in a vertical smearing of all the
  x-rays absorbed during this time.

23
OOTEs continued

• The MOS chips use a more complicated readout
  strategy
• Each chip has in fact twice as many pixels as
  advertised.
• The extra pixels (which can be made much smaller,
  since they dont have to detect x-rays, just hold
  charge) are located behind an x-ray absorbing
  shield.
• The readout phase is divided into 2 parts
• a quick phase during which all the exposed rows
  are shunted into this frame store
• a slow phase during which the frame store is read
  out to the ADC.
• Result MOS have far fewer OOTEs.

24
OOTEs continued
Bright pn OOTEs
Faint MOS OOTEs
25
Pileup

• Earlier it was said that, in order to preserve
  the relation between charge size and x-ray
  energy, the frame time had to be short enough for
  the probability of 2 x-rays landing on the same
  pixel, same frame to be small.
• It does happen, however... and obviously the
  brighter the source, the more likely it is.
• The phenomenon is known as pileup.

26
Pileup

• Because of patterns, interaction between 2 events
  is difficult to calculate (but has been done
  however).
• Broadly speaking, 2 piled-up photons look like a
  single photon of the sum of their energies.
• This mucks up the spectrum of the source.
• Many piled-up events generate cosmic ray-like
  patterns and are thus discarded.
• MOS diagonal doubles are a good diagnostic.
• Heavy pileup leads to the event energy being
  greater than the accepted cutoff these events
  are then also discarded.
• The result is that holes are seen at the
  centres of very bright sources.

27
Other modes of operating the CCDs

• So far what has been described is full-window
  imaging mode.
• But there are at least 2 other modes
• Small-window imaging mode.
• If were prepared to sacrifice some imaging area,
  we can have a shorter frame time.
• A way to image very bright sources while avoiding
  pileup.
• See timing diagram next slide...
• Timing mode. In this mode, the CCD is read out
  continually ? much finer time resolution. This
  only works where the x-ray flux is dominated by a
  single bright source.
• The pn has an additional burst mode which can
  give time resolution down to 7 µs.

28
100x100 MOS small window mode example
Shift and discard rows 1 to 250 (quick)
Shift and read rows 251 to 350 (slow)
Shift and discard pixels 1 to 250 (quick)
Integrate
Shift to ADC pixels 251 to 350 (slow)
Shift and discard pixels 351 to 600 (quick)
Shift and discard rows 351 to 600 (quick)
29
Windowed imaging examples