Makextalevc: determine optimal integration parameters
Usage
The makextalevc program can be called as follows:
makextalevc [shoe box file] [reflection number]
When the "makextalevc" program is started, it will present a dialog box
with all parameters for the integration process:
The variables are:
- Shoe file name
-
The name of the shoe box file we are currently examining. Using the
"select" button a file can be selected using a file name dialog.
- Reflection number
-
The number of the currently displayed reflection out of the shoe box
file. Normally about 1000 reflections will be stored in a shoe box
file. Using the "next reflection" button, the next reflection will be
studied. Using the "next problem" button, the program will run in
batch mode through all reflections in the shoe box file, until one is
encountered with an integration problem. If the "Show image" is
clicked, the ndisp program will be brought
up showing the central image from the shoe box, with a box marking the
current reflection.
- Peak quality for strong reflections
-
The "peak quality" is a property that is higher for better defined
peaks. If your goal is to improve the description of the experimental
parameters you can use the "next strong peak" button to browse through
good peaks in the shoe box file. The default selection criterion is
that the peak quality should be higher than 20; if you want only
stronger peaks, increase this number, if you want to allow weaker
peaks, decrease this number.
- Crystal form and dimensions
-
Select the shape (sphere or box) of the crystal, and give the
approximate dimensions. For a sphere the diameter is given, for a box
the three orthogonal box lengths. If a crystal description was made,
these numbers are filled in automatically when the "makextalevc"
program is run for the first time.
- Crystal orientation angles
-
For a box-shaped crystal, these are the orientation angles of the box
in space when the goniostat is in its "all-zero" position. If a
crystal description was made, these numbers are filled in
automatically when the "makextalevc" program is run for the first
time.
- Isotropic mosaicity
-
The real mosaicity of the crystal (not the same as what is called
"mosaicity" in HKL programs!).
When the "More options" button is activated, a third and fourth panel of
questions appears:
- Anisotropic mosaicity
-
Some crystals are more like a pack of mikado sticks, where there is a
large freedom of orientation in one direction, but less in the
others. It looks like the mosaics are all rotated a tiny bit around a
fixed vector.
If you have such a crystal, 4 numbers need to be filled in under
"anisotropic modaicity":
- The additional perpendicular mosaicity
- The rotation vector in terms of a*, b* and c*.
- Incommensurate vector
-
For incommensurately modulated structures, the 0th order approximation
to the structure uses the intensities of the sub-lattice reflections only.
The 1st order approximation uses the sum of the sub-lattice reflections plus
their satellites. For this approach, fill in the "q" vector, and the
total extra length of the integration in the direction of the "q" vector as
the number of orders.
- Bad background treshold
-
Criterion to determine when the background behind a reflection can not be
trusted.
- Minimum Zeta
-
The lowest value of zeta where reflections should still be
integrated.
- expansion of reflections
-
Minimal number of pixels and minimal number of frames that a
reflection will take. This is to take effects like point spread into
account.
- Position and size of focus
-
The setup of the X-ray focus.
- Minimal "observed" peak quality
-
Minimal value of the peak quality to make a peak be considered "observed"
- I/sigma limit for background test
-
Minimal value of I/sigma required to attempt a background quality assessment.
- I/sigma limit for contour shifts
-
Minimal value of I/sigma required to attempt peak shifting.
- Remove from background pixels
-
Number of standard deviations from the mean before background pixels
are rejected from the least squares hyperplane and not used for
background subtraction.
- Maximal contour shifts
-
The maximal number of contour shifts, and the maximal total shift vector
allowed in three dimensions.
- Split overlap fraction
-
Overlapping peaks are split if the overlapping fraction is smaller than
the "split overlap fraction".
- Sum overlap fraction
-
Overlapping peaks are summed if the overlapping fraction is larger than
the "split overlap fraction".
- Split peak quality ratio
-
For peaks with an overlap between the "split overlap fraction" and the
"sum overlap fraction" the intensity is split if the ratio of the peak
quality (approximately the ratio of the intensities) is greater than the
"split peak quality ratio".
All variables can be changed either by typing in a new value and
pressing the "Return" key, or by clicking on the left/right arrows to
change the value by fixed increments.
If a shoe box file is specified (either specified from the command
line, or by typing the name in the appropriate entry box), two more
windows will show up: a text window showing the integration information of
a single reflection, and a reflection visualisation window.
The visualisation window has a number of components that are worth
discussing.
At the top-right, there is a big window representing the detector. The
center of the detector is represented by a small grey cross. The
position of the primary beam is shown as a small green circle. The
hourglass shape that extends to the top and the bottom of the beam stop
circle encloses the (uninteresting) region of reflections with a zeta
value below the zeta limit (normally 20 degrees). Reflections that are
studied are drawn to scale into the detector area with integration
problems indicated in blue. The current reflection under study is
highlighted by a large black cross.
To the left of the detector view, there are three projections of the
shoe box. In it you can see reflection outlines. The colors of
the reflection contours indicate their function:
- brownish-red is the predicted reflection contour
- bright red is the final location of the reflection contour
- purple (two different shades for predicted and final location) indicate
reflection contours of reflections from an interfering lattice.
- orange (two different shades for predicted and final location)
indicate the reflection contours of neighbouring reflections from
the primary lattice. There may be very many of these contours, not
all of them really contain at least one pixel (i.e. center of
gravity of the pixel is inside all three projected contours).
Pixels are considered to be part of the reflection if their center is
inside the drawn reflection contour. Pixels are thus allowed to
fractionally "stick out" the drawn contours.
Below the three projections, the three projections are repeated, but now
all pixels that belong to a "foreign" reflection are not used, and thus the
image can look much cleaner.
In the bottom row, the rightmost image is a histogram of the
difference of all background pixels from the least squares
hyperplane. Background pixels that differ by more than a specified
number of standard deviations (normally 3) from the hyperplane are not
used in the calculation of the reflection background.
The rest of the bottom row are the constituent images of the shoe box.
Represented in red are the pixels belonging to the current reflection.
In purple, pixels that belong to a reflection of a different
lattice. In orange, a neighboring reflection. In green, any overlap of
the current reflection with interfering reflections. Represented in
blue are all background pixels that do not fit the background
hyperplane.
The text output window shows a detailed log of the integration process.
Most of the output can be considered debugging output: only in
very rare cases will one need this information. The one important
line is the last one before the prompt: it gives the integrated
intensity (Int), the standard deviation (SigPois), and a series
of 8 flags which can be unset (a "." is shown) or set to:
- 'x': reflection touches the left side of the shoebox
- 'X': reflection touches the right side of the shoebox
- 'y': reflection touches the bottom of the shoebox
- 'Y': reflection touches the top of the shoebox
- 'z': reflection touches the front side of the shoebox in the rotation
direction.
- 'Z': reflection touches the back side of the shoebox in the rotation
direction.
- 'B': reflection has a "bad background". This means that although the
reflection seems to have a reasonable intensity, in fact its integration
may be very inaccurate due to unspecified anomalies in the background
distribution (in this case the line marked "Intensity" a bit higher up
could give you a hint on what is happening).
- 'O': reflection is overloaded.
If the no intensity can be calculated at all, the last line of the
text window will show the reason for this failure in words. This could
e.g. be because there are unreliable pixels from the CCD inside the
main peak contour.
If the peak is overlapping with other peaks, the text window will provide
information on whether this is being resolved by "splitting" the common
part of the intensity over the two contributing reflections, or by
"summation" of the two peaks to one combined intensity.
Operation
Using the dialog window with all parameters, small adjustments can be
made to improve the fit of the predicted reflection form with the
observed reflections. On first invocation of the "makextalevc"
program, an existing face index description is used to get an initial
crystal form and orientation. If an accurate face index description
existed, this is sufficient, and the crystal form and orientation do
not need to be touched any more. The only parameter that needs
adjustment in such a case is the mosaicity. If no face index
description was present, the program will start with a standard
spherical crystal description. In such a case a number of reflections
will need to be examined to see whether this model is sufficiently
accurate.
Two very useful buttons are "next strong reflection", which will skip to
the next reflection that is clearly observable, and "next problem", which
will skip to the next reflection that can not be reliably integrated.
At the start, you should click "next strong reflection" a number of
times. Check whether all the pixels that seem to belong to the
reflection in the top row of three projections are actually part of
the integration. Also make sure that the prediction is not much too
big. You may encounter very strong reflections that show "borders of
blue pixels" in the bottom row of images; this is not a cause of concern
if there are not many of these. If many reflections show such borders,
you may need to change parameters such that the reflection contours are
wider (e.g. increase the mosaicity).
Once the fit is OK, you should click on "next problem" a number of
times. Verify that most reflections integrate without problems. Some
"bad background" reflections will show up. If many reflections (even
those not at the edge of the detector) have "x"/"X" or "y"/"Y" flags,
the shoe box size will need to be increased. If many reflections have
"z" flags, the shoe box depth might need to be increased (but some "z"
flags are allowed at the beginning and at the end of the scan; and you
might also see some at low zeta values (close to the rotation axis)).
To make sure that the integration will work as desired, it is a good
idea in general to look at reflections from other shoe box files too.
When you are happy, the program should be closed using the
"Save and Exit" button.
