XAS
Tutorials
XAS
Data Pre-Processing
Once you have successfully
installed EXAFSPAK
(and an X-Windows server),
you are ready to start processing raw XAS data files collected at SSRL.
Within the Scott group over the years, many students and postdocs have
processed data and we have archived not only the raw data but the processed
data as well. For their work to be of use to you (and for your work to
be of use to future students and postdocs), we have developed a standardized
directory hierarchy within which both raw and processed data files reside.
This is described in detail in the XAS Data Handling protocol (pdf
download) and you should familiarize yourself with this procedure. In
summary, copies of raw data files are brought to your working directory,
pre-processing and reduction are carried out there, then copies of the
resulting processed files are stored in a defined location. Details of
this processing are then logged in our data index and data log forms.
The input raw data files are
obtained from XAS-COLLECT,
the current data collection software in use at SSRL. These files are binary,
but contain single scans of XAS data separated into arrays for monochromator
position and individual detector outputs (details below). They have filenames
like FX12B_044.001, in which FX12B
is a user-assigned filename that collects all the scans of a given run,
044 is the run number that is assigned uniquely
by XAS-COLLECT, and 001 is the scan number
for this run. In this and subsequent protocols, a generic raw data file
will be specified as file_rrr.sss.
Since the raw data files are
binary, they cannot be viewed without software designed to read this particular
format. EXAFSPAK is a suite of programs designed to perform these functions
and the data reduction and analysis. More information is available online
(http://www-ssrl.slac.stanford.edu/~george/exafspak/exafs.htm).
MLIST
C:\dir>mlist F*.*
C:\dir>mlist FX12B_044.001 /array
C:\dir>mlist F*.* /find="FdxB"
/array specifies
that output should be both header information and a summary of array definitions
/find="string" searches
the files specified for any whose header contains "string" (case-sensitive)
[Throughout this and subsequent
tutorials, text in this Sans Serif font (and indented) will be comments,
while text in Serif font
will be text seen in your Command Window. If the text is black,
the computer put it there; if it is red,
you put it there. C:\dir>
represents a generic prompt. In Command Windows, it is assumed that
each command is followed by pressing the Enter/Return key.]
This program provides a way
to extract information about a given raw data file or files. Minimally
this is the header information that was created at data collection time.
Another useful set of information is the structure of the arrays of
data in the file. There is also a search function that allows the user
to find a data file with a given text string in the header.
Use of mlist
file_rrr.sss with no
qualifiers provides just the header of that file (or files if wildcards
are used). Use mlist *
to list the headers of all raw data files in the current directory (this
works like a dir
command but only looks at XAS raw data files). The following screenshot
shows a result with no qualifiers as well as the result of using the
/array
qualifier.
MVIEW
C:\dir>mview /num=10-39 /den=4 /dev=2 FSBRA_006.001
C:\dir>mview /num=10-39 FSBRA_006.001
/den defines
which channel to use as denominator (default /den=4)
/num defines
which channels to use as numerator. A range of channels will be displayed
one by one, as num/den.
/dev defines
what output channel is used: 2
for terminal display, 8 for
PS file, 10 for
X-window (and others). Try not specifying /dev
first; the default often works.
This program is used to check
visually each detector channel of each scan for glitches, anomalies,
noise, etc. A given data file contains all detector channels of a single
scan. In the above example, Array positions 4, 5, 6 contain I0, I1,
I2, while Array positions 10-39 contain 30 channels of "SCA1"
(fluorescence data) and Array positions 40-69 contain 30 channels of
"ICR" (incoming count rate) for the same 30 detector elements.
(The ICR data may or may not be present in the data file.)
As an example, all fluorescence
data channels for a given scan can be viewed on a single plot with the
following command: mview /num=10-39 fsbra_006.001.
[Successful execution of this command (and many others in EXAFSPAK)
will establish a new graphics window for display of the result. When
finished examining the result, hit the Enter/Return key to return to
the CLI terminal emulator (the Command Window). If this new graphics
window does not appear, make sure that you have EXAFSPAK installed correctly
to use either inherent terminal graphics (e.g., Regis on a VAX) or X-windows
and make sure that your X-windows server is operating for the latter
and that the System Variables are set properly as discussed in the EXAFSPAK
Installation protocol].
It may be hard to compare
30 channels at a time, in which case you can display any range of these
channels you like with the /num
qualifier. You also can use additional qualifiers /xmin,
/xmax,
/ymin,
/ymax
to specify an explicit range of data, rather than the extent of all
channels (the default). (It will be useful to leave these off the first
time to get a feeling for the appropriate values.)
If there is a question about
whether a given channel may contain an anomaly or noise, the ICR channel
associated with the same physical detector element is often more sensitive
to electrical noise, crystallite diffraction artifacts, etc. These can
also be examined, for example with mview
/num=40-69 fsbra_006.001:
With the default assignment
(SCA: 10-39; ICR: 40-69), it is useful to remember that for a given
physical detector element (el = 1-30), SCA = el + 9 and ICR = el + 39.
So the ICR for a given SCA is SCA + 30. Should you decide that a particular
SCA channel should be discarded, make a note of it for use later on
in the data processing. Be conservative in discarding data; you should
be absolutely convinced that a given channel will contribute more noise
than signal if included in the weighted average. For example, a channel
having an edge jump that appears smaller than the others is not a
priori a reason to discard that channel. Also, problems in the
ICR trace (e.g., above) should only be used to confirm suspicions about
an SCA trace since some of the problems observable in ICR ratio out
very well in the SCA/I0 plot.
Another useful trick is that
the /num
qualifier can have a complicated set of channels specified. For example,
in the above plot, you might want to convince yourself that the noisiest
channel is in fact channel 62. You can redo this plot leaving out only
that channel by using mview /num=40-61;63-69
fsbra_006.001. If you are using Cygwin/X, this command may not work since some Unix shells (e.g., sh, bash) trap semicolons. In that case, precede each semicolon with a backslash: mview /num=40-61\;63-69
fsbra_006.001
MCALIB
C:\dir>mcalib /elem=Fe /edge=K /peak=0.01 FSBRA_006.001
/elem
is the two-letter element symbol for the edge being calibrated (default
/elem=Fe)
/edge is
the edge type (default /edge=K)
/peak gives
a lower limit; used to ignore a derivative peak lower than a given fraction
of the largest peak (useful for noisy data)
This program calculates the
first derivative in an edge region of an internal reference spectrum
and searches for first-derivative peak positions as inflection points.
Usually, the first peak position in the first derivative of the edge
region (the first inflection point) will be used to calibrate the energy
scale for experimental data. MCALIB will
output a file named CALIBRATE.DAT to accumulate a list of data file
calibration positions. This file will be read by MAVE,
so this is the first point at which you need to consider the set of
individual scans you plan to include in your averaged file. You should
calibrate all the files you could imagine including in the average in
one session (creating one CALIBRATE.DAT file); when MAVE
is used, you can change your mind and exclude some. MCALIB
should be run repeatedly on each data file until all files for a given
sample (average) have been examined.
For example, mcalib
/el=fe fsbra_006.001
generates the following graphical window:
In this window, if necessary,
you can use left or right arrow keys to move the selected peak position
(the uniquely colored vertical marker); press the Return/Enter key to
confirm your choice (always pick the first inflection point). Then repeat
the process for all other data files in your list.
It is worth examining the
contents of the resulting CALIBRATE.DAT file. You can display it in
the CLI window with the type calibrate.dat
command or you can use the NotePad application from a CLI command: notepad
calibrate.dat, which
generates the following NotePad window:
In the ideal circumstance,
all the calibration energy values will be very similar (as in this example).
If this is the case, it implies that any differences across scans may
simply be arithmetic uncertainties in the peak finder; thus, a single
averaged energy should be used to perform the calibration (done in MAVE).
If, on the other hand, these numbers are not the same (e.g., a shift
in the calibration occurred in the middle of the series), then you may
choose to calibrate scans individually (see MAVE)
or you may choose to break the series of scans into two groups, that
are separately calibrated with the average energy of that group (essentially
two executions of MCALIB and MAVE).
Then a decision can be made down the line of whether to average these
two averaged files together.
MAVE
C:\dir>mave /out=fsbra.ave <input_file_spec>
/out specifies
an output filename (default = mave.ave)
<input_file_spec> is
a wildcard input file specification to denote a set of input files (if
it is not specified, this list will be taken from a CALIBRATE.DAT file;
note that this can be a single file)
This program is used to calculate
the weighted sum over (30) fluorescence channels (or ln(I0/I1) or any
other ratio you choose) and to average all data files into a single
averaged data file (now a new format, with extension .AVE by convention).
In the most common usage, it obtains the set of raw data files to use
as input from a CALIBRATE.DAT file and also calibrates the energy scale
using the values read from this file. The output file, *.AVE, is ready
for PROCESS (discussed in the XAS Data Reduction
protocol). Before running MAVE, confirm
that the most recent version of CALIBRATE.DAT in the default directory
lists the files you want. REMEMBER to fill in the data log sheet with
correct information about each average (see XAS Data Handling Protocol).
An example dialog is provided
here, using the CALIBRATE.DAT file created above:
C:\dir>mave /out=fsbra.ave
8 Files found :
1 FSBRA_006.001
2 FSBRA_006.002
3 FSBRA_006.003
4 FSBRA_006.004
5 FSBRA_006.005
6 FSBRA_006.006
7 FSBRA_006.007
8 FSBRA_006.008
Do you want to remove files from this list? [y/n] [N] :y
(in case you have decided
that one of the files should not be included in the average)
Enter filenumber to remove [1] :8
8 FSBRA_006.008
remove this file? [y/n] [Y] :
7 Files found :
1 FSBRA_006.001
2 FSBRA_006.002
3 FSBRA_006.003
4 FSBRA_006.004
5 FSBRA_006.005
6 FSBRA_006.006
7 FSBRA_006.007
Do you want to remove files from this list? [y/n] [Y] :n
Do you want to change file weights? [y/n] [N] :
(This would allow you to
change from unit weights for every file; very seldom used)
Enter array position for eV [3] :
Enter array position for rtc [1] :
Enter array position range for numerator [10, 39] :
(Be sure that this is the
correct range for your data file; use MLIST
as above)
Enter array position range for denominator [4, 4] :
Do you want to take logs? [y/n] [N] :
(You would use this if you
were analyzing transmission data, for which you want log(I0/I1), so
the response for numerator
would be 4,4 and for
denominator would be
5,5. You can also extract
the transmission data for the calibrant by choosing numerator
5,5 and denominator
6,6.)
Do you want to change array position weights? [y/n] [N] :
Do you want to perform deadtime corrections? [y/n] [Y] :n
Statistical weights :
Enter 1 to calculate, 2 to use weights from file, 3 for unit weights [2]
:1
[Proper weighting of fluorescence
channels adjusts for more or less edge (signal) in channels. See brief
discussion in SSRL EXAFSPAK Manual]
Calculate statistical weights :
Enter ev-low, ev-high [7070.00, 7160.00] :
(These are the energy positions
just before and just after the edge of interest, respectively)
Do you want to exclude specific array positions? [y/n] [N] :y
(Sometimes some channels
are bad and need to be excluded from average)
Enter file-number [0] :1
(You can choose a specific
file with a particular bad channel)
1 FSBRA_006.001
Enter array position to exclude [0] :13
(This is the number from
the array position range you have specified (10-39
here))
Enter file-number [0] :3
2 FSBRA_030.002
Enter array position to exclude [0] :16
Enter file-number [0] :
(0 ends the channel selection.
We have now excluded channel 13 of scan 1 and channel 16 of scan 3.
It is more common to exclude an array position from all scans if it
shows problems in any scan.)
Recalibration ...
Do you want to shift individual spectra? [y/n] [N] :
(This choice will average
energies and perform the same calibration for all scans; y
would result in separate calibration of each individual scan)
Enter apparent eV0, true eV0 [7111.06, 7111.30] :
(apparent eV0
is read from CALIBRATE.DAT; true ev0
is looked up as first inflection for this edge. A selected list of values
is available in the XAS
Element & Energy Table)
Enter d-space, steps per degree [1.92016, 20000.00] :
(These values were inserted
into the data file at data collection; they should be correct)
Enter Za, E0 [26.0000, 7130.00] :,7120
(We have chosen to use E0
values -10 eV compared to the lookup table in EXAFSPAK. A selected list
can be found in our XAS
Element & Energy Table. Notice the use of the initial comma
to leave the first default value unchanged)
File FSBRA_006.001
calculating statistical weights ....
File FSBRA_006.002
calculating statistical weights ....
File FSBRA_006.003
calculating statistical weights ....
File FSBRA_006.004
calculating statistical weights ....
File FSBRA_006.005
calculating statistical weights ....
File FSBRA_006.006
calculating statistical weights ....
File FSBRA_006.007
calculating statistical weights ....
File FSBRA_006.008
calculating statistical weights ....
Hit the Return/Enter key
to exit the MAVE window. You have now created
your first averaged data file. You will be able to see the result when
you use PROCESS, the first step of the XAS
Data Reduction tutorial.
scott@chem.uga.edu
Department of Chemistry
University of Georgia
Athens, GA 30602-2556
706 542-2240 | FAX: 706
542-2295 |
