Quantification
of XPS spectra is performed using the information prepared on the Regions
Property Page of the Quantification Parameters dialog window.
Regions can
be created using:

Figure 1: Regions Property Page
Integration
regions define the parameters used to identify the signal associated with a
peak in the spectrum. These parameters are tabulated on the Regions Property
Page in columns where each item in a column is either an editable input or a
read-only value determined from the data. Editable items, when selected with
the left mouse button, become text entry fields and may be altered using
keyboard entry. These include
|
Name |
The name used to identify the region in quantification tables. |
|
RSF |
Relative Sensitivity Factor for a transition. |
|
Start |
Start energy for the integration region. |
|
End |
End energy for the integration region. |
|
BG Type |
Background type used as a lower bound to the peak. |
|
Av Width |
Determines the number of points used to tie the background to the data. Actual number of points averaged = 2 * Av Width + 1. |
|
Start Offset |
Percentage offset applied to the start of the background after the initial tie point is determined from the Av Width. |
|
End Offset |
Percentage offset applied to the end of the background after the initial tie point is determined from the Av Width. |
|
Cross-section |
Parameters used to determine the background. Meaning depends on the chosen background type. |
|
Tag |
A string used to identify the region for computational purposes. |
Table 1: Editable items
available on the Regions Property Page.
Figure 2
shows the list of background types on offer in CasaXPS. The dialog window in
Figure 2 is invoked by holding the Control Key down at the same time as
selecting the BG Type field within a Region. The BG Type text-field is updated
from the selected background when the OK pushbutton is pressed. The region is
updated with the chosen background once the Return Key is pressed on the
Regions Property Page.
The
read-backs include:
The
reported area is determined from the background-subtracted data after
normalisation with respect to the total acquisition time (dwell-time times
number of sweeps), energy step size and is also adjusted for transmission
correction using the Intensity Calibrations fields marked in Figure 1. If the
VAMAS file contains transmission correction and the Automatic checkbox
is ticked, then the reported area includes the transmission adjustment. There
is also an energy dependent exponent text-field for compensating for
mean-free-path and analyser variations. The measured area is multiplied by the
kinetic energy of the ejected electron raised to the power of the value entered
in this text-field. The energy dependent adjustments are typically required
when Scofield cross-sections are used to compensate for the relative transition
probabilities for the measure photoelectric lines and, for many instruments, a
value of 0.5 would be typical (however this value is not necessarily
universally applicable).
The
standard deviation in the Area parameter read-back is zero unless the Calculate
Error Bars pushbutton (Figure 1) is pressed following any adjustments to
the regions parameters. The standard deviation in the area is computed for each
region defined on a spectrum using a
The

Figure 2: Background Types offered on the
Regions Property Page. To invoke this dialog window, hold the Control Key down
when selecting the BG Type field.
CasaXPS
2.1.8 offers backgrounds based upon the existing CasaXPS background types
Shirley, Linear, Tougaard and (so called) None, but
where the curve is decided by a set of cubic spline
polynomials rather than the underlying functional forms. These new background
types Spline Linear (abbreviation sl), Spline Shirley (ss), Spline Tougaard (st) and Spline None (sn) are flexible backgrounds and can be adjusted under
mouse control or optimised as part of a synthetic peak model.
A spline is a mechanical device used by draftsman to draw
aesthetically pleasing curves using pen and paper: the draftsman fixes a set of
points (knots) on the drawing, then bends a flexible strip of plastic or wood
(the spline) around the knots and traces the shape
onto the paper. Spline interpolation is the
mathematical equivalent of this process and is achieved using a piece-wise
cubic polynomial approximation to replace the mechanical spline.
The smooth curve is achieved by requiring that adjacent cubic polynomials take
on the value of the knots and that the first derivatives of these polynomials
at the knots are equal. These conditions together with two end point
conditions, such as assuming the second derivative at the two extreme knots is
zero, allow a unique solution for the set of cubic polynomials for a given set
of knots.
CasaXPS
uses six knots to define five cubic polynomials cross a region, which determine
the background shape for that region. The knots are evenly spaced between the
start and end points of the region, while the intensity at a knot may be
adjusted under mouse control. This is achieved on the Components Property Page,
where selecting a point on the background and dragging the mouse to a new
position repositions the nearest knot and therefore adjusts the shape of the spline background. If the knot is an internal knot then the
spline is bent to a new shape, while if the knot
corresponds to the end-points, the start and end offset region-parameters are adjusted resulting in the background gradient altering.
A new
option on the Components Property Page allows the internal knots within a spline background to be optimised. Please note, the chi-square goodness-of-fit knows nothing about the
physical world and therefore optimising a background, while improving the
goodness-of-fit will probably have nothing to do with the true mechanisms
unless the model optimised is perfectly defined and there is no noise in the
data. Well-defined rigid models are the only models worth optimising.
Backgrounds are the least-well understood shapes in XPS and as such
optimisation is not recommended. The inclusion of the spline
background type in CasaXPS is intended for use where all else fails: for
example a peak positioned on a strong plasmon loss
feature is not open to a Shirley background approximation, however a Spline Linear with a few judicious modifications may allow
an analysis to proceed.
While
optimisation of peaks and background is not recommended, optimising a
background for a region without peaks does have a useful benefit. The spline background, for a well-chosen interval, will create
a smooth approximation to the signal and so provides a means of estimating the
noise in the data. The estimate for the noise will appear along with the
residual trace in the form of the standard deviation in the normalised
residual. If Poisson statistics are at work, then the value for the noise
should be close to unity. With knowledge of this statistic for a given piece of
data the target value for a peak-fit can be established.