101
was ever accumulated for more than 20 seconds without
interrupting the analysis, checking the probe position,
and if contamination were noted, repositioning the probe
to an area that was not contaminated.
The data related variables are variables over which
there was little or no experimental control. These
effects are primarily related to various data reduction
schemes, i.e., background subtraction, peak deconvolution,
peak modeling, etc. These effects on quantitative
analysis are documented by Zalusec (1979) The background
fitting routine was the source of most difficulty in the
data reduction. In this routine, developed by Zalusec
(1978), the background is modeled to fit a 4th order
polynominal with three operator selectable regions of the
background to provide input data for the fitting routine.
Figure 4.12b is a spectrum of the gamma phase of RSR 209
solution heat treated, quenched and aged at 870 C for one
hour. A background fit in the low energy region of this
spectrum is very difficult due to overlapping Ni-L, Al-K,
W-M, and Mo-L (in order of increasing energy) peaks in
this region. The consequence of this overlap is that no
isolated background region exists in this part of the
spectrum for entry into the modeling routine.
It was generally not possible to fit the whole
spectrum successfully. A good fit in the low energy
region did not necessarily mean a good fit at higher