Recently Julien Allaz at ETH Zurich discovered (and we fixed) a subtle bug in Probe for EPMA that occurred when several specific features were utilized together.
In software development it is one thing to implement a software feature and get it working accurately and reliably. And, it is fairly easy to test these new features with appropriate methods and data. However, it is an entirely different affair to test for all possible interactions *between* software features. The following is an example of this "feature interaction" testing difficulty.
Many of you already utilize the Time Dependent Intensity (TDI) correction in Probe for EPMA to correct for changes in x-ray intensity over time, usually due to ion migration (e.g., alkali glasses) or volatilization (from specimen beam damage). The TDI correction utilizes replicate intensity measurements that are time stamped to characterize these changes in intensity as described here:
https://probesoftware.com/smf/index.php?topic=11.msg234#msg234So whereas we normally simply record the total integrated intensity, with TDI we not only record the total intensity using sub time increments that are summed together to get our time integrated intensity, we also record the individual sub interval intensities, each with their own time stamp. This is so that the slope of the sub interval intensities over time can be applied to the total summed intensity to correct for an increase or decrease in the intensities over time. So far so good.
Next we have the Nth point off-peak acquisition method. With this acquisition method we acquire the on-peak and off-peak intensities as usual on our first acquisition point (and every Nth point from there on), but for subsequent points we only acquire the on-peak intensity and simply utilize the previously acquired off-peak intensities as described here:
https://probesoftware.com/smf/index.php?topic=71.msg7875#msg7875If anyone interested in seeing how the Nth point off-peak compares to normal off-peak and also to the MAN background correction, see this white paper from a few years ago:
https://epmalab.uoregon.edu/reports/Comparison%20of%20Off-peak,%20MAN%20and%20Nth%20Point%20Backgrounds.pdfThe bottom line on Nth point acquisition is that statistically it compares well to the MAN method, but it does not handle situations when the sample composition changes (due to changes in average Z and hence continuum production). However this is a situation the MAN background method handles easily. Again so far so good, the TDI method and the Nth point method are often utilized at the same time and they work well with each other.
Then we have the "On Peak Time Fraction" parameter option in the Nth point method which is where we can reduce the count time on the Nth point measurement so as to spend a proportionally larger amount of time characterizing the off-peak intensities before the specimen has been beam damaged as described here:
https://probesoftware.com/smf/index.php?topic=806.msg8036#msg8036Originally the allowable range of values for this On Peak Time Fraction was from 1.0 (the default) to 0.1 (10%). For those of you who happened to make use of the TDI and Nth Point methods with this range of On Peak Time Fraction values everything worked just fine and dandy.
But then, the guys at U Mass Amherst (Mike Jercinovic and Mike Williams), wanted a modification that allowed a *zero* On Peak Time Fraction value, so that the entire on-peak intensity measurement would be skipped altogether, thereby only measuring the off-peak intensities for each Nth point. The idea being to measure the background intensities before any sample damage had begun to occur. Also, they wanted the software to automatically specify each Nth point as being "disabled", so that these Nth points would not be included in any quantitative calculations, since there were no on-peak intensity measurements performed for them this makes sense.
Now we need to take a moment and discuss how the intensity data in Probe for EPMA is saved to the MDB database file. For intensities such as on and off-peak intensities, these values are always stored (even if they are not measured). For example the off-peak intensities when performing MAN background acquisitions are not measured but are stored as zero values. In such cases these zero intensities will be accessible in the database for queries, but not be very useful.
On the other hand, some intensities are only stored if they are acquired. For example, the TDI sub interval intensities are only acquired if they are actually acquired. These TDI database fields are simply not present if no on-peak measurement was acquired.
So the bug that Julien found (you were wondering when I would get around to that!) is related to all of the above. Because when the On Peak Time Fraction for the Nth Point method is greater than zero, every data line will have these TDI intensities stored, and we have no problems. But if the On Peak Time Fraction is zero, then the on peak acquisition is skipped and therefore no TDI intensities are acquired or stored for every Nth noint.
So if you utilize both TDI and Nth point acquisitions *and* you specify a *zero* ON Peak Time, this latest bug fix will take care of it. As always, please update from the Help menu in Probe for EPMA.
Please let me know if you have any questions at all.
To protect your privacy, you must first log in before you can view forum statistics and user profiles
