News:

:) All Electron Probe Micro-Analysts are welcome to register and post!

Main Menu

Multiple background "sharing"

Started by Karsten Goemann, July 04, 2013, 07:19:02 PM

Previous topic - Next topic

John Donovan

#15
OK, I had to tweak a few things for the "shared" background method, but it looks good to go now.

Here are some screen shots attached below and this is a listing of the data after the sample was searched for shared bgds; note the lines highlighted in red:

Un   33 Tin metal
(Magnification (analytical) =  20000),        Beam Mode = Analog  Spot
(Magnification (default) =      400, Magnification (imaging) =    400)
Image Shift (X,Y):                                          .00,   .00
Pre Acquire String :                                        mag 100000
Post Acquire String :                                          mag 400
Number of Data Lines:   3             Number of 'Good' Data Lines:   3
First/Last Date-Time: 09/03/2014 03:20:36 PM to 09/03/2014 03:27:39 PM

Stage (or Beam Deflection) Coordinate Positions:
          X          Y          Z                 X          Y          Z
  170G  -11896.00   2673.000  -18.00000   171G  -11884.00   2673.000  -19.00000
  172G  -11872.00   2673.000  -18.00000

Sample Coordinates Referenced to Fiducial Set  13 F:\Userdata\StandardPOSData\johnson-metal_pos3.pos
1  -12502. 4057.00 .000000
2  -6053.0 -4729.0 6.00000
3  -20437. -4894.0 -10.000

On and Off Peak Positions:
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
ONPEAK 41034.0 47355.0 49789.0 61652.0 39303.9 27621.9 48115.4
OFFSET 114.098 131.043 134.418 146.051 .007813 118.668 -.01172
HIPEAK 42377.8 48062.6 50818.5 62497.5 43763.9 28953.7 48658.5
LOPEAK 39585.9 45607.2 49098.9 60640.7 34843.9 26290.1 47572.3
HI-OFF 1343.80 707.602 1029.47 845.535 4460.00 1331.80 543.102
LO-OFF -1448.1 -1747.8 -690.10 -1011.3 -4460.0 -1331.8 -543.10

Multi-Point Background Positions and Parameters:
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
MULHI: 42377.8 48062.6 50818.5 62497.5    ----    ----    ----
MULHI: 48062.6 50818.5 62497.5    ----    ----    ----    ----
MULHI: 45607.2 49098.9 60640.7    ----    ----    ----    ----
MULHI: 50818.5 62497.5    ----    ----    ----    ----    ----
MULHI: 49098.9 60640.7    ----    ----    ----    ----    ----
MULHI: 62497.5    ----    ----    ----    ----    ----    ----
MULHI: 60640.7    ----    ----    ----    ----    ----    ----

MHIOFF 1343.80 707.602 1029.47 845.535    ----    ----    ----
MHIOFF 7028.60 3463.47 12708.5    ----    ----    ----    ----
MHIOFF 4573.22 1743.90 10851.7    ----    ----    ----    ----
MHIOFF 9784.47 15142.5    ----    ----    ----    ----    ----
MHIOFF 8064.90 13285.7    ----    ----    ----    ----    ----
MHIOFF 21463.5    ----    ----    ----    ----    ----    ----
MHIOFF 19606.7    ----    ----    ----    ----    ----    ----

ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
MULLO: 39585.9 39585.9 39585.9 39585.9    ----    ----    ----
MULLO:    ---- 42377.8 42377.8 42377.8    ----    ----    ----
MULLO:    ---- 45607.2 45607.2 45607.2    ----    ----    ----
MULLO:    ----    ---- 48062.6 48062.6    ----    ----    ----
MULLO:    ----    ---- 49098.9 49098.9    ----    ----    ----
MULLO:    ----    ----    ---- 50818.5    ----    ----    ----
MULLO:    ----    ----    ---- 60640.7    ----    ----    ----

MLOOFF -1448.1 -7769.1 -10203. -22066.    ----    ----    ----
MLOOFF    ---- -4977.2 -7411.2 -19274.    ----    ----    ----
MLOOFF    ---- -1747.8 -4181.8 -16045.    ----    ----    ----
MLOOFF    ----    ---- -1726.4 -13589.    ----    ----    ----
MLOOFF    ----    ---- -690.10 -12553.    ----    ----    ----
MLOOFF    ----    ----    ---- -10834.    ----    ----    ----
MLOOFF    ----    ----    ---- -1011.3    ----    ----    ----

ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
MACQHI       7       5       3       1    ----    ----    ----
MACQLO       1       3       5       7    ----    ----    ----
MUITHI       7       5       3       1    ----    ----    ----
MUITLO       1       3       5       7    ----    ----    ----
MULFIT       2       2       2       2    ----    ----    ----

PHA Parameters:
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
DEAD:     5.13    5.13    5.13    5.13    5.83    5.49    4.78
BASE:      .32     .32     .32     .32     .40     .30     .50
WINDOW    4.00    4.00    4.00    4.00    4.00    4.00    4.00
MODE:       -1      -1      -1      -1      -1      -1      -1
GAIN:     255.    255.    255.    255.    240.    122.     95.
BIAS:    1995.   1995.   1995.   1995.   1470.   1450.   1930.

Last (Current) On and Off Peak Count Times:
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
BGD:       OFF     OFF     OFF     OFF     OFF     OFF     OFF
BGDS:     MULT    MULT    MULT    MULT     LIN     LIN     LIN
SPEC:        4       4       4       4       1       2       3
CRYST:     PET     PET     PET     PET     PC1     TAP     LIF
ORDER:       1       4       2       3       1       1       1
ONTIM:   10.00   10.00   10.00   10.00   10.00   10.00   10.00
HITIM:    5.00    5.00    5.00    5.00    5.00    5.00    5.00
LOTIM:    5.00    5.00    5.00    5.00    5.00    5.00    5.00

Miscellaneous Sample Acquisition/Calculation Parameters:
KILO:    15.00   15.00   15.00   15.00   15.00   15.00   15.00
ENERGY   3.444   2.984   2.839   2.293    .525   1.740   6.400
EDGE:    3.929   3.352   3.174   2.521    .532   1.839   7.112
Eo/Ec:    3.82    4.47    4.73    5.95   28.21    8.16    2.11
STDS:      550     547     546     542      12     162     162

Combined Analytical Condition Arrays:
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
TAKE:     40.0    40.0    40.0    40.0    40.0    40.0    40.0
KILO:     15.0    15.0    15.0    15.0    15.0    15.0    15.0
CURR:     30.0    60.0    30.0    60.0    30.0    30.0    30.0
SIZE:       .0      .0      .0      .0      .0      .0      .0

Faraday/Aperture Beam Currents:
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
  170G  30.097  60.048  30.097  60.048  30.097  30.097  30.097
  171G  30.095  60.195  30.095  60.195  30.095  30.095  30.095
  172G  30.103  60.246  30.103  60.246  30.103  30.103  30.103

AVER:   30.098  60.163  30.098  60.163  30.098  30.098  30.098
SDEV:     .004    .103    .004    .103    .004    .004    .004

On-Peak (off-peak corrected) or MAN On-Peak X-ray Counts (cps/1nA) (and Faraday Current):
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
BGD:       OFF     OFF     OFF     OFF     OFF     OFF     OFF
SPEC:        4       4       4       4       1       2       3
CRYST:     PET     PET     PET     PET     PC1     TAP     LIF
ORDER:       1       4       2       3       1       1       1
  170G  327.82    -.18    -.09     .04    -.20   -1.20    -.06
  171G  327.82    -.09    -.07     .03     .03   -1.40     .01
  172G  329.21    -.11    -.11     .00     .15   -1.46    -.02

AVER:   328.28    -.13    -.09     .02    -.01   -1.35    -.02
SDEV:      .80     .05     .02     .02     .18     .14     .04
1SIG:     1.02     .04     .05     .02     .08     .11     .06
SIGR:      .78    1.21     .35     .84    2.28    1.31     .64
SERR:      .46     .03     .01     .01     .10     .08     .02
%RSD:      .24  -38.01  -20.02   80.13-2260.58  -10.28 -158.99

Off-Peak (calculated) X-ray Counts (cps/1nA):
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
TYPE:    MULTI   MULTI   MULTI   MULTI  LINEAR  LINEAR  LINEAR
COEF1:  4.2314  4.2322  4.2314  4.2322    ----    ----    ----
COEF2:  -.0001  -.0001  -.0001  -.0001    ----    ----    ----
COEF3:   .0000   .0000   .0000   .0000    ----    ----    ----
  170G    1.95    1.10     .88     .30    1.95    4.71     .93
  171G    1.91    1.09     .88     .30    1.84    4.79     .97
  172G    1.92    1.10     .89     .32    1.72    4.70     .92

AVER:     1.93    1.10     .88     .31    1.84    4.73     .94
SDEV:      .02     .01     .01     .01     .12     .05     .03

Raw Hi-Peak X-ray Counts (cps/1nA):
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
  170G    1.86     .85     .70     .31    1.04    2.45     .98
  171G    1.83     .91     .71     .33    1.16    2.38     .96
  172G    2.02     .90     .74     .33     .93    2.33     .89

AVER:     1.90     .88     .72     .32    1.04    2.39     .94
SDEV:      .10     .03     .02     .01     .11     .06     .05
1SIG:      .11     .05     .07     .03     .08     .13     .08
SIGR:      .91     .62     .31     .39    1.36     .45     .61

Raw Lo-Peak X-ray Counts (cps/1nA):
ELEM:    sn la   ag la   pd la   mo la    o ka   si ka   fe ka
  170G    2.67    1.15     .97     .33    2.87    6.97     .87
  171G    2.72    1.09     .87     .33    2.52    7.19     .99
  172G    2.49    1.14     .86     .36    2.51    7.06     .96

AVER:     2.63    1.13     .90     .34    2.63    7.07     .94
SDEV:      .12     .03     .06     .01     .20     .11     .06
1SIG:      .13     .06     .08     .03     .13     .22     .08
SIGR:      .90     .50     .80     .44    1.53     .53     .78
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#16
If anyone would like to "play" with a demonstration file with multi-point "shared" backgrounds, attached below is a PFE MDB file containing a number of standards with K and Cl already scanned for "shared" bgds.

When you open the MDB file in PFE, the app will warn that some standards are different. Usually it will recognize that you are using a different standard composition database than mine and offer to create temporary standard database. CLick Yes for this, and you'll be able to open the run and play around in it.

If necessary I've also attached my standard database, but you shouldn't need it as the standard compositions are now stored in the PFE MDB file as well.

Check out the improved MPB dialog from the Run menu as seen here:



Remember, you have to login to see and download attachments!

Warning: it would be best to click "no" after opening this file in Probe for EPMA (when asked whether to connect to the instrument hardware), as the default PFE install is for JEOL instruments, and this amphibole-demo2.mdb file was acquired using a Cameca instrument configuration.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Philipp Poeml

Dear John,

I use this feature and I like it a lot! Could I ask for some improvements?

* If I press "search for shared backgrounds" it finds almost all elements on the same spectro, but I do not want to share backgrounds for everything. Could it be possible to somehow select for wich elements it should detect the shared backgrounds?

* Let's say I have 5 actinides on one spectro. And I want to measure 3 hi and 3 low background points and iterate 3 points. However, for 5 elements I have 5 hi and 5 low backgrounds. Could it be possible to select before searching the shared backgrounds that I want to iterate 3 high and 3 low? That would save me some clicking later, when selecting all the parameters for each elements...

Cheers
Philipp

Probeman

Quote from: Philipp Poeml on April 20, 2015, 07:17:56 AM
Dear John,

I use this feature and I like it a lot! Could I ask for some improvements?

* If I press "search for shared backgrounds" it finds almost all elements on the same spectro, but I do not want to share backgrounds for everything. Could it be possible to somehow select for wich elements it should detect the shared backgrounds?

* Let's say I have 5 actinides on one spectro. And I want to measure 3 hi and 3 low background points and iterate 3 points. However, for 5 elements I have 5 hi and 5 low backgrounds. Could it be possible to select before searching the shared backgrounds that I want to iterate 3 high and 3 low? That would save me some clicking later, when selecting all the parameters for each elements...

Cheers
Philipp

Hi Philipp,
This is a good idea and already requested by Karsten Goemann.   

You could, from the Run | Display Multi-Point background Intensities dialog, edit the "Never Use" and "Always use" check boxes to configure your multi-point backgrounds anyway you want, but you'll have to do it one sample at a time as seen here:



Therefore, I intend (as soon as I get a few days free), to add the "Never Use" and "Always use" checkboxes to the Elements/Cation dialog so one can set them for all selected samples.  This will be cool, I agree.
The only stupid question is the one not asked!

Philipp Poeml

Yes, it would be something like allowing multi select here:



The other two points would be also helpful, like a short dialog about what elements it should consider before detecting. Also the how many points you would like to consider. As seen from the screen shot that would be valid for all the actinides and could be set easily before starting the multipoint detection.

John Donovan

#20
Ok, I added the MPB "NeverUse/AlwaysUse" manual flags to the Elements/Cations dialog in PFE as seen here:



It seems to work great with my testing, but give it a try yourself. Ready to download now.

The advantage being of course that once you figure which MPB is problematic you can manually set them for selected samples from the Elements/Cations dialog.
john
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Karsten Goemann

Hi John,

Thanks for adding this, looks promising!

This might also make it easier to specify always include/exclude positions in sample setups with shared backgrounds, because it'll remember those settings in spite of the samples being acquired as off-peak? Or will it overwrite those settings when you hit the "Search for Shared Backgrounds" button? I'll have to try it out!

Cheers,
Karsten

John Donovan

#22
This modified dialog should allow one to edit "shared" bgds just as you do multi-point backgrounds (MPBs).  See here for the parameters that are now editable for selected samples from the new Elements/Cations dialog for this "normal" MPB acquired channel:



Here by the way, is how a "shared" bgd channel appears in the Elements/Cations dialog:



It was originally two separate off-peak channels that were "shared" using the Search/Remove "Shared" Bgds buttons in the Analyze! window as seen here:


John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#23
Karsten Goemann recently noticed something interesting regarding the use of "shared" backgrounds. In a run analyzing some rutile minerals, Karsten acquired Ti using EDS (a good idea since it's not a trace element), and the remaining minor and trace elements using WDS and normal off-peak backgrounds.  Without "shared" bgds feature turned on, we see the following example of (intensity data) output:

On-Peak (off-peak corrected) or EDS (bgd corrected) or MAN On-Peak X-ray Counts (cps/1nA) (and Faraday/Absorbed Currents):
ELEM:    Ti ka   Si ka   Ca ka    V kb   Cr ka   Mn ka   Fe ka   Zr la   Nb la   Sn la   Ta la    W ma   BEAM1   BEAM2
BGD:       EDS     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF
SPEC:        0       1       4       5       5       2       2       3       3       4       2       1
CRYST:     EDS     TAP    PETL    LIFL    LIFL    LiFL    LiFL    PETL    PETL    PETL    LiFL     TAP
ORDER:       1       2       1       1       2       3       2       1       2       2       1       1
  108G   71.48     .28     .77     .20     .81     .01    6.37     .33     .49     .05     .05    -.03 200.200 200.200

AVER:    71.48     .28     .77     .20     .81     .01    6.37     .33     .49     .05     .05    -.03 200.200 200.200
SDEV:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00    .000    .000

Note that a number of elements "share" the same spectrometer and crystal, therefore they are candidates for the "shared" background method which converts normal off-peak measurements into multi-point background (MPB) measurements using only software. This feature is activated in the Analyze! window using the following buttons as seen here:



However, after clicking the "Search For "Shared" Bgds" button, Karsten noticed that the software did not seem to find any elements with off-peak bgds that were capable of being "shared".  The following output still shows normal (unshared) off-peak backgrounds as seen here:

On-Peak (off-peak corrected) or EDS (bgd corrected) or MAN On-Peak X-ray Counts (cps/1nA) (and Faraday/Absorbed Currents):
ELEM:    Ti ka   Si ka   Ca ka    V kb   Cr ka   Mn ka   Fe ka   Zr la   Nb la   Sn la   Ta la    W ma   BEAM1   BEAM2
BGD:       EDS     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF
SPEC:        0       1       4       5       5       2       2       3       3       4       2       1
CRYST:     EDS     TAP    PETL    LIFL    LIFL    LiFL    LiFL    PETL    PETL    PETL    LiFL     TAP
ORDER:       1       2       1       1       2       3       2       1       2       2       1       1
  108G   71.48     .28     .77     .20     .81     .01    6.37     .33     .49     .05     .05    -.03 200.200 200.200

AVER:    71.48     .28     .77     .20     .81     .01    6.37     .33     .49     .05     .05    -.03 200.200 200.200
SDEV:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00    .000    .000
1SIG:      .05     .02     .03     .01     .01     .02     .04     .01     .01     .01     .03     .01
SIGR:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00
SERR:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00
%RSD:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00

Off-Peak (calculated) or EDS (bgd) X-ray Counts (cps/1nA):
ELEM:    Ti ka   Si ka   Ca ka    V kb   Cr ka   Mn ka   Fe ka   Zr la   Nb la   Sn la   Ta la    W ma
TYPE:     ----  LINEAR  LINEAR  LINEAR  LINEAR  LINEAR  LINEAR  LINEAR  LINEAR  LINEAR  LINEAR  LINEAR
  108G     .00    1.77    3.39     .72     .71     .96    1.41     .26     .33    2.60    3.41    1.84

When we looked further into it, we found the problem: the software had decided that the elements that would seemingly be candidates for "shared" bgds did not qualify because they had different PHA bias values!

Now we probably should discuss whether having different PHA bias values is a valid concern here, because we really don't know if sharing off-peak measurements for elements with the same spectrometer and crystal (and keV of course!) is all that needs to be checked for.  Karsten and I (and Julien Allaz and Mike Jercinovic) originally decided to err on the side of caution and make the default to also check for different PHA biases, but we could use some testing to investigate this question, and I imagine it would, at the very least, be a question of exactly how different the PHA bias values are... it would make a nice little research project if anyone is interested.

That said, there is a flag in the Analysis Option dialog to "turn off" checking for different PHA bias values for "shared" bgds as seen here:



So now that we've unchecked the flag to check for elements with different PHA values, we again run the search for "shared" backgrounds and low and behold, we now obtain the "shared" backgrounds as we originally expected:

On-Peak (off-peak corrected) or EDS (bgd corrected) or MAN On-Peak X-ray Counts (cps/1nA) (and Faraday/Absorbed Currents):
ELEM:    Ti ka   Si ka   Ca ka    V kb   Cr ka   Mn ka   Fe ka   Zr la   Nb la   Sn la   Ta la    W ma   BEAM1   BEAM2
BGD:       EDS     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF
SPEC:        0       1       4       5       5       2       2       3       3       4       2       1
CRYST:     EDS     TAP    PETL    LIFL    LIFL    LiFL    LiFL    PETL    PETL    PETL    LiFL     TAP
ORDER:       1       2       1       1       2       3       2       1       2       2       1       1
  108G   71.48     .33     .81     .20     .82     .01    6.40     .34     .49     .05     .06     .00 200.200 200.200

AVER:    71.48     .33     .81     .20     .82     .01    6.40     .34     .49     .05     .06     .00 200.200 200.200
SDEV:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00    .000    .000
1SIG:      .05     .02     .03     .01     .01     .02     .04     .01     .01     .01     .03     .01
SIGR:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00
SERR:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00
%RSD:      .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00     .00

Off-Peak (calculated) or EDS (bgd) X-ray Counts (cps/1nA):
ELEM:    Ti ka   Si ka   Ca ka    V kb   Cr ka   Mn ka   Fe ka   Zr la   Nb la   Sn la   Ta la    W ma
TYPE:     ----   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI
COEF1:    ----  3.3610  4.7098  4.8158  4.8158  4.5182  4.5182  3.4195  3.4195  4.7098  4.5182  3.3610
COEF2:    ----  -.0365  -.0325  -.0324  -.0324  -.0312  -.0312  -.0247  -.0247  -.0325  -.0312  -.0365
COEF3:    ----   .0000   .0000   .0000   .0000   .0000   .0000   .0000   .0000   .0000   .0000   .0000
  108G     .00    1.71    3.35     .71     .71     .96   

So all is good now, but just because we were both fooled by this extra PHA bias check, we decided to add a warning to the "shared" bgd search code, to let the user know if the off-peak bgd elements could be "shared", but for the fact that the PHA bias values are different!  Here is an example of that warning output to the log window:



Slightly cryptic maybe, but at least it's a reminder to jog our memory!  Anyway, now we can utilize these "shared" bgd intensities in the multi-point bgd (MPB) dialog for high accuracy trace element analysis as seen here:



These new "shared" bgd different PHA bias warnings are available in Probe for EPMA v. 12.1.6 and is ready to download now.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

Earlier this year I posted some measurements I did on two NIST SRM metal alloy standards looking at their micro homogeneity:

https://smf.probesoftware.com/index.php?topic=1067.msg7105#msg7105

Recently I realized that this run might be a nice candidate for showing off the "shared bgd" method (pioneered by Karsten Goemann), in which one takes normal off-peak measurements (when two or more elements use the same WDS crystal and spectrometer) and loads these off-peak background intensities from each element, into the multi-point-background (MPB) arrays for more accurate fitting of background models, particularly where the continuum is curved or interference by secondary peaks on the background positions may occur (a method pioneered by Jullien Allaz, Mike Jercinovic and Mike Williams). A full paper on both of these techniques is currently in review.

The elemental setup for this particular run is shown here:



And it can be seen that except for oxygen all the other elements could benefit from this shared bgd treatment.  That is, these elements were all acquired using normal off-peak bgds as seen here:

On-Peak (off-peak corrected) or EDS (bgd corrected) or MAN On-Peak X-ray Counts (cps/30nA) (and Faraday/Absorbed Currents):
ELEM:    ni ka   mo la   cr ka   fe ka    w ma   co ka   si ka   mn ka    v ka   cu ka    o ka   BEAM1   BEAM2
BGD:       OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF     OFF
SPEC:        5       4       5       3       2       3       4       5       2       3       1
CRYST:     LIF     PET     LIF    LLIF    LPET    LLIF     PET     LIF    LPET    LLIF     PC1
ORDER:       1       1       2       1       1       2       2       3       2       3       1
  521G  2717.2      .5     3.4 10597.0     -.7    14.5    23.4    20.6     6.3    35.1    34.5  49.883  49.883
  522G  2725.7      .3     4.7 10602.9     -.5    16.7    23.9    19.5     9.5    41.7    37.5  49.898  49.913
  523G  2730.6     1.1     3.4 10586.4      .0    18.7    23.1    19.6     6.4    42.7    40.6  49.928  49.913
etc.


But after utilizing the Search For "Shared" Bgds button seen here:



The MPB arrays are automatically loaded and assigned as MPB background fits as seen here:

Off-Peak (calculated) or EDS (bgd) X-ray Counts (cps/30nA):
ELEM:    ni ka   mo la   cr ka   fe ka    w ma   co ka   si ka   mn ka    v ka   cu ka    o ka
TYPE:    MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI   MULTI  LINEAR
COEF1:  7.8540  6.9217  7.8539-29.7384 -3.3149-29.7384  6.9218  7.8540 -3.3149-29.7384    ----
COEF2:  -.0002  -.0001  -.0002   .0006   .0000   .0006  -.0001  -.0002   .0000   .0006    ----
COEF3:   .0000   .0000   .0000   .0000   .0000   .0000   .0000   .0000   .0000   .0000    ----
  521G    25.6     4.6     8.3    61.3     7.5    75.1     3.3    11.6   143.9   105.8    77.4
  522G    27.7     4.7     8.0    60.3     7.0    73.7     3.2    11.5   142.2   103.3    76.0
  523G    27.5     3.9     8.6    59.3     6.7    73.0     3.1    12.1   143.4   103.6    76.9
etc.


So all well and good, but a problem then appeared.

This run had a lot of data points as these were long traverses in order to document the micro-homogeneity, and after loading the MPB arrays with these "shared" bgds , the program slowed way down when loading data from these samples.  Everything would work, but everything got slow as molasses!

We finally tracked the problem down to the fact that I (smack on forehead) had coded the saving of the MPB intensities inside the datarow and element loops, so the MPB intensities got written numerous times causing the database to rapidly grow in disk size.  This had not been noticeable in previous uses of the "shared" bgds feature but once you start getting hundreds of data points it starts to multiply.   :-[

Anyway, we think we've fixed it all, but Karsten Goemann, Owen Neill and Gareth Seward are testing the new code as we speak.  But in the meantime I can still "show off" the results. First here is a screenshot of the MPB display after the shared bgds are searched for and loaded:



Note that depending on the element location, there will be various numbers of MPB backgrounds on either side of the peak (shown by the vertical red line). And for reference here are the quant results, first without the shared backgrounds (normal off-peaks):



and here using the shared (MPB) bgds:



None of these elements are really trace level so the application of shared MPB fitting in this situation is not really necessary (though it does validate the new and improved code!). But when one starts looking at concentrations below a few hundred PPM, this becomes a big deal because what really matters for trace element quantification is the accuracy of the background measurement!  And having more than two background measurements comes in really handy for determining that.

Because, after all, one can fit a line to any two points and call it an interpolation...  but the question is: it is the *correct* line?

Anyway, if anyone else is interested in beta testing this improved shared bgd code, drop us an email and we will invite you to the PFE Dropbox where you can grab this latest Probewin.exe version.  Once we hear back from our beta testers we will release the full 12.5.0 version of Probe for EPMA. The good news is that the new code seems to speed things up considerably, especially in large probe databases with lots of analyses.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

BenjaminWade

Hi John
Just wanted to say I am looking forward to the bug fix being uploaded. I recently did some shared background work recently on some glass standards of which I had a few hundred points under each sample heading...and I think my beard grew another inch when I had to process it.

Cheers

John Donovan

#26
Quote from: BenjaminWade on October 28, 2018, 04:30:54 PM
Hi John
Just wanted to say I am looking forward to the bug fix being uploaded. I recently did some shared background work recently on some glass standards of which I had a few hundred points under each sample heading...and I think my beard grew another inch when I had to process it.

Cheers

Hi Ben,
I'll tell you a little secret: if you update using the third update method (download from probesoftware), you can get the update now.  You might need to start with the original file and run the "search for shared bgds" method for best results.  Or maybe just run the "remove shared bgds" first.  Both methods run faster now.

That's what we're testing.
john
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

BenjaminWade

Hi John
Great thanks for that. I have updated it and it does seem to run a lot faster now analysing the data

cheers

John Donovan

#28
Mike Jercinovic and Mike Williams (U Mass Amherst) recently contacted us with a feature request for the "shared" background method. As many of you know, the "shared" background method, which is a variation of the multi-point background method, was originally suggested by Karsten Goemann (U Tasmania). Let's quickly review these various background methods.

When setting up a quantitative analysis in Probe for EPMA, one can select three different methods for background calibration: traditional off-peak, mean atomic number (MAN), and multiple-point background acquisition methods.

Everyone is familiar with the traditional off-peak method where one acquires a background measurement on each side of the emission peak, and then interpolates the background intensity under the peak (or alternatively one can acquire both background measurements on the same side of the peak and then extrapolate to the background intensity under the peak).

The MAN background method, described in various topics throughout the forum, e.g., here:

https://smf.probesoftware.com/index.php?topic=4.0

https://smf.probesoftware.com/index.php?topic=307.0

has advantages in saving time and improving precision, by constructing an MAN calibration curve that can be applied to both standards and unknowns of diverse compositions, by making measurements at the *on-peak* position, versus average atomic number (Kramer's Law),  in standard materials that *do not* contain the element of interest. A recent variation on the MAN method, originally suggested by Ben Hanson (Corning Glass) and further suggestions by Phil Orlandini (U Texas, Austin), utilizes interpolated *off-peak* measurements on standards to construct an MAN calibration curve for unknowns, where only the on-peak intensities are measured.  Thus avoiding on-peak contamination effects (from impure standards), on the on-peak measurements on the standards, and also off-peak interferences on the unknowns. More on this can be found here:

https://smf.probesoftware.com/index.php?topic=987.0

Please note that these topics go back a long way, so you might want to skip to the end of the topics, to start with the most recently posted material. A nice summary of the original MAN method can be found in Donovan et al., (2016), Amer. Min.

The multi-point background method (originally suggested by Mike Jercinovic, Julien Allaz (ETH Zurich) and Mike Williams) is similar to the off-peak method, but it utilizes more than one background measurement on each side of the peak. The idea being to characterize the background with greater rigor, particularly in cases where extreme accuracy is required and/or the shape of the background is significantly curved.  Here is an example of a typical MPB acquisition for U mb by Mike Jercinovic:



The MPB method is especially useful for complex matrices (e.g., monazites) where the composition can change unexpectedly, thus causing unanticipated off-peak interferences. And because the MPB has multiple background measurements, it can quickly iterate to find the best background interpolation as seen here:



One can also specify these MPB adjustments manually using the so called "manual" over-ride flags if necessary in post processing.

The "shared" background method is a variation of the MPB method, where one acquires traditional off-peak backgrounds on more than one element, which "share" the same spectrometer and Bragg crystal as seen here:



and described in more detail here:

https://smf.probesoftware.com/index.php?topic=9.0

and then utilizes the MPB arrays to process these off-peak data *as though* they had been acquired using the MPB acquisition method. The curious aspect to the "shared" bgd method is that because the off-peak measurements are *shared* for all the elements using the same spectrometer and Bragg crystal, the background fits for all these shared elements are exactly the same!

What Mike Jercinovic and Mike Williams wanted us to do was to allow the user to assign the same background fit obtained for one shared element, to all the other shared elements on that same spectrometer and crystal.  Which is perfectly reasonable of course, but it turned out to be a bit of a hassle to code for a couple of reasons.

So when we first designed the MPB method we decided to limit the acquisition to 12 off-peak measurements on each side of the emission peak for a total of 24 off-peak measurements. No one would ever need to acquire more than 12 backgrounds on each side of an emission peak, right? 

Not a bad assumption when utilizing multi-point backgrounds of course. But when using the shared background method, the first and last element on that spectrometer/crystal can contain 2 * n -1 backgrounds on one side of the peak (where n is the number of elements). This means that one could have up to seven elements per spectrometer/crystal combination and all the background measurements would be available when the limit is 12 backgrounds. But Mike and Mike had a monazite run where they had *nine* elements on the same spectrometer and crystal, not seven!

https://www.youtube.com/watch?v=xwKaXcTPKjE

So that means the elements at the top and bottom of the spectrometer range could have up to 17 background on one side of the peak. Oh, god.  The problem wasn't so much just increasing the size of the MPB arrays, but finding enough space to fit all the additional manual over-ride flags in the GUI:



So we settled on adding 6 more backgrounds to each side of the peak for a total of 36 off-peak backgrounds per elements. Yes, that's right, thirty-six off-peak measurements.   :o

So, in practical terms what does this mean? Well looking at the above plot, we can adjust our "iterate to" parameters to get the MPB background fit we want, as seen here:



and then we utilize this handy new button seen here, to assign this fit to all the other "shared"background elements as seen here:



The other less visible difficulty was figuring out how to keep track of which backgrounds corresponded to each of the elements on that spectrometer/crystal, as the load order will be different for each element depending on all kinds of details in the acquisition. In the end, we figured out a way to base the background selections on the actual off-peak spectrometer positions, because, after all, they are "shared" backgrounds!

:)

Version 12.8.4 of Probe for EPMA is now available for downloading using the Help menu as usual.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Philipp Poeml

Quote from: John Donovan on February 20, 2020, 11:06:29 AM
But Mike and Mike had a monazite run where they had *nine* elements on the same spectrometer and crystal, not seven!


So we settled on adding 6 more backgrounds to each side of the peak for a total of 36 off-peak backgrounds per elements. Yes, that's right, thirty-six off-peak measurements.   :o


Great! Happens to us all the time. Was on my wishlist for long, but would not have dared to ask.