Author Topic: Short guide to Multipoint Background  (Read 4979 times)

Julien

  • Professor
  • ****
  • Posts: 76
  • The truth is out there...
    • Geoloweb - J. Allaz personal website
Short guide to Multipoint Background
« on: December 21, 2013, 10:01:34 am »
Hi folks!

Here is a short guide to the use of the Multipoint background acquisition method. This feature allows us to set up to 12 points on each side of a peak, in order to accurately determine the background value and its curvature. This is a recommended method for trace element analysis in complicated background (i.e., when numerous background interferences are present).

Feel free to contact me for details about this method!

Enjoy,

Julien
« Last Edit: December 21, 2013, 11:07:23 am by John Donovan »

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 2632
  • Other duties as assigned...
    • Probe Software
Re: Short guide to Multipoint Background
« Reply #1 on: February 02, 2014, 10:28:22 pm »
Also see this thread for more details and discussion on the multi-point background method:

https://probesoftware.com/smf/index.php?topic=9.0
« Last Edit: September 23, 2019, 10:50:52 am by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Philipp Poeml

  • Professor
  • ****
  • Posts: 213
Multipoint Bgd Acquitistion Properties
« Reply #2 on: September 20, 2019, 02:57:59 am »
Hi John,

Could you explain a little bit the implications when I sent the background type to multi-point here in this window? What changes in the software and how it acquires and calculates things?

Thanks

« Last Edit: September 21, 2019, 08:48:18 am by John Donovan »

Probeman

  • Emeritus
  • *****
  • Posts: 2079
  • Never sleeps...
    • John Donovan
Re: Multipoint Bgd Acquitistion Properties
« Reply #3 on: September 20, 2019, 09:52:16 am »
Hi Philipp,
It changes the background type to multi-point backgrounds (MPB). The dialog you show is from the Acquisition Options dialog, but you'll need to go to the Elements/Cations dialog to specify specific parameters for the MPB. Here's a short guide Julien Allaz wrote up:

https://probesoftware.com/smf/index.php?topic=131.0

The reason these background options are here, is so the user can specify the background acquisition types differently for standards and unknowns. Normally (though not always!) one would like the unknowns to utilize MPBs, but not the standards.
« Last Edit: September 21, 2019, 08:48:36 am by John Donovan »
The only stupid question is the one not asked!

Philipp Poeml

  • Professor
  • ****
  • Posts: 213
Re: Multipoint Bgd Acquitistion Properties
« Reply #4 on: September 23, 2019, 08:14:51 am »
Hi John,

you wrote to me "It's just a subtle point about the difference between acquired MPB and shared MPBs." and I think this might be what I was actually wondering about. So I am curious what the explanation is.

My original message to you was:
Quote
Hi John,

well, then I guess I am missing something still... Maybe I am too stupid for the concept.

So, let's say I have 6 REEs on a Lif. That would make 12 off peak backgrounds. Now I want to use MPB, let's say 4 backgrounds to get a good fit.

Normally what I do is, I have those 6 elements marked as off peak. Then I acquire counts. I get 12 off peak measurements. Then I have PfE search for "shared" backgrounds. The software sets those elements to MPB then. Then I define for each element which background it should use for the MPB. The rest 12-4 = 8 are kind of discarded.

Ah, maybe now I have an idea what it does: So it just means that for subsequent samples measured it will set the backgrounds to off peak or MPB depending if it is an unknown or standard. Is that correct? Is this really a useful feature?

Will it also search automatically for shared backgrounds?

Actually I was thinking to use off peak for standards and MPB for unknowns. But something stopped me:
Especially for REE and ACT, those lines are really close together and overlapping. So my 3-4 MPB backgrounds are different from the off peak BG I would use for the standards.

I know I could solve this by using sample setups or so. But here it comes. Our measurements take very long and we 21 standards to measure for our sample. That also takes very long. So we kind of measure standards once per week and check peak PHA and some secondary standards on a daily basis. But when we do the peak PHA these are only automatically updated in the current unknown sample. Not in any saved sample setup we would use.

And to do this fresh every day, with changing between MPB and off peak and typing all these numbers also scares me off. So at the moment we use MPB for everything.

Cheers
Philipp

Probeman

  • Emeritus
  • *****
  • Posts: 2079
  • Never sleeps...
    • John Donovan
Re: Multipoint Bgd Acquitistion Properties
« Reply #5 on: September 23, 2019, 10:40:01 am »
Hi John,

you wrote to me "It's just a subtle point about the difference between acquired MPB and shared MPBs." and I think this might be what I was actually wondering about. So I am curious what the explanation is.

My original message to you was:
Quote
Hi John,

well, then I guess I am missing something still... Maybe I am too stupid for the concept.

So, let's say I have 6 REEs on a Lif. That would make 12 off peak backgrounds. Now I want to use MPB, let's say 4 backgrounds to get a good fit.

Normally what I do is, I have those 6 elements marked as off peak. Then I acquire counts. I get 12 off peak measurements. Then I have PfE search for "shared" backgrounds. The software sets those elements to MPB then. Then I define for each element which background it should use for the MPB. The rest 12-4 = 8 are kind of discarded.

Ah, maybe now I have an idea what it does: So it just means that for subsequent samples measured it will set the backgrounds to off peak or MPB depending if it is an unknown or standard. Is that correct? Is this really a useful feature?

Will it also search automatically for shared backgrounds?

Actually I was thinking to use off peak for standards and MPB for unknowns. But something stopped me:
Especially for REE and ACT, those lines are really close together and overlapping. So my 3-4 MPB backgrounds are different from the off peak BG I would use for the standards.

I know I could solve this by using sample setups or so. But here it comes. Our measurements take very long and we 21 standards to measure for our sample. That also takes very long. So we kind of measure standards once per week and check peak PHA and some secondary standards on a daily basis. But when we do the peak PHA these are only automatically updated in the current unknown sample. Not in any saved sample setup we would use.

And to do this fresh every day, with changing between MPB and off peak and typing all these numbers also scares me off. So at the moment we use MPB for everything.

Cheers
Philipp

Hi Philipp,
As you mentioned there actually are two types of multi-point-backgrounds (MPBs). The first MPB method was developed by Julien Allaz, Mike Jercinovic, Mike Williams and myself for improving trace element accuracy for U-Th-Pb in monazite and trace elements in general, for use where the background is complex and/or curved. In this method, implemented into the Probe for EPMA software, multiple background measurements are made on each side of the peak (in contrast to the traditional off-peak method in which a single background is measured on each side of the peak). This MPB method in Probe for EPMA is in addition to the traditional off-peak method and the MAN method, where the background is measured at the peak position, but in standards that do *not* contain the element of interest.

So whether the backgrounds are measured using off-peak, MAN or MPB methods, this is an acquisition parameter that is specified prior to data acquisition.

The other type of MPB, invented by Karsten Goemann is the so called "shared" MPB background method which was subsequently implemented in the Probe for EPMA software. This method utilizes the acquisition of traditional off-peak backgrounds, when more than one element is measured on a specific spectrometer and Bragg crystal combination. So in this MPB method, only off-peak acquired elements are utilized. Subsequently, in post processing, the user has the software search all off-peak elements to see if any of the measurements can utilize the MPB "shared" data processing method. If such off-peak measurements are found in the data file, the program automatically loads them into the multipoint background arrays for processing them *as though* they were acquired using the MPB method.

So to sum up, the acquired MPB method is an background acquisition method (hence is specified by the user prior to data acquisition), and therefore applies to the Acquisition Options dialog you showed earlier in which the user would specify the multipoint bgd method for acquisition (separately for standard or unknown samples).

I found that there is no topic specifically discussing to the MPB acquisition method so I merged this discussion into the topic of Julien's "short guide to the multipoint background method" topic.

While the "shared" MPB method is a post processing method, where traditional off-peak background measurements are acquired by specifiying (also) in the Acquisition Options dialog you showed earlier in which the user would specify the off-peak bgd method for acquisition (separately for standard or unknown samples).  A discussion of the MPB "shared" method is here:

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

I hope that is more clear, please let me know.

See Julien's multipoint background paper attached below.
« Last Edit: September 23, 2019, 10:54:53 am by John Donovan »
The only stupid question is the one not asked!

Philipp Poeml

  • Professor
  • ****
  • Posts: 213
Re: Short guide to Multipoint Background
« Reply #6 on: September 23, 2019, 01:33:10 pm »
Ok, I get it now.
So essentially we have
* a pre-acquisition MPB (Allaz)
* a post-acquisition / shared MPB (Goemann).

Good. Here it comes:
My sample is very complicated. Like a Monazite, just more complicated. On top we also have more actinides, lots of platin group elements, Zr, Y, Xe, Cs, and so on and so forth.
To save time I do the post-acquisition mode, so share the backgrounds for each spectrometer. Something like a 4 point exponential fit models the background curve pretty well for a certain range of a spectro / crystal.
Out of 24 elements 20 are trace, so their peaks are very small.
However on the standards, like Ru metal I have huge wide peaks and all the smaller lines (beta, gamma, ...) you could think of.

Now my question:
Would there be a way in PfE, to use
* normal off-peak (left right) for standards
* shared / post acquisition mpb for unknowns?

It means the standards would have different BG positions than the unknwons. I think it would be possible using sample setups, but this is a bit complicated, because we check peak positions and pha every day . This means that either:
* We would check all the peak PHAs and then update our saved sample setups manually, or
* we save new sample setups, but then we would need to "program" the off peaks manually for standards / unkowns.

So for our case (and probably we are the only ones as usual) it would be nice to use "normal" off peaks for standards and shared / post-acquisition mpb for unknowns. Any chance..... ? :)

Probeman

  • Emeritus
  • *****
  • Posts: 2079
  • Never sleeps...
    • John Donovan
Re: Short guide to Multipoint Background
« Reply #7 on: September 23, 2019, 01:56:34 pm »
Now my question:
Would there be a way in PfE, to use
* normal off-peak (left right) for standards
* shared / post acquisition mpb for unknowns?

This is easy. Just specify off-peak acquisition for everything, then in post processing from the Analyze! window, select all the unknowns and click the Search for Shared Bgds button.  The standards will still be using traditional off-peaks, but the unknowns will now be using "shared" MPBs.

It means the standards would have different BG positions than the unknwons. I think it would be possible using sample setups, but this is a bit complicated, because we check peak positions and pha every day . This means that either:
* We would check all the peak PHAs and then update our saved sample setups manually, or
* we save new sample setups, but then we would need to "program" the off peaks manually for standards / unkowns.

So for our case (and probably we are the only ones as usual) it would be nice to use "normal" off peaks for standards and shared / post-acquisition mpb for unknowns. Any chance..... ? :)

For different background positions between the unknown and standard samples you will have to use sample setups with the different background positions. But to update is easy. Just make a new unknown sample using the original unknown sample setup, update it, then add this updated sample as the sample setup for new unknowns.  Do the same for the standard.

I'm sure you have a good reason, but for the life of me I cannot imagine why one would need to update their peak positions and PHAs *every day*.   Is the instrument that unstable?   Why is it so unstable?  This is not good.
The only stupid question is the one not asked!

Philipp Poeml

  • Professor
  • ****
  • Posts: 213
Re: Short guide to Multipoint Background
« Reply #8 on: September 23, 2019, 02:30:36 pm »
For different background positions between the unknown and standard samples you will have to use sample setups with the different background positions. But to update is easy. Just make a new unknown sample using the original unknown sample setup, update it, then add this updated sample as the sample setup for new unknowns.  Do the same for the standard.

Yes, I am familiar to this method. But it involves lots up manual updating. And sometimes we make mistakes when typing...
This is then not a preferred for us.
I will then remain with my method where I find shared MPBs that fit to all unknowns and standards.

I'm sure you have a good reason, but for the life of me I cannot imagine why one would need to update their peak positions and PHAs *every day*.   Is the instrument that unstable?   Why is it so unstable?  This is not good.

No, it is just very small changes and just to verify the machine does not have problems. Verifying peak + PHA is done quickly.
Where I come from we normally measure our standards every day. For some garnet or so this is done in no time.
Now, we have 21 standards to measure. And we cannot use quick standards, because we are using the shared MPB... One point on a standard is thus 6 minutes, times 6 points.... Takes a long time.
I can think about the sample setups... Let's see.

Probeman

  • Emeritus
  • *****
  • Posts: 2079
  • Never sleeps...
    • John Donovan
Re: Short guide to Multipoint Background
« Reply #9 on: September 23, 2019, 02:35:12 pm »
No need to type anything. When you re-peak the spectrometer, the new peak position is already loaded for that sample, which then becomes the new sample setup using the Add to Setups button in Analyze!  Same applies to adjustments to the PHA settings.

I should also mention that instead of re-peaking and re-PHAing, another strategy is to simply re-standardize. The automatic standard drift correction in Probe for EPMA is really good, so long as the drift is relatively linear and each drift correction curve is on an element by element basis:

https://probesoftware.com/smf/index.php?topic=168.0

It works even for labs where the temperature control is not stable.
« Last Edit: September 23, 2019, 04:18:27 pm by Probeman »
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2079
  • Never sleeps...
    • John Donovan
Re: Short guide to Multipoint Background
« Reply #10 on: September 23, 2019, 05:19:01 pm »
I don't know why I didn't think of this sooner.

For your situation the easiest way to do this is to go to the Acquisition Options dialog and set your standards to off-peak background acquisition (and specify the off-peak positions for the standards), and for the unknowns set the background acquisition to multi-point background (and set the MPB bgd positions for the unknown sample). Then you'd have one sample setup for both your standards (off-peak) and unknowns (MPB).  I think this could work for you.

No need for the "shared" multipoint bgds in this case.
The only stupid question is the one not asked!

Philipp Poeml

  • Professor
  • ****
  • Posts: 213
Re: Short guide to Multipoint Background
« Reply #11 on: September 25, 2019, 08:46:40 am »
Quote
For different background positions between the unknown and standard samples you will have to use sample setups with the different background positions. But to update is easy. Just make a new unknown sample using the original unknown sample setup, update it, then add this updated sample as the sample setup for new unknowns.  Do the same for the standard.

No need to type anything. When you re-peak the spectrometer, the new peak position is already loaded for that sample, which then becomes the new sample setup using the Add to Setups button in Analyze!  Same applies to adjustments to the PHA settings.

Well, I think there is still a lot of typing, unless I am missing the point. So I have two sample setups. One with all the backgrounds for the standards and one with all the backgrounds for the unknowns.

New day, new luck: So I load the sample setup for the unknowns into a new unknown and update all the peak / PHA. So all the new data is there. But then, how do I get this new data into the sample setup for the standards?

I can make another new unknown, which will also contain the new peak / PHA data, but it will NOT contain the backgrounds setup. I would need to modify this and change all the backgrounds from the unknown setup to standard setup.

Or, I could load the sample setup for the standards, which would contain the correct background setup. But then it would contain the old peak / PHA data.

So whatever I do, I will have to change lot of things manually. Unless there is a third option I am not aware of.

For your situation the easiest way to do this is to go to the Acquisition Options dialog and set your standards to off-peak background acquisition (and specify the off-peak positions for the standards), and for the unknowns set the background acquisition to multi-point background (and set the MPB bgd positions for the unknown sample). Then you'd have one sample setup for both your standards (off-peak) and unknowns (MPB).  I think this could work for you.

No need for the "shared" multipoint bgds in this case.

Hm, wouldn't it acquire millions of backgrounds then? For example for each element 4 multi point backgrounds? For 6 elements that would mean 24 backgrounds? Or am I not getting something?

Actually for my case the best would be pre-acquisition shared MPB per spectrometer for unknowns.
1) For the standards I specify normal off-peak backgrounds in Acquisition options.
2) For the unknowns I select the elements for SP1 LiF, let's say 6 elements, a forest of peaks.
3) I make a spectrum over the whole range of the 6 elements of interest to find good background spots.
4) I select let's say 4 background positions over the whole range of interest and enter them into some window. In Acquisition Options I select "Shared pre-acquisition backgrounds".
5) Now PfE will go and measure the peaks of my 6 elements and just the 4 backgrounds specified and makes them directly available as shared backgrounds for all 6 elements.
6) Happy. Standards measured traditionally, also quick standards possible. Unknowns need just to measure 4 backgrounds. Only one sample setup necessary.

Probeman

  • Emeritus
  • *****
  • Posts: 2079
  • Never sleeps...
    • John Donovan
Re: Short guide to Multipoint Background
« Reply #12 on: September 25, 2019, 10:00:00 am »
Hi Philipp,
I still don't understand why you need to re-peak and re-PHA every day, but fine, let's move on. Basically (I'm doing this from memory):

Assuming you already have an unknown sample with the correct off-peak positions set using the Plot! window and Model Backgrounds dialog.  That is you used the Assign To Samples button in Model Backgrounds dialog to save the off-peak and multipoint background positions to both the last wavescan and a new unknown sample. So now your new unknown sample has the correct off-peaks and multipoint bgd positions already specified:

1. Update the peaks and PHAs as usual by running the peaking and PHAs from the the Automate! window. This updates the off-peak positions also.

2. Browse through the new peaks and PHAs from the Run | Display, Fit, Export peaking and PHA Scans menu. Confirm all is good.

3. From the Acquire! Acquisition Options dialog, make sure the background acquisition method for standards is off-peak and the acquisition method for unknowns is multipoint.

4. That's all folks!

As for multi-point backgrounds having lots of backgrounds to measure, that just depends on what you define. You can have up to 12 backgrounds on each side or a little as two backgrounds on each side. That's your decision. You can set these from the Plot! Model Backgrounds dialog.

I usually first set the number of MPBs for each side, then select each MPB (it gets highlighted in light blue), then I use the +/- buttons to move them around until they look good. Then assign these settings to the current (empty) unknown and the wavescan sample I'm utilizing for the plot.

I guess I don't understand where all the manual typing comes in.
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2079
  • Never sleeps...
    • John Donovan
Re: Short guide to Multipoint Background
« Reply #13 on: September 26, 2019, 09:39:22 am »
Hi Philipp,
It was nice to chat by Skype with you yesterday.

Did you get a chance to try the idea we came up with for your probe runs? That is, where you run the unknowns (with their off-peak positions) as normal off-peaks. So you can later apply the "shared" background method in post-processing.

And also utilize multipoint backgrounds for your standard (with their different off-peak positions), but specify just one MPB position on each side of the peak for standards?

And using only one sample setup for both by using the Acquisition Options dialog to declare off-peaks for unknowns and MPBs for standards?
The only stupid question is the one not asked!

Philipp Poeml

  • Professor
  • ****
  • Posts: 213
Re: Short guide to Multipoint Background
« Reply #14 on: September 26, 2019, 10:01:07 am »
...and I would even be able to use quick standards then.

No, I haven't tried that path yet, today I made a complete new file with better background positions fitting hopefully everything and no duplicate elements any more. I had already selected those backgrounds yesterday.

For the method we came up with yesterday, I will need to go through the wavescans to find good off peak backgrounds for the standards, that also do not interfere with those elements measured for interference corrections. I am correcting for 29 interferences in that file.

Thanks anyway for the good idea. I will check it out and report!