Author Topic: Drift correction  (Read 6911 times)

Edeltraud_Macmillan

  • Student
  • *
  • Posts: 2
Drift correction
« on: February 05, 2014, 10:02:23 PM »
Hello!

I just need some clarification, I have run primary standards as part of a large package for both calibration and interference corrections. I have also run some of these standards as "unknowns" to check the results as I am running (i.e. not as type "standard"). Is it possible to now use these standards which were run as "unknowns" to conduct a drift correction? Or do these need to be set as "standards" before acquisition to be able to use the drift correction feature?

This package is quite large and has 28 elements in it + standards are mounted on two different shuttles so not possible to run all standard checks all the time. The data looks fairly consistent in terms of the values attained for the standards as unknowns, but it was run over a 1-2 week period so I wanted to test if drift correction was required. Is this now possible?

A couple of other datasets have been run for a few days, then a few weeks off and more data appended to the same database. I assume that when I recalibrated for the datasets that were appended to an older set - that drift correction in this case will occur to all unknowns that were run in between the standard sets? Is this correct?

Thanks,

Edeltraud

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 3223
  • Other duties as assigned...
    • Probe Software
Re: Drift correction
« Reply #1 on: February 06, 2014, 11:09:40 AM »
I just need some clarification, I have run primary standards as part of a large package for both calibration and interference corrections. I have also run some of these standards as "unknowns" to check the results as I am running (i.e. not as type "standard"). Is it possible to now use these standards which were run as "unknowns" to conduct a drift correction? Or do these need to be set as "standards" before acquisition to be able to use the drift correction feature?

Hi Edeltraud,
Unfortunately, it is not possible to convert unknowns into standards at this time. Though standards *are* treated as unknowns in the Analyze! window for the purposes of evaluating accuracy as you probably know. The main difference is that because the program knows these samples are actually standards, it will automatically include elements present in the standard but missing from the acquisition, into the matrix correction automatically.  Also, the program will print out the "PUBL:" or published composition of the standards so one can compare your secondary standard measurements with the published values for accuracy.

Quote from: Edeltraud_Macmillan
This package is quite large and has 28 elements in it + standards are mounted on two different shuttles so not possible to run all standard checks all the time. The data looks fairly consistent in terms of the values attained for the standards as unknowns, but it was run over a 1-2 week period so I wanted to test if drift correction was required. Is this now possible?

Remember this: the "List Report" button in the Analyze! window is "your friend"!   :)

Clicking this button for a selected sample will force the software to list all standard intensities that correspond to the assigned standards of the selected sample.  For example here is the List Report output from a large run:



Each intensity listed in the primary standard intensity section is the average of multiple standard points for each standard set number (1, 2, 3, etc...). In this particular run the variations in intensities seem to be mostly statistical.

Let's examine it in more detail, for example the warning about the zero or negative Ni counts for the specified Ca interference correction is so close to zero that it may not need to be specified at all. It essentially is having no effect on the analysis.



Meanwhile the primary standard intensities look pretty consistent between standard sets. Note that if you delete a standard set completely, the List Report button will *not* list the standard intensity in the List Report output.  The the first listed standard intensity is usually Set 1, but if there was a previous standard set acquired before that and subsequently deleted, the next (Set 2) undeleted standard set will be listed first.

Quote from: Edeltraud_Macmillan
A couple of other datasets have been run for a few days, then a few weeks off and more data appended to the same database. I assume that when I recalibrated for the datasets that were appended to an older set - that drift correction in this case will occur to all unknowns that were run in between the standard sets? Is this correct?

The standard drift correction (unless turned off in the Analytical | Analysis Options menu dialog) is automatically applied.  Basically it searches for the last standardization prior to the selected sample (being analyzed) and also the first standardization after the selected sample (being analyzed).  If a prior or subsequent standardization is *not* available, the program will utilize the one standardization available.

Here's a couple more useful tips:

1. The Run | List Standard Counts (Intensities) menu will list the average standard intensities for *every* standard acquisition as seen here:



2. But my favorite feature for evaluating a large number of standard intensities is also available from the Run menu. This is the output from the Run | List Anomalous Intensity Data From Standards Or Unknowns menu. I clicked the No button to see standard intensities listed. Simply click the Yes button for testing unknown sample intensities:



Simply note that for standard intensities.

Black color indicates an element channel that is *not* utilized for any calculations and has point intensity deviations  that are greater than 4 times the predicted variance of that channel.

Blue color indicates an element channel that is *not* utilized for any calculations and has point intensity deviations that are greater than 10 times the predicted variance of that channel.

Magenta color indicates an element channel that is utilized for standard calculations and has point intensities that are greater than 4 times the predicted variance of that channel.

Red color indicates an element channel that is utilized for standard calculations and has point intensities that are greater than 10 times the predicted variance of that channel.

For unknown samples the key is simpler:
Black color indicates an element channel that has point intensity deviations  that are greater than 4 times the predicted variance of that channel.

Blue color indicates an element channel that has point intensity deviations that are greater than 10 times the predicted variance of that channel.

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

"Not Absolutely Certain, Yet Reliable"

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 3223
  • Other duties as assigned...
    • Probe Software
Re: Drift correction
« Reply #2 on: February 06, 2014, 03:04:07 PM »
A few additional points about standard drift.

When you analyze a sample, the "STCT:" output shows the interpolated standard intensities based on the standard drift arrays. So for example, this run here:

Selected Samples...
Un   12  FeS-Lb at 15.00 keV

Assigned average standard intensities for sample Un   12  FeS-Lb

Drift array background intensities (cps/1nA) for standards:
ELMXRY:    fe ka    s ka   br la   cl ka   se la
MOTCRY:  3  LLIF 2  LPET 1   TAP 5   PET 4   TAP
INTEGR:        0       0       0       0       0
STDASS:      526     730     829     285     534
STDVIR:        0       0       0       0       0
            2.50     .71    2.24     .25    2.45
            2.48     .70    2.77     .29    2.38

Drift array standard intensities (cps/1nA) (background corrected):
ELMXRY:    fe ka    s ka   br la   cl ka   se la
MOTCRY:  3  LLIF 2  LPET 1   TAP 5   PET 4   TAP
STDASS:      526     730     829     285     534
STDVIR:        0       0       0       0       0
          737.23  461.39   87.24   30.17  679.89
          729.94  464.51   81.52   28.92  679.96

Drift array interference standard intensities (cps/1nA):

1st assigned interference elements
ELMXRY:    fe ka    s ka   br la   cl ka   se la
INTFELM:                     se                 
INTFSTD:                     534               
                            2.85               
                            2.76               


Which shows the standard drift arrays (from the List Report button) highlighted in red, while the analysis itself has the interpolated standard intensities based on the actual time of acquisition:

Un   12 FeS-Lb
TakeOff = 40.0  KiloVolt = 15.0  Beam Current = 30.0  Beam Size =    0
(Magnification (analytical) =  40000),        Beam Mode = Analog  Spot
(Magnification (default) =      400, Magnification (imaging) =    800)
Image Shift (X,Y):                                          .00,   .00
Number of Data Lines:   3             Number of 'Good' Data Lines:   3
First/Last Date-Time: 02/26/2013 08:00:57 PM to 02/26/2013 08:11:06 PM
WARNING- Using Low Counts Only Off-Peak correction for br la
WARNING- Using Exponential Off-Peak correction for se la
WARNING- Using Time Dependent Intensity (TDI) Element Correction

Average Total Oxygen:         .000     Average Total Weight%:   99.845
Average Calculated Oxygen:    .000     Average Atomic Number:   31.802
Average Excess Oxygen:        .000     Average Atomic Weight:   70.800
Average ZAF Iteration:        5.00     Average Quant Iterate:     4.00

Un   12 FeS-Lb, Results in Elemental Weight Percents
 
ELEM:       Fe       S      Br      Cl      Se
BGDS:      LIN     LIN     LOW     LIN     EXP
TIME:   120.00  120.00  120.00  120.00  120.00
BEAM:    29.99   29.99   29.99   29.99   29.99

ELEM:       Fe       S      Br      Cl      Se   SUM 
   241  26.740    .229   -.025   -.001  72.665  99.608
   242  26.611    .222   -.033   -.003  73.805 100.601
   243  26.916    .448   -.030   -.006  71.998  99.326

AVER:   26.755    .300   -.029   -.004  72.823  99.845
SDEV:     .153    .129    .004    .002    .914    .670
SERR:     .088    .074    .002    .001    .528
%RSD:      .57   42.97  -14.43  -61.27    1.26
STDS:      526     730     829     285     534

STKF:   1.0000   .5061   .0809   .0601  1.0000
STCT:   733.50  462.81   84.78   29.50  679.92

UNKF:    .2877   .0023  -.0001   .0000   .5763
UNCT:   211.02    2.08    -.13    -.01  391.86
UNBG:     2.53     .37    1.17     .42    1.78

ZCOR:    .9300  1.3158  2.3881  1.2156  1.2636
KRAW:    .2877   .0045  -.0015  -.0005   .5763
PKBG:    84.30    6.71     .89     .97  220.96
INT%:     ----    ---- -106.06    ----    ----

TDI%:     .109   -.937    .765    .000   -.013
DEV%:       .4     3.1     2.9      .0      .2
TDIF:   LINEAR  LINEAR  LINEAR  LINEAR  LINEAR
TDIT:   148.67  149.33  148.67  149.67  149.67
TDII:     214.    2.35    3.28    .403    394.


The interpolated standard intensities for the analysis are, again, highlighted in red above.

Finally, the software will automatically let you know if any standard interpolation shows more than 2% drift per hour by printing a warning to the log window...

Hopefully your long runs will look similar to this week long run where only the standard intensities drifted down only slightly, but are automatically and fully drift corrected using a linear interpolation in the quant analyses:

Selected Samples...
St  730 Set   7 Pyrite UC # 21334 at 12.00 keV

Assigned average standard intensities for sample St  730 Set   7 Pyrite UC # 21334

Drift array background intensities (cps/20nA) for standards:
ELMXRY:    si ka   fe ka    s ka   cl ka
MOTCRY:  1   TAP 3   LIF 4   PET 2   PET
INTEGR:        0       0       0       0
STDASS:       14     730     730     285
STDVIR:        0       0       0       0
            64.4    28.5    26.7     9.1
            60.5    28.7    26.3     8.1
            61.5    30.3    26.9     9.7
            62.0    29.2    27.6      -
            63.8    29.6    25.8      -
            61.1    30.2    26.4      -
            64.3    29.1    25.2      -

Drift array standard intensities (cps/20nA) (background corrected):
ELMXRY:    si ka   fe ka    s ka   cl ka
MOTCRY:  1   TAP 3   LIF 4   PET 2   PET
STDASS:       14     730     730     285
STDVIR:        0       0       0       0
         30321.4  2188.0  6518.8   787.0
         30735.5  2171.9  6528.5   756.6
         30486.9  2158.2  6446.4   676.9
         30205.8  2150.0  6397.9      -
         30454.6  2133.9  6391.5      -
         30433.9  2109.7  6379.4      -
         30257.2  2106.2  6377.0      -
« Last Edit: February 06, 2014, 05:15:13 PM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Edeltraud_Macmillan

  • Student
  • *
  • Posts: 2
Re: Drift correction
« Reply #3 on: February 06, 2014, 07:51:18 PM »
Thank you for the prompt reply! Makes sense now. I had been lead to believe that the drift correction was a post-measurement correction that was done, and you could select which unknowns/standards you wanted to use for the drift correction to be conducted – hence the confusion!
Unfortunately after calibration I ran all subsequent standards as “unknowns” (except when I noticed significant drift & re-ran the calibration for certain elements). I was unaware of how the drift correction was conducted by the software, but will DEFINATELY run more “standards” throughout my long runs now.
In terms of dealing with elements contained in standards (but not in my element list) I have been adding these using “Specified Concentrations” to each standard run as an unknown. So this is how I have been tracking how my calibration is holding out. At least that’s something I was kind of doing correctly!
Thanks for your in depth reply – I am slowly learning some of the intricacies of Probe for EPMA!
I will have more of a play with my datasets and see how I go!
« Last Edit: February 06, 2014, 07:55:08 PM by Edeltraud_Macmillan »

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 3223
  • Other duties as assigned...
    • Probe Software
Re: Drift correction
« Reply #4 on: February 06, 2014, 09:08:50 PM »
Thank you for the prompt reply! Makes sense now. I had been lead to believe that the drift correction was a post-measurement correction that was done, and you could select which unknowns/standards you wanted to use for the drift correction to be conducted- hence the confusion!

Yes, that is true. It is a post-measurement correction applied at "analysis time" (conversion of intensities to concentrations by clicking the Analyze button, for example) to the standard intensities utilized for the k-ratio based on the time of acquisition of the unknown (and standard) samples.

And yes, one can change the standard intensities that are drift corrected, first by simply changing the standard assignments (and hence the intensities displayed in the List Report output), and second by deleting and undeleting the standard samples since deleted samples are ignored, that is until they are ever undeleted.

Quote from: Edeltraud_Macmillan
Unfortunately after calibration I ran all subsequent standards as "unknowns" (except when I noticed significant drift & re-ran the calibration for certain elements). I was unaware of how the drift correction was conducted by the software, but will DEFINATELY run more "standards" throughout my long runs now.

Yes, the frequency of re-standardization depends on instrument stability. Which is mostly a function of room temperature control. That means (for example), that elements on PET crystals might need to have their standards re-run more often, for best accuracy, if the room temperature varies significantly in the lab.

And I know, that no one (including myself!), looks things up except for a last resort, but... if you go to page 319 of the User's Reference manual, there is a nice explanation of the standard drift correction details.

http://probesoftware.com/download/PROBEWIN.pdf#page=334
« Last Edit: July 29, 2015, 11:40:23 AM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Mike Matthews

  • Global Moderator
  • Professor
  • *****
  • Posts: 141
Calibration and recalibration
« Reply #5 on: November 15, 2018, 12:28:48 AM »
Hi John,

I’m playing around with different standards for U analyses and have recalibrated several times (e.g. standard uncounted then coated). If I’ve acquired analyses on my sample then recalibrated and post process the acquired sample again does it automatically pick up the latest calibration of the standards? Is there a way I can see which version of the set of calibrations is being used? Having recalibrated can I revert to an earlier calibration?

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 3223
  • Other duties as assigned...
    • Probe Software
Re: Drift correction
« Reply #6 on: November 15, 2018, 09:23:06 AM »
Hi Mike,
I merged your topic with an existing topic on the standard drift correction because I think the above posts mostly answer your questions.  If not please ask more questions. But basically the software automatically performs an interpolation between standard calibrations based on the elapsed times of the standard acquisitions and the unknown acquisitions between the standard calibrations.

However one has lots of flexibility. For example one can disable a standard sample (from the Analyze! window), and then that standard will no longer get utilized in the quant calculations.

In addition one can turn off the standard drift correction feature from the Analytical | Analysis Options menu dialog, and then the software will only utilize the preceding standard calibration *without* a drift correction to any subsequent standard calibrations.  That is, if the standard drift correction is turned off, the program will only utilize the standard sample acquired prior to the unknown sample, and not interpolate the standard intensity to the standard acquired after the unknown sample.

So depending on the exact circumstances of your difficulties, one of the above suggestions should be helpful.  Let us know what works for you!
« Last Edit: November 15, 2018, 11:57:48 AM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Mike Matthews

  • Global Moderator
  • Professor
  • *****
  • Posts: 141
Re: Drift correction
« Reply #7 on: November 15, 2018, 12:13:43 PM »
Thanks John,

I knew there would be a way of doing it :). Can I add a request, though? Would it be possible to add which calibration set has been used when reporting quant results? This could be a line below the one which gives the standard number. For drift corrected calibrations more than one set number would be listed.

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 3223
  • Other duties as assigned...
    • Probe Software
Re: Drift correction
« Reply #8 on: November 15, 2018, 02:46:02 PM »
Thanks John,

I knew there would be a way of doing it :). Can I add a request, though? Would it be possible to add which calibration set has been used when reporting quant results? This could be a line below the one which gives the standard number. For drift corrected calibrations more than one set number would be listed.

Hi Mike,
It never hurts to ask!    :)

This might be possible but I would need to look in it.  I don't think these parameters are available at the sample level. Let me ask another question though: would it suffice to just look at the samples listed in the Analyze! window when the "All Samples" option is clicked?  In that case the standard that appears before the sample is the first standard (assuming it hasn't been disabled), and the standard that appears after the sample is the second standard (assuming it hasn't been disabled), which are utilized for the standard intensity drift correction.

Would that work for you?
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Mike Matthews

  • Global Moderator
  • Professor
  • *****
  • Posts: 141
Re: Drift correction
« Reply #9 on: November 16, 2018, 03:48:58 AM »
Thanks John,

That should work.

Michael Lance

  • Post Doc
  • ***
  • Posts: 16
Re: Drift correction
« Reply #10 on: March 04, 2020, 12:30:55 PM »
Sorry if these questions have already been addressed:

Within the same PfE .MDB file, I have collected two runs on two different days with two standardizations (both before and after each run). Based upon the discussion above, I see that PfE should automatically use the correct standardizations for each run when I crunch the data, correct?

What is the difference between Analyze within the Analyze! window and Output (Save user specified output) in the main window? Do i need to click Analyze first before I can output to an excel file?

Will PfE analyze the correct standards if I just select my unknowns for analysis or do I need to select both my standards and unknowns within Analyze!?

Will all of the settings for my last sample created through the New Sample/Setup button apply to the samples I create through the Digitize button? (I think the answer is yes).

Why is there "Calculation Options" in Analyze and then "Analysis Options" in the main window? There is also "user specified custom output" under Output which gives even more options. Why not put all of these options together?

Anette von der Handt

  • Global Moderator
  • Professor
  • *****
  • Posts: 344
    • UMN Probelab
Re: Drift correction
« Reply #11 on: March 04, 2020, 02:30:49 PM »
Sorry if these questions have already been addressed:

Within the same PfE .MDB file, I have collected two runs on two different days with two standardizations (both before and after each run). Based upon the discussion above, I see that PfE should automatically use the correct standardizations for each run when I crunch the data, correct?

Each standardization and analysis has a time stamp. If analyzes are bracketed by calibrations (one before and after) it will apply the drift correction between them. If you have two different analyses sessions that were acquired quite far apart but each session had calibrations at the beginning and end, you would be good. If you only had a calibration at the beginning of each day, you should disable the second calibration for the first set of analysis data, export the data, then re-enable the second calibration and export the second set of data.

I usually keep each session in its own mdb file to avoid problems. You can easily load in analytical setups with or without the calibration intensities through the "Load file set-up" option in "Acquire" - "New unknown" dialog.


What is the difference between Analyze within the Analyze! window and Output (Save user specified output) in the main window? Do i need to click Analyze first before I can output to an excel file?

No, you can go directly to Output if you want. The "Analyze" in the Analyze window just let's look at the data which is useful if you want to make changes to your standard assignments, calculation options, specified concentrations etc. We often analyze new materials and want to develop new methods and this helps. Or the user thought it is one phase and it really was a different one.

Will PfE analyze the correct standards if I just select my unknowns for analysis or do I need to select both my standards and unknowns within Analyze!?

You assign the standards that you want to use in the "Standard Assignments" dialog either before your analyses in the "Acquire" dialog or after the acquisition in the "Analysis" dialog. The "Analysis" window just displays the data for you to look at.

Will all of the settings for my last sample created through the New Sample/Setup button apply to the samples I create through the Digitize button? (I think the answer is yes).

Yes

Why is there "Calculation Options" in Analyze and then "Analysis Options" in the main window? There is also "user specified custom output" under Output which gives even more options. Why not put all of these options together?

The Calculation Options in Analyze are applied on an "unknown" to "unknown" basis. We often analyse multiple different phases within the same run and it is important that I can have some calculated as oxides and others as elemental for example or have different stoichiometry assigned.

The Analysis Options in the main window are global settings and allow me to switch settings for all samples at the same time. This allows me to check the impact of interference corrections, aggregation of intensities or beam drift corrections for example without having to enable or disable that on an analysis-by-analysis basis.

Having both options is very useful. The "user specified custom output" is just to select what data will be included for the output and uses for processing whatever is set in "Calculation Options" and "Analysis Options". I run a multi-user facility and depending on the user and project, I elect for output very different types of data.

Hope this helps.
Against the dark, a tall white fountain played.

Michael Lance

  • Post Doc
  • ***
  • Posts: 16
Re: Drift correction
« Reply #12 on: March 04, 2020, 02:48:18 PM »
Thanks for your help!