Re:RE: NIST SRM656 Analysis

[iangie]

Hi Elena,


Internal standard method means physically mixing sample with the internal 
standard. 
Adding two data together won't work as their mass absorption coefficients are 
different, hence cannot be cancelled out.


>"Why I'm getting negative amorphous in Topas using IS internal method?" 
If correct weighings and homogenous mixing have been done (scaresly by 
students), "negative amorphous" generally mean your sample is more crystalline 
than your internal standard.


>In DIFFRAC.EVA, any powder will show some %-Amorphous if we compute 
>crystallinity using the Default Bg model parameters  (Curvature=1,  
>Threshold=1 and No Enhancement).
EVA's automatic background if just a boundary line between sharp peaks and 
board humps, for the purpose of Phase ID only. 
It is neither true instrument background, nor amorphous hump. Please do not 
abuse this background for Rietveld.  


>However, even if the powder is known to have an amorphous phase, the 
>%-Amorphous would be =0% if we use a flat Bg with curvature=0.
The amorphous hump area are between your "boundary line" and the flat Bg with 
curverture=0. 
If you adjust the "boundary line" to the "flat Bg with curvature=0", you are 
basically telling EVA that you have 0% amorphous, which is what you get. 
I don't know why do you so though, as EVA tutorial tells users to adjust the 
"boundary line" to cover the amorphous hump. 
This Degree of Crystalline function is just a simple ratio between area under 
sharp crystalline peaks and Amorphous board humps. It can be used in pharma or 
ploymer samples, in which the amorphous part and crystalline part have similar 
chemistries. It cannot be used in multi-phase samples like geological samples, 
becasue their scattering form factors are different. 


>refinement is instrument sensitive...run reference materialsmodel it as an 
>hkl phase
Refinement results from data measured on different machines can be compared if 
both machine's geometric contributions are accounted correctly. 
I would not use "calibrated hkl_Is" model for internal standard, which will 
give others another degree of uncertanity. 
In RR assessment, people even tend to use same .cif structure to compare 
results.


Hope these helps.



--

Dr. Tony Wang
Senior Research Infrastructure Specialist (XRD)
Central Analytical Research Facility (CARF)   |  Queensland University of 
Technology
Address: Level 6, P Block, Gardens Point campus, 2 George St Brisbane QLD 4000
Tel: +61 7 3138 1904  |   Mob: 0452 571 680
Email:   tony.w...@qut.edu.au   |   Web: www.qut.edu.au/ife/carf




At 2022-05-02 18:39:08, "Elena Lounejeva"  wrote:

Good day everybody,

 

I agree with Holger. I'd be happy to help organise a RR for amorphous content 
quantification if somebody takes over. Count me in.

 

Hi Matthew,

 

I came back to your question after your last message on April 30th. I only know 
how to estimate amorphous using the internal calibration in Topas; therefore, I 
made an approximate estimation. I summed both of your XY files below and put 
them into the attached .pro file. Doing that, I have noticed a couple of 
things. First, I have noticed that the Bg is a bit convex, and you probably 
would like to double-check your instrumental setup (was it the silicon holder? 
0.5 or 3 mm air screen? A narrow incident slit? Or something else). Second, 
please review the instrumental parameters in the attached *.pro file as I use 
D2 and am not familiar with D8 details. Third, I'm not sure how to improve 
refinement for Corundum, but it is definitively deficient, as you can see on 
the residual curve and high Rwp ~7.

I firstly ran refinement of the mixture with three structures and got ~55% of 
Corundum which is logical given the uncertainty introduced by summing the BGs. 
Then I fixed it as an internal standard at 55%, and obviously, Amorphous 
content was very low (try and see the result). So I tried several values down 
to 50% for Corundum and got the best match for Corundum ~52.5% with Rwp~6:

 

 

|

Phase

|

Wt% in mixture

|

Wt% in original

|
|

Cor

|

52.5

|

 

|
|

Alpha

|

41.47

|

87

|
|

Beta

|

1.16

|

2.5

|
|

Amor.

|

4.87

|

10

|

 

You probably can improve it. 
https://www.dropbox.com/s/r0stzrl8so0h43x/ROW.pro?dl=0 

A proper calculation should be much better after running a physical mixture 
with ~10-20 wt% of Corundum by weight rather than simply summing the data and 
using each simple structure refined for the instrument setup (I agree to DO NOT 
refine Beq) before refining the mixture. I'd be relatively happy with the 
result, but the problem remains with the "unknown" samples…

 

Please review and share your thoughts. Bets wishes,

Elena

 

PS

 

"Why I'm getting negative amorphous in Topas using IS internal method?" - is 
the most off-putting question from our students. In DIFFRAC.EVA, any powder 
will show some %-Amorphous if we compute crystallinity using the Default Bg 
model parameters  (Curvature=1,  Threshold=1 and 

Re:Re: NIST SRM656 Analysis

[iangie]

Dear Jon,


It might be just me, my current QPA practise is to micronise all samples down 
to 10-40 micron range and make a flat sample surface for the machine to scan. 
I trust the SRMs are already fine enough and Matthrew packed them well.
There are roughness correction models but they may be sample (particles size) 
dependant hence used scaresly in my QPA practise. 
I am not aware of any guideline of roughness models based on controlled QPA 
experiments, so I am just worry the roughness models could either under- or 
over- correct the QPA results.


Cheers!

--

Dr. Xiaodong (Tony) Wang
Senior Research Infrastructure Specialist (XRD)
Central Analytical Research Facility (CARF)   |  Queensland University of 
Technology
Address: Level 6, P Block, Gardens Point campus, 2 George St Brisbane QLD 4000
Tel: +61 7 3138 1904  |   Mob: 0452 571 680
Email:   tony.w...@qut.edu.au   |   Web: www.qut.edu.au/ife/carf





At 2022-04-13 18:09:53, "Jonathan WRIGHT"  wrote:
>Maybe a silly question: are you assuming you have the data on an absolute 
>scale 
>in order to do these calculations? Do things like surface roughness somehow 
>not 
>matter?
>
>Best
>
>Jon
>
>
>
>
>On 13/04/2022 11:56, Matthew Rowles wrote:
>> Thanks Tony
>> 
>> When I add the absorption edge correction to the silicon nitride model (and 
>> add 
>> beta-silicon nitride), it becomes -2.8 wt% amorphous; up a little due to the 
>> added correction and down a little due to the extra phase.
>> 
>> If I change your al-SN thermals from 1 to those given in ICSD 77811, I get 
>> +4.8 
>> wt% amorphous. Your atoms were already neutral, so I left them as-is.
>> 
>> If I make all the Si-nitride phases' thermal parameters == 0, then I can get 
>> up 
>> to 8.5 wt% amorphous.
>> 
>> In all of this, corundum is staying as charged atoms, with Al and O beqs 
>> fixed 
>> at 0.334 and 0.278.
>> 
>> I'll have to have a go at applying the absorption edge correction; I always 
>> forget which parameter is which thing and have to rederive my understanding 
>> of 
>> it everytime...
>> 
>> 
>> Thanks
>> 
>> Matthew
>> 
>> On Wed, 13 Apr 2022 at 13:58, iangie > > wrote:
>> 
>> Dear Matthew,
>> 
>> I tried your data and get ~-1% amorphous. My .pro is in below link.
>> 
>> https://www.dropbox.com/s/xuw13c91l9gq5m5/ROW.pro?dl=0
>> 
>> 
>> I normally do not refine Beq, which I believe gives biggest source of 
>> error
>> in QPA...
>> 
>> 
>> Cheers!
>> 
>> --
>> 
>> *Dr. Xiaodong (Tony) Wang*
>> *Senior Research Infrastructure Specialist (XRD)*
>> Central Analytical Research Facility (CARF)   |  Queensland University of
>> Technology
>> *Address:* Level 6, P Block, Gardens Point campus, 2 George St Brisbane 
>> QLD 4000
>> *Tel:*+61 7 3138 1904  | *Mob:*0452 571 680
>> *Email:* tony .
>> w...@qut.edu.au
>>    | *Web:* www.qut.edu.au/ife/carf
>> 
>> 
>> 
>> At 2022-04-13 13:00:44, "Matthew Rowles" > > wrote:
>> 
>> Hi all
>> 
>> Thanks to those that have replied off-list.
>> 
>> I've managed to jiggle things around and get various answers. If you
>> want an answer between -14 and +6 wt% amorphous, I can make it 
>> happen. I
>> can either use charged atoms or not, or use thermal parameters or 
>> not.
>> Combining those between the corundum and Si3N4, you get the 
>> following:
>> 
>> 
>> image.png
>> This is using the scattering factors defined with 11 gaussians.
>> 
>> If you use the ones defined by 9 gaussians, you get
>> image.png
>> 
>> The structures I used are below. In the certificate for 656, the
>> structures are referenced, but those structures have no thermal
>> parameters. Does anyone know what was used in the NIST determination?
>> 
>> 
>> 
>> phase_name "Aluminium_oxide_alpha_10425_icsd"
>> Hexagonal( 4.759355, 12.99231)
>> space_group "R -3 c H"
>> site Al1 num_posns  12x 0  y 0 z 0.14772 occ Al+3 1. beq  0.318
>> site O1  num_posns  18x 0.3064 y 0 z 0.25occ O-2  1. beq  0.334
>> 
>> phase_name "ALPHA_Marchand_ICSD_26191"
>> Hexagonal( 7.75411, 5.62034)
>> space_group "P31c" 'atom positions from Marchand. Thermals from ICSD 
>> 77811
>> site Si1 num_posns  6x 0.0806 y 0.5095 z 0.3020 occ Si+4 1   beq 0.25
>> site Si2 num_posns  6x 0.1675 y 0.2560 z 0.0070 occ Si+4 1   beq 0.29
>> site N1  num_posns  2x 0  y 0  z 0  occ N1.  beq 0.88
>> site N2  num_posns  2x =1/3;  y =2/3;  z 0.3500 occ N1.  beq 0.46
>> site N3  num_posns  6x 0.0390 y 0.3860 z 0.0310 occ N1.  beq 1.06
>> site N4  num_posns  6x 

Re:Re: NIST SRM656 Analysis

[iangie]

Dear Matthew,


I tried your data and get ~-1% amorphous. My .pro is in below link.

https://www.dropbox.com/s/xuw13c91l9gq5m5/ROW.pro?dl=0 

I normally do not refine Beq, which I believe gives biggest source of error in 
QPA...




Cheers!

--

Dr. Xiaodong (Tony) Wang
Senior Research Infrastructure Specialist (XRD)
Central Analytical Research Facility (CARF)   |  Queensland University of 
Technology
Address: Level 6, P Block, Gardens Point campus, 2 George St Brisbane QLD 4000
Tel: +61 7 3138 1904  |   Mob: 0452 571 680
Email:   tony.w...@qut.edu.au   |   Web: www.qut.edu.au/ife/carf




At 2022-04-13 13:00:44, "Matthew Rowles"  wrote:

Hi all



Thanks to those that have replied off-list.


I've managed to jiggle things around and get various answers. If you want an 
answer between -14 and +6 wt% amorphous, I can make it happen. I can either use 
charged atoms or not, or use thermal parameters or not. Combining those between 
the corundum and Si3N4, you get the following:






This is using the scattering factors defined with 11 gaussians.


If you use the ones defined by 9 gaussians, you get




The structures I used are below. In the certificate for 656, the structures are 
referenced, but those structures have no thermal parameters. Does anyone know 
what was used in the NIST determination?






phase_name "Aluminium_oxide_alpha_10425_icsd"

Hexagonal( 4.759355, 12.99231)
space_group "R -3 c H"
site Al1 num_posns  12x 0  y 0 z 0.14772 occ Al+3 1. beq  0.318
site O1  num_posns  18x 0.3064 y 0 z 0.25occ O-2  1. beq  0.334


phase_name "ALPHA_Marchand_ICSD_26191"

Hexagonal( 7.75411, 5.62034)
space_group "P31c" 'atom positions from Marchand. Thermals from ICSD 77811

site Si1 num_posns  6x 0.0806 y 0.5095 z 0.3020 occ Si+4 1   beq 0.25
site Si2 num_posns  6x 0.1675 y 0.2560 z 0.0070 occ Si+4 1   beq 0.29
site N1  num_posns  2x 0  y 0  z 0  occ N1.  beq 0.88
site N2  num_posns  2x =1/3;  y =2/3;  z 0.3500 occ N1.  beq 0.46
site N3  num_posns  6x 0.0390 y 0.3860 z 0.0310 occ N1.  beq 1.06
site N4  num_posns  6x 0.3190 y 0.3210 z 0.2660 occ N1.  beq 0.17


phase_name "BETA_Billy_ICSD_35566"

Hexagonal( 7.60633, 2.90778)
space_group "P 63/m" ' structure from Billy, thermals from ICSD 170004
site Si1 num_posns  6x 0.2323 y 0.4096 z 0.25 occ Si+4   1  beq 0.231 
site N1  num_posns  2x =1/3;  y =2/3;  z 0.25 occ N  1  beq 0.326 
site N2  num_posns  6x 0.3337 y 0.0323 z 0.25 occ N  1  beq 0.314 












On Tue, 12 Apr 2022 at 17:13, Matthew Rowles  wrote:

Hi all



I've collected some more data, and am still getting spurious results, and by 
spurious, I mean -5 wt% amorphous in SRM-alpha-656 when quantified by the 
external method against SRM 676a.


We had some SRM-656alpha (couldn't find any of the beta) stored in a drying 
oven, and some SRM676a stored in a cupboard. I collected some data using a D8 
with Ni-filtered Cu and a lynx-eye detector (0.25° fixed divergence, 250 mm 
radius, 2x2.5° sollers). The patterns were collected consecutively (using the 
same program), with a single peak from SRM1976 (b, I think), acting as an 
intensity calibrant (the intensity didn't appreciably change), collected 
before, after, and between.


Does anybody want to have a look at the data and see what I'm doing wrong? Data 
available at:
https://raw.githubusercontent.com/rowlesmr/pdCIFplotter/changing-str-to-float-conversion/data/row_Cu_676a.xy
 

https://raw.githubusercontent.com/rowlesmr/pdCIFplotter/changing-str-to-float-conversion/data/row_Cu_al656.xy
 



Thanks


Matthew










On Wed, 16 Mar 2022 at 21:13, Matthew Rowles  wrote:

Hi List People


Do any of you use NIST SRM656 in your quantitative analysis quality control?


I've recently started at a new lab, and am finding it impossible to make a 
physically realistic model (in Topas) that gives results anywhere near correct 
(or at least, close to the certificate values).


As an example, using the external std approach with SRM676, I've managed to 
calculate there is -11 wt% amorphous in the beta-656 standard.


I've tried using the silicon nitride structures given in the SRM certificate, 
but the papers and the ICSD entries don't list any thermal parameters.


I can get the same results as given on the certificate using a siroquant model, 
but I don't know the provenance of the HKL files used in the analysis.




Thanks in advance




Matthew Rowles++
Please do NOT attach files to the whole list 
Send commands to  eg: HELP as the subject with no body text
The Rietveld_L list archive is on http://www.mail-archive.com/rietveld_l@ill.fr/
++