Thin Film quantitative analysis

[Pointman]

Hello, Nicholas

Just recently I asked a question about about thin film quantification. I give you the link:
 http://probesoftware.com/smf/index.php?topic=726.msg4478#msg4478

What is the best way of doing thin film quantification?

Pointman

Modify message

[Nicholas Ritchie]

DTSA-II doesn't support quantification of supported thin film samples - only bulk samples or unsupported thin films.
However, you can take the k-ratios from the report and input them into another program (maybe CalcZAF or GMRFilm) to extract the composition. 

Nicholas

[Probeman]

DTSA-II doesn't support quantification of supported thin film samples - only bulk samples or unsupported thin films.
However, you can take the k-ratios from the report and input them into another program (maybe CalcZAF or GMRFilm) to extract the composition. 

Nicholas

Hi Nicholas,
Unfortunately as I posted to Pointman here:

http://probesoftware.com/smf/index.php?topic=726.msg4474#msg4474

CalcZAF also only performs quantification on unsupported thin films!   But as I also again pointed out to Pointman, CalcZAF does a pretty good job on thin films so long as the film is not fluoresced by the substrate.

Otherwise, as you mention above, one has to use STRATAGem or GMRFILM.  One is expensive and the other painful. 

I sure wish someone would offer a relatively inexpensive alternative with a nice GUI for thin films with substrates...   I know personally that Probe Software would be interested in marketing such an app, so a bright young person could make some decent money from an effort to develop such an app...

[Ben Buse]

I may have misread your question but I think you need a slightly different approach. For the example you gave - you can do this with just k-ratio not weight pct.

Lets consider 170 nm. thick TiN film (5,4 g/cm3) on Si substrate, which is being penetrated by 10 kV beam. How should i figure out the quantitative analysis results for this particular sample?

If you simulate TiN film on Si substrate (using which ever monte carlo package you like). Then you simulate the standards you used for the EDS analysis - e.g. simulate Ti metal for Ti, Si metal for Si and what ever N standard.

Then you calculate simulated kratios - simulated Ti intensity in TiN/Si divided by simulated Ti in Ti metal.

Then you can compare the simulated kratios with the measured kratios. If thickness and composition correct should agree.

You can then do simulated kratios for different thicknesses and compare to measured kratio and see which gives best agreement.

Ben

[DrSciman]

Hi everyone - I wasn't sure if starting a new thread on the same topic is what you do on this forum (some places prefer you not create new threads), so please forgive me if I've got this wrong here!

Using NIST DTSA-II, I have been working towards getting some compositions of thin films from EDX measurements on those films, using reference spectra of elemental standards that I have measured with the same system and conditions. So I now have EDX data of my Co2MnAl films (where I know the thicknesses with a reasonable accuracy, these are ~50-100 nm) deposited on MgO, and separately I have EDX spectra from Co, Mn and Al standards, as well as from a blank MgO substrate.   

I have tried to use the NIST-DTSA-II quantification alien to fit the film composition using the measured reference spectra.  This does not work - an error is reported after choosing the spectrum to fit. That may be unsurprising since there appears to be no specific quantification option for fitting to a model of a film on a substrate.  But does anyone know what user Pointman would have done in this thread http://probesoftware.com/smf/index.php?topic=726.msg4478#msg4478 that seemed to get it to work?

In case that's a fool's errand, my second question relates to NicholasRitchie's remark from a year ago:

DTSA-II doesn't support quantification of supported thin film samples - only bulk samples or unsupported thin films.
However, you can take the k-ratios from the report and input them into another program (maybe CalcZAF or GMRFilm) to extract the composition. 

I have installed CalcZAF and started playing with inputting k-ratios to then get a composition from them, but what report are these k-ratios from? Is it the report from the successful fit that Pointman managed above, or something else?

And lastly, about Ben Buse's most recent post:

If you simulate TiN film on Si substrate (using which ever monte carlo package you like). Then you simulate the standards you used for the EDS analysis - e.g. simulate Ti metal for Ti, Si metal for Si and what ever N standard.
Then you calculate simulated kratios - simulated Ti intensity in TiN/Si divided by simulated Ti in Ti metal.
Then you can compare the simulated kratios with the measured kratios. If thickness and composition correct should agree.
You can then do simulated kratios for different thicknesses and compare to measured kratio and see which gives best agreement.

To get the measured k ratios, I need the intensity of the measured peaks from the film and from the standards (to find their ratio, and compare to the ratio of the simulated peak intensities).  Can the measured intensities be found using the NIST-DTSA-II software somehow, or do I have to get the integrated intensity of these peaks from a separate program that can do it, like Originlab?

Thanks in advance for the help - hopefully these questions need only pretty simple answers, and then I can give the standards-based quantification a go!

[Ben Buse]

Hi Dr Sciman,

The easiest method is to use gmrfilm http://probesoftware.com/smf/index.php?topic=927.msg5955#msg5955

Gmrfilm will calculate the composition for you.

If you want to use DTSA-II, to calculate the simulated k-ratios you need to take the values from the report (characteristic + fluorescence). Dividing sample by pure metals.
Ben

[Mike Matthews]

As Ben says, GMRFilm can spit out compositions from input k-ratios, or visa versa. Be aware, though, that it requires the k-ratios for each layer separately so there is a problem if you have any elements in common between the layers.
To use DTSA-II you need to calculate k-ratios, as Ben outlined, for known coating thicknesses above and below your suspected thickness. If the range is small enough you can use linear interpolation to generate k-ratio versus thickness correlations and back-calculate your measured k-ratios to thicknesses. This certainly works for 5-20nm mono-elemental films, but your 50-100nm may need a 2nd order fit. Fortunately GMRFilm is very quick to generate each result.

[John Donovan]

I have installed CalcZAF and started playing with inputting k-ratios to then get a composition from them, but what report are these k-ratios from? Is it the report from the successful fit that Pointman managed above, or something else?

The k-ratios you enter into CalcZAF can be from any source. Integrated net intensities, peak minus background intensities, etc.  I finally got around to documenting how to perform thin film analysis in CalcZAF here:

http://probesoftware.com/smf/index.php?topic=1007.0

But it has the same limitations as DTSA in that it only deals with unsupported thin films or thin films on a low Z substrate where fluorescence from the substrate isn't critical. 

But that said, it can do a decent job analytically, as long as there aren't substrate to film fluorescence issues.
john

[DrSciman]

Thanks everyone for your help!  Over the holidays I managed to work out how to get the intensities for calculating k-ratios from NIST-DTSA-II, and to get something suitable working in GMRFilm and CalcZAF for modelling the compositions.  I may have more specific questions later, but your advice has been really valuable in getting me to the point I can work on my measurements.  Cheers all!