Hi

Probably mass spectroscopy will be the best buy. However if one can compromise somwhat on the accuracy ~ 15%, then Auger spectroscopy may also be used.

JP

I think XPS will detect it very easily.

a couple of things you need to make sure, which you did not mention in your question, before talikng about the accuracy: the continuity of the coating, the depth of the electron beam and escape length of the characteristic X-Ray from your sample, in short you have to make sure that the signal comes solely from your film in the first place.

The depth of penetration of the incident electrons.

I wonder whether you are forming a layer of FeSn2 at the interface with a copper-rich layer above it. Certainly when tin coatings on steel are reflowed (i.e. melted) a layer of FeSn2 is formed that is of the order of 0.1um thick. It does not form unless you apply heat, however.

The other issue is that the EDX analysis will be sampling to a much greater depth than your coating, and so it may be getting the ZAF correction wrong.

Did you use a moderate tilt angle and optimum electron energy? I suggest 45 degrees tilt angle and 20 to 25 keV electrons.

If the results are still poor then you should analyse the film with a method which is able to detect concentration gradients.

Greatings from Leipzig, Germany

The thickness of your coating is lower than x-ray emission depth (about 1 um).

You have two ways:

1) use more surface sensitive technique such as XPS, AES, static SIMS;

2) exclude matrix composition from EDX results. This way will provide less accurate result than the first one.

I think Mr. Hirsch oppinion is the best. With tilt angle & optimum energy selection.

Your problem is the thickness, therefore I suggest to apply XPS investiogation of the surface

I think another crucial point is that you are not able to excite the Sn K-alpha line with a common SEM, where the kinetic energy of the electrons is limited to about 20-25 keV, while you need about 30 keV to excite the K-Apha fluorescence, So you are stucked to the L-alpa and L-beta lines, which are less intense. I would suggest to use a quite small primary electron energy, so that i) your electron penetration depth is smaller, and ii) the efficiency of Sn L excitation increases, so you are more sensitive! Good luck, Dirk

In addition to what Dirk have suggested, vary the voltage and compare SE and BSE output.

Anyway, EDX can not give you right result, others make sense.

Chemical analysis is the most valid way and then XPS also help you make a success

You can try to take in consideration other area of analysis from surface.

Do cross section analysis ...and take EDS from your film....

Other way are .....XPS, SIMS ..AES .... Apart from these techniques .... You can do GIXRD ...which can tell you the different phase formation of your film. Keep Grazing angle according to your film thickness.

And You can easily differentiate, weather your film has Sn, FeSn2 ...or any other phase.

hi I don't think XPS surface analysis makes a whole lot of sense for this particular bulk problem since the top 5-10 nm will likely not reflect bulk composition. Judging from your materials system description, RBS should do the trick, since it has great cross-section for heavy Sn, gives absolute dose (quantitative), and may be able to give information on concentration gradient. Not sure about other additives in your bronze films (with mass in close proximity to Sn), which could overlap with the Sn-peak.

regards,

As the coating thickness is very small, i would like to suggest you to go for lateral characterization like AES and / or XPS so that the error due to the steel background should be minimal (as coating thickness in micron level)

XPS gives you the composition at the surface, typically 5-10 nm deep. But if you suspect the composition might be different at the surface from the bulk, you can probably run a step profile to access the profile of the composition vs. thickness.

I think that the HRTEM/EDX is useful to analyze your samples. To do that you need cut your samples to thin slices by FIB and then use the HRTEM / EDX analysize the structure and element compositions. With the HRTEM / EDX analysis you also can get an element map of the sample. so it is very useful, you can get more valuable information than what you can image.

I'm not suggest to use XPS, since it can only give the information deep to several nm.

XPS is only valid for for 5-10nm. This composition might vary from bulk composition.

Lower energy in SEM might help to reduce the EDX analysis to 200-300nm. It might be possible to create a calibration curve by comparison of chemical analysis and EDX.

Making a cross section for SEM/EDX analysis is a great suggestion. Also, make sure you take some time adjusting various parameters in the EDX software - by getting the ideal settings you can resolve peaks that may be slightly overlapping.

TEM/EDX is another great suggestion, but preparing the sample can take some time, but I think in the end, you will get more accurate results.

Yuxin, you said "........coat steel with bronze at a thickness of ~0.2-0.3 um. It is very difficult to accurately detect Sn in bronze via SEM-EDX, it is always much lower then chemical analysis results." What is the expected % composition of Sn in the alloy or is it electrolessly coated? Anyway if SEM/EDS has not given you what you expected, then try TEM/EDS or GIXRD for such thin film phase/thickness analysis, ok? All the best!

Not just to add a new technique, but also to increase understanding of your specimen, I would like to propose TXRF, i.e. X-ray excited fluorescence spectroscopy with grazing incidence angle. Thereby you can control the penetration (=information) depth, and you do not have the problems in quantification as in EDX! If you measure the XRF spectra for various incidence angles, you can calculate a depth profile of the Sn and Cu concentrations!

Another EDX trick is tilting the sample to keep more of the interaction volume located within the film thickness of interest. This sometimes provides additional flexibility in case a certain voltage is necessary for EDX peak excitation. Monte Carlo modeling is useful for tuning all of this. Make sure the sample tilt is used in the quant routine as well.

Decreasing the interaction volume using EDS, could be the easy way or using XPS.

I think EDS is OK. You can change the tilt angle as proposed by Matt Nowell or vary accelerating voltage. If you make several measurements at different voltages you can calculate thickness and composition of your bronze coatings. We use a special software for it (Stratagem), but if you can obtain thickness form cross-section observations (SEM), you can make some calculations with Casino, too. However it is necessary to check if your coating are homogeneous. If they contain precipitates of dimensions comparable or larger than the information zone dimensions (calculated using Casino), it could be an additional source of your problems.

Try a modeling simulation with an x-ray analysis program such as DTSA-II developed by NIST or with CASINO (as mentioned earlier). The model should be set up with a beam going through the top layer of 0.2 or 0.3 um steel (as you mentioned) on a brass substrate (with whatever %Sn you believe should be present) and then you can compare results using different beam energies experimentally. As the beam energy increases, your interaction volume increases and you will be generating more x-rays from the brass. The Sn x-rays will be attenuated by both metallic regions. By comparing the model and the experiment under different conditions, it should be possible to determine, if you should be able to see Sn in the first place and the possible level. However, as others have suggested the most accurate method is to prepare a cross-section.

HRTEM/EDX is really a good technique for your work.

But the problem is the sample preparation ... The sample preparation for cross section analysis is not very easy. After number of trail only you will get good result. But once you will get, then that will be the best result.....

@ Volker Klemm .... In TEM we can do Magnetic samples .... infact ferromagnetic materials with ~80-90um thick samples.... only thing is that ....we need magnetic sample holder ......

and there is nothing with good machine .... even bad machine also give good result ...if operator has experience and more focused .......

@ Dr. Klemm ... completely agree ....number of trials only give good sample ....

How about scanning Auger electron microscopy (SAM)?

You can do the depth profile by sputtering. Sn is near to the upper limit for AES.

You are lucky:)

By the price-effectiveness relation you could use:

1. RBS - the number one technique for thin film analysis. The principal advantage is a really quantitative in depth analysis without standards with a good accuracy.

2. Depth profiling by AES or by XPS.

3. SIMS

4. HRTEM/EDX

EDX is qualitative. you may try using wavelength dispersive spectroscopy(WDS) analysis in a electron probe microanalyzer (EPMA) which does not require additional sample preparation except that the sample surface must be flat for optical focussing for WDS spectrometer. You have to run the standard analysis of Sn first and then run your sample to get quantitative information on the tin content in your bronze coating.