Just an idea:

When your film is very thin compared to your EDX excitation volume Sr could be under your detection limit.

How much Sr do you expect in relation to La and Mg?

Calculate ratio of intensities of La (or Mn, whatever is higher) and Sr peaks from spectra obtained from your bulk specimen. You may expect the same ratio for your thin film, so you can calculate expected height of Sr peak using existing intensity of La (thin film) and ratio you calculated earlier. If expected peak is substantially higher than background of thin film spectra, then you can suspect that you technique of film deposition is changing its chemical composition.

I agree with Dr Simons. The most likely reason is the thickness of your film. If it is too small you have not enough atoms on the spot. The simplest way for check this hypothesis is increase the thickness of film by increasing of the laser pulses number.

Good luck!    

the thickness of the film is 103 nm and the ratio is La0.8Sr0.2MnO3 .all elements appear in TEM/EDX of target and Sr ratio disappear in SEM/EDX of the film on SI substrate .I can not understand what the reason behind despite all XRD diffraction for target and film confirmed Sr content 

Can you post your spectra? It will be much easier to discuss. 

Ok the attached two images contain EDX spectra of target and film

Yes, now it's much easier to see what was a problem.

Too thin film, too short acquisition time, too weak Sr peaks to detect. You can see that even relatively bigger La M-peaks were not identified on your film (around I Kev). And your EDS from film was taken at too low voltage to detect Sr K-peaks (around 16 Kev). I'd repeat EDS on film a bit differently.

Use specimen tilt, at least 60 degrees, toward EDS detector (to increase effective film thickness).

Make spectra acquisition at much longer time and at two different voltages:

5 kV - better for detection of low-energy Sr L-peaks at around 2 Kev.

25-30 kV - better for detection of K-peaks at about 16 kV.

if possible, you could try to put a copper filter between source and sample. The Cu Ka and Cu escape radiation might cause additional excitation of your Sr L lines.

@Florian Ströbele

Method under discussion is EDS, not XRD. In EDS sample is the "source". Cu lines are from a grid (specimen support).

Your Sr is probably there, but you can not resolve it in EDS. The Sr L-alpha is at 1.8 keV but it is hidden by the the large Si K-alpha peak at 1.74 keV. No EDS can resolve 60 eV. As Vladimir suggested, you need to look for Sr K-alpha at 14.16 keV to confirm the presence of Sr. Since it is at low concentration, acquire the spectrum for several minutes using 25 kV accelerating voltage.

Dear Vladimir,

I am aware of that. Probably I did not express myself in a good way. I wanted to say put a filter between x-ray tube (I have written source for this in my previous post, sorry for that inaccuracy) and sample. Of course this only works if your "source" is an x-ray tube and not an electron beam...

The substrate you are depositing on is presumably Si or a Si Oxide (Glass) as there is a large Si peak in the spectrum you posted of the film. The Sr L line is at 1.8keV and Si at 1.73keV which means the two lines will overlap. The presence of Sr will change the shape of your Si peak from the substrate but your EDS system may not be picking this up depending on whether its automated element ID is based purely on peak position or also on peak shapes. For a proper analysis of a film on a substrate with EDS I would suggest using a dedicated thin film analysis software to separate the contribution of the film from the contribution of the substrate. The company I work for makes a product called AZtec LayerProbe but other products are available.