Please put some more data, table or the spider graph to compare your results. before that, I can suggest you to consider some analytical limitations and artifact.

I concur with Hasan. The energy resolution of EDX is often insufficient to resolve closely spaced peaks, leading to erroneous element concentrations and identifications. For instance, you measure %-levels of Tc, yet the half-live time of its most stable isotope is 4.2 Ma, so that primordial Tc is nearly extinct. Basically, nearly all Tc that currently exists is Tc that's recently produced in U fission reactions, and it's present in ng/kg levels in uranium ores.

I'd suggest to look again at your spectra to determine which peaks don't show significant overlap. For the other elements, you'll have to use another method, for instance LA-ICP-MS.

David is absolutely right about interpreting SEM-EDX analyses. Automatic peak detection and identification software (if you are using any) often suggest some rather exotic elements for ordinary minerals. REE usually have multiple closely spaced peaks (L-peaks) that are close to and similar with elements such as Ag, Te, Sb. Below is an asdress for web-site that has some EDS spectras for some common sulfides and also REE-bearing minerals:

http://www.sfu.ca/~marshall/sem/mineral.htm

If ypu don't ahve access to LA-ICP-MS, you could try electron microprobe, which are more common, and possibly available in Iran?

Besides that sulphides are much more soft in comparison with allanite. So sulphides able to incorporate into their surface submicron particles of REE silicates during preparation of polished sections.

From other hand if you had use for polishing abrasive "polarit" (pink synthetic REE2O3) it also able to dirten your preparates.

As third possibility, I can to analyse shallow inclusions of REE-minerals under surface of polished preparate.

Of course, all these three possibilities unable to explain high contents of Eu in your analyza. I investigate REE mineralogy more than 25 years, and met significant amounts of Eu (accompanied high concentrations of Sm and Gd) in natural minerals only twice. So your high Eu contents are connected with some error or misunderstanding in routine microprobe measurements.

Perhaps a better method would be for you to analyse a known standard material by the same technique that you have used. In that way you will recognise whether or not you have instrumental interference, as seems likely. You should have no difficulty obtaining a standard to analyse. You can also calculate the stoichiometry to verify crystallographic site occupancy, including the charge balance issues you correctly identified in your initial question. The method is well described in a classic textbook on Mineralogy by Deer Howie and Zussman.

I rather agree with Tuomo Törmänen, it's most probably interference from X-rays lines of other metals/metalloids, since you've been working with L-lines of REE. I think that it's safer to try a mass spectrometry-based method...

Hi Saeid, It is definitely likely that it is do to peak overlap. However another possibility is that you have microinclusions of minerals like alanite. Do you have other enrichments of elements that you may be ignoring because they are not as interesting as REE? I would check to see if microinclusions are a possibility. It is a shame you cannot send any samples out for LA-ICPMS analysis as that would answer the question very quickly. Is there any way you could send out a sample or two? We have had a number of Iranian researchers coming through UTas in recent years....

Allanite inclusions able to explain presence of La? Ce and Nd in analysa. But unable to explain Eu, Gd and Dy. Existence of significant (measureable) contents of Gd and Dy in natural minerals is nonsence without much more significant and well recognizable Y contents. Usually Y:SumHREE ratio is on level 3:1 or even more. Only very rare and in local places (for example in Fennoscandia or Ploskay Mt, Aryskan and Tommot massifs in Siberia) we observe significantly lower ratios 1:1 and even 1:2 with formation of Yb- and Gd-dominant phases. But all these are extremely rare cases.

In your case, I am think, that LREE peaks connected with admixture of some LREE mineral (allanite, monazite), while HREE and Eu peaks are result of software error of your equipment. Otherwise you shoult to observe significant Y peaks in your spectra.

Dear Saeid, please be specific on your question then probably you will get more precise answer to your question.

1- Based on SEM-EDX analysis, amount of Eu, Gd and Dy is higher than 1% ( upto 4%).

are you talking about whole rock geochemistry or analysee on ore? (e.g. magnetit iron skarn ?) 

These elements are not chalcophile and are 3+ but S is 2+ or +. How can we explain these higher amounts of HREEs in sulfide minerals?

Please identify your mineral association or REE-bearing minerals if you have any information on that.

explanation on REE distribution can be depend to many factors (e.g. mineralogy, REE-bearing minerals, method of analyses, extractability of REE based on your analyses techniques etc.