Please take just a Cr filter. The K-edge of Cr is at 5,99keV. The Cr  K-edge pratically acts as a kind of 'bandpath filter' for 'killing' the Fe K X-ray fluorescence lines.  

So the X-ray  K-alphas and K-betas of Fe will be strongly reduced compared to the K  lines of Pb, Au and Ag and the L-lines of Pb and Au. The thickness of the Cr-filter should be around 10µm or even a bit more. But please calculate the thickness according to your demands.  You can find the x-ray attenuation values by NIST.

http://physics.nist.gov/PhysRefData/XrayMassCoef/tab3.html

Dear Dr. Martens

Thanks for your suggestion. Actually, we are also interested to know the Fe concentration in the sample along with Pb, Ag, Au etc. Since, Fe X-ray lines from sample and Fe X-ray line from the chamber (Steel, Fe) mixed up, we are getting more concentration (Intense) as expected. Actually, we are not interested to suppress the total Fe X-ray line. So, we are using light materials like graphite collimator to avoid those problems.

The K-Edge Cr filter does not completely 'kill' the Fe lines. They will be reduced according to the attenuation coefficient of Cr at Fe K Alpha and the thickness of the filter. So you can go to lower thicknesses than 10µm, e.g. 5µm or even 2µm, depending on the suprresion factor you want to have. My understanding of your question was that for example you suffer from excessive detector dead time and pile-up and therefore you will preferably reduce the amount of Fe-K X-rays falling onto your detector.

But I agree with you;  separating the Fe line coming from the sample and  from those coming from the chamber cannot be done by a filter but by diaphragm collimation and shielding, e.g. by a stack of collimators in such a way that the detector only 'sees'  the sample but nothing else. But you cannot avoid that  the detector 'sees' the collimator(s) as well. Therefore I would prefer a collimator material of much higher Z than that of Carbon, just to prevent an excessive Compton scatter of the Fe x-rays coming from the chamber walls and being scattered at the collimator diaphragms onto the detector. Cr is also here a good candidate followed by V and Ti.

Remark: I must admit that the Compton scatter cross-section is very low for Fe-K radiation, but due to many reasons (you know them) the chamber  produces a very, very  lot of them. 

Altough you will have collimated your system, as suggested above, you may have X-ray contamination of your detector signal due to unsufficient all-around  shielding (side(s) and back) of your detector. This all-around shielding is some times not really considered when setting up a detector (by the experimentalist and/or even the detector manufacturer). But for Fe-K radiation (~6,4keV ) this shielding is much easier to realize as compared to W-K radiation (~60keV) in the case of X-ray tube applications. I think you have tested the shielding by using a beam stop at the entrance port of the dectector.

In addition please be sure that there is no iron in the very close vicinity of the detector element inside the detector housing/shield. Otherwise some of the x-ray photons passing the detector window (or perhaps the shielding) and having energies larger than the K-edge threshhold of Fe may produce the Fe-K radiation too.

Unfortunately I had such an incident/'accident' with Ag. 

A further aspect came up to me with respect of separating the Fe radiation coming from the sample from those coming from the surrounding: a reference measurement.

If you are able to take a similar sample having the same geometry and content but without  the Iron ( or at least a much lower content) so you will have a good estimation of the Fe radiation contamination from the surroundings (sample holder /chamber). I must admit that such a reference measurement has to be carried out very carefully.

A second thing is that  with this measurement you get a feeling of how perfect your collimation is or has to be set up and you have a measure to impove it.