you can do XRD and examine with EVA command

You can use AAS is the best option  due to detection limit of this spectroscopy. It's very useful for determine stoichiometry.

For the element ratio in QDs, people usually use Rutherford Back Scattering (RBS) spectroscopy or Induced Couple Plasma Mass Spectrometry (ICP-MS).

Rutherford Back-scattering Spectrometry (RBS) is good technique for your purpose. If you have access of XRF thats very nice.

The elemental composition may also be found using EDX. This can be done together with an SEM of your sample.

I agree Ivan Culaba. I think Energy-dispersive X-ray spectroscopy (EDX) is the best way. If you have time and the equipment you can do Wavelength-dispersive X-ray spectroscopy (WDX) which gives better results.

Rutherford Back Scattering (RBS) and Energy Despersive X ray spectroscopy (EDX)  both are good technique to find out the composition of the sample. u can use anyone.

Thank you every one for your valuable suggestions.. 

The EDS analysis equipped with both SEM or TEM instruments provides the exact composition of your materials.

I am aware of EDS but its not so precise as it deals with only the surface. That is the reason behind the idea of going for either XRF or AAS techniques

EDX (or EDXs or EDS, depending on whether you are in England or the US) analyzes the X-rays emerging from the sample when excited by fast electrons in a scanning electron microscope (SEM) or transmission electron microscope (TEM). As the penetration of fast electrons depends on the energy but is in the range of perhaps 100nm to a few microns the X-rays do NOT come from the surface. If you have nanoparticles  then you can get signals from single or multiple particles and so can check homogeneity, which neither AAS nor MS. Drawback: you will have to disperse the nanoparticles onto a carbon film with with support grid and will get stray X-rays from both. If the standard Cu or Ni grids interfer with you Zn line, then use Al instead. You may also see a faint Fe line from the stainless steel of the objective pole piece. And you will have to trust the sensitivity k-factors from your manufacturer for quantification, if you don't calibrate them yourself. 

Thomas, is EDX better then RBS or ICP-MS? They are as accurate as ~2% and show average element ratio. What size of NC can be analyzed by EDX? Certainly 5-10nm, but what about nanorods of ~30nm in diameter?

Danylo,  you are not comparing like with like.  "EDX" (properly "EDS": energy dispersive X-ray spectrometry) is usually done on an SEM and looks only at the X-ray signal which has no (or little,  and quite hard to get at) depth information.  The trouble is that you need to know from what depth the X-rays are coming,  and through what material,  to properly apply the absorption correction.  Your estimate of 2% accuracy is wildly optimistic for this problem - if you got within 10% you would be being lucky.

ICP-MS (inductively-coupled-plasma mass spectrometry) is a bulk technique,  not a surface one.  To apply it to inhomogeneous materials,  again,  you have to apply a series of assumptions.  In this case,  if you know that the substrate has no Zn or S,  you might be able to dissolve the whole sample and get a pretty good measurement of the stoichiometry,  provided your standards were good.  There are a series of examples in the literature of standards labs using ICP-MS successfully in this way.  But be warned.  There are lots of pitfalls!

RBS (Rutherford backscattering spectrometry) is a near-surface technique just like SEM-EDS but in this case the depth information is direct.  And the traceability is heaps easier than EDS so that it is quite easy to get a real accuracy of 5% or better.  In favourable cases you can get 1% absolute traceable accuracy as we showed recently (see link). The disadvantage with RBS is that you do not have the lateral resolution of the SEM.  In this case this may not be so bad since you probably want an average over an ensemble of QDs.

If you do RBS then you can also simultaneously do PIXE (particle-induced X-ray emission) which is also an EDS method but uses an ion beam (say 2 MeV He+) rather than the (say) 5 keV electron beam used in the SEM.  PIXE usually is not sensitive to the L lines since the backscattered particles have to be excluded from the X-ray detector with a filter which also cuts off the light X-rays.  But the excitation cross-section for the K lines is quite high,  and the fundamental parameters are in any case best known for the K lines.  So PIXE accuracy is usually significantly better than SEM-EDS accuracy,  when the fundamental parameters method is used for quantification.  (Note,  where it is not used you usually need sample-matched standards for good quantification!)

As a footnote,  Thomas meant of course that in SEM-EDS the X-rays do not come ONLY from the surface.  But where the ZnS particles are dispersed on a surface then of course the Zn and S signals do come from the surface.  But calibrating still has to be done carefully if you do not have sample-matched standards.  Answering Danylo's question,  I guess that nanorods of 30 nm can be treated quite accurately as a thin surface film.

http://dx.doi.org/10.1039/c3ay41398e

Thanks, Chris, for your detail explanations. In the previous post 2% referred to the RBS and ICP (in case of experienced user).

Can we conclude that RBS, PIXE and ICP-MS give the best element-ratio accuracy?

In this case RBS is the simplest technique which readily gives high accuracy without sample-matched standards with an easily constructed uncertainty budget,  provided the ZnS is on a lighter substrate.  If the substrate is not lighter or the sample structure is complicated then simultaneous PIXE may be needed to resolve ambiguity,  using "Total-IBA" techniques (see link).

PIXE is comparable to SEM-EDS,  and can be interpreted the same way.  But the excitation volume is much simpler than for electron excitation so that quantification using fundamental parameter methods is more secure (that is,  more accurate)..

I am not an ICP-MS expert so cannot comment in detail.  But it is clear that the applicabiliity of ICP-MS depends on the details of the system.  But the traceability chain is certainly much longer than for RBS so that there are more pitfalls in this method.

http://dx.doi.org/10.1016/j.nimb.2011.09.020