In principle AFM actually can measure chemical information by measuring the difference in force between tip and individual atoms, see e.g. Nature 446, 64-67 (2007, http://www.nature.com/nature/journal/v446/n7131/full/nature05530.html). This is however by no means routinely done - you will need atomic resolution and a set of force distance curves above the individual atoms in question (with the same tip!) to be able to discriminate between atoms. Then to determine which atom is which you will also need theory (e.g. DFT) to compare the force distance curves.

Ideally this is done in non-contact AFM and UHV/LT conditions.However, as can be read in the reference added above it can be done at RT with a regular cantilever AFM.

In tapping mode the phase can sometimes reveal chemical contrast. But this is by no means trivial or something you will get in any condition, as I think it matters on the water adhesion on the different surfaces; I would think you need to use a rather high feedback of your PLL to be sure that what you see is not just the tip out of phase, so to say you would need very good topography images before you can trust the 'phase image'.

It is not possible to find the chemical composition of thin film from an AFM image. AFM simply gives a topography information of any surface and phase image may provide the contrast between hard and soft regions of topography. You can acquire several other force dependent parameters in form of images for the surface analysis. 

For chemical composition you need spectroscopy in general. 

Better use XPS if you need chemical composition of your films accurately.

thank you very much for your answer to above query. 

Dear  Kotarisrinu Srinivasarao,

XPS , AES , SIMS to find chemical composition .

Bette go with EDX than XPS. In my opinion it gives more reliable results faster.  The EDX often is calibrated for all elements, but with XPS you alwas need to confirm the accuracy of the sensitvity factors with a reference sample of known composition. Especially if you do depth profiles by sputtering.

For chemical analysis, I will recommend you XPS, whichis the best technique to know the chemical composition at the surface. if you don't have XPS facility, you may use EDS/EDX.

In principle AFM actually can measure chemical information by measuring the difference in force between tip and individual atoms, see e.g. Nature 446, 64-67 (2007, http://www.nature.com/nature/journal/v446/n7131/full/nature05530.html). This is however by no means routinely done - you will need atomic resolution and a set of force distance curves above the individual atoms in question (with the same tip!) to be able to discriminate between atoms. Then to determine which atom is which you will also need theory (e.g. DFT) to compare the force distance curves.

Ideally this is done in non-contact AFM and UHV/LT conditions.However, as can be read in the reference added above it can be done at RT with a regular cantilever AFM.

Yes this is the paper I would also have recommended as a reference. It's very, very cool but so special. Monolayer adsorbate with three different (but known!!) kind of adatoms. But a cute example of tuning a method to its extreme limits.

I agree with the recommendations of Mr. Mark Pieter Boneschanscher  as well. Hopefully it will help a lot.

Accourding to my knowladge, you could not only depend on AFM for chemical composition unless you have definate composition from other technique, therefore, AFM can be complementry.

I guess that you're trying to check whether the composition of the WO3 has been maintained after sputtering. The other answers here are correct, some form of x-ray spectroscopy will be needed to get detailed information. However, if you're set on using an AFM, then you could try calibrating your force curves against reference samples, such as your sputtering target. Unfortunately your tip is unlikely to stay as sharp as it is for very long and the blunter it gets the higher adhesion force will get regardless of the surface, making your life tough and your measurements harder. Non-contact force curves are possible too, which would reduce the blunting that might occur but I think these give you a less accurate force measurement (check that).

As a better option, could you run Kelvin Probe microscopy on your AFM? This would measure the work function of the surface matter and as WO3 is quite conductive (I think) it should give decent measurements. You'd need another material with a stable work function like gold next to your sample in the scan to get a good contact potential difference. Again, you'd need to compare this to the original WO3, which would be difficult if you're comparing it to your sputtering target as you can't go depositing gold on that. It's also possible to calibrate the work function of your tip before scanning your WO3 reference.

However, any AFM method would only tell you about the surface composition and nothing about the bulk, so I would recommend x-rays too.

For compositional analysis, there are more direct methods to use like RBS, XPS, XRF, EDS, PIXE, ERDA, SIMS. If such techniques are unavailable, then you need good experience with your AFM to obtain quantifiable data, which will definitely be only surface composition.