There are several methods for ZLP subtraction. Mostly, it depends on the setup you are using and quality of the data (signal to noise ratio). Basically, if you are using a TEM featuring a monochromator, you can simply subtract the right-hand side of the ZLP by a reflection of the lefft-hand side. This is because the ZLP should be fairly symmetric. For cold-FEG TEMs, you can not do it because the ZLP is not symmetric. In this case, you can then use some deconvolution method (like Richardson–Lucy) to get the ZLP narrower. Then you can fit the narrow ZLP (result of the deconvolution) with some function and subtract. If you are using a Schottky-FEG, the same procedure applies, but it can be harder due to the larger ZLP. In all cases, the signal to noise ratio should be as high as possible. This means you should acquire many spectra (typically 50), align and sum then. You should also use a very large dispersion and check the spectrometer calibration with some sample with know low energy plasmon peak. A nice description of ZLP subtraction and deconvolution is given in doi:10.1038/nphys575.

I did this work in the following paper, Ultramicroscopy 96 (2003) 559–564

Have a look.

What method and software do you use now and what is your energy resolution ? If it's let's say 0.8 or 1 eV I would tend to say that it's hard to be confident of what is happening in the 3 eV energy region. Maybe you can also try the Richardson-Lucy algorythm. You can also try the software of RH French, for the few times I used it it was really good for this (http://www.lrsm.upenn.edu/~frenchrh/elec__struc__tools.htm)

From the litterature I would tend to answer that there is not a best way to do this. There are a lot of methods available and everyone use the one that seems the best for their own experiments.

Also you really need to check the importance of Cerenkov effect on your experiments (see the papers of Stoger-Pollach) if you want to measure the gap.

I'm using a JEOL ARM200 equipped with a cold FEG. I'm acquiring the spectrum with a dispersion of 0.1 eV/channel and the FWHM of the ZLP is 0.78 eV, so probably, as Luc said, I may need to focus the zero loss better . For the analysis we use basically Digital Micrograph.

I've also being searching the web and I may have problems with Cherenkov radiation, so I may not be able to measure the gap properly.

Thank you all for the answers and the references.

Using an ARM200 with a Cold-FEG you should be able to have a ZLP with less than 0.3 eV (typically 0.27 is achievable). Still, to do it, you should use the highest energy dispersion available. To avoid Cherenkov radiation, you can select only higher angle electrons (by using an aperture or selecting a specific CCD area of the spectrometer). I think it should be feasible to determine a 3eV band-gap using a Cold-FEG, but it still depends on the sample.

In order to get to ~0.3 eV (typical for cold FEG) you may need to play with focus X and focus Y and the aberration correctors SX,SY in the GIF filter control panel (assuming you have a GIF). Also try using a very short acquisition time and see if your resolution improves. If so, you need to adjust the AC compensators to cancel external fields from jittering your ZLP. Also to minimize Cherenkov effects your sample should be very thin (

Luiz Zagonel: [To avoid Cherenkov radiation, you can select only higher angle electrons (by using an aperture or selecting a specific CCD area of the spectrometer)]

Sorry but I don't understand what you mean, for what I know, if you use an aperture you are selecting lower angle electrons and stopping the high angle ones. Also, how much is a high angle for this measure?

Thanks

To avoid Cherenkov loss using an aperture, the aperture is positioned off axis. It would block all the ZLP and all Cherenkov. The exact angle depends on the angle of the Chrenkov loss in your material. Indeed, Cherenkov loss has a well defined loss cone and can be observed very clearly in Energy vs. Angle figures (that is what is observed in the GIF CCD when not binned). Check http://personal.teknik.uu.se/teknikvetenskaper/elmin/eels/talks%5CSigle.pdf and references within.

Luiz Zagonel: [To avoid Cherenkov radiation, you can select only higher angle electrons (by using an aperture or selecting a specific CCD area of the spectrometer)]

Lowering the voltage is the way to prevent Cherenkov radiation. depending on the material the spped of the electron is below the speed of light, when going to 80,60 or 40keV. moreover the energy resolution becomes better since there is a contribution coming form the accelartor, which broadens the enegy resolution of the ZL. An overview of applications for high energy resolution applcations is given in the attachment. some of the literture is exactly dealing with you problem e.g. the ZnO band gap measurments.

Bert

Forgot the attachment in my last filing :-)