I'm using a pretty standard lab photoelectron spectroscopy system to measure the work functions of metals with UPS. However, I systematically find 0.7eV lower values for the work functions than those reported in literature.
I determine the work functions in the regular way, subtract from the photon energy (21.2eV) the width of the spectrum (onset of the low kinetic energy part of the spectrum until the centre of the Fermi step). I bias the sample to -4V to see a clear low Ek onset.
Does anyone have an idea where such a shift might come from? I don't think it's contamination of the samples since an annealed gold sample shows interface states and still has a low (4.4eV) work function.
Do your measured UPS spectra also exactly reproduce reference spectra for the clean metal? Are the reference work functions averages over a bulk or are they specific to a crystallographic orientation? What is the crystallographic orientation of your sample -- is it single crystal or is it polycrystalline? What method was used to obtain the reference work functions ... UPS or Kelvin probe or ...?
So, eliminate contaminates, crystal orientation, and method. When you still see differences in your measurements from otherwise identical reference measurements, I might suspect bulk composition (i.e. your metal is not a pure as you expected) or instrumentation (i.e. your voltage controller has a bias offset of 700 mV).
Do your measured UPS spectra also exactly reproduce reference spectra for the clean metal? Are the reference work functions averages over a bulk or are they specific to a crystallographic orientation? What is the crystallographic orientation of your sample -- is it single crystal or is it polycrystalline? What method was used to obtain the reference work functions ... UPS or Kelvin probe or ...?
So, eliminate contaminates, crystal orientation, and method. When you still see differences in your measurements from otherwise identical reference measurements, I might suspect bulk composition (i.e. your metal is not a pure as you expected) or instrumentation (i.e. your voltage controller has a bias offset of 700 mV).
Dear Thijs, what about the bias voltage - can you increase it and the entire spectrum shifts in energy? or is there a shift? In which mode do you use your photoelectron spectrometer? Slow electrons are very sensitive towards electric or magnetic stray fields - in our old VG-spectrometer, we had the so-called "magnetic trim" to correct for this. And in fact, this turns out to be an important issue. And, as Jeffrey mentioned above, try to clean your sample, adsorbates always influence the measured WFs! Good luck, Dirk
Following Dirk's excellent suggestion, another option might be to use a different photon lamp with higher energy. The stray magnetic field effect that he suggests will trim the high BE side. As you increase the photon energy, that effect should diminish.
Try the He II line as your source. Do you get the same value?
I guess you measure only gold? If so, what does mean "systematic error"? When you try to reproduce such a sensitive value like WF, you hardly receive the same value - something always remains unaccounted. Esspecialy with such not enough straightforward for this purpose technique like UPS.
Following Jeffreys idea it could also be of interest to go to the photoemission threshold: We have a tuneable laser with excitation energies from about 2 eV to 6 eV - so you can got to the point where emission just starts ... Anyhow as far as I have learned when we built up our spectrometer (check Rev. Sci. Instrum. 83 (2012) 013302; doi: 10.1063/1.3673475 and RSI 86 (2015) 043307; doi: 10.1063/1.4919011), there are not some many lasers used for photoemission studies as one may assume...
Thanks a lot for the suggestions, I'll check especially the magnetic trim and ionisation energy. I didn't consider any stray fields yet.
Some further information:
The spectra do rigid shifts when I apply bias voltages.
I've measured samples of polycrystalline gold and copper, as well as gold 111. Furthermore, also the ionisation potential of C60 also shows the lower value.
These spectra do reproduce the peaks of the references.
I'll get back once I've checked these possibilities. Thanks again!
A workfunction of 4.4eV looks like carbon! All statements by Dirk and Jeffrey are correct. Have a look at the spot of your He source on your sample, do you see any parts of the sample holder illuminated? Try to adjust your lamp by tilting that only your sample is illuminated and adjust your sample with respect to the analyzer to get the steepest onset of the cutoff. The cutoff should have no structure and should be very steep. If your sample is very clean, the true photo elektrons from the d-electrons might have more intensity than the cutoff
- Badges
- Science topic
- Similar topics
- Engineering
- Materials Engineering
- Annealing
- Metallurgy
- Metals