I just need to calculate the valence band maxima for a semiconductor which is actually in a heterojunction with another semiconductor. The motivation is to find the exact energy level of VB and CB of different semiconductors, which are actually in a heterojunction.
Do you also have UPS available? Because that will give you more accurate results since the S/N will be better than with XPS.
In both methods, you need to apply a negative bias -UB to your sample or you'll just measure the work function of the spectrometer.
Then the fermi energy EF will be found at h*nu+e*UB in your spectrum (there's no signal there, it's just your reference point).
You will observe a cutoff of your signal somewhere below that energy, that's your VB maximum and the energy difference from EF to this cutoff is its binding energy.
If you want to get an ionization potential additionally, you have to add the work function to the binding energy. With XPS you may have problems to do that because in XPS mode some detectors can't handle the low binding energies. At the low kinetic energy end of your spectrum your will find another cutoff, that's the so-called secondary electron edge. Subtract your bias energy from the kinetic energy of that edge, that gives you the work function.
The CB level cannot be determined from XPS since that probes only occupied states, you can either perform inverse photoemission (IPE, pretty rare) or you can measure the band gap by optical spectroscopy/tunneling spectroscopy/whatever you have available. Please be aware that optical bandgaps can be larger than electronic bandgaps.
Jürgen Weippert I think you mean that the optical bandgap is usually smaller than the electronic bandgap, right?
Regarding the determination of the VB, I think here it is not necessary to apply a bias. This is usually only needed for the measurement of the work function.
I'm afraid that VB edge can not be reliably measured with XPS, because of too low cross section for X-ray absorption by valence electrons... this is the reason for which UPS was developed several years ago. This is even more difficult with semiconductors having small electronic gap, and one seeks for BE resolution of few tenths of eV..
Giuseppe Curro I agree with you, that the resolution of XPS is usually too bad to reliably determine the band gap of one material. Here UPS (combined with IPES) is the measurement of choice. However, for the determination of valence band offsets in semiconducting heterojunctions XPS is quite reliable, sometimes more than UPS.
The following paper and the references therein demonstrate that quite well I hope.
Article Band offsets at κ-([Al,In] x Ga 1-x ) 2 O 3 /MgO interfaces
Thorsten Schultz : Why would the optical bandgap be smaller than the electronic bandgap? If the electronic bandgap is indirect it cannot be excited optically (phonon coupled exceptions aside) so in that case the optical bandgap will be larger since the excitation won't go "diagonal" through k-space but just "straight up".
The optical band gap is usually smaller for organic or low dimensional semiconductors, due to their significant exciton binding energy. Same holds also true for bulk inorganic semiconductors, although here the exciton binding energy is typically much smaller.
Regarding indirect band gaps: Isn't the indirect band gap in silicon also considered the optical band gap? Therefore I'd find this definition a bit ambiguous.
Thorsten Schultz , generally speaking the extraction of a reliable position for any structure in a XPS spectrum is limited by usual instrumental contribution, more or less represented by a gaussian HWHM to be convoluted with the expected physical signal. This works better for larger signals (I mean larger S/N edges or peaks). Of course, using highly performing detectors or highly brilliant X-ray sources (namely synchrotron radiation) results may be obtained even with small cross section transitions, but how good such results are considered depends on the accepted uncertainty for a specific experiment. My suspicion is that when small BE variations are the target of a measurement, comparable wity instrumental uncertainty, the result reliability may be detrimentally low for low S/N structures like a VB edge. Of course, what about is a general view and, again, acceptable results can be obtained in some special cases
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science
- Materials
- Semiconductor