Kelvin Probe - measurements with monochromatic light

KrzysztofBienkowski

How we can interpretated results of CPD with and without monochromatich light? Light with energy higher and lower of energy of bandgap of semiconductor?

zbyszek

CPD and monochromatic light

Although the interpretation of Contact Potential Difference (CPD) measurements is usually a difficult task because it involves variety of effects like

sample inhomogeneity,
surface roughness,
surface dipole effects,
light induced surface reconstruction,
surface embedded impurities,
sample thickness,
back contact properties
and many more,

it is important to have an understanding of basic Surface Photovoltage mechanisms.

In this answer I will use nomenclature from L. Kronik, Y. Shapira / Surface Science Reports 37 (1999) 1:206.
Photons with energy higher than bandgap will be called super-bandgap photons and those with lower energy: sub-bandgap photons.

Assumptions

Here is the list of assumptions made in what follows:

sample is homogeneous with no surface roughness
it is thick enough, so that light does not interact with the carries on the back contact interface
there are no surface impurities
there no trap states,
the only states with energies in the gap are the surface states
there are no molecules adsorbed at the surface
work function of the sample is smaller than the work function of gold (tip)
light intensity is large enough to lower the surface potential with super-bandgap photons.

I will also, unlike Kronik and Shapira, avoid using a misleading term of “band bending” which should be best abandoned altogether.

Sub-bandgap photons

N-type semiconductor

Here is the energy diagram for n-type semiconductor. It has some surface states in the energy gap.
They are mostly occupied by electrons, up to the Fermi level [imath]E_F[/imath]. The yellow line represents an energy potential function for electrons. Negative carriers, i.e., mobile electrons in conduction band, are repelled from the surface by trapped surface charges and the Space Charge Region is formed.
When low energy photons are shined upon the surface:

they cannot excite a valence band electron to the conduction band because they don’t have enough energy
they cannot excite a valence band electron to the surface state because the surface states are already occupied
they can excite the surface electrons to the conduction band, where they will be immediately pushed to the bulk by potential in the SCR
if the electrons are removed from the surface, it becomes less negatively charged
if the surface is less negatively charged, it becomes more difficult to remove an electron from thus the work function increases and CPD increases or becomes less negative. For example, CPD will change from -100 mV to -50 mV.

P-type semiconductor

For p-type semiconductor the surface states in the energy gap will be mostly empty. The surface charge will be less negative than in the n-type semiconductor. That would indicate much lower sensitivity to light, because SCR will be smaller, too. For that reason SCR is not drawn in the diagram.
When low energy photons hit the semiconductor:

they cannot excite an electron in valence band to the conduction band, because they have insufficient energy
they cannot excite surface states to the conduction band for the surface states are empty,
however, they can excite valence band electrons to the surface states
thus, the surface will be charged negative
work function will decrease, so the CPD will decrease. For example CPD will go from -100 mV to -200 mV.

Super-bandgap photons

N-type semiconductor

Here are the processes for super-bandgap photons:

they have enough energy to excite valence band electrons to the conduction band
both bands have much larger state density than surface states. That means that the inter-band excitation will be way more probable than band-surface excitation
the valence band electrons excited to the conduction band will be numerous and they will destroy the Space Charge Region and the surface potential
no surface potential will increase surface electron density hence work function will decrease and CPD will decrease. For example, CPD value could change from -100 mV to -200 mV.

P-type semiconductor

This scenario will involve the following analysis

photons will couple the valence and the conduction band
in comparison, the direct coupling with surface states can be neglected
there will two processes removing conduction band electrons: recombination and deexcitation to the surface states
recombiantion does not affect the surface charge density, deexciation does
the deexciation will charge the surface negative
if surface becomes negatively charged, the work function will decrease, so the CPD will decrease as well. For example, CPD could shift from -100 mV to -200 mV.

Summary

In the four cases considered CPD will increase only for low energy illumination of an n-type semiconductor.
Otherwise CPD decreases.

KrzysztofBienkowski

Could you also explain cpd increase (value)?
And time constant of changing of of cpd during ilumination?

zbyszek

KrzysztofBienkowski
I don’t think it is possible to do without knowing more about a sample in the consideration.

Increase of numerical value of CPD, i.e., as increase from -100 mV to 0 mV , is related to the increase of work function of the sample. That would mean charging the surface more positive.

As to the time scales, they usually fall into three categories:

Slow, when trap/surface states are involved, because those states tend to be long living
Fast, when inter-band transitions take place. For example, recombination belongs to that group
Very fast, when intra-band transitions are considered. Like in conduction band thermalization processes.

In real live experiments, you have most likely all three types present. However, very fast processes are instantaneous for most experimental circumstances.

Martin

Dear Zbyszek,
could you please describe the processes on real measurements example? I’m attaching print screen of measurement with notes when irradiation was turned on and off. The attached measurement is not whole due to a LED wheel issue. I started the measurement and the CPD was +25 mV (it was stable for 8000s) than I irradiated the sample with sub-bandgap photons (530 nm) and the CPD decreased to approximately -120 mV. Then I wanted to change the LED but the LED wheel got stuck. So I had to stop the measurements, disconnect and reconnect and start the measurement again that’s where the print screen starts. I believe it would be very beneficial to all users if you could explain/describe the processes happening in the sample. Thank you very much for you time.

zbyszek

Hi Martin,

First of all, please make sure the signal that you recorded comes from the sample and not from a contaminated tip.
Check out the most recent post by Asia.

Second, cool example of temperature and relative humidity relation visible in the plot!

If your tip is ok, and it is does not give signal on its own when illuminated, here I will present my take on your results.

In the plot above, I have depicted what you described in your post. The part with the light blue background is what is shown in the printscreen from your measurement results.
I have divided the CPD curve into six consecutive segments denoted with roman numerals.
Without knowing what the sample is, one can deduce the following: It is a type-n optically thick sample with plethora of trap states, and the mobility gap between 370 nm and 530 nm. Moreover, the sample has not been kept in dark before the KP measurements were performed.

Here is what might have happened:
I. sample is not in its ground state because it has been unintentionally(?) excited with ambient light, so its trap states above Fermi level are partially occupied at expense of surface states which are partially empty. It looks stable only because the trap states are long lived.
II. After 530 nm light is turned on the empty surface states (low density) are coupled with valence band states (high density) and the electrons come to the surface. The CPD decreases.
Some of the surface states assist in promoting electrons to the conduction band from valence band (a two photon process) and surface potential gets somewhat decreased.
III. After 530 nm light is turned off, the surface potential is recovering as soon as conduction band electrons deexcite.
IV. Now, the high densities of valence and conduction band states are coupled. This is why the change in CPD is so rapid. The surface potential is destroyed, because there are many electrons now in conduction band where they can move. More electrons can flow towards the surface. CPD decreaces.
V. The UV radiation will penetrate only near-surface part of the sample. Electrons in the conduction band can freely move from the surface towards bulk where they can recombine not only with the holes in valence band but also with empty trap states. So, the number of electrons available for the excitation from valence to conduction band decreases. So, does the total number of electrons close to the surface. Thus, the CPD increases.
VI. The UV light is turned off and the surface potential is recovering repelling conduction band electrons from the surface even more.

As you can see, that is just one speculation based on the information that you provided. There are more scenarios possible. But first, make sure your KP tip isn’t playing games with us.