About photoelectron spectroscopy, work function, ionization potential

[linwenzi]

Dear all,

I learn that photoelectron spectroscopy can measure the work function of solid semiconductors.
However, some research papers reported it measured the ionization potential of solid semiconductor nanoparticle films.
As we know there's obvious difference between the work function and the ionization potential of solid semiconductors.
Which should be the results measured by photoelectron spectroscopy?

Many thanks!

 

[ZapperZ]

Dear all,

I learn that photoelectron spectroscopy can measure the work function of solid semiconductors.

Only with "calibration" or reference to material with known work function, and a bit more work. Work function is not trivially extracted out of a typical photoemission spectroscopy experiment without the use of various photon energies.

However, some research papers reported it measured the ionization potential of solid semiconductor nanoparticle films.

This is insufficient. You must include proper references here or else we are groping in the dark.

As we know there's obvious difference between the work function and the ionization potential of solid semiconductors.
Which should be the results measured by photoelectron spectroscopy?

Many thanks!

The problem here has many different folds: is it really more of the issue with the property of this particular "solid semiconductor nanoparticle films", or is it more to do with how one defines "work function"?

Note that work function sometime is loosely used to mean different things to different people. While it is unambiguously clear what it is for metals (i.e. energy difference between the Fermi energy and the vacuum level), it can something be unclear for a semiconductor. Some people STILL use the same energy difference (i.e. Fermi energy and vacuum level) in a semiconductor, but now, this is no longer the photoemission threshold energy (which is now the energy difference between the top of the valence band and the vacuum level), which other people have also called as the "work function".

So without knowing what is what in the context of your references, there's no way to know where the source of confusion here is.

Zz.

 

[gadong]

(A) First, I mention that the vacuum level plays no role in photoelectron spectroscopy or in the definition of the work function. This is clear even in elementary texts, such as the chapter in Ashcoft & Mermin's Solid State Physics. For more serious analyses I recommend W.F. Egelhoff Jr., Surface Science Reports, Volume 6, Issues 6–8, May 1987, Pages 253-415, and if the discussion becomes very deep, see C. Herring and M. H. Nichols, Rev. Mod. Phys. 21, 185 (1949).

(B) Work function shift measurements in PES involve observation of the position of the secondary electron cut-off point (near KE = 0). Since the secondary electrons have to travel far from the sample to get to the detector, and thus traverse macroscopic electric fields outside the sample, the measured shifts only approximate (at best) shifts in the true work function (which is specifically defined to exclude macroscopic fields, including those from nearby crystal faces). However, this does provide a poor man's method for monitoring changes in a quantity that is connected to the work function. Many have used it, probably without thinking deeply about what they are actually measuring (contact potential shifts).

(C) PES can also be used to measure ionization energies for particular electronic levels, and this is its normal use. For metals, these ionization energies are referenced to the Fermi level. For semiconductors, the Fermi level may not be a good reference, due to band bending.

To go back to the original question, PES can be used to do both types of measurements, in a qualified sense.

 

[ZapperZ]

(A) First, I mention that the vacuum level plays no role in photoelectron spectroscopy or in the definition of the work function. This is clear even in elementary texts, such as the chapter in Ashcoft & Mermin's Solid State Physics. For more serious analyses I recommend W.F. Egelhoff Jr., Surface Science Reports, Volume 6, Issues 6–8, May 1987, Pages 253-415, and if the discussion becomes very deep, see C. Herring and M. H. Nichols, Rev. Mod. Phys. 21, 185 (1949).

Er... which chapter in Ashcrof and Mermin's text?

This also doesn't make any sense. If the vacuum level plays no role, then bend bending and negative electron affinity should never happen and should not affect the photoemission threshold. Yet, they do!

(B) Work function shift measurements in PES involve observation of the position of the secondary electron cut-off point (near KE = 0). Since the secondary electrons have to travel far from the sample to get to the detector, and thus traverse macroscopic electric fields outside the sample, the measured shifts only approximate (at best) shifts in the true work function (which is specifically defined to exclude macroscopic fields, including those from nearby crystal faces). However, this does provide a poor man's method for monitoring changes in a quantity that is connected to the work function. Many have used it, probably without thinking deeply about what they are actually measuring (contact potential shifts).

This is puzzling. Secondary electrons? Since when are secondary electrons present in photoemission spectroscopy? My avatar is the ARPES measurement on overdoped Bi-2212 high Tc superconductor. Can you point out where the secondary electrons are?

The term "secondary electrons" are used in mutipactor phenomenon where the primary electron induce secondary electron yield of greater than 1. Using it here is highly puzzling, especially when there are no primary electron source. I am also puzzled by the mention of "traverse macroscopic electric fields outside the sample" in photoemission spectroscopy. Such fields (including magnetic fields) are a no-no in photoemission experiment around the sample since this will clearly mess up the energy (and momentum spectrum in ARPES experiments) spectrum of the photoelectrons.

(C) PES can also be used to measure ionization energies for particular electronic levels, and this is its normal use. For metals, these ionization energies are referenced to the Fermi level. For semiconductors, the Fermi level may not be a good reference, due to band bending.

Again, a head scratcher. Not all semiconductors undergo band banding. Stick a piece of silicon in there, and you will observe no band banding at all. So this band bending behavior is NOT a universal behavior of semiconductors.

To go back to the original question, PES can be used to do both types of measurements, in a qualified sense.

Not sure what you mean by both types, but measurement of the work function using photoemission spectroscopy is not trivial, be it a metal or a semiconductor.

Zz.

 

[gadong]

Er... which chapter in Ashcrof and Mermin's text?

This also doesn't make any sense. If the vacuum level plays no role, then bend bending and negative electron affinity should never happen and should not affect the photoemission threshold. Yet, they do!
Zz.

The analysis of a photoemission experiment isn't easy and may indeed not make much sense at first, but there are benefits to doing it properly. Popularity is not one of them. The references that I provided will explain all.

This is puzzling. Secondary electrons? Since when are secondary electrons present in photoemission spectroscopy? My avatar is the ARPES measurement on overdoped Bi-2212 high Tc superconductor. Can you point out where the secondary electrons are?
Zz.

Show me the energy distribution, and I'll show you the secondary electrons - it's the biggest peak in the energy spectrum (near zero KE).

The term "secondary electrons" are used in mutipactor phenomenon where the primary electron induce secondary electron yield of greater than 1. Using it here is highly puzzling, especially when there are no primary electron source. I am also puzzled by the mention of "traverse macroscopic electric fields outside the sample" in photoemission spectroscopy. Such fields (including magnetic fields) are a no-no in photoemission experiment around the sample since this will clearly mess up the energy (and momentum spectrum in ARPES experiments) spectrum of the photoelectrons.
Zz.

The "primary" electrons are of course the photoelectrons. The secondary electrons arise from interband transitions and possibly plasmon decay (initiated by inelastic scattering of the photoelectrons). There is really nothing puzzling about this terminology.

Again, a head scratcher. Not all semiconductors undergo band banding. Stick a piece of silicon in there, and you will observe no band banding at all. So this band bending behavior is NOT a universal behavior of semiconductors.
Zz.

Not universal, no.

Not sure what you mean by both types, but measurement of the work function using photoemission spectroscopy is not trivial, be it a metal or a semiconductor.
Zz.

Not trivial, no. In fact, I think this method stinks, but I know some people swear by it.

 

[ZapperZ]

Again, many of these things you are saying are very puzzling.

The analysis of a photoemission experiment isn't easy and may indeed not make much sense at first, but there are benefits to doing it properly. Popularity is not one of them. The references that I provided will explain all.

I have both references, and more. Can you cite specifically where in those two references are arguments that support what you were saying earlier?

Show me the energy distribution, and I'll show you the secondary electrons - it's the biggest peak in the energy spectrum (near zero KE).

The "primary" electrons are of course the photoelectrons. The secondary electrons arise from interband transitions and possibly plasmon decay (initiated by inelastic scattering of the photoelectrons). There is really nothing puzzling about this terminology.

This is very strange, because ALL of the electrons in the EDC being detected are photoelectrons! The sharp peak "near zero KE", which corresponds to the Fermi edge, is often called the coherent peak, whereas the rest of the tail of the EDC are often call the incoherent electrons, NOT the "secondary electrons". This is why the second hump in the EDC that one often sees in the High-Tc superconductors spectra is often called the incoherent peak. Your terminology of using "secondary electrons" is something that I've never come across in many photoemission literature.

Not universal, no.

Then don't make it sound as if it is.

Not trivial, no. In fact, I think this method stinks, but I know some people swear by it.

Who does? Practically everyone that I know of who does photoemission spectroscopy (including me) would never use this technique as a primary technique to obtain work function values. In fact, the more primitive "photoelectric effect" experiment with varying photon energy has more accurate info on the work function value than the standard photoemission spectroscopy experiments.

Please address my objection to your assertion that the vacuum level has nothing to do with the work function/photoemission threshold, especially in light of physical characteristics such as negative electron affinity.

Zz.

 

[gadong]

I wrote: "the vacuum level plays no role in photoelectron spectroscopy or in the definition of the work function".

(a) In the PES measurement, the photoelectron never travels to a region of zero potential. Its energy can only be referenced to the Fermi level of the spectrometer, whose position with respect to the vacuum level is not generally known to a useful degree of precision. Since you do PES measurements yourself, you are surely aware of this.

(b) The work function of a metal is defined as the energy required to extract an electron at the Fermi level to a point just outside the metal surface - not to the vacuum level (see Ashcroft and Mermin's chapter on surface effects). This is why PES, which extracts the electron to a detector, cannot measure the true work function shift.

I ran across the following document just now that may be of interest to the original poster, if he/she is still around:

http://rsl.eng.usf.edu/Documents/Tutorials/PEScalibration.pdf

 

[ZapperZ]

I wrote: "the vacuum level plays no role in photoelectron spectroscopy or in the definition of the work function".

(a) In the PES measurement, the photoelectron never travels to a region of zero potential. Its energy can only be referenced to the Fermi level of the spectrometer, whose position with respect to the vacuum level is not generally known to a useful degree of precision. Since you do PES measurements yourself, you are surely aware of this.

I don't understand where the "photoelectron never travels to a region of zero potential" comes from. I never say such a thing. Theory of photoemission spectroscopy doesn't say such a thing.

The work function in metal is defined to be the energy between the Fermi energy and the vacuum level. The degree of how well the vacuum level is measured is irrelevant to how it is defined. Is it useful in photemission spectroscopy? Nope! However, this is different than saying it plays NO ROLE in the definition!

(b) The work function of a metal is defined as the energy required to extract an electron at the Fermi level to a point just outside the metal surface - not to the vacuum level (see Ashcroft and Mermin's chapter on surface effects). This is why PES, which extracts the electron to a detector, cannot measure the true work function shift.

But the vacuum level DOES include the image charge potential (see, for example, the chapter of photoemission in Modinos text). So in essence, it encompasses not just the surface states effects! The experimental measurement that determines the vacuum level doesn't care what causes what. It only cares what comes out and when!

I've doped semiconductors surfaces with Cs. ALL the surface calculations show a description of band banding (conduction band, vacuum level, etc.) but NOT the Fermi level. The consequences of such vacuum level banding can be measured in experiments, especially in terms of the increase in quantum efficiency! This is a clear example where it falsifies your claim that the vacuum level has no effect photoemission!

Zz.

 

[gadong]

"I never say such a thing"

I thought this was a serious forum, but perhaps I'm mistaken.

 

[ZapperZ]

"I never say such a thing"

I thought this was a serious forum, but perhaps I'm mistaken.

What does that have anything to do with anything?

You brought up something that came out of nowhere and which I consider to be irrelevant to what we have been discussing. Somehow, that invoke an utterly non sequitur question on whether this is a "serious" forum.

Zz.

 

[linwenzi]

Dear Zapper and gadong,

Thank you for your discussion and suggestion on my questions.

For photoelectron spectroscopy measurement, Seki and co-authors have reviewed its use in the determination of work function and ionization energy. (see Adv. Mater. 1999, 11, 607. DOI: 10.1002/(SICI)1521-4095(199906)11:8<605::AID-ADMA605>3.0.CO;2-Q. http://onlinelibrary.wiley.com/doi/...906)11:8<605::AID-ADMA605>3.0.CO;2-Q/abstract.)

Could you give a comment on this review?

Many thanks!

 

[gadong]

linwenzi,

OK, I would like to draw your attention to the key distinction made between VL(surface) and VL(infinity) in sect. 2.2. The term VL(s) is not standard, but the paper makes the correct distinction and notes that "there has often been misunderstanding about this point"... VL(s) corresponds to the FL+WF (the true WF in this case).

However, there is scope for confusion later in section 3.1.1 and afterwards when the authors refer only to VL in their description of the WF measurement. But it is clear (e.g. Eq(9)) that the paper at this point is referring to a VL referenced to the FL.

A key theme in the paper is that there *may* be misalignment between the FLs of the metal and organic layer FLs. This does not make life any easier.

Fig. 9 shows clearly the secondary electrons at the low KE end of the spectrum (high BE). While the analysis of the experiment seems to be OK, the authors have implicitly introduced yet another version of the VL (at the position of the detector, rather than just outside the surface). One consequence of this is that the crystallographic dependence of the WF shown in Fig. 3 will vanish (this is explained by Ashcroft and Mermin). Maybe they mentioned this somewhere, but I did not see it.

Unfortunately, I do not have time to do the authors justice by reading the paper in depth, but it seems that they have tried to do a serious review of the subject. However, using the term VL to refer to different physical quantities is probably not a good idea.