You can firstly get at% of elements from XPS spectra with n=I/S, where n means number of atoms per cm2 area, I is area of XPS spectra and S mean atomic sensitivity factor (NOTE: S is different for different chemical states of atoms). You can find an atomic sensitivity factor table by checking the keyword on Google. After that, just transfer at% to wt% easily.

Dear Guoran, Thanks a lot for this. This is extremely useful.

Regards

Patro

You also have to take into account the analyzer transmission function (the amount of electrons reaching the detector depends on their initial KE). Just check out the analyzer manual for that. Also, if the binding energies of the different CL are quite different, you should correct the spectral weight you measured with respect to the electron inelastic mean free path.

Best,

Patrick

Another thing to note from the above suggestions be advised that the above formula is an empirical formula and the sensitivity factor is also empirically determined. If possible, it is always best to use internal references with different elements to calibrate your instrument. Also note that the answer you get will be very dependent upon how you quantify your peak area. Figure out the best background subtraction and peak fitting method for your samples.

Thank you very much all for the useful suggestions.

Comment to Paul Desario: I checked and found that the height of the peak in the wide spectra can also be used for quantifying the elements instead of area which depends on peak fitting methods, hence error prone.

Comment to T.U.P.:

Be careful when using just the heights of the peaks in a survey spectrum.

For example, p-, d- and f-orbitals form doublets that can (in some cases) not be

resolved in a wide scan spectrum. These orbitals exhibit therefore a larger width

than e.g. s-orbitals. When using just the peak heights, you completely ignore

the different widths that also determine intensity.

Sincerely, FM

In general, we use the general spectrum to identify the regions of interest for the high resolution spectra acquisition, and to obtain a fast and rough estimation of the elemental composition of the specimen.

For a more precise quantification, the high resolution spectrum of each element is needed.

Dear FM and S.Jr., you are absolutely right. But still I did not understand how do I find at% from individual spectrum. Can the areas of individual elements be compared with each other?

Best,

Patro

Dear Patro:

You may use the following approximation

Cx = (Ix/Sx)/(∑〖Ii/Si〗)

where Cx is the atomic concentration of the element, Ix is the intensity (height) or area of the peak of the element x and Sx is the sensitivity factor for the element.

You may obtain useful sensitivity factors in function of the area or the height of the peak in the following direction

http://www.uksaf.org/data/sfactors.html

Kind Regards.

S.

Dear Sir,

when taking the formula provided by S.Jr. the energy dependence of electron

mean free path as well as the energy dependence of the analyzer transmission

is not taken into account (see answer by P. Amsalem) .

The latter effects can have a distinct impact especially

in the case that the orbitals used for analysis strongly differ in binding energy

(i.e. they also strongly differ in kinetic energy!).

The photoemission intensity for obital a from element z is given (using the

symbols from S.Jr´s answer)

I(Z,a) ~ C(z) x S(hw,z,a) x T(Ekin) x L(Ekin)

with

S(hw,z,a)

the element specific, orbitla specific AND photon energy (i.e. hw) specific photoemission cross section. These value are listed in the tables by Yeh and LIndau, see: Atomic Data and Nuclear Data Tables 32 (1985) 1

T(Ekin)

the energy dependent transmission that can be often described

T(Ekin) ~ Ekin^apha (alpha depends on the particular spectrometer, see manual)

L(kin)

the enegy dependent electron mean free path that can be approximated

L(Ekin) ~ Ekin^beta, with beta ~ 0.5, see: M.P. Seah, W.A. Dench, Surf. Interface Anal. 1 (1979) 1.

When comparing the intensity I(z1,a1) and I(z2,a2) from orbital a1 and a2 of elements

z1 and z2, the ratio of both elements (in at%) is therefore

C(z1):C(z2) =

[ I(z1,a1) x S(hw,z2,a2) x T(Ekin2) x L(Ekin2) ] : [ I(z2,a2) x S(hw,z1,a1) x T(Ekin1) x L(Ekin1) ]

or

C(z1):C(z2) =

I(z1,a1) : I(z2,a2) x S(hw,z2,a2) : S(hw,z1,a1) x [Ekin2:Ekin1]^(alpha+beta)

As you can see, S.Jr.´s formula is an approximation when the orbitals used for analysis

have nearly the same binding energy (i.e. also nearly the same kinetic energy).

Then, Ekin2:Ekin1 is close to 1 and the exponents alpha and beta have no distinct impact since (~1)^(alpha+beta) ~ 1

Sincerely, FM

http://www.lasurface.com/accueil/index.php

I woud recommand you to have a look at this website.

Cheers,

Patrick

Dear Prof F.M:

Thank you for your enriching comment.

In our previous studies, we have used the formula and sensitivity factors given in my previous answer to obtain the XPS atomic percentages of Zn, Mg and Al content in the surface of a commercial magnesium alloy. For quantification, we used the area of high-resolution Mg 2p, Al2s and Zn2p3/2 peaks. Comparing the results obtained by XPS with other independent techniques, such as bulk chemical analysis or EDX,, the error in the atomic percentages obtained by XPS was below 1%.

Only as a curiosity and from a technological point of view, which is the typical difference in atomic percentages obtained by XPS between one or other approximation.

Kind Regards.

S.

Dear S.Jr.,

fortunately, the parameters alpha and beta can - depending on setup -

exhibit different signs. In our case, we have beta ~ +0.5 for the electron mean free path

vs. alpha ~ -0.7 for the transmission, resulting in a correction of (Ekin1/Ekin2)^0.2

which can be small even for large differences in kinetic energy.

The error concerning analysis with and without this correction is hard to estimate,

especially when comparing the surface sensitive XPS technique with more bulk sensitive techniques such as EDX. The results obtained by both techniques cannot

be compared directly, because in many cases, the surface must not be considered

as a cut from the bulk. Just think of relaxation or reconstruction phenomena or

oxidation that affects rather the surface than the bulk.

Since our (alpha+beta) is small, we also ignore this correction in most cases

especially when comparing intensities from signals that are close in binding

energy (if possible).

Regards, FM

Dear Prof F.M.

Again, thank you very much for your enriching comments.

In the specific case of polished commercial magnesium alloys in contact with the laboratory atmosphere, we have obtained by XPS Al/(Al+Mg) x 100 atomic ratios on the specimen surfaces after short periods of sputtering ( only 2-3 minutes) practically identical to those of the aluminium content in the bulk of these materials obtained by EDX or chemical analysis.

Kind regards.

S.