Chemical State Spectra



Where do Chemical State spectra come from?



After you have collected a Survey Spectrum, you see there are several elements present.  Carbon and Oxygen are almost always present.  In addition to Carbon and Oxygen, you should see other peaks that belong to the elements that you know are part of the chemistry of the surface being analyzed.

If you made a survey spectrum from a sample of Magnesium Sulfide (MgS) then you should also find peaks that belong to Magnesium and also Sulfur.  Magnesium has 3 XPS signals and a lot of Auger signals.  Sulfur has 2 XPS signals and a small Auger signal at high BE.

Each element has a Principle XPS Peak that is analyzed when we need Chemical State information from that particular element.  For Mg, we normally analyze the Mg (2p) XPS signal using high energy resolutions settings when we want to know the chemical states that exist for Magnesium.  For Sulfur, we would analyze the S (2p) XPS signal.  In this website you can find various Periodic Tables of BEs.  Those BEs were collected from the Principle XPS Signal.

Each element has several signals, but we do not analyze the weaker XPS signals.  We analyze only the stronger XPS signals for any one element.  The Principle XPS signal (peak) has the most narrow FWHM which is important because we want to resolve chemical states that overlap.  If we used a peak with a large FWHM, then we would have more difficulty to resolve the presence or absence of various chemical states.



In the example below, we want to measure the chemical states that exist for the Si (2p) signal.  We have used a BE table to identify the various peaks in the Survey spectrum.  Please find the peak labelled as Si (2p) printed on the survey spectrum near 100 eV BE.  As we look at this signal we can barely see a shoulder at the upper BE region of the Si (2p) peak.  From theory we know that Silicon should have a Si (2p3/2) peak and also a Si (2p1/2) peak that are separated by <1 eV.  The peak area ratio for those two peaks is 2:1 based on theory.  As we look at the survey spectrum, the ratio of the two peaks is not 2:1. Why?

The energy resolution of most survey spectra is low (~2 eV) because we use a large Pass Energy (150-200 eV) to Survey what elements are present or absent within the detection limit of XPS and our measurement settings.  Such a poor energy resolution (2 eV) is almost never used when wanting to measure the presence or absence of different chemical states which are usually separated by only 1 eV.

We normally need much higher energy resolution (0.5-1.0 eV) to be able to see different chemical states. We use small Pass Energy values (50-90 eV) to measure normal everyday Chemical State spectra.  We can use smaller pass energies (5-20 eV) but the signal is very very weak and the energy resolution is only a little bit better.  It takes too much time to use very small pass energies and they are only useful in a few cases when measuring conductive materials, never insulators.  It is best to use Pass Energies that produce a FWHM of ~1 eV for freshly, ion etched Ag (3d5) signal.

The image below shows a Survey Spectrum from a native oxide of Silicon (wafer).  The survey spectrum is useful to measure elements, not chemical states. A survey spectrum is normally designed to run from -10 to 1,100 eV with a step size of 0.8-1.0 eV/data point (step) and using a large Pass Energy (150-200 eV).

To measure the chemical state spectrum for the Si (2p) signal we normally measure a much smaller spectrum window (range) – usually 10-20 eV wide spectrum.  The step size should be 0.08-0.12 eV/data-point (step) and the Pass Energy should be between 50-90 eV for normal chemical state spectra.

The Chemical State Spectrum shown here is a rare special case.  The energy separation between the Si (2p3/2) and the Si (2p1/2) peaks is only 0.602 eV.  This sort of spectrum is used by instrument makers to show the highest performance of the XPS instrument.

To be able to resolve the two peaks at ~99.5 eV, we need to use a very small pass energy, PE=10-20 eV which is smaller than the everyday PE we discussed above.  Special case.

The time needed to measure this ultra high resolution chemical state spectrum can be 20 minutes or 60 minutes.  The time depends on the X-ray power being used.

And, then there is a small trick to get such nice data. Do NOT ion etch this native oxide of silicon.  If you ion etch, then the surface becomes rough and the roughness causes FWHM to increase by  ~2x.

And another trick is to measure the freshly cleave edge of the silicon wafer.  Again, do not ion etch.  Various single crystals of semiconductors will give nice looking data if your do NOT ion etch.  If you ion etch the normal surface, then the peaks will broaden due to the roughness from ion etching.



A Chemical State spectrum is produced by focusing on one of the elements that are found in a Survey spectrum of the sample.



This plot shows the various chemical states and chemical state labels used for the native oxide of silicon.

If you study these sub-oxides, you will learn that the true chemical structure is much more complicated then this simple picture.



This plot shows three chemical states and chemical state labels used to identify the chemical states that exist for the C (1s) in PET (Mylar).

If you study this materials, you will learn that there is one more chemical state that you can peak-fit if you have the time and interest.  That extra peak is part of the benzene ring and have a BE ~285.5 eV.

There is another peak at 292 eV which is due to a Shake-up of a pi electron in the benzene ring. To properly quantify PET you need to include the pi-pi* peak for the C (1s) peak area.  The same is true for the O (1s) spectrum of PET.




Chemical States usually have a 1 eV difference between the different states in many, but not all,  of the Chemical State spectra.

Note:  A chemical shift is always made with reference to the pure element that is not ionized.