Crist Empirically Measured SFs – Scofield Base – IP eSFs – PROJECT
To improve the usefulness and accuracy of Atom% produced by XPS systems, we started a major project to improve or refine the usefulness of theoretically defined Scofield SFs. This project uses high purity inorganic compounds, most of which are non-conductive and produce a band gap that allow us to better locate the upper BE that is used to define the peak area. With this goal in mind, we are using man-made single crystals, natural single crystals, and very high purity inorganic chemical compounds. We favor the use of single crystals of chemical compounds.
When a single crystal is used, we cleave or fracture the crystal in lab air to expose fresh bulk which is soon loaded into the instrument. The expected adventitious carbon is considered to be negligent. Only inorganic materials are being used. A separate project for organic compounds will start later. There are difficulties when using chemical compounds so we are using 2-3 chemically related compounds to cross-check each SF value. We are using the Survey Spectra to generate the peaks areas used to generate improved SFs which is the process most commonly used around the world. The average laboratory uses Survey Spectra to measure Atom%s. Our main objective is to produced useful (reliable) atom%s from survey spectra.
When we integrate each peak, we use the most obvious Inflection Points (IP) that are as close as possible to the peak being integrated. We have captured Screen-Shots of our choices for Inflection Points, and have made them available as Screen-Shots and BE ranges (start and end BEs) in the PDF at the end of this section.
We used the Scofield theoretically calculated SFs (sigma) values as the base value (starting value) for each empirically derived SF. Scofield SFs have been normalized to give a C (1s) SF = 1.0, so the C (1s) SF = 1.0 serves as a natural and obvious reference value. To begin the task of developing new empirically defined SFs that are defined by obvious “Inflection Points” at the baseline, we needed a second obvious and natural reference value so we chose to work mainly with man-made and natural single crystals having just 2 elements to simplify the process. We focused on the O (1s) signal as our secondary point of reference. (In later work, a 3rd element has recently been added to the process of refining and cross-checking our empirical SFs.)
Wagner and others chose to use F (1s) as a secondary reference SF mainly because Metal Fluorides are normally anhydrous, and because he had ready access to many different metal fluorides.
We opted to imitate Wagner’s choice, but also chose to use O (1s) as a secondary reference SF because so many commonly used compounds contain oxygen, and because there are many sources of man-made single crystals that are simple 2 element metal oxides.
We opted to purchase man-made single crystals and to deliberately break or cleave them in half in the lab air to expose the fresh bulk, which allowed us to avoid the need to clean the surface of the crystals. By using single crystals, we assumed that the supplier had established the true stoichiometry of the single crystals.
We opted to purchase hundreds of natural crystals from various vendors which should have known stoichiometry assuming that the vendor had accurately identified the natural crystal. The presence of an unexpected element in a natural crystal forced the elimination of those natural crystals.
At the very start, we analyzed various natural, anhydrous, Inorganic Carbonate crystals (ie CaCO3, PbCO3, MgCO3, ZnCO3, BaCO3) to determine the best SF for O (1s). The result is an SF=2.88 for O (1s), which is only 1.8% smaller than the Scofield 2.93 SF. The natural 1:3 ratio of C:O was used to determine this SF.
With a useful (valid) SF for O (1s) and C (1s) we then generated useful (valid) SFs for Ca 2p and Ca 2s. Those are shown above in Red boxes.
The Crist modified SFs for secondary Ca SFs are: Ca (2p) = 6.15, and Ca (2s) = 2.16. The corresponding original Scofield SFs are Ca (2p) = 5.07 and Ca (2s) = 2.59. When comparing these SFs, we note that Ca (2p) for Crist SF is ~20% larger than the Scofield SF, while the Ca (2s) for Crist SF is ~17% smaller than the Scofield SF. These differences between our IP eSFs and Scofield SFs are most likely due to our preferred Inflection Point choice of baseline endpoints. This is an accepted difference because this Project is not trying to establish endpoints for Scofield SFs quantification, but rather we are developing a new method for quantification that is based on the “commonly used Inflection Point choices” made by most XPS analysts.
Rows 1 and 2 of the Periodic Table
The Crist Modified SF is located just to the right of the original Scofield SF. For example Li2CO3 has a Scofield SF of 0.057 and Crist SF is 0.065 for the Li (1s) SF.
Words highlighted with Green lettering are from man-made Single Crystals (s-xtal) and Natural Crystals (n-xtal). Words highlighted with White in Red boxes are the base reference SFs with C (1s) SF = 1.0 which is the normal SF for C (1s) in Scofield tables. The next chemical that extends the chain of correlation is CaCO3 (Calcite). A natural crystal easy to obtain and has very high purity. The next step in our effort is to define a Second SF that is bound in a crystal to Carbon, which explains our choice of CaCO3. The ratio of C:O is 1:3 in CaCO3.
At the start of this project, we opted to work with Metal Fluorides, just as C. D. Wagner did in his effort to produce SFs. Metal Fluorides are useful because many elements form Metal Fluorides and most of them form small crystals and are anhydrous. Metal Carbonates are also useful, but are not so plentiful and are normally powders. Metal Sulfates are useful because many elements form Metal Sulfates, and mos of them them can be bought in anhydrous form, or else we dry them in a small vacuum oven. As work began, we realized that each of these compounds can substitute a Hydrogen (H) for one of the metals or the halides, making it necessary to cross-check more compounds.
The next step is to extend the chain of correlation to other elements such as Li, Na, K, Mg, Al, Si, F, Cl, Br, I, and S. We chose CaF2 (Fluorite) another natural crystal that is readily available with high purity. From CaF2, we generated a modified F (1s) SF. The Crist modified F (1s) SF = 3.95, which is smaller than the Scofield F (1s) SF = 4.43, by ~11%.
Crist Modified SFs – Scofield Base: “IP eSFs”
This Screen-Shot shows most of our efforts so-far. If you squint, you can read the SFs, but be careful, as this is a work in progress for which we welcome lots of help!
Example spectrum and data table of actual spectrum used, and all the key data analysis parameters used to produce the IP eSF values.
Data Analysis Parameters: baseline endpoints used, choice of IMFP KE 0.6 exponent, and choice of Iterated Shirley Background