Report 186 · Lab Science
In work on copper catalysts, sensors and corrosion, it is common to read the oxidation state off one XPS peak. For metallic copper and copper(I) oxide, that peak can land at exactly the same energy. A second number, the modified Auger parameter, separates them by about 2 eV, and it has a property the main peak lacks: sample charging cancels out of it.
In X-ray photoelectron spectroscopy (XPS), you identify a chemical state by where a core-level peak sits on the binding-energy axis. For copper, the peak everyone reaches for is Cu 2p3/2. A shift of a volt or so is supposed to tell you whether copper has given up electrons. Often it does. For the one comparison people care about most in copper chemistry, metal versus Cu2O, it frequently does not.
In Report 122 I wrote about the 284.8 eV carbon reference, and ended on a recommendation from the surface-science community: when you can, check your answer with the Auger parameter. This is the worked example of why.
The NIST X-ray Photoelectron Spectroscopy Database (Standard Reference Database 20) compiles evaluated line positions from the published literature. Pull up the records from a 1985 Journal of Catalysis study by Strohmeier, Leyden, Field and Hercules, which measured copper metal and a series of copper compounds in one set of experiments. The Cu 2p3/2 binding energies it lists:
Metal and Cu2O are identical to two decimal places. CuO stands apart (and has strong shake-up satellites that make it easier to spot), but the zero-valent and +1 states are indistinguishable on this axis.
Another entry shows the problem is not a quirk of one lab. A 1973 Surface Science study by Schoen, also in the NIST database, lists copper metal at 932.20 eV and Cu2O at 932.60 eV. So in one study the oxide sits 0.4 eV above the metal, and in the other the gap is zero. Mark Biesinger's 2017 Surface and Interface Analysis review of copper spectra names the issue in its first lines: interpretation is hard because of "overlapping binding energies for Cu metal and Cu(I) species." A few tenths of an eV is about the size of the uncertainty you get from charge referencing alone, which is exactly the problem from Report 122.
When an X-ray knocks out a 2p electron, the atom is left with a hole. One way it relaxes emits a second electron, an Auger electron, with its own kinetic energy. For copper the relevant one is the L3M45M45 line, usually written Cu LMM. It shows up in the same XPS spectrum. Its kinetic energies in the Strohmeier records:
Now the metal and Cu2O are 2 eV apart. In the semi-empirical framework the arXiv paper below relies on (Moretti and Beck), the Auger parameter is governed mainly by final-state effects, meaning how the atom's surroundings respond once the hole is created, which is how it can separate states that the photoelectron peak blurs together.
The modified Auger parameter, written α′, combines the two:
α′ = Ebinding(Cu 2p3/2) + Ekinetic(Cu L3M45M45)
That is the definition as written in a 2026 arXiv study of sputtered copper-tungsten oxide films by José Montero-Amenedo, which I use here because it states it cleanly. Add the numbers from the same 1985 records and you get the α′ values NIST lists for them: 1851.10 eV for the metal, 1849.10 eV for Cu2O, 1851.60 eV for CuO. The Schoen records give 1851.20, 1849.60 and 1851.30 eV. A 1980 calibration study by Bird and Swift, also in the database, puts copper metal at 1851.25 to 1851.32 eV across three measurements. Across all three sources, metal versus Cu2O differs by 1.6 to 2.0 eV in α′, compared with 0 to 0.4 eV in the binding energy alone.
This is the property that makes α′ more than a tiebreaker. On an insulating or poorly grounded sample, positive charge builds up at the surface and slows every electron that leaves it by the same amount. The photoelectron comes out with less kinetic energy, so its calculated binding energy goes up. The Auger electron comes out with less kinetic energy too, so its kinetic energy goes down by the same amount.
Here is a simple illustration (my arithmetic, not a measurement). Say a Cu2O sample charges by 3 eV. Its 2p peak now reads 935.3 eV instead of 932.3 eV, which looks like nothing in the table above. Its LMM peak reads 913.8 eV instead of 916.8 eV. The sum is still 1849.1 eV. The shift added to one term and subtracted from the other, so α′ never saw it.
That is why Biesinger, in his 2022 Applied Surface Science review of 1,237 samples from a multi-user facility, made it one of his closing recommendations:
"The XPS community needs to embrace the use of the Auger parameter which is free of charging issues."
His same review found adventitious carbon referencing gave satisfactory results in 95% of 522 assessed cases, but only when combined with cross-checks, and the Auger parameter was the first one he listed after binding energies themselves.
Look again at the three α′ values: metal around 1851.1 to 1851.3 eV, CuO around 1851.3 to 1851.6 eV. Those overlap almost as badly as the binding energies did for metal and Cu2O. Montero-Amenedo's films are a live example: their α′ of 1851.67 ± 0.03 eV was, in the paper's words, "consistent with reported values for solid copper Cu (s) (α′ ≈1851.24 eV) and Cu2+ compounds such as CuO, CuF2, and CuSO4." The parameter alone could not say metal or Cu(II); the binding energy and line shapes did that work.
So the useful tool is the pair, not either number alone. Plot kinetic energy against binding energy and each compound becomes a point; that is the Wagner (chemical state) plot, and it is how the arXiv paper and Biesinger's review both present copper data. Metal and Cu2O separate vertically, CuO separates horizontally, and charging moves a point along a fixed diagonal instead of scattering it.
Two more cautions. The Cu LMM line is broad and has structure, so its "position" depends on how you fit it; Biesinger's 2017 paper is substantially about curve-fitting that Auger region. And the database values come from different instruments and calibrations (the NIST records list them: one-point corrections against gold and copper, a dedicated calibration study, and so on). Compare your α′ against reference values measured carefully, preferably your own standards on your own instrument, not against a single number from a table.
My own published research is in microwave spectroscopy, not XPS, and none of the data here is mine. But the habit this teaches carries across every kind of spectroscopy I have worked with: a single peak position is a measurement plus a reference plus an instrument state. When two of those three can drift, the most valuable thing you can do is find a quantity in which the drift cancels. In XPS, for copper, that quantity exists, and it comes out of the same measurement as the peak you were already reading.
If a paper assigns copper as metallic or Cu(I) from the Cu 2p3/2 position alone, it has not shown which. The two can sit at the same 932.3 eV. Ask for the Cu LMM kinetic energy and the modified Auger parameter, which separates them by about 2 eV and is immune to sample charging. Then check the binding energy and satellites as well, because α′ in turn struggles to separate metal from Cu(II). One peak is a hint. Two peaks, read together, are an assignment.
I could not open the IUPAC Gold Book entry for the modified Auger parameter (the site blocks automated access), so the definition here is cited to the arXiv paper that states it. I did not read Biesinger's 2017 copper paper in full; I quote only its abstract. I did not open the original 1973, 1980 or 1985 papers; their values come from the NIST database records, which I read directly. The 3 eV charging example is my illustration of the arithmetic, not a reported measurement.
Scope note: the 3 eV charging example and the 0 to 0.4 eV versus 1.6 to 2.0 eV comparison are the author's arithmetic on the database values above.
U.S. Army combat veteran (Counter-IED / Electronic Warfare), peer-reviewed researcher in microwave spectroscopy, and founder & CEO of Shroombiosis. Consults on laboratory operations, AI, and supplement formulation.