Hello Vince!

You have an interesting adventure before you. I am curious what you mean by problems. Are you asking about finding values that are inaccurate or ambiguous or missing? IOW, fundamental problems existing in the database records. Or are you asking about not be able to find values because the site itself is cumbersome to use?

I do not use the database enough to recognize any of the former issues. I would however hope that folks would use the citation links as a way to follow up on such issues. OTOH, when I do use the database, I find its implementation cumbersome.

Hope this gives a reference for your report.

• Hello Jeff,
• I just submitted my review to JESRP titled "XPS in Industry - Problems in BEs".
• There are various problems that I discuss. The sources of the problems are several. The problems persist today.
• I suspect that experienced XPS scientists tried to use the various numerical data-banks that exist, found uncertainties that are too large and gave up to use various numerical data-banks, including NIST. But young XPS users don't know the problems and reviewers are not being diligent.
• Dr Wagner built the first numerical data-bank based on peer-reviewed literature and used his best judgement in listing the BEs. He gave his collection to PHI and then NIST. He probably never expected future users to believe those BEs were accurate or very reliable. He made very few of the BEs in that collection.
• The BEs in his collection are from hundreds of different instruments, using many different calibration energies, using instruments that have not been calibrated for months or years, using various "assumed" C 1s BEs for the hydrocarbon moiety, and graduate students who did not care or did not know how to check calibration energies. On top of that most publications in those numerical data-banks do NOT provide a high BE calibration energy (e.g. Cu 2p3/2).
• Unfortunately different instrument makers used different calibration energies, different energy scales and promoted different C 1s BEs which authors blindly accepted and used and continue to use. Some authors did not know or care that the high BE end of different instruments vary by as much as 0.6 eV (i.e. Cu 2p3 at 932.2 vs 932.8 eV). Chemical shifts for a AuCu alloy series has shifts that are
  1 votes 1 thanks

Your history and background report on the NIST database is informative. I certainly hope it will get published and referenced accordingly. Could it even go so far as to spur NIST to redesign the database? I have to think perhaps along the lines of implementing ISO guidelines to control how peak values are posted. But then, is this not what Surf Sci Spectra is doing? Has this journal essentially overtaken the NIST database as a more reliable authority on XPS spectra?

I've always respected your long-standing diligence to be both accurate and precise in your reports of XPS information. It is an approach near and dear to my own in some regards. I'm glad to see that you have started this effort, and I'd be curious to know whether I could support it in some way beyond this thread in this forum.

Finally, the number of posts in this forum that ask about the basics of XPS--questions that demonstrate levels of understanding from total ignorance to serious confusion (and I'd say the former type dominate)--support your final statements. As a colleague of mine has said, history seems to be repeating itself in the world of materials characterization. Researchers seem to have to be relearning lessons that we learned a few decades ago. You would think by now those lessons were well established. This provides opportunities for reports such as yours that can correct the misconceptions and abuses that have grown out of neglect. It also gives rise to frustration as you note.

Best regards in your efforts. I'd be pleased to hear when the report is published.

You are very kind. Thank you. 

I urged Cedric to add histograms or weighted means to SRD-20, but my words fell on deaf ears.  And, Cedric has carefully crafted the definitions in the database such that I doubt any improvements are forthcoming. Sadly a government standards system should be the vanguard of highly reliable data. 

SSS was supposed to be the guardian of reliability, but due to the great lack of contributions, design issues, and the difficulty in getting authors to measure calibration energies on the same day as the material of interest, SSS relinquished to accept contributions that used various calibration energies and various C 1s BEs, and did not required same day calibration.  They also did not require Cu 2p3/2 BE for all pubs.  They also allowed many to contribute just 2 calibration energies, e.g. Au and Ag, both low BE values.

And to my chagrin, SSS does not supply C 1s spectra unless it is an intergral part of the chemical.

Today, new people control SSS.  The new data-sets have the same problems, but some of the contributing editors are contributing more calibration energies.

Bottomline, SSS is not much better than NIST.

Dear Prof. Crist,

In my humble opinion, unfortunately this Database is old - it has no fresh XPS Data...

XPS systems had been changed significantly during last decade so some included into NIST results seems doubtful, because it is not refreshing with actual XPS results.

Kindest wishes,

Dmitry

this article/post three existing labels 'writing, xps, databases' is most inefficient to group this post with other relevant research posts. I wonder what's new in XPS hardware systems? and how new hardware components affect ontology of their descriptive analysis or interpretive results, and how theoretical hypothesis on what to look for also impact data... ?

Dear Vince,

I think the NIST database actually does more harm than good to the XPS community, especially to the younger generation. With the exception for metals majority of reported BE values exhibit unacceptably large spread, often 1 eV or more.

As you write improper instrument calibration is one of the factors responsible. I would also add insufficient sample characterization by techniques complementary to XPS. First thing that comes to my mind is the determination of sample composition. How many of the compound samples included in the data base have been characterized by EDX or ERDA for major elements, as well as for contamination? I would guess a few percent in the best case.

Second thing is how the analyzed surface was treated prior to XPS analysis. Often used Ar+ etch obviously affects the top surface layers which contribute most to the signal intensity. Lighter elements sputter away faster which impacts the composition of the analyzed volume.

To summarize my point: I think that in majority of cases reported BE values are not representative of the claimed compound.

Dear Dr. Greczynski,

You are correct on all counts. NIST SRD 20 is in many ways a problem, but to some extent it is useful as a simple guide, not as definitive results.

Sadly EDS (EDX) has its' own set of problems with sensitivity factors which are increased now that large area detectors are common.

Yes, Argon ion etching, if done excessively, does more harm than good. But beginners do not know this. It is clear that universities are not able to provide the necessary training to help XPS users to perform reliable measurements. Why? Well, there are various reasons.

Unfortunately, many users think XPS is a mature and reliable method. I wish that were true.

Let's work together to improve things. Yes?

Vince

I agree!