A work sample from a physics and catalysis project: an account of a collaboration, written so a reader can judge the level of the work. It is not peer-reviewed. The 2014 paper is cited below as published literature. Who sent the data and gave the review is not identified on this page, and permission to identify her is pending as of 2026-10-22; I will say who she is only if she agrees. No table values are shown, and the experimental numbers stay with her until she agrees otherwise. The companion chapter, Chapter 5, runs the decision rule on invented numbers. This page is checked by book26/ch06-verify.py.
The work this builds on is a published comparison of two zeolites, HZSM-5 and HMCM-22, converting ethanol to hydrocarbons: Sousa et al., Catalysis Today 234, 2014. Its abstract reports that HZSM-5 favors propene with coke strongly inhibited, while HMCM-22 forms coke in larger channels and favors ethene. The paper itself is paywalled and was not read in full by me; conditions and characterization figures come from a collaborator’s summary of it.
The law being tested is in my paper, “The Self-Trapping Selectivity Principle”, preprint V6 (doi:10.5281/zenodo.22929142), also on its own page: one pore-size law with a single self-trapping radius r* explains both materials. If so, a ratio R built from each material’s selectivity and pore radius should equal 1. A collaborator sent product distributions versus time on stream for both materials. I computed R with every definition of selectivity and pore dimension I tried (ethene fraction, ethene against propene, and three choices of dimension for HMCM-22). R was never close to 1. It ran from about 1.4 (ethene fraction, cavity size, earliest time) to about 13 (smallest channel dimension, latest time), and grew with time on stream when compared at the same times. The direction agrees with the paper (more ethene in the larger-cavity material).
That is not a test. I looked at the numbers before fixing the definitions, so I could have found a definition that fit, and the fact that none did is informative but not decisive. No definition I tried makes the law fit, which is a reason to doubt it as stated. It is not a verdict: pass or fail is not called at this stage, and the comparison also assumes both materials differ only in pore size, which she pointed out is not so (their acid properties differ). There are no uncertainties on the figures, and the transcription of two table images was by hand, with the assignment of images to catalysts and two rows that do not sum to 100 still to be confirmed with her.
She said plainly that the mathematics is outside her field. She reviewed how the experiment and the data are defined; she has not reviewed or endorsed the mathematics or the law, which are mine alone. She gave seven specific points, which I paraphrase. Most of them are ordinary practice in experimental catalysis, as I read the zeolite literature; none is exotic. What they changed is this project: the plan went from “compute and see” to “define first, then test”, and two points were specific to this paper (the premise sentence, and the assumption of the same acid chemistry).
She also said the paper’s premise sentence overstated things: the two zeolites do not have “nominally identical acid chemistry” but similar nominal Si/Al ratios with different acid properties and pore structures. She pointed out a mislabeled line in a 2016 manuscript, asked that two experimental conditions be kept as separate datasets, and said her advisor must be a co-author on anything that uses her thesis data. Each of these is now in the plan. She also asked that the experimental paper and the mathematical paper be kept separate. I read that as a way to find the right audience for each, not as a retreat from either: the experimental work goes to catalysis readers without the mathematics, and the law stays in my own paper. A draft of the experimental paper, written for Catalysis Science & Technology (Royal Society of Chemistry) on its free subscription route, now exists. She has not yet reviewed it, it has not been submitted, and whether it is accepted is not known. The proposed test fixes definitions and a decision rule on paper first, as described in Chapter 5.