Despite the arguments of Rioux and DeKock (1), I continue to maintain (2) that their statement that "interpreting the ionization of any atom or molecule requires quantum chemical tools and a consideration of both kinetic and potential energies" (3) is too restrictive. Carlton has also criticized this statement (4). Authors of general chemistry texts often interpret ionization energies as evidence for the shell structure of atoms. First ionization energies across a series of atoms are most often used for this purpose (which Rioux and DeKock appear to find acceptable), and successive ionization energies for a single atom are sometimes used also (5-7).

The point raised independently by Carlton and by me is that the virial theorem guarantees that the change in total energy for such a process involving atoms is equal to twice the change in potential energy and also to the negative of the change in kinetic energy. Thus, if we know the ionization energy, we also know the potential energy change accompanying the ionizing process as well as the kinetic energy change. We can use our knowledge of potential energy changes, inferred from measured ionization energies, to draw conclusions about the electronic structure of atoms. When we do this, the shell structure is revealed. However, for this approach to work, we must have some sort of relationship in mind that connects the data (ionization energies) and the structure. That relationship, which I have referred to as a "model", is the mediator between the data and the structure. As mediator, a model should work in both directions: It should allow one to infer the structure from the data, and it should allow one to predict the data from the structure. If a model is acceptable for use in one direction, it ought to be acceptable for use in the other direction.

The model I have been discussing is referred to in my paper as a potential-energy-only (PEO) model. One could also seek to focus on the kinetic energy changes, creating a kinetic-energy-only model (not yet done successfully, to my knowledge), or one could seek a model that deals directly with the total energy change. According to their statement quoted above, Rioux and DeKock are of the opinion that a PEO model is inappropriate. However, according to certain other statements they make, they seem to think it is appropriate. In their penultimate paragraph they make the comment, "Evidence for electronic shells and subshells is easily seen in a graph of the first ionization energies of the elements found in any general chemistry text." This statement, along with the subsequent sentences, indicates to me that Rioux and DeKock are themselves using a PEO model, perhaps without realizing it. How else does any of us get from a graph of ionization energies to a shell structure than by supposing that there is a relationship between the energy needed to extract an electron and the distance from the nucleus from which it starts its journey? Isn't that a potential energy model?

Rioux and DeKock raise a number of criticisms of the PEO model and, in the process, often refer to it as "Lowe's model". This is misleading. I certainly cannot lay claim to having discovered this model, and I am far from the first person to use it to infer shell structure from ionization energies. In my paper I cite related approaches going back to 1973. Interestingly, one of the earlier users of this model is DeKock himself, who coauthored a textbook (7) that uses a similar PEO model. Using the symbols defined above in Rioux and DeKock's response (1), the formula given by DeKock (7) for the ith ionization energy is I_i = i/r_i. Comparing this with the formula given above by Rioux and DeKock, we see that it is in error by a factor of two. Apparently the distinction between potential and total energy changes was overlooked when DeKock's textbook was written. Despite Rioux and DeKock's attacks on PEO models, I have so far not seen an attack on, or retraction of, the model used by DeKock himself, which is the only one I have seen in print that explicitly posits the relationship confusing total with potential energy that they find so objectionable. Because of this history, it would be more appropriate for them to label the model they are criticizing as "the corrected version of DeKock's PEO model" rather than "Lowe's model".

Of course, the factor of two doesn't lead to errors if one uses the PEO model only for purposes of comparison, as I did in my paper (2), and as DeKock himself does in his textbook. In his Problem 8 of Chapter 2, for example, he asks the student to calculate I_i /i for the 12 ionization energies of Mg (8). Thus, DeKock uses observed successive ionization energies to demonstrate patterns in r_i values that suggest shell structure. I described the alternative of estimating ionization energies from assumed shell structure. If one of these approaches is suitable for use in an introductory course, then so is the other. They both rely on the PEO model, but they simply use it in opposite directions.

Rioux and DeKock imply in their first paragraph that I have misled readers by not being up front in my discussion. I leave it to readers to examine my paper and draw their own conclusions. I put considerable effort into making the discussion clear, and I believe that I pointed out the approximate nature of the model and certain pedagogical advantages that accompany such approximate and (yes) simple models.

Turning to specific criticisms:

Rioux and DeKock claim that the model "considers only electron-nuclear potential energy and ignores electron-electron potential energy." This simply isn't true. If only electron-nuclear potential energy were included, then every electron would "see" the full nuclear charge, and the model would become I_i = Z/2r_i, where Z is the atomic number. This would be an absurd model, and would not even come close to reproducing trends in observed ionization energies.

Rioux and DeKock claim that known ionic radii are incompatible with the PEO model, but their data are not relevant for the claims they make. They claim that the radius of S⁴⁺ tells us something about the size of the 3p subshell, but this ion doesn't have any 3p electrons, and S⁶⁺ has no 3p or 3s electrons. Their comparisons of the radii of S⁴⁺ and S⁶⁺ come from data on coordinate complexes having different coordination numbers. Inspection of the table from which these data are taken (9) reveals that even the same ion (same charge) in complexes having different coordination numbers is tabulated with rather different radii.

Finally, Rioux and DeKock question why anyone would use I₂ instead of I₁ to predict I₃, or I₄ of sulfur. The answer, which I alluded to in my paper, is that I₃ and I₄ are processes where an electron is being removed from a half-filled p orbital. I₂ is similar in this regard, hence it makes a good basis for comparison. But I₁ is different, since, in that case, an electron is being removed from a filled orbital. My students seem to get it.

Literature Cited

1. Rioux, F.; DeKock, R. L. J. Chem. Educ. 2002, 79, 429.
2. Lowe, J. P. J. Chem. Educ. 2000, 77, 155-156.
3. Rioux, F.; DeKock, R. L. J. Chem. Educ. 1998, 75, 537-539.
4. Carlton, T. S. J. Chem. Educ. 1999, 76, 605, 606-607.
5. Moore, J. W.; Stanitski, C. L.; Wood, J. L. The Chemical World: Concepts and Applications, 2nd ed.; Saunders: New York, 1998; pp 337-340.
6. Brown, T. L.; LeMay, H. E.; Bursten, B. E. Chemistry, The Central Science, 8th ed.; Prentice Hall: Upper Saddle River, NJ, 2000; pp 233-236.
7. DeKock, R. L.; Gray, H. B. Chemical Structure and Bonding; University Science Books: Mill Valley, CA, 1989; pp 74-81.
8. DeKock, R. L.; Gray, H. B. Op. cit., p 129.
9. CRC Handbook of Chemistry and Physics, 80th ed.; Lide, D. R., Ed.; CRC: Boca Raton, FL, 1999; Section 12, p 15.