p. 807 “Most chemists think about bonding improperly” – What an opening salvo for this Chapter — “Advanced Concepts in Electronic Structure Theory”. I think A&D’s reasons for this are correct (at least for me). They can be found on p. 813 (see the note) and p. 838 (ditto)

p. 808 — “These wavefunctions contain al the observable information about the system.” – A huge assumption, and in fact a postulate of quantum mechanics. OK, actually, since QM has never made an incorrect prediction.

p. 809 — “In classical mechanics, the forces on a system create two kinds of energy — kinetic and potential”. Hmm. How does force ‘create’ energy? It does so by doing work. Work is force * distance, and if you do a dimensional analysis, you find that force * distance has the dimensions of kinetic energy (mass * velocity^2) — It’s worth working through this yourself, if you’ve been away from physics for a while. Recall that potential energy is the general name for energy which has to do with location relative to something else.

After reading Lawrence Krause’s biography of Feynmann (which goes much more into the actual physics than other biographies including Gleick’s), I cracked open the 3 volumes of the Feynmann lectures on physics and have begun reading. It’s amazing how uncannily accurate his speculations were. particularly about things which weren’t known in the 60’s but which are known now.

Feynman lists 9 types of energy (lecture 4 page 2)

l. gravitational (a type of potential energy)

2. kinetic

3. heat

4. elastic

5. electric

6. chemical

7. radiant (??)

8. nuclear

9. mass

He says that we really don’t know what energy is (even though we know 9 forms in which it appears) just that it’s conserved. Even so, the conservation law allows all sorts of physics problems to be solved. To really get into why energy is conserved, you have to read about Noether’s theorem — which I’m about to do, using a book called “Emmy Noether’s Wonderful Theorem” by Dwight E. Neuenschwander.

Later (Lecture 4 page 4) Feynmann defines potential energy as the general name of energy which has to do with location relative to something else.

p. 809 — The QM course I audited 2 years ago, noted that the Schrodinger equation really can’t be derived, but is used because it works. However, the prof then proceeded to give us a nice pseudo-derivation based on the standard equation for a wave propagating in space and time, Einsteins E = h * nu, and De Broglie’s p = h/lambda, and differenating the wave equation twice with respect to position, then twice with respect to time and equating what he got.

However, to get the usual Hamiltonian, he had to arbitrarily throw in a term for potential energy (because it works).

p. 810 “The energy E is simply a number” – should have said “The energy E is simply a **real** number” which is exactly why the complex conjugate must be used. If you *really* want to know what’s going on see — the 10 articles in the category == Linear Algebra Survival Guide for QM.

p. 811 – 812 — The qualitative discussion of the Laplacian is great — it also explains why higher frequency light has higher energy. Worth the price of the book. Localizing an electron increases its energy by the Heisenberg uncertainty principle, which was the reasoning I’d grown up with.

One point for the unitiated (into the mysteries of quantum mechanics) to consider. “The more nodes an orbital has, the higher is its energy. Recall from Chapter 1 that nodes are points of zero electron density, where the wavefunction changes sign.” Well, a point of zero electron density, or a point at which the wavefunction equals zero, means the electron is never (bold) found there. So why is the probability of finding an electron on both sides of the node not zero. You need to abandon the notion that an electron has a trajectory within an atom. Having done so, what does angular momentum mean in quantum mechanics?

p. 813 — Interesting that the electrostatic arguments for bonding (shielding nuclei frm each other, etc. etc. which I’ve heard a zillion times) are incorrect. This probably explains the opening salvo of the chapter. (See also the note on p. 838).

p. 814 — “This is the fundamental reason that a bond forms; the kinetic energy of the electrons in the bonding region is lower than the kinetic energy of the electrons in isolated atomic orbits” – this is because the wave function amplitude changes less between the nuclei. However, since we’ve had to abandon the notion of a trajectory — what does kinetic energy actually mean in the quantum mechanical situation (see note on pp. 811 – 812).

p. 814 — “The greater the overlap between to orbitals the lower the kinetic energy” – to really see this you have to look at figure 14.6 on p. 813 — The greater the overlap, the shallower the depression of the wave function amplitude between the nuclei, which implies less change in amplitude with distance, whch implies a smaller Laplacian (a second derivative) and less kinetic energy for the electrons here. So this is why overlapping atomic orbitals result in lower kinetic electron energy at sites of overlap – e. g. why bonds form (bold). Great stuff ! ! ! !

Continuing on, the next paragraph explains where Morse potentials (p. 422) come from, and why populating antisymmetric orbitals causes repulsion (the change in orbital sign increases the Lagrangian greatly along with it the electron kinetic energy, despite the fact the the potential energy of the antisymmetric orbitals is favorable for keeping the atoms close — e.g. bonding).

p. 815 — What does the Born Oppenheimer approximation (which keeps internuclear distances fixed) do to the calculation of vibrational energies — which depend on nuclear motion? The way the energy of the solutions of the Schrodinger Equation using the BO approximation is gradually approached (moving the nuclei around and calculating energy) clearly won’t work for CnH2n+2 with n > 2. There will be more than a single minimum. What about a small protein? Clearly these situations the Born Oppenheimer approximation is hopeless. Because of the difficulty in understanding A&Ds discussion of the secular equation (see comments on p. 828), I’ve taken to reading other books (which have the advantage of devoting hundreds of pages to A&Ds 60 to computational chemistry), notably Cramer’s “Essentials of Computational Chemistry” – He notes that lacking the Born Oppenheimer approximation, the concept of the potential energy surface vanishes.

p. 816 – 817 — Antisymmetric wave functions and Slater determinants are interesting ways to look at the Pauli exclusion principle. The Slater determinant is basically a linear combination of orbitals — why is this allowed? – because the orbitals are the solution of a differential equation, and the differential of a sum of functions is the same as the sum of differentials of the functions and orbitals are the solution to a differential equation.

p. 823 — What’s a diffuse function? Also polarization orbitals strike me as a gigantic kludge. I suppose the proof of the pudding is in the prediction of energy levels, but there appear to be an awful lot of adjustible parameters lurking about.

p. 824 — “the motions of the electrons are pairwise correlated to keep the electrons apart” — but electrons don’t really have trajectories — see note on pp. 811 – 812. I got all this stuff from Mark P. Silverman “And Yet It Moves” Chapter 3

p. 824 — “the c(i)’s are incrementally changed until capital Psi approximates a correlated wavefunction.” More kludges.

p. 825 — Nice to see why electron correlation is required, if you want to study Van der Waals forces between molecules, The correlation energy could be considered a intramolecular Van der Waals force.

What is a single point energy? — I couldn’t find it in the index.

p. 826 — The descriptions of the successive kludges required for the ab initio approach to orbitals are rather depressing. However, there’s no way around it. You are basically trying to solve a many body problem when you solve the Schrodinger equation. It’s time to remember what a former editor of Nature (John Gribbin) said “*It’s important to appreciate, though, that the lack of solutions to the three-body problem is not caused by our human deficiencies as mathematicians; it is built into the laws of mathematics* “

p. 828 — I was beginning not to take all this stuff seriously, until I found that the Hartree Fock approach produces results agreeing with experiment. However, given the zillions of adjustable parameters involved in getting to any one energy, it better produce good results for n^2 molecules, where n is the number of adjustable parameters. Fortunately, organic chemistry can provide well over n^2 molecules with n carbons and 2n+2 hydrogens.

p. 828 — The discussion of secular determinants and the equations leading to them is opaque (to me). So I had to look it up in another book “Essentials of Computational Chemistry” by Christopher J. Cramer (2nd Edition). Clear as a bell (although, to be fair, I read it after slowly going through 20 pages of A&D), and done in 10+ pages (105 –> 115).

Along these lines, how did the secular equation get its name. Is there a religious equation?

What can you do with an approximate wavefunction produced by any of these methods. The discussion in A&D so far is all about energy levels. However, unlike wavefunctions, operators on wavefunctions are completely known, so you can use them to calculate other properties (Cramer doesn’t give an example).

p. 830 – 831 — Even so, given the solutions of the secular equation for a very simple case, you see why the energy of a bonding orbital is less than two separate atomic orbital — call the amount B (for bonding). More importantly the energy of the antibonding oribtal his greater than the two separate atomic orbitals by a greater amount than B — explaining why filling a both a bonding and the corresponding antibonding orbital results in repulsion. This is rule #8 the rules of Qualitative Molecular Orbital theory (p. 28).

p. 834 — An acronym festival — CNDO, INDO, PNDO, MINDO 1- 3, AMI, PM3 etc. at least they tell you that they aren’t much used any more.

It’s amazing that Huckel theory works at all, ignoring as it does electron electron repulsion.

p. 836 — If everything in Density Functional Theory (DFT) depends on the electron density — how do you ever find it? Isn’t this what the wavefunctions which are the solutions to the Schrodinger equation actually are? I’m missing something here and will have to dig into Cramer again.

p. 838 — Most energy diagrams of molecular orbitals made from two identical atomic orbitals have the bonding and antibonding orbitals have them symmetrically disposed lower and higher than the atomic orbitals. This is from use of the Huckel approximation, which simply ignores overlap integrals. The truth is shown in the diagram on p. 831.

p. 838 — Another statement worth the price of the book — the sigma amd pi orbitals are of opposite symmetry (different symmetry) and so the sigma and pi orbitals don’t mix. The sigma electrons provide part of the potential field experienced by the pi electrons.

p. 839 — Spectacular to see how well Huckel Molecular Orbital Theory works for fulvene — even if the bonding and antibonding orbitals* are* symmetrically disposed.

p. 841 — Fig. 14.13 — I don’t understand what is going on in part B in the two diagrams where not all the atoms have an associated p orbital.

p. 843 — With all these nice energy level diagrams, presumably spectroscopy has been able to determine the difference between them, and see how well the Huckel theory fits the pattern of energy levels (if not the absolute values).

p. 846 — Table 1.1 should be table 1.4 (I think)

p. 849 — How in the world was the bridged [10] annulene made?

p. 853 — Why is planar tetracoordinate carbon a goal of physical organic chemistry? The power of the molecular orbital approach is evident — put the C’s and H’s where you want in space, and watch what happens — sort of like making a chemical unicorn. Why not methane with 3 hydrogens in an equilateral triangle, the carbon in the center of the triangular plane, and the fourth hydrogen perpendicular to the central carbon?

Dilithiocycloproane has been made — presumably as a test bed for MO calculations, being able to predict something is always more convincing of a theory’s validity than justifying something you know to be true.

p. 854 — What are the observations supporting the A < S ordering of molecular orbitals in 1, 4 dihydrobenzene? The arguments to rationalize the unexpected strike me as talmudic, not that talmudic reasoning is bad, just that no one calls it scientific.

p. 856 — Good to see that one can calculate NMR chemical shifts using ab initio calculations (hopefully without tweaking parameters for each molecule). Bad to see that it is too complicated to go into here. More reading to do (next year) after Anslyn (probably Cramer), with a little help from two computational chemist friends.

p. 858 — How in the world did anyone ever make Coates’ ion?

p. 861 — Have the cyclobutanediyl and cyclpentanediyl radicals ever been made?

p. 863 — “Recall that the 3d orbitals are in the same row of the period(ic) table as the 4s and 4p orbitals’ — does anyone have an idea why this is so? Given the periodic table, the 4s orbitals fill before the 3d, which fill before the 4p (lowest energy orbital fill first presumably). The higher energy of the 4p than 3d may explain by d2sp3 orbitals are higher in energy than the leftover 3d orbitals — d(xy), d(yz) and d(zx). Is this correct? However the diagram in part B. of Figure 14.33 on p. 864 still shows that (n+1)s is of higher energy than nd, even though the periodic table would imply the opposite.

It’s not clear why the t(2g) combinations of d(xy), d(yz) and d(zx) orbitals don’t interact with the 6 ligand orbitals, since they are closer to them in energy. Presumably the geometry is wrong? Presumably d(z2) and d(x2 – y2) are used to hybridize with the p orbitals because they are oriented the same way, and d(xy), d(yz) and d(zx) are offset from the p orbitals by 45 degrees. Is this correct?

This is the downside of self-study. I’m sure a practicing transitional metal organic chemist could answer these questions quickly, but without these answers it’s back to the med school drill — that’s the way it is, and you’d best memorize it.

p. 865 — Frontier orbitals have only been defined for the Huckel approximation at this point — the index has them discussed in the next chapter. On p. 888 they are defined as HOMO and LUMO which have been well defined previously.

p. 866 — The isolobal work is fascinating, primarily because it allows you to predict (or at least rationalize) things.

This section does elaborate a bit on organometallic bonding details, lacking in chapter 12. However, no reactions are discussed, and electrons are not pushed. Perhaps later in the book, but I doubt it.

It’s clear from reading chapter 12 that organometallics have revolutionized synthesis in the past 50 years. I’ll need to read further next year, in order to reach the goal of enjoying new and clever syntheses as they come out.

Does anyone out there have any thoughts about Cramer’s book? Any recommendations for other computational chemistry books? I’ve clearly got to go farther.

Merry Christmas and Happy New Year