Thermal vs Photochemical Control of Pericyclic Reactions

Photochemical pericyclic reactions governed by this HOMO/LUMO reversal are central to vitamin D biosynthesis: 7-dehydrocholesterol undergoes a UV-driven (photochemical, conrotatory-type) ring-opening electrocyclic reaction in skin to give previtamin D3, a transformation whose stereochemical course is dictated by exactly the node-counting logic developed here, and which would proceed with the opposite stereochemical outcome if driven thermally instead of photochemically. The reverse process — thermal, disrotatory-type ring closures and [1,7]-sigmatropic shifts — completes the conversion to vitamin D3 in a subsequent thermal step, making the whole biosynthetic pathway a real-world demonstration of thermal/photochemical stereodivergence occurring inside living skin. In synthetic methodology, photochemical 4π-electrocyclizations are used deliberately to access ring stereochemistry unobtainable thermally — for example in the synthesis of substituted cyclobutenes as strained building blocks for subsequent thermal ring-opening (retro-electrocyclization) cascades used to generate reactive dienes in natural product total synthesis. The endiandric acid cascade (Nicolaou, 1982) strings together an $8\pi$ photochemical/thermal electrocyclization sequence followed by a $6\pi$ thermal electrocyclization and an intramolecular Diels–Alder step, using precisely the con/dis rules derived here to control which of several possible polycyclic diastereomers is formed at each stage — a single molecule undergoing four consecutive pericyclic steps whose combined stereochemical outcome was predicted, and then confirmed, entirely from orbital symmetry. Beyond synthesis, femtosecond time-resolved spectroscopy now directly observes the excited-state electron populating the former LUMO during a photopericyclic reaction, watching the con/disrotatory ring closure of model dienes and trienes unfold on picosecond timescales — turning what began as a symmetry argument on paper into a directly measurable dynamical event, and confirming that the frontier-orbital-reversal picture is not merely a bookkeeping device but a description of the actual electron dynamics.

Ask the tutor: How do these same con/disrotatory rules extend to sigmatropic hydrogen shifts, where there is no ring being formed at all?

Consider a linear conjugated polyene with $n$ contiguous $p$-orbitals, one on each sp²-hybridized carbon, all parallel and available for $\pi$-overlap. Solving the Hückel secular equations for such a chain gives $n$ molecular orbitals, denoted $\psi_1$ through $\psi_n$, ordered by increasing energy and increasing number of nodes: $\psi_1$ has zero internal nodes (all lobes same phase), $\psi_2$ has one node, and in general $\psi_k$ has $(k-1)$ internal nodes between adjacent atoms. This is exactly the same node-counting logic as a particle in a box — more nodes, higher energy, faster sign alternation along the chain. For a neutral polyene with $n$ $\pi$ electrons occupying $n/2$ orbitals (assume $n$ even for now), the HOMO is $\psi_{n/2}$. Its terminal lobes — the $p$-orbitals on the first and last carbon of the chain — have a definite relative phase relationship that depends on $n/2$: for $\psi_{n/2}$ with an even number of internal nodes the terminal lobes are symmetric (same phase, as if reflected through the molecular mirror plane), and with an odd number of internal nodes they are antisymmetric (opposite phase). This terminal-lobe phase relationship is the entire content of orbital-symmetry control for electrocyclic ring closure: to form the new $\sigma$-bond, the two terminal $p$-orbitals must rotate so that lobes of matching phase point toward each other and overlap constructively. There are exactly two ways two $p$-orbitals at the ends of a chain can rotate to bring lobes together: conrotatory, where both termini rotate in the same sense (both clockwise or both counterclockwise, when viewed along the chain axis), and disrotatory, where they rotate in opposite senses (one clockwise, one counterclockwise — equivalently, one rotates "up-and-back" while the other rotates "down-and-back"). Which of these two motions brings matching phases together depends only on whether the terminal lobes of the relevant orbital are symmetric or antisymmetric with respect to the molecule's local mirror plane. If the terminal lobes have the same phase on the same face (symmetric, "top lobe/top lobe" already matching), disrotatory closure connects them without any sign mismatch. If the terminal lobes have opposite phase (antisymmetric), a conrotatory motion is required so that the top lobe of one carbon rotates to meet the bottom lobe of the other, which — because that lobe is the opposite-phase lobe — now matches.

Butadiene, the four-carbon case shown above, is the textbook example: the ground-state HOMO is $\psi_2$, with one internal node, and its terminal lobes are antisymmetric — top-phase on C1, bottom-phase (opposite) on C4 on the same molecular face. Closing this to cyclobutene thermally therefore requires conrotatory rotation. Absorb a photon and promote an electron from $\psi_2$ to $\psi_3$: the new singly-occupied orbitals are $\psi_2$ (now half-filled) and $\psi_3$ (now half-filled), and the orbital that has become the highest singly occupied orbital, $\psi_3$, has terminal lobes that are symmetric. Ring closure from this excited state is therefore disrotatory. This is why photochemical electrocyclic ring closures and openings run opposite to their thermal counterparts, and it is exactly why the classic hexatriene/cyclohexadiene interconversion, worked out in detail below, reverses stereochemistry under UV light. The same logic, applied to the pair of $\pi$ systems that come together in a cycloaddition (rather than the two ends of one chain), gives the suprafacial/antarafacial rules for reactions such as the Diels–Alder cycloaddition. There the diene HOMO and the dienophile LUMO (thermal case) or an excited-state configuration (photochemical case) must overlap constructively across both new $\sigma$ bonds simultaneously; whether "both bonds form on the same face of each component" (suprafacial) or "opposite faces" (antarafacial) is allowed follows the identical node-counting argument.

The controlling relationship is a node-parity statement. For a linear polyene with $n$ $\pi$-centers, orbital $\psi_k$ has $(k-1)$ internal nodes, and its terminal-lobe symmetry alternates with $k$:

where: $k$ is the orbital index (1 = lowest energy), "symmetric" means the two terminal $p$-lobes on the same molecular face have the same sign, and "antisymmetric" means they have opposite sign. For a neutral polyene with $n$ π electrons ($n$ even), the thermal HOMO is $\psi_{n/2}$ and the photochemical controlling orbital (former LUMO, now singly occupied) is $\psi_{n/2+1}$:

where: $n$ is the total number of π electrons in the conjugated chain undergoing electrocyclization. Because these two indices differ by exactly 1, and terminal-lobe symmetry alternates with every unit change in $k$, thermal and photochemical control are always opposite:

where: parity refers to even/odd-ness of the index, which by the rule above fixes symmetric/antisymmetric terminal lobes. The stereochemical consequence is then read off directly:

where: "allowed" means the pathway proceeds via a favorable, node-preserving overlap in a concerted, synchronous transition state, at an energy accessible under the given conditions (thermal ground state or photochemical excited state). Finally, this whole apparatus is equivalent to (and historically was reformulated as) an aromaticity count in the cyclic transition state — the Möbius–Hückel method: a transition state with $4n+2$ electrons and zero (or an even number of) phase inversions in the interacting-orbital array is aromatic and thermally allowed, while a transition state with $4n$ electrons requires one phase inversion (a Möbius array) to be allowed. Photochemical excitation adds/effectively changes the electron count and orbital symmetry so that the roles trade:

where: the subscript $s$ denotes "Hückel" (suprafacial, zero sign inversions around the cyclic array), and the counts refer to the total number of electrons in the cyclic transition-state array.

Example 1 (foundational) — butadiene ↔ cyclobutene, counting electrons and nodes explicitly. Butadiene has $n=4$ π electrons. Thermal HOMO index: $n/2 = 2$, so $\psi_2$ controls; $k=2$ is even, so by the parity rule terminal lobes are antisymmetric → conrotatory thermal closure/opening. Photochemical SOMO index: $n/2+1 = 3$, so $\psi_3$ controls; $k=3$ is odd, so terminal lobes are symmetric → disrotatory photochemical closure/opening. This is the textbook base case, and every other electrocyclic system is analyzed by the identical two-line calculation — only $n$ changes.

Example 2 (applied) — hexatriene ↔ cyclohexadiene, and why the mnemonic table works. (2E,4Z,6E)-octatriene-type systems (a hexatriene π framework, 6 π electrons, $n=6$) close to cyclohexadienes. Thermal HOMO index: $n/2 = 3$; $k=3$ is odd → terminal lobes symmetric → disrotatory thermally. Photochemical SOMO index: $n/2+1 = 4$; $k=4$ is even → terminal lobes antisymmetric → conrotatory photochemically. Concretely: thermal, disrotatory ring closure of (2E,4Z,6E)-2,4,6-octatriene brings the two terminal methyl groups, which point "outward" in the extended $s$-cis,$s$-cis conformer, together via a motion where both ends rotate so their substituent-bearing faces rotate toward each other on the same side — the two methyls end up cis on the newly formed sp³ centers of cis-5,6-dimethyl-1,3-cyclohexadiene. Under UV light, the same starting triene closes conrotatory, and now one methyl group must rotate "up" while the other rotates "down" relative to the ring plane, delivering the trans diastereomer, trans-5,6-dimethyl-1,3-cyclohexadiene. The stereochemical labels (cis/trans product) are a direct geometric readout of the con/disrotatory motion, which is itself a direct readout of the HOMO or photochemical-SOMO node count — nothing is assumed. Notice the alternation with $n$: $n=4$ gives thermal-conrotatory, $n=6$ gives thermal-disrotatory — this reflects the fact that $\psi_{n/2}$'s parity alternates as $n$ increases by 2 (index increases by 1 each time), which is a genuine, checkable limiting-case pattern rather than a coincidence.

Example 3 (exam/competition level) — predicting product stereochemistry for a triene with defined alkene geometry, including a trap. Consider (2E,4Z,6E)-3,4-dimethylocta-2,4,6-triene-type substrate (methyls at the internal 3,4-positions rather than the termini — a common trap since substituent position, not just count, must be tracked through the rotation). The reacting π system is still the same 6-electron hexatriene framework ($n=6$, since the electrocyclization uses C1–C6 termini and the C1–C2–C3–C4–C5–C6 backbone); the thermal/photochemical con-/disrotatory assignment from Example 2 is unchanged (thermal disrotatory, photochemical conrotatory) because that assignment depends only on the number of π electrons in the chain undergoing ring closure, not on where substituents sit. What changes is which specific diastereomer forms, because now the substituents at C3–C4 remain sp² alkene substituents in the product (they are not at the termini being rotated) while the terminal substituents at C1 and C6 (here, say, H and a vinyl or methyl group) are what get delivered cis or trans by the rotation. Under thermal conditions: disrotatory closure of the $E,Z,E$ geometry places the two terminal substituents that were cis to each other across the triene (as dictated by the $Z$ central double bond) onto the same face of the ring — giving the cis-disubstituted cyclohexadiene at C5–C6. Under photochemical conditions, conrotatory closure of the identical starting geometry gives the trans-disubstituted product at C5–C6. The key exam trap is assuming that "more substituents" changes the con/disrotatory assignment — it never does; only the total π-electron count $n$ in the pericyclic array sets con/dis, while alkene geometry ($E/Z$) sets which specific enantiomer/diastereomer results from a given rotational sense.

As a limiting-case check: setting $n=4$ in the general "$4n$ electrons" row recovers exactly the butadiene result above (conrotatory thermal), and setting $n=1$ in the "$4n+2$" row ($4(1)+2=6$) recovers the hexatriene result (disrotatory thermal) — the general formulas correctly reduce to the two worked base cases, confirming the alternation is not an artifact of small-$n$ bookkeeping.