Every chapter so far has run forwards: here is a substrate, here is a reagent, what is the product. Synthesis runs the other way. You are given a target and asked how to build it, and the number of possible routes is far too large to search by trying reactions and seeing where they lead.
The technique that makes the problem tractable is retrosynthesis: start at the target and work backwards, one bond at a time, toward something you can buy.
The notation, and why it is different
A retrosynthetic step is written with a double-lined open arrow, ⇒, and it means "could be made from" — not "reacts to give". It is deliberately a different arrow from the ordinary reaction arrow, because it points in the opposite direction in time.
Breaking a bond in that backwards direction is a disconnection. It is a thought, not a reaction: nothing in a flask breaks the bond you just crossed out. You are asking which two pieces, joined by a reaction you know, would give this bond.
Synthons and their synthetic equivalents
When you disconnect a bond you get two idealized fragments, usually charged, called synthons. Disconnecting the C–C bond of a secondary alcohol next to the carbinol carbon gives, say, a carbanion R⁻ and an aldehyde.
R⁻ is not a bottle on a shelf. The reagent that behaves like it is a synthetic equivalent — here, a Grignard reagent RMgBr. The synthon is the idea; the synthetic equivalent is what you actually order.
| Synthon | Synthetic equivalent |
|---|---|
| R⁻ (carbanion) | RMgBr, RLi |
| R–C≡C⁻ | terminal alkyne + strong base (NaNH₂) |
| ⁻CN | NaCN |
| ⁻CH₂–CO–R (enolate) | the ketone + base |
| R⁺ (carbocation) | alkyl halide, or alcohol + acid |
| R–CO⁺ (acylium) | acyl chloride + AlCl₃ |
Notice the pattern: nucleophilic synthons are things that attack and electrophilic ones are things that get attacked. Every disconnection produces one of each, because that is what a polar bond-forming reaction is.
Where to cut
Not all disconnections are equally useful. The ones worth trying first:
- Next to a functional group. Functional groups are where reactivity lives, so bonds near them are the ones you have reactions for. A C–C bond to a carbinol carbon suggests a Grignard; a C–C bond β to a carbonyl suggests an aldol.
- At a branch point. Branches are where a molecule was assembled, and disconnecting one usually splits the target into two comparably sized, simpler pieces.
- At a ring, if the ring is six-membered and has an alkene in it. That is the retro-Diels–Alder pattern, and it turns one ring into two open-chain pieces in a single step.
- Symmetrically. If the target is symmetric, a disconnection through the middle halves the problem.
A structural pattern that signals a particular disconnection is called a retron. A β-hydroxy ketone is the retron for an aldol; a cyclohexene is the retron for a Diels–Alder; a 1,3-dicarbonyl is the retron for a Claisen. Learning to see retrons is most of what makes an experienced chemist fast at this.
Target: 2-phenylbutan-2-ol, PhC(CH₃)(OH)CH₂CH₃.
Spot the retron. A tertiary alcohol with three different groups on the carbinol carbon. That is the Grignard retron: a C–C bond to a carbinol carbon can be disconnected.
Disconnect. Cut the bond to the ethyl group. The synthons are CH₃CH₂⁻ and the ketone PhCOCH₃ (acetophenone).
Name the equivalents. CH₃CH₂MgBr and acetophenone — both commercially available, so the route is one step.
Check forwards. CH₃CH₂MgBr attacks the ketone carbonyl; aqueous workup protonates the alkoxide; product is 2-phenylbutan-2-ol. ✓
Note there were three possible disconnections at that carbinol carbon — the ethyl, the methyl, or the phenyl. All three are legitimate Grignard disconnections, and the one to prefer is whichever leaves the most available starting materials. Having several valid answers is normal in synthesis, and a question asking for "a" synthesis rather than "the" synthesis is telling you so.
Knowing when to stop
You keep disconnecting until every piece is something you can reasonably buy: simple alkyl halides, small alcohols, benzene and its common derivatives, acetylene, simple ketones and esters. A question usually tells you the permitted starting materials, and that constraint is the real definition of "done".
The other stopping rule is practical. Each step loses material, so a route of ten steps at 80% each returns about 11% overall. Fewer steps is genuinely better, and a shorter route with a slightly awkward step usually beats a longer elegant one.
What carries forward
Retrosynthesis is a way of using everything else in the course rather than a new body of facts. The next three sections supply what it draws on: how to change one functional group into another, the short list of ways to make a carbon–carbon bond, and what to do when a reagent cannot tell two groups apart.