Retrosynthesis Step by Step: Spotting Retrons and Choosing the Strategic Disconnection

You have a target structure on the canvas — a substituted secondary alcohol, say, with an aryl ring and a branched chain hanging off the carbinol carbon — and the cursor is blinking while you try to decide where to break it. Every bond looks like a candidate. That paralysis is the real bottleneck in route planning, and it is exactly what a disciplined retrosynthesis procedure fixes. This guide walks through how to do retrosynthesis step by step: a repeatable decision procedure for finding the bond worth cutting, naming the fragments it produces, and translating those fragments into reagents you can actually order.

Most online “retrosynthesis tutorial” pages stop at the vocabulary — here is the retrosynthetic arrow, here is a synthon — and leave you to guess which of a dozen bonds to disconnect. The hard part is the choosing, and that is what gets the most attention below.

The vocabulary you need before step 1

Retrosynthetic analysis, the disconnection approach formalized by E.J. Corey (Nobel Prize in Chemistry, 1990), runs the synthesis in reverse. You take a target molecule (the TM) and work backward to simpler precursors. The notation is the retrosynthetic arrow, a double-lined open arrow written ⇒, which means “is made from” — not “reacts to give.”

Three terms do the heavy lifting:

• Disconnection — the imagined breaking of a bond in the target, the reverse of a real bond-forming reaction.
• Synthon — an idealized fragment, usually charged, that results from a disconnection. Synthons are bookkeeping devices, not bottles on the shelf.
• Synthetic equivalent — the real reagent that supplies a synthon’s reactivity. An acyl anion synthon, for example, has no stable bottled form; you reach for an umpolung reagent instead.

A retron is the structural pattern that signals a particular disconnection is available — the substructure a named reaction leaves behind. A β-hydroxy carbonyl is the retron for an aldol disconnection; an allylic alcohol or a defined alkene geometry points toward a Wittig. Learning to read retrons on sight is most of the skill. For the deeper treatment of why a fragment is a synthon and a reagent is its stand-in, see synthons vs synthetic equivalents.

How to do retrosynthesis step by step

The procedure below is the order experienced chemists actually run, even when they do it fast enough that it looks like intuition.

Step 1 — Map the target: number the skeleton and mark every functional group

Before cutting anything, inventory what you have. Number the carbon skeleton, then mark each functional group: hydroxyl, carbonyl, amine, halide, the heteroatom-bearing positions. Functional groups are where reactivity lives, and almost every productive disconnection sits adjacent to one. If you want a clean structure to annotate, paste your SMILES into the SMILES-to-structure converter and work from the rendered skeleton rather than a hand sketch.

Step 2 — Identify the retrons and the strategic bonds

Scan the map for retrons. A carbon bearing both a hydroxyl and, two carbons over, a carbonyl is an aldol retron. A carbon–carbon bond joining two aromatic rings, or an aryl to a vinyl, is a cross-coupling retron (Suzuki, for one). A heteroatom–carbonyl bond — an amide, an ester — is almost always a disconnection point because the forward reaction (acylation) is reliable.

Not all bonds are equal. A strategic bond is one whose disconnection simplifies the target the most — typically a bond that breaks the molecule into two pieces of comparable size, sits near the middle of the carbon framework, or lies at a ring fusion. Cutting a terminal methyl off the end buys you almost nothing; cutting the bond that splits a 14-carbon chain into a 7 and a 7 buys you a lot.

Step 3 — Make the disconnection and name the synthons

Draw the retrosynthetic arrow and split the target at your chosen bond. Assign charges to the two fragments based on which polarity the forward reaction would use. The carbon that was electrophilic in the forward sense becomes a cation synthon; the nucleophilic partner becomes an anion synthon. This is bookkeeping, but it is the step that tells you what kind of reagent you need next.

Step 4 — Translate synthons into synthetic equivalents

Now leave the idealized world and pick real reagents. A carbanion synthon adjacent to nothing stabilizing is supplied by an organometallic — a Grignard or organolithium. An enolate synthon is supplied by the parent carbonyl plus base. A cation synthon at a carbonyl carbon is just the aldehyde or ketone itself. When the natural polarity is wrong — when you need an acyl anion or a nucleophilic carbon where the structure says electrophile — you have to invert it, which is the domain of umpolung and reversed-polarity synthons.

Step 5 — Apply functional group interconversion where the disconnection won’t come

Sometimes no clean disconnection presents itself because the functional group in the target is not one a good bond-forming reaction produces directly. Functional group interconversion (FGI) is the move here: transform the group in your retrosynthetic notation into one that does open a disconnection. A primary amine that resists a clean C–N disconnection can be drawn back to a nitro group or a nitrile, either of which traces to a more obvious precursor. FGI does not simplify the skeleton; it changes the chemistry so a skeletal disconnection becomes possible.

Step 6 — Iterate back to buyable, simple precursors

Repeat steps 2 through 5 on each fragment. Stop when a fragment is a commercially available building block or a textbook one-step precursor. The endpoint is judgment: a chemist with catalog access stops at different points than a student working a problem set. There is no single correct stopping depth — stop when the next disconnection no longer buys real simplification.

Step 7 — Sanity-check the forward direction and selectivity

Flip each disconnection back to a forward reaction and ask whether it would actually run as drawn. Three checks matter most: chemoselectivity (will the reagent hit the bond you want and ignore the others?), the need for protecting groups (does a competing functional group have to be masked first?), and stereochemistry (does the forward step set the configuration the target requires, or give a mixture?). A disconnection that is elegant on paper but demands a protecting-group dance at every step may lose to a less clever route with fewer steps.

Where this procedure pays off — and where to lean on tooling

The seven steps generalize. For a β-hydroxy ketone the retron points to an aldol disconnection back to two carbonyl partners; for a 1,3-diene plus a dienophile pattern, to a Diels–Alder. The discipline is identical — map, find the retron, cut the strategic bond, name synthons, translate to reagents, iterate, sanity-check.

Choosing which of several valid disconnections to pursue is its own decision, driven by building-block availability, step count, and how many protecting groups each route forces. That trade-off is worth its own treatment — see the deeper guide on choosing the strategic disconnection.

For multi-step routes, software can surface disconnections you did not consider. ChemStitch includes an AI suggest_retrosynthesis tool whose output is labeled “AI Suggested” rather than computed — a deliberate distinction, because a generative suggestion is a starting point to verify against the literature, not a settled route. If you are weighing how far to trust generated chemistry, the post on whether AI reaction prediction is reliable covers the failure modes. For a broader look at the planning tools, see retrosynthesis software medicinal chemists use and reaction-mode synthesis planning.

Once a route is chosen and you move to the bench, the bookkeeping shifts from synthons to equivalents and limiting reagents. The post on building a reagent table covers that, and the stoichiometry calculator handles the reagent-equivalents math for each step.

The disconnection approach is established organic-chemistry method, not a recipe with one right answer. Treat any specific route — whether you drew it or a tool suggested it — as a hypothesis to confirm against primary literature before you order reagents.