Chemistry · Lesson

Retrosynthetic thinking and synthetic planning

Work backward from a precisely specified target, then check the forward chemistry and material balance. Compare serial and convergent routes using separate branch inventories and explicit yield bases.

Back to the Chemistry map

By the end, you can

Distinguish a retrosynthetic disconnection from electron movement.
Check a proposed precursor relationship in the forward direction.
Calculate serial yield on one material basis.
Count a convergent route’s limiting branch, longest sequence and total operations separately.

Define the target before suggesting a route

A synthetic target includes connectivity, required configuration and the form of material needed. A molecular formula alone leaves many possible structures. Work backward by asking which immediate precursor relationships might produce the specified target, then ask how those precursors could be obtained. Retrosynthesis is a planning argument, rather than the physical process of running a reaction backward.

A planning disconnection selects a bond or structural feature to simplify. It is distinct from a curved electron-flow arrow, a resonance relation and an equilibrium arrow. The imagined fragments are useful structural roles; they are not automatically free ions or isolable bottles of reagent. A forward reaction model, compatibility evidence and practical verification must support any route proposal.

Trace the carbon frameworks across a proposed cut

For the original target CH₃CH(OH)CH₂CHO, 3-hydroxybutanal, consider cutting the C–C bond between CH(OH) and CH₂. This partitions the carbon skeleton into two fragments with two carbons each. The oxygen-bearing carbon in one fragment could be connected with a carbon donor role in the other through a selected carbonyl-addition relationship.

A symbolic aldol-addition net from two ethanal units conserves C₄H₈O₂ and supplies that connectivity. This checks one possible structural relationship. It does not establish a real isolated yield, the product’s configuration, competing reactions or the literal reagent forms represented by the backward cut. Forward bond changes and proton transfers still need their own ledger.

A serial yield product needs a common basis

For an invented four-step one-to-one route starting with 0.0100 mol limiting material, take isolated molar yields 0.900, 0.850, 0.800 and 0.750. Successive amounts are 0.00900, 0.00765, 0.00612 and 0.00459 mol. Multiplying all four yields gives 0.459, or 45.9% on that original material basis.

This multiplication assumes each reported stage yield uses the incoming material from the previous stage, with the stated stoichiometry and no recycling. A route with changing limiting reagents, branching, different stoichiometric coefficients or recovered material requires a more explicit inventory. The plotted stage index counts operations rather than elapsed time or a reaction coordinate.

Serial isolated amounts on one stated limiting-material basis

Retained amount / mmol

Completed operation count

Invented serial inventory

Underlying data

Completed operation count	Retained amount / mmol
0	10 · Invented serial inventory
1	9 · Invented serial inventory
2	7.65 · Invented serial inventory
3	6.12 · Invented serial inventory
4	4.59 · Invented serial inventory

Original discrete four-step teaching inventory with initial amount 10.0 mmol, one-to-one stoichiometry and stipulated stage yields 90.0%, 85.0%, 80.0% and 75.0%. Markers are retained amounts after each complete operation. The horizontal axis is operation count, not time; no continuous reaction trajectory or measured yield curve is implied.

Convergent branches must meet at a limiting inventory

In a separate invented network, branch A starts with 0.0100 mol and passes through yields 0.800 and 0.900, supplying 0.00720 mol A. Branch B starts independently with 0.0120 mol and has yield 0.750, supplying 0.00900 mol B. For a one-to-one A+B→T join, the smaller branch amount limits the maximum T capacity to 0.00720 mol. A stipulated join yield 0.900 gives 0.00648 mol T.

On the initial-A basis that is 64.8%. Multiplying B’s 75.0% yield into A’s serial yield would mix independent material bases. The separate B input remains part of the total resource inventory. The available branch amounts and join yield do not determine how much B is recovered or which practical losses occur.

A shorter longest sequence need not mean fewer operations

A four-operation linear route has longest sequence four. A convergent plan with two A-branch operations, one B-branch operation and one joining operation also has four transformations, but its longest sequence to T is three. The join has two incoming material connections representing one transformation. Counting those connections as two separate joins would overcount operations.

Longest sequence can describe route depth while total operations describe a different burden. Neither alone proves shorter elapsed time, lower cost, higher yield or lower environmental impact. Actual operation durations, compatible processing, purification, supply and resource use determine those comparisons.

Worked examples

Count carbons after a selected disconnection

Cut the CH(OH)–CH₂ bond in CH₃CH(OH)CH₂CHO.

One fragment contains CH₃ and CH(OH): two carbons.
The other contains CH₂ and CHO: two carbons.
Reconnecting these roles needs a valid forward model, not literal isolation of the cut fragments.

A two-carbon plus two-carbon framework proposal, with configuration and actual feasibility still unestablished.

Multiply four yields without changing the basis

Use initial 10.0 mmol and serial yields 0.900, 0.850, 0.800 and 0.750.

Overall fraction = 0.900 × 0.850 × 0.800 × 0.750 = 0.459.
Final amount = 10.0 × 0.459 = 4.59 mmol.

45.9% on the stated initial limiting-material basis.

Find the join’s limiting branch

Use the separate branches and 90.0% A+B→T join.

A available = 10.0 × 0.800 × 0.900 = 7.20 mmol.
B available = 12.0 × 0.750 = 9.00 mmol.
Join capacity = min(7.20,9.00) = 7.20 mmol.
T = 7.20 × 0.900 = 6.48 mmol.

6.48 mmol T in this stipulated inventory; 64.8% relative to initial A.

Count transformations rather than incoming edges

Compare four linear operations with two A operations, one B operation and a join.

Linear route: four transformations, depth four.
Convergent route: 2 + 1 + 1 = 4 transformations.
Its longest path has two branch operations plus the join: depth three.

Equal total operation counts, different longest sequence lengths. Timing and cost were not supplied.

Try it yourself

Is a backward planning disconnection a curved electron-flow arrow?
Show the answer and reasoning
No.
They express different relationships: precursor planning versus an electron-pair or single-electron ledger.
Can independent branch yields always be multiplied into one overall serial yield?
Show the answer and reasoning
No.
Each branch has its own material input, followed by a stoichiometric limiting inventory at the join.
Does a retrosynthetic charged fragment prove that a free ion exists as an isolable reagent?
Show the answer and reasoning
No.
It can represent a structural role whose actual reagent form needs separate justification.
Does a shorter longest sequence establish fewer total transformations?
Show the answer and reasoning
No.
The number of branch operations and joins must also be counted.

What to carry forward

Specify structure and required configuration first.
Use backward planning and forward feasibility together.
Keep serial yield on one named material basis.
Find the limiting inventory before a convergent join.
Count route depth and total operations separately.

References

Teaching resources and reference material supporting this lesson’s explanations and examples.