Retrosynthetic analysis, pioneered by E.J. Corey (Nobel Prize 1990), is a systematic method for planning organic syntheses. The chemist works backward from the target molecule, applying strategic disconnections to reveal simpler precursor structures. Each disconnection corresponds to a known chemical reaction in the forward direction. The process continues until all fragments are commercially available or readily prepared starting materials. Synthons are idealized fragments (often charged species) that represent the reactive form; synthetic equivalents are the actual reagents used in the laboratory.
Disconnections and Functional Group Interconversions
One-group C-X disconnections remove a functional group to simplify the molecule. For example, an alcohol can be traced back to a carbonyl compound through reduction; an alkyl chloride traces back to an alcohol through chlorination. Two-group C-X disconnections involve relationships between functional groups, such as 1,2-, 1,3-, 1,4-, and 1,5-difunctional patterns. The 1,3-dicarbonyl motif suggests a Claisen condensation disconnection; a 1,5-dicarbonyl suggests a Michael addition followed by aldol reaction. Functional group interconversions (FGI) transform one functional group into another without changing the carbon skeleton, providing alternative synthetic pathways.
Protecting Group Strategy
A protecting group temporarily masks a reactive functional group to prevent interference during a transformation at another site in the molecule. Alcohols are commonly protected as silyl ethers (TMS, TBS, TIPS) using the corresponding silyl chloride and imidazole; deprotection is achieved with fluoride (TBAF) or mild acid. Carbonyl groups are protected as acetals or ketals using ethylene glycol and p-toluenesulfonic acid; deprotection requires aqueous acid. Amines are protected as carbamates (Boc, Cbz, Fmoc) with orthogonal deprotection conditions. An ideal protecting group is introduced in high yield, stable to the planned reaction conditions, and removed selectively without affecting other functional groups.
Convergent vs Linear Synthesis
Linear synthesis proceeds stepwise from starting material to product: A → B → C → D → E → F. Each step reduces the overall yield multiplicatively; a ten-step linear synthesis with 80% yield per step gives only 10.7% overall yield. Convergent synthesis assembles fragments independently and joins them late in the sequence, giving dramatically higher yields. For example, two five-step fragments (80% yield each) joined in one final step yield 0.8⁵ × 0.8⁵ × 0.8 = 13.4%, versus 0.8¹¹ = 8.6% for a linear sequence. Convergent approaches are the preferred strategy for complex molecule synthesis.
Selectivity in Synthesis
Chemoselectivity requires distinguishing between different functional groups — for instance, reducing a ketone in the presence of an ester using NaBH₄. Regioselectivity controls which position in a molecule reacts, such as electrophilic substitution at the ortho, meta, or para position of an aromatic ring. Stereoselectivity encompasses diastereoselectivity (preferential formation of one diastereomer) and enantioselectivity (preferential formation of one enantiomer, achieved through chiral auxiliaries, chiral catalysts, or enzymatic resolution). The control of selectivity is often the central challenge in complex molecule synthesis.
Key Named Reactions in Strategy
Several reactions are staples of retrosynthetic planning due to their reliability and stereochemical predictability. The aldol reaction constructs 1,3-diol and β-hydroxy carbonyl motifs with control of stereochemistry (Evans auxiliaries). The Wittig reaction provides alkenes with defined double-bond position. The Diels-Alder reaction assembles six-membered rings with up to four stereocenters in a single step. Palladium-catalyzed cross-couplings (Suzuki, Heck, Negishi, Sonogashira) form C-C bonds under mild conditions with high functional group tolerance. Computer-aided retrosynthesis (e.g., the LHASA program, more recently the work by Bartosz Grzybowski and IBM) uses databases of known reactions to propose synthetic routes automatically.
