The proposal and the hidden assumption
The plan is to add 2 eq nBuLi and stir overnight so that the excess reagent consumes the residual nBuBr left over from the Li–Br exchange. That plan assumes nBuLi reacts with nBuBr much faster than with anything else in the flask. In this mixture, that assumption fails.
nBuLi is a strongly polarized carbanion reagent: the C–Li bond is highly polarized, so for practical purposes it behaves as a butyl anion plus a lithium cation. That makes it simultaneously a strong base and a strong nucleophile. It will react with whichever partner is available and kinetically accessible, not with a chosen target.
The competing partners present
- Residual nBuBr — the intended target. Reaction here is a halogen–metal exchange, giving nBuLi back plus a new butyl bromide, so this pathway is largely self-cancelling rather than a net sink.
- The metal carbonyl — a strong electrophile at carbon. nBuLi adds to a coordinated CO ligand, exactly as the aryllithium does, generating an acyl–lithium species instead of the desired acylate.
- The oxy-carbene carbon — electrophilic because the metal carbonyl withdraws electron density and the alkoxy oxygen donates through its lone pair. nBuLi alkylates this carbon, giving the same class of butyl–carbene adduct that residual nBuBr produces.
- The aryllithium and the solvent — nBuLi can deprotonate acidic sites and slowly attack ethereal solvent, consuming reagent and degrading the desired intermediate.
Why the flux splits
The exchange between nBuLi and nBuBr is an equilibrium, not an irreversible sink. Once the aryllithium has formed, the mixture already contains nBuBr and nBuLi in a ratio set by that equilibrium, so adding more nBuLi pushes the equilibrium but does not remove the butyl bromide from the system. Meanwhile the addition and alkylation pathways are effectively irreversible, so they accumulate product over time.
That is the key asymmetry: the pathway you want is reversible and self-limiting, while the pathways you do not want are irreversible and time-dependent.