Buchwald–Hartwig Conditions: How to Choose Ligand, Base, and Solvent

Overnight at 100 °C, and the LCMS shows mostly aryl bromide, a growing peak for the dehalogenated arene, and a trace of product. You used Pd₂(dba)₃, BINAP and NaOtBu in toluene because that was the last Buchwald run in the group notebook — on a different amine. Picking Buchwald–Hartwig conditions (ligand, base, and solvent) starts with the nitrogen nucleophile and the functional groups on both partners, not with the last set that worked for someone else.

Below is a decision path for choosing a first set of conditions, and for recognizing when a small screen beats a single guess. Treat every recommendation here as a starting point to confirm on your substrate.

How to choose Buchwald–Hartwig conditions: the factors that matter

Four questions, roughly in order of how much they constrain the choice:

1. What is the N–H partner? Primary aliphatic amine, secondary amine, aniline, or a weak N-nucleophile (amide, carbamate, sulfonamide, some heterocycles). Ligand choice follows mostly from this.
2. Can the substrates survive a strong alkoxide base? Esters, enolizable ketones, and nitro groups often cannot survive NaOtBu at 100 °C, and that rules out the fastest base.
3. Are there protic groups (free OH, a second NH) that could compete or get deprotonated?
4. What is the electrophile? Aryl bromide, chloride, or triflate, and whether it is a heteroaryl with a nitrogen that can coordinate palladium.

The ligand recommendations below follow the dialkylbiaryl phosphine “user’s guide” from the Buchwald group (Surry & Buchwald, Chem. Sci. 2011), which is still the most practical single reference for first conditions.

Path A — Secondary amine: start with RuPhos

If the nucleophile is a secondary amine (morpholine, a piperazine, an N-methyl aniline), RuPhos is the standard first ligand. Pair it with NaOtBu in toluene or 1,4-dioxane at 80–100 °C if the substrates tolerate a strong base.

Cyclic secondary amines are usually the easy case. Acyclic, hindered ones (diisopropylamine-like) are harder, and that is where a small ligand screen starts to pay for itself.

Path B — Primary amine or aniline: start with BrettPhos

For primary aliphatic amines and anilines, BrettPhos is the usual first choice. One reason is selectivity: primary amines can arylate twice, and BrettPhos favors the monoarylated product.

NaOtBu is the default base. If the molecule carries a free OH or another protic group, switch to LiHMDS. The Buchwald group’s BrettPhos/RuPhos work reports LiHMDS as the base that lets substrates carrying protic functional groups couple.

Path C — Base-sensitive substrate: switch to a weak inorganic base

If either partner carries an ester, an enolizable ketone, or another group that won’t survive NaOtBu, keep the ligand from Path A or B and change the base to Cs₂CO₃ or K₃PO₄. These bases are weaker and only partly soluble, so reactions are slower. Expect to run hotter or longer, and use a solvent that helps: 1,4-dioxane, or tBuOH / t-amyl alcohol, which are common with carbonate and phosphate bases.

Path D — Amide, carbamate, or sulfonamide: use a ligand built for weak nucleophiles

Amides and related weak N-nucleophiles need different ligands from amines. Two common starting points are Xantphos, a wide-bite-angle bisphosphine with a long track record in Pd-catalyzed amidation, and the bulkier tBuBrettPhos. Pair either with Cs₂CO₃ or K₃PO₄ in dioxane. NaOtBu is usually too harsh for these substrates.

Path E — Aryl chloride or coordinating heteroaryl halide

Aryl chlorides are routine with the dialkylbiaryl ligands above. Older triarylphosphine and BINAP-type systems often stall on them. Some practical notes:

• Use a palladacycle precatalyst (third- or fourth-generation “G3/G4” versions of the ligand above). Precatalysts generate the active L–Pd(0) species cleanly. Pd(OAc)₂ has to be reduced in situ first, and dba from Pd₂(dba)₃ can slow the reaction. Both are common, quiet sources of irreproducibility.
• Heteroaryl halides with a nearby ring nitrogen (2-halopyridines, many azoles) can coordinate palladium and slow the catalysis. Expect to need higher loading or a screen.
• Aryl iodides are not automatically the best partner in C–N coupling. Iodide can inhibit the catalyst, and a bromide often behaves better.

When to screen instead of guessing

A single guess is fine for a cyclic secondary amine and an aryl bromide with no sensitive groups. Run a small screen instead when any of these apply:

• The electrophile is a heteroaryl halide with a coordinating nitrogen.
• The substrate is both base-sensitive and hindered.
• The first set of conditions gave hydrodehalogenation or no conversion.
• You have limited material and can’t afford a series of sequential single attempts.

A useful minimum is 2–3 ligands × 2 bases in one solvent: 4–6 vials at 20–50 mg each, analyzed by LCMS the next morning. That is the same idea as a high-throughput screening (HTE) plate, at a scale a single chemist can set up in an afternoon. Software condition prediction can propose candidates for the screen. Treat those as suggestions to test, not answers — our post on how reliably AI predicts reaction outcomes covers why.

Summary table

Setting up the first run

Typical discovery-scale starting amounts are 1–2 mol% Pd as a precatalyst, 1.1–1.2 equiv of amine, and 1.2–1.5 equiv of NaOtBu. Use more for the weaker carbonate and phosphate bases (around 2 equiv is common). Degas the solvent and run under nitrogen or argon. The active Pd(0) catalyst is air-sensitive.

Catalyst loading at milligram scale is where arithmetic errors creep in. Our walkthrough of calculating mol% catalyst loading covers the conversion to milligrams. The stoichiometry calculator will build the full reagent table from the limiting aryl halide. For the broader layout, see building a reagent table for organic synthesis.

For the wider scope of the reaction, including heteroaryl and late-stage examples, the Buchwald group’s Chem. Rev. 2016 review of Pd-catalyzed C–N coupling applications is a good place to find precedent close to your substrate.