A paper published in Nature Chemistry on 5 October 2026 joins alkyl carboxylic acids to alkylboronic acids by electrolysis, with no metal catalyst. The boronic acid takes part only after fluoride has converted it into a fluoroborate, and the fluoride source the authors chose is anhydrous tetramethylammonium fluoride (TMAF). Their supporting information shows two things that every user of the method needs to know: the amount of TMAF decides which product forms, and water suppresses the boron side of the reaction.

What the paper reports

Jessica Zhong, Maxime Boudjelel and co-workers in Christian Malapit's group at Northwestern University couple an alkyl carboxylic acid with an alkylboronic acid.^[1] The acid loses CO₂, the boronic acid loses its boron, and the two alkyl fragments join through a new C(sp³)–C(sp³) bond.

Scheme 1. The cross-coupling of an alkyl carboxylic acid with an alkylboronic acid. The new C–C bond is shown in gold. Conditions from the general procedures of the SI.^[1]

The general conditions are short:

• an undivided cell with two carbon-foam electrodes (reticulated vitreous carbon);
• constant current, with the polarity of the electrodes reversed every 50 ms (a frequency of 10 Hz);
• acetone at room temperature, under nitrogen;
• anhydrous TMAF as the activator, with boron oxide (B₂O₃) as an additive;
• no transition-metal catalyst. The authors analysed the reaction mixture by ICP-MS and found transition metals only at trace (ppb) levels.

The boronic acid can also be made from a terminal alkene by hydroboration and used without purification, so in practice an alkene can be coupled with a carboxylic acid. Similar conditions without B₂O₃ couple a boronic acid with itself. An aryl chloride in the product survives the electrolysis, so the crude product can go directly into a Suzuki–Miyaura coupling or a Buchwald–Hartwig amination.

In the examples of the supporting information (SI), both partners are primary alkyl groups. The acids include protected aspartic and glutamic acids, levulinic acid and linoleic acid, and the cross-coupled products were isolated in 38 to 67 % yield.

Scheme 2. Three products from the SI, new bond in gold. A: from Boc-Glu-OBzl and the boronic acid made from styrene, 41 %. B: from linoleic acid and propylboronic acid, 55 %. C: the cross product of 3-(4-chlorophenyl)propionic acid and propylboronic acid, taken without isolation into a Buchwald–Hartwig amination with morpholine, 61 % over two steps.^[1]

Two abundant classes of compound, carboxylic acids and boronic acids, give an alkyl–alkyl bond without a metal catalyst.

Why two alkyl radicals are hard to cross-couple

The classic reaction of this kind is the Kolbe electrolysis, in which carboxylic acids lose CO₂ at the anode and the alkyl radicals combine. With two different acids, the radicals combine without preference, so the cross product comes mixed with two homodimers. The authors note that the direct cross-coupling of radicals from two different precursors remains challenging.

The new method takes the two radicals from two different classes of compound and controls how each one is activated. The authors describe this as redox-matched alternating-polarity electrolysis with controlled activation of redox-active species.^[1] On the boron side, the activation is done by fluoride.

The product depends on the balance between the two radical sources, and on the boron side that balance is set by fluoride.

The boronic acid reacts only after fluoride activation

An alkylboronic acid on its own is hardly oxidised. In the authors' cyclic voltammetry, phenethylboronic acid in acetone gives almost no oxidation current up to 1.5 V against ferrocene. When TMAF is added, oxidation waves appear and grow with each further equivalent (SI, Fig. 8).

The reason is that fluoride adds to the boron atom and forms negatively charged fluoroborates. The ¹¹B NMR titration in the SI shows the conversion: as 0.5 to 3 equivalents of TMAF are added, the boronic acid signal falls and an upfield signal close to that of the potassium trifluoroborate salt appears (SI, Fig. 5). The fluoroborates drawn in the authors' scheme are R–BF(OH)₂⁻, R–BF₂(OH)⁻ and R–BF₃⁻, and the authors propose that these anions are oxidised at the anode to give the alkyl radical (SI, Fig. 7).

Scheme 3. Fluoride activation of the boronic acid. TMAF converts it into fluoroborate anions, which are oxidised at the anode to the alkyl radical; water shifts the equilibrium back. Redrawn from SI, Figs 5 and 7; Me₄N⁺ counter-ions are omitted.^[1]

The choice of activator matters (SI, Table 1):

• With the potassium trifluoroborate as the boron source, ammonium salts without fluoride (TBABF₄, TMABF₄, TMAPF₆) gave 22 to 30 % of the boron homodimer. CsF with 18-crown-6 or with HFIP gave 47 and 50 %, and anhydrous TMAF gave 52 %.
• With the free boronic acid and anhydrous TMAF, the same product formed in 65 % with 1 equivalent and in 82 % with 1.5 equivalents.
• Tetramethylammonium hydroxide, the same cation with hydroxide in place of fluoride, gave no cross-coupled product at all.

The best activation in the study is the free boronic acid with anhydrous TMAF, converted to the fluoroborate in the reaction flask.

A precedent: TMAF and B₂pin₂ in König's borylation

TMAF had already been shown to turn a neutral boron reagent into a reactive fluoroborate anion. In 2021, König and co-workers reported a photocatalytic borylation that replaces C–F, C–O, C–S and C–N bonds of arenes with a pinacol boronic ester, using bis(pinacolato)diboron (B₂pin₂).^[2] A fluoride base was needed, and TMAF gave the best result: 74 % (NMR yield), against 56 % with CsF and 19 % with KF, which the authors attributed to its better solubility in acetonitrile.^[3]

By ¹⁹F NMR, the authors observed an adduct between TMAF and B₂pin₂. They proposed that this gives the anion [B₂pin₂–F]⁻, which activates the substrate and delivers the boryl group to the aryl radical.^[2,3] An insightful 2022 review of TMAF chemistry had already described König's result as the first case in which TMAF is the key to an anionic boron species of this kind, and noted that it "may open new avenues in reaction development".^[3]

Scheme 4. König's photocatalytic borylation of arenes with B₂pin₂ and TMAF, and the fluoroborate anion [B₂pin₂–F]⁻ proposed from ¹⁹F NMR. Drawn from the description in ref. 3.^[2,3]

The two reactions use different boron partners and different sources of energy. König used B₂pin₂ and light; the Northwestern group uses alkylboronic acids and an electrode. But the sequence is the same in both. A neutral boron compound does little on its own. Fluoride from TMAF turns it into a fluoroborate anion. Light or an electrode then drives radical chemistry from that anion, or from a species that it forms.

Neutral boron compound + fluoride from TMAF + light or an electrode = new radical chemistry. Two reactions now follow this sequence.

The amount of TMAF decides which product forms

In the model reaction, 4-phenylbutyric acid is coupled with five equivalents of phenethylboronic acid. Three products are possible: the cross product, the dimer of the acid-derived radical and the dimer of the boron-derived radical. The SI reports all three at three amounts of TMAF.

Scheme 5. The model reaction of the optimisation study and its three possible products, new bonds in gold.^[1]

At 1.25 equivalents the cross product is the main product. At 3 equivalents the main product is the boronic acid dimer, and nothing else in the reaction was changed.

Our reading of these data, which is not a statement from the paper, is as follows. More fluoride converts more of the boronic acid into the oxidisable fluoroborate, as the ¹¹B titration shows. The boron-derived radicals then form faster than the radicals from the acid, so they combine more often with each other than with the acid-derived radicals.

This has a practical consequence. TMAF that has taken up water contains less fluoride per gram than the dry salt. If it is weighed as dry TMAF, the reaction receives less fluoride than intended, and water as well.

The TMAF must be dry and weighed exactly, because its stoichiometry sets the selectivity.

What water does to the reaction

The SI contains two direct tests of water.

• Anhydrous TMAF against the tetrahydrate. With 1.5 equivalents of anhydrous TMAF, the boronic acid dimer formed in 82 %. With the same amount of TMAF·4H₂O, it formed in 26 % (SI, Table 1A, entries 9 and 10).
• Water added to the reaction. Without B₂O₃, the cross product formed in 37 %. When 0.5 mL of water was also added to the 4 mL of acetone, the cross product fell to 5 % and the boronic acid dimer fell from 18 to 2 %. The dimer of the carboxylic acid rose from 16 to 62 % (SI, Tables 4 and 5).

So water suppresses the boron side of the reaction, while the Kolbe-type coupling of the acid continues.

The authors propose an explanation (SI, Fig. 7). Water shifts the fluoroborates back to the free boronic acid. At the same time, acetone is reduced at the cathode and couples to pinacolate. The free boronic acid is captured by this pinacol as its pinacol ester (R–Bpin), which is not oxidised under the conditions, so boronic acid captured in this way is lost from the reaction.

Scheme 6. The unproductive pathway proposed by the authors. Water returns the fluoroborate to the boronic acid. Pinacolate, formed by reduction of acetone at the cathode, captures the boronic acid as its pinacol ester, which is not oxidised. Redrawn from SI, Fig. 7.^[1]

The procedures also exclude water throughout. The reactions are set up in a glovebox, and the TMAF is dried and stored there. Boronic acids made by hydroboration, and the boroxines, are dried by evaporation with toluene and under high vacuum before use.

In both tests, water lowered the yield from the boron partner, and the authors propose that it does so by reversing the activation step.

Anhydrous TMAF is slow to make from the hydrate

The authors made their anhydrous TMAF from the tetrahydrate, TMAF·4H₂O, in a similar manner to a procedure published for production scale.^[4] They evaporated the hydrate five times with isopropanol and five times with toluene at 45 °C, dried the powder under high vacuum overnight and kept it in a glovebox.

The production-scale work shows why this effort is needed. In an SNAr fluorination in DMF, undried TMAF hydrate gave no conversion, while rigorously dried TMAF gave 79 area % of product. The drying took repeated distillations with isopropanol and then DMF, and removing the isopropanol took several days under high vacuum. When the dry salt was isolated as a solid, the water content of the DMF reaction solution rose to 7.8 %, which the authors suspect came from handling the very hygroscopic solid.^[4]

Anhydrous TMAF is slow to make from the hydrate, and the dry solid is very hygroscopic.

The short route: the methanol adduct

Removing a bound solvent under vacuum is an established way to reach anhydrous TMAF. Christe and co-workers isolated anhydrous TMAF in 1990 from its isopropanol solvate, under dynamic vacuum with heating.^[4,5] A route through methanol is also published: TMAF made by anion exchange in methanol was dried under vacuum at 130–140 °C for 12–16 hours, to 0.06 wt % water, in place of the weeks of drying that older methods needed.^[3]

The methanol adduct, TMAF·MeOH, is a key intermediate in the preparation of anhydrous TMAF.^[3] It contains methanol in place of the water of the hydrate, so there is no water to remove. In our experience, high vacuum removes the methanol and gives anhydrous TMAF in 3–4 hours at 110–120 °C, or in several days at room temperature. In our set-up, 120 °C was the optimum temperature (note: TMAF itself starts to decompose slowly only at about 160 °C).^[3,5] The exact drying time depends on the pump, the experimental set-up and the amount of material. The step needs no solvent, and we produce TMAF at kilogram scale under these conditions.

Scheme 7. Two routes to anhydrous TMAF. Top: the drying used by the authors (SI, p. 10). Bottom: removal of methanol from TMAF·MeOH under high vacuum, at 110–120 °C or at room temperature, from NorrChemica's experience. Drying times depend on the pump, the set-up and the scale.

This route also makes scale-up possible. The coupling needs dry TMAF in an exact amount: water lowers the yield from the boron partner (Scheme 6), and the amount of TMAF sets the selectivity (Table 1). At production scale, drying the hydrate is the difficult step, and the isolated dry solid takes up water during handling.^[4] Drying the methanol adduct produces the fluoride activator at large scale with control of its moisture content, and therefore with less variation in yield between batches.

The bound methanol also changes the reactivity. In Sanford's SNAr test, TMAF·MeOH in DMF at room temperature gave no conversion, where anhydrous TMAF had given 99 % under the same conditions.^[3,6] TMAF alcohol adducts are, in turn, much less hygroscopic than anhydrous TMAF.^[3,6] So for a reaction that needs the full reactivity of fluoride, such as this coupling, the adduct is the most convenient form to store and the anhydrous salt is the form to use.

Our TMAF·MeOH contains at least 0.91 fluoride per tetramethylammonium ion (91 mol%), measured by quantitative ¹H and ¹⁹F NMR on every batch. In our published comparison, this was within 1 mol% of the best commercial anhydrous TMAF we measured under the same conditions.

One point for users of the new method: the authors used anhydrous TMAF, not the methanol adduct. Remove the methanol first (high vacuum, 3–4 hours at 110–120 °C or several days at room temperature, and not above 120 °C), then weigh and handle the dry salt under nitrogen, as the paper does.

No water in, no water to take out. In our experience, TMAF·MeOH is the cheapest, simplest and fastest route to the anhydrous TMAF that this coupling needs, and it gives the dry activator at kilogram scale with controlled moisture.

© 2026 NorrChemica. All rights reserved.
Content may not be reproduced without written permission.