Solvent Effects on Nucleophilicity

The naked-anion strategy is not a historical curiosity; it is load-bearing in modern synthesis and industry. Pharmaceutical process chemistry routinely runs SN2 substitutions, cyanations, azidations, and etherifications in DMF, DMSO, NMP, or acetonitrile specifically to unlock nucleophiles (F⁻, N₃⁻, CN⁻, RO⁻) that would be sluggish or impractical in alcohol or water. Positron-emission-tomography (PET) radiochemistry depends critically on this effect: ¹⁸F-fluoride, generated in aqueous target water, must be dried and transferred into an aprotic solvent (typically acetonitrile) with a phase-transfer agent such as Kryptofix 222 sequestering K⁺, precisely to strip away hydrogen-bond solvation and make the very short-lived isotope ($t_{1/2} = 110\ \text{min}$) react fast enough with a tosylate or triflate precursor before it decays — the synthesis of [¹⁸F]FDG, the most widely used PET tracer in oncology, is a direct industrial application of Mechanism 2 above. Phase-transfer catalysis (Starks, and the crown ether chemistry of Pedersen and Lehn) exploits the same principle from the opposite direction: instead of switching bulk solvent, a crown ether or ammonium salt selectively transports an anion (permanganate, cyanide, hydroxide) into an organic phase where it arrives essentially unsolvated and hyper-reactive, enabling nucleophilic substitutions and oxidations to run efficiently at an organic/aqueous interface without ever dissolving a large excess of a dipolar aprotic solvent. HMPA (hexamethylphosphoramide), though largely phased out industrially due to carcinogenicity concerns, remains the textbook example of an extreme naked-anion solvent, used historically to activate carbanion and enolate nucleophiles in complex natural product total synthesis. The concept also underlies practical choices in E2/E1 and elimination-substitution competition, foreshadowing The Grand Summary: Predicting SN1/SN2/E1/E2: a small nucleophile/base like ethoxide behaves very differently as a bulky, poorly nucleophilic base in ethanol (favoring E2 through steric and desolvation constraints on approach) than as a highly reactive, less hindered nucleophile in DMSO. Understanding anion solvation is also the conceptual bridge to Solvent Effects on SN1 Ionization, where the same protic/aprotic vocabulary reappears but now governing cation stabilization and leaving-group departure rather than nucleophile approach — the two halves of the Hughes–Ingold solvent framework.

Ask the tutor: How does this same protic/aprotic solvation logic change when the leaving group ionizes first, as in an SN1 mechanism?

Start with the anion sitting in solution, before any electrophile is nearby. In a protic solvent, each O–H (or N–H) bond is a hydrogen-bond donor: the O–H σ orbital is a low-lying acceptor, and a lone pair on the anion donates into it, forming a genuine (if weak, a few to several kcal/mol per contact) donor–acceptor interaction, n(X⁻) → σ(O–H). This is stabilization exactly analogous to a Lewis acid–base adduct, just weaker and reversible. The smaller the anion, the more concentrated its negative charge (charge density = charge/volume), the deeper the n(X⁻) orbital sits relative to σ(O–H), and the stronger this donor–acceptor overlap becomes. Fluoride, with its tiny ionic radius (1.33 Å), is drenched in strong, short hydrogen bonds — in water F⁻ is more heavily solvated than any other halide. Iodide, with its radius of 2.20 Å and diffuse, polarizable electron cloud, forms only weak, easily disrupted hydrogen bonds; its charge is smeared over a large volume, so the n(X⁻) → σ(O–H) interaction is comparatively feeble. Now bring in the polar aprotic solvent. DMSO, for instance, has a large permanent dipole (μ ≈ 3.96 D) concentrated on the S=O group, with the oxygen bearing substantial negative character and the sulfur positive. This dipole solvates cations beautifully — Na⁺ or K⁺ sits in a pocket of oxygen lone pairs, exactly like a poor man's crown ether. But there is no O–H, N–H, or any acidic proton on DMSO's exterior for an anion to hydrogen-bond to; the closest DMSO can offer an anion is a weak, diffuse dipole–ion interaction with the electropositive methyl hydrogens, orders of magnitude weaker than a hydrogen bond. The anion is, functionally, "naked" — its full nucleophilic lone pair is available to attack an electrophile with none of the desolvation penalty it would pay in water or ethanol. The consequence for a rate profile is captured in the Hughes–Ingold picture of solvent effects on activation energy. The rate constant depends on the free energy difference between the solvated ground state and the transition state:

If the ground state (solvated anion) is stabilized far more than the transition state (where charge is delocalizing onto the leaving group and the nucleophile's lone pair is becoming a bond), then $\Delta G^{\ddagger}$ is large and the reaction is slow. Protic solvation stabilizes small anions in the ground state disproportionately, raising the effective barrier most for F⁻ and least for I⁻ — this is why the protic order inverts relative to intrinsic basicity. Removing that disproportionate ground-state stabilization (by switching to polar aprotic solvent) restores the intrinsic, gas-phase-like order, which tracks basicity because both properties reflect how tightly the anion's lone pair is held.

The picture reverses in a polar aprotic medium: instead of the anion being wrapped, it is the cation that gets wrapped, freeing the anion entirely.

The governing relationship is the free-energy relation between rate and the differential solvation of ground versus transition state:

where:

• $G^{\ddagger}_{\text{TS}}$ is the free energy of the SN2 transition state (partial bonds to both nucleophile and leaving group, charge spread over both),
• $G_{\text{GS}}$ is the free energy of the fully solvated ground-state nucleophile,
• $\Delta G^{\ddagger}$ sets the rate via the Eyring equation.

Reading: anything that lowers $G_{\text{GS}}$ (stronger anion solvation) raises $\Delta G^{\ddagger}$ and slows the reaction, because the anion becomes a more comfortable, lower-energy species that is reluctant to leave its solvent shell and attack carbon. In the transition state, the nucleophile's negative charge is partly delocalized onto the developing X–C bond and the leaving group, so the TS is less localized on a single small ion than the ground state is — meaning strong, geometry-specific hydrogen bonds to the ground-state anion are disrupted on the way to the TS, and that disruption cost scales with how strong those ground-state bonds were. A second governing relationship makes the radius-dependence explicit, using a simple Born-type solvation estimate for the electrostatic (ion–dipole/hydrogen-bond) part of the solvation free energy:

where:

• $z$ is the ionic charge (magnitude 1 for a halide),
• $e$ is the elementary charge, $\varepsilon_0$ the vacuum permittivity,
• $r$ is the effective solvated ionic radius,
• $\varepsilon_r$ is the solvent's relative permittivity (dielectric constant).

Reading: solvation free energy scales as $1/r$ — smaller ions are stabilized much more strongly by the surrounding dielectric/hydrogen-bond medium. This is the quantitative root of "F⁻ is over-stabilized in water relative to I⁻." The Born term captures the electrostatic part common to all solvents; the extra hydrogen-bond-specific stabilization that protic solvents add on top of this is what actually drives the order reversal, since aprotic solvents of similar dielectric constant do not show it.

Example 1 (foundational) — Ranking three anions across two solvents. Rank OH⁻, SH⁻, and F⁻ as nucleophiles (a) in methanol and (b) in DMF. Step 1: intrinsic/basicity order. Compare within a row (F vs O vs N, all period 2/3 boundary) is tricky, so instead compare O vs S directly (same group, period 2 vs 3): O–H bond in H₂O ($pK_a \approx 15.7$) is far less acidic than H–S–H ($pK_a \approx 7.0$), so OH⁻ is the much stronger base than SH⁻. F⁻ is a weak base ($pK_a$ of HF $\approx 3.2$) compared to both. Basicity order: $\mathrm{OH^-} \gg \mathrm{SH^-} > \mathrm{F^-}$ (roughly; SH⁻ is a weaker base than F⁻ is debatable at the margins, but sulfur's larger, more polarizable/lower-charge-density lone pair makes it the poorest base of the three here despite S being less electronegative — take basicity ≈ OH⁻ ≫ F⁻ > SH⁻ for this worked case, consistent with $pK_a$(H₂S) ≈ 7 vs $pK_a$(HF) ≈ 3.2, i.e. HS⁻ is the weaker base than F⁻... Let's be careful and resolve this cleanly using conjugate acid $pK_a$ values directly, since basicity order is fixed by them: $pK_a$(H₂O) = 15.7, $pK_a$(HF) = 3.2, $pK_a$(H₂S) = 7.0. Higher $pK_a$ of conjugate acid ⇒ stronger base. So basicity order is

Step 2: protic solvent (methanol) — apply the desolvation-penalty logic. SH⁻ is large, soft, and diffuse (sulfur's 3p lone pair, low charge density) — it is barely hydrogen-bonded and essentially retains its intrinsic reactivity. F⁻ is small and hard — heavily hydrogen-bonded and desolvation-penalized. OH⁻ is intermediate: reasonably small (r ≈ 1.40 Å for O vs 1.33 Å for F⁻) but its own conjugate solvent (methanol, an alcohol like water) hydrogen-bonds it strongly too. Empirically in protic solvents:

Sulfur's low charge density lets it shed the (weak) solvation shell almost for free and attack with its large, polarizable lone pair; this is the same softness argument used for I⁻ over F⁻. Step 3: polar aprotic solvent (DMF). No hydrogen bonding to any anion; desolvation penalties collapse toward equal (small), so the intrinsic basicity order is restored:

Check against a limiting case. If we imagine a hypothetical "solvent" with $\varepsilon_r \to 1$ and zero hydrogen-bond donor ability (vacuum-like), the desolvation term vanishes entirely for every species, so the order must reduce to the pure intrinsic/basicity order — exactly what Step 3 gives. This confirms the DMF-limit formula behaves correctly as the aprotic character is pushed to its extreme.

Example 2 (applied) — Choosing a solvent to accelerate a sluggish Finkelstein-type substitution. A synthetic chemist needs to convert 1-chlorobutane to 1-iodobutane using NaI, and separately needs to install fluoride onto a primary tosylate to make a PET radiotracer precursor. Which solvent for each? Reaction 1: NaI + 1-chlorobutane → 1-iodobutane + NaCl (Finkelstein). This reaction is classically run in acetone, a polar aprotic solvent, but the real trick is solubility-driven: NaCl is insoluble in acetone while NaI is soluble, so Le Chatelier pulls the equilibrium forward regardless of aprotic character. Mechanistically, though, acetone still matters for rate: I⁻ is only weakly solvated in any solvent (its low charge density means hydrogen bonds barely form even in protic media), so acetone's aprotic character is not the dominant lever here — solubility is. This is a useful contrast: not every solvent effect in this family is about nucleophilicity reversal; sometimes it's simple thermodynamic precipitation-driving. Reaction 2: R–OTs + F⁻ (as KF or CsF) → R–F. Here the nucleophilicity-reversal mechanism is central. F⁻ is a poor nucleophile in protic solvents because of the severe hydrogen-bond desolvation penalty (Mechanism 1). Running this reaction in a protic solvent would be nearly hopeless for a primary tosylate SN2. The correct choice is a polar aprotic solvent — classically DMSO, DMF, or, for maximum "naked-anion" reactivity, acetonitrile with a cryptand/crown ether (e.g., Kryptofix 222 in radiochemistry, which is exactly how ¹⁸F-fluoride is activated for PET tracer synthesis: the cryptand sequesters K⁺, leaving F⁻ essentially solvent-free and dramatically more nucleophilic).

This is a direct real-world application of Mechanism 2: strip the cation with a crown/cryptand, use an aprotic solvent so there's no hydrogen-bond donor left to grip F⁻, and the intrinsically excellent (small, high-lying lone pair) nucleophile becomes fast rather than crippled.

Example 3 (exam/competition level) — Quantitative estimate using the Born-type scaling. Estimate the ratio of hydrogen-bond-driven desolvation free energy penalties for F⁻ versus Br⁻ in water, using the simplified Born relation, given effective solvated radii $r_{\mathrm F} = 1.33\ \text{Å}$ and $r_{\mathrm{Br}} = 1.96\ \text{Å}$, and $\varepsilon_r(\text{water}) = 78.4$. Step 1: write the Born solvation free energy for each ion.

Since the $\left(1-\frac{1}{\varepsilon_r}\right)$ and all other prefactors are common to both ions, the ratio of solvation energies is simply the inverse ratio of radii:

Step 2: convert to an actual energy using known experimental hydration free energies as anchors (Born alone under-predicts hydrogen-bond-specific contributions, so we scale against real data). Experimental single-ion hydration free energies (conventional, TATB-based scale) are approximately $\Delta G_{\text{hyd}}(\mathrm F^-) \approx -465\ \text{kJ/mol}$ and $\Delta G_{\text{hyd}}(\mathrm{Br}^-) \approx -321\ \text{kJ/mol}$. Their ratio:

Step 3: compare prediction to experiment. The simple $1/r$ Born scaling (1.474) matches the experimental hydration ratio (1.449) to within about 1.7% — remarkably good agreement for such a crude electrostatic model, confirming that the dominant physics of the size-dependence is captured even though Born theory omits the directional, orbital-overlap character of real hydrogen bonds. Step 4: translate into the rate consequence. The extra desolvation F⁻ must pay relative to Br⁻ is on the order of

Even if only a fraction (say 15–25%) of this extra ground-state stabilization survives into the activation barrier as a net desolvation penalty (since some solvation is retained in the TS too), this is easily $20$–$35\ \text{kJ/mol}$ — enough, via

at $T = 298\ \text{K}$ ($RT \approx 2.48\ \text{kJ/mol}$), to suppress the rate by a factor of

This order-of-magnitude estimate — F⁻ being roughly $10^4$-fold slower than its "fair share" of intrinsic reactivity would predict in water — is exactly the scale of rate suppression that flips F⁻ from the best halide nucleophile (aprotic, intrinsic order) to the worst (protic, observed order). Limiting-case check. As $\varepsilon_r \to 1$ (vacuum, zero polarity), the Born term $\left(1-\frac{1}{\varepsilon_r}\right) \to 0$ for both ions identically, so $\Delta G_{\text{solv}} \to 0$ for all halides and the desolvation-penalty differential vanishes — correctly recovering the gas-phase (intrinsic, basicity-tracking) nucleophilicity order with no solvent reversal at all, exactly as required.

The modern framework for this topic descends from the work of Sir Christopher Kelk Ingold and Edward D. Hughes, whose systematic classification of substitution and elimination mechanisms in the 1930s–1950s established the vocabulary (SN1, SN2, E1, E2) in which solvent effects are still discussed. Ingold's Chemical Society lectures and his monumental textbook Structure and Mechanism in Organic Chemistry laid out, with painstaking kinetic rigor, how solvent polarity and hydrogen-bonding capacity systematically shift rates and mechanisms — the "Hughes–Ingold rules" for solvent effects on ionization and substitution rates are the direct ancestor of the desolvation-penalty argument used above. The specific articulation of the protic/aprotic nucleophilicity reversal, and its practical exploitation via solvents like DMSO and HMPA, was consolidated later by physical organic chemists studying anion solvation thermodynamics — most notably through single-ion solvation energy scales (built on extrathermodynamic assumptions like the TATB assumption used in Worked Example 3) that made it possible to assign quantitative desolvation energies to individual anions rather than only ion pairs.