Liquid Ammonia as a Solvent System

Drop a coin-sized lump of sodium into water and you get a violent hiss. There's a flash of hydrogen flame, and scorched glassware follows. Drop the same lump into liquid ammonia at −33 °C, and something different happens. The metal simply dissolves, turning the solvent a brilliant, ink-blue color. That color looks like nothing else in inorganic chemistry. That blue solution — sodium metal, intact as Na⁺ cations — has its valence electrons swimming free in cavities of solvent molecules. This is the signature phenomenon of liquid ammonia chemistry. It has no counterpart in water at all. Understanding why ammonia can do this is the subject of this article. So is understanding why so much of the rest of its chemistry runs in parallel to aqueous chemistry, with the labels changed. Liquid ammonia matters because it is the paradigm non-aqueous ionizing solvent. It is the system against which the generality of the Autoixonization and the Solvent-System Concept was first tested. It was proven using ammonia as the key test case. Ammonia autoionizes, and has its own acids and bases. These are defined relative to the ammonium/amide pair. It supports solvolysis reactions (ammonolysis) that are exact structural analogs of aqueous hydrolysis. Yet it differs from water in fundamental ways. It is a much weaker Brønsted acid. It is a much better electron acceptor into antibonding orbitals of its own solvated cavities. It is a poorer solvator of anions. However, it is a far better solvator of large soft cations and neutral molecules. These differences are not incidental. They follow directly from ammonia's smaller dipole moment and weaker hydrogen bonding. They also follow from its higher-lying antibonding orbitals, and the polarizability contrast between N and O. The big picture is that liquid ammonia is simultaneously the best illustration and the sharpest limit-test of the solvent-system concept. Reactions look bizarre if you insist on defining acid and base by reference to H₃O⁺ and OH⁻. They become perfectly systematic once you substitute NH₄⁺ and NH₂⁻. Metathesis reactions ("double decompositions"), acid–base neutralizations, and solvolysis of covalent halides and nitrides all fall into place. So does the entire electride/solvated-electron chemistry of the alkali and alkaline-earth metals. All of this falls into a coherent scheme once the analogy is set up correctly. The places where the analogy breaks are equally instructive. These include weaker self-ionization, and no true "ammono-hydronium" chemistry of comparable strength. They also include the metal–ammonia solutions, which have no aqueous parallel at all. These gaps tell you exactly what water's uniqueness actually consists of. This article builds the parallel systematically. It covers the physical properties that set ammonia apart, and the autoionization equilibrium. It covers the acid/base/salt classification this equilibrium generates. It also covers ammonolysis as the ammonia analog of hydrolysis. It also covers the physics of solvated electrons and metal–ammonia solutions. This spans the full concentration range, from dilute blue solutions to concentrated bronze, metallic liquids. Along the way it connects forward to other exotic solvent systems. These include anhydrous Reactions in Anhydrous Sulfuric Acid and BrF3. These extend the same solvent-system logic to acidic and superacidic media.

Key takeaways

Liquid NH₃ autoionizes as $2\,\text{NH}_3 \rightleftharpoons \text{NH}_4^+ + \text{NH}_2^-$ , with $K_{\text{auto}} \approx 10^{-33}$ at −50 °C — far smaller than water's $K_w$ .
Ammono-acids donate NH₄⁺ (or raise its concentration); ammono-bases donate NH₂⁻ (amide) — the exact structural analog of the aqueous H₃O⁺/OH⁻ scheme.
Ammonolysis is solvolysis by NH₃ in place of H₂O: covalent halides and metal ions are converted to amides, imides, or nitrides exactly as they hydrolyze to hydroxides/oxides in water.
Alkali and heavier alkaline-earth metals dissolve reversibly in NH₃(l) to give solvated electrons — deep blue, paramagnetic, highly conducting dilute solutions that turn bronze and metallic above ~3 mol % metal.
Ammonia solvates large, soft cations (Ag⁺, transition-metal ammines) and neutral Lewis acids far better than it solvates small anions, because N lone pairs are strong σ-donors but poor H-bond acceptors/donors compared to O.
Metathesis ("double decomposition") reactions in NH₃(l) are driven by precipitation of an insoluble salt, just as in water — solubility rules are simply different because ammonia is far less polar and far less able to hydrogen-bond to anions.
Key idea: every acid–base and solvolysis reaction familiar from water has a structurally identical liquid-ammonia counterpart built on NH₄⁺/NH₂⁻ instead of H₃O⁺/OH⁻ — but ammonia's unique ability to stabilize free electrons in solvent cavities produces a whole class of chemistry (solvated electrons, metal–ammonia solutions) that water simply cannot support.

The core insight

Liquid ammonia is water's structural mirror-solvent: replacing H₃O⁺/OH⁻ with NH₄⁺/NH₂⁻ regenerates the entire acid–base and solvolysis logic of aqueous chemistry, while ammonia's weaker H-bonding, smaller autoionization constant, and much more negative electron affinity for solvation together enable a distinctive chemistry — solvated electrons and metal–ammonia solutions — that has no aqueous analog.

Where this lives today

Metal–ammonia solutions remain a standard synthetic reagent: sodium or lithium in liquid ammonia is the classic Birch reduction system, converting aromatic rings to 1,4-cyclohexadienes via sequential solvated-electron transfer and protonation steps, a transformation used throughout steroid and terpenoid total synthesis and in industrial routes to intermediates for pharmaceuticals. Dissolving-metal reductions with Na/NH₃ or Li/NH₃ are also used to reduce alkynes selectively to trans-alkenes, a stereochemical outcome that follows directly from the radical-anion mechanism enabled by solvated-electron transfer. Liquid ammonia itself remains an industrially vast commodity (the Haber–Bosch product), and much of its downstream chemistry — urea and nitric acid manufacture, amide/nitride ceramic synthesis (boron nitride, silicon nitride, aluminum nitride via high-temperature ammonolysis of the corresponding halides or oxides) — is literally ammonolysis chemistry run at industrial scale, following the same amide→imide→nitride condensation ladder worked out in the derivation above. Solvated electrons in ammonia and related amine solvents are also studied as model systems in ultrafast spectroscopy for understanding electron solvation dynamics more generally, informing work on radiation chemistry, electron transfer in biological systems, and the physics of expanded/metallic fluids near a metal–insulator transition. Electrides and alkalides — crystalline solids in which alkali metal cations are complexed by cryptands or crown ethers while the corresponding anion is a "bare" trapped electron or alkali metal anion — are synthesized by crystallizing concentrated metal–ammonia or metal–amine solutions, extending the ammonia solvated-electron concept into isolable, structurally characterized solid-state materials of interest for low-work-function and highly reducing applications. The acid–base language developed here — defining acidity and basicity relative to a solvent's own autoionization rather than to water specifically — is exactly the conceptual tool later needed to make sense of superacidic and highly oxidizing media such as anhydrous sulfuric acid and BrF₃, where the aqueous H₃O⁺/OH⁻ framework is entirely inapplicable but a solvent-specific ammono-style scheme works cleanly.

Ask the tutor: Why does the solvated electron's absorption band shift and change shape as a metal–ammonia solution is concentrated toward the metallic regime?

The physical picture

Ammonia is a small, strongly-dipolar, hydrogen-bonding molecule, structurally similar to water but chemically different in ways that matter enormously for solvent behavior. Its dipole moment (1.47 D) is markedly smaller than water's (1.85 D), and each nitrogen carries only one lone pair rather than two, so NH₃ can act as a hydrogen-bond acceptor much better than it can act as a donor, and its overall hydrogen-bonded network is weaker and less three-dimensionally connected than water's. This shows up directly in the bulk liquid properties: ammonia boils at −33.3 °C and freezes at −77.7 °C (a 44 °C liquid range at 1 atm, extendable enormously under pressure — ammonia is often handled in sealed tubes or pressure vessels well above room temperature), has a dielectric constant of only about 22 at −33 °C (versus water's 78 at 25 °C), and a much lower density and viscosity than water. The weaker hydrogen-bond network and lower dielectric constant mean ammonia solvates small, hard anions (F⁻, OH⁻) far more poorly than water does, while its strong, exposed nitrogen lone pair makes it an excellent σ-donor ligand toward soft, polarizable, or highly charged cations — hence the vast ammine chemistry of transition metals and the ready solubility of many covalent and organic compounds that are insoluble in water. The defining reaction of the pure solvent is self-ionization, directly parallel to water's $2\text{H}_2\text{O} \rightleftharpoons \text{H}_3\text{O}^+ + \text{OH}^-$ :

2\,\text{NH}_3(l) \rightleftharpoons \text{NH}_4^+(am) + \text{NH}_2^-(am)

Here "(am)" denotes ammonia solvation, exactly as "(aq)" denotes aqueous solvation. The equilibrium constant for this process, conventionally written $K_{\text{auto}} = [\text{NH}_4^+][\text{NH}_2^-]$ in the pure solvent, is extremely small — estimates from conductivity place it near $10^{-33}$ near the boiling point, some 19 orders of magnitude smaller than $K_w = 10^{-14}$ for water. This single number is the quantitative root of much of ammonia's distinct behavior: an "ammono-neutral" solution has $[\text{NH}_4^+] = [\text{NH}_2^-] \approx 10^{-16.5}\,\text{M}$ , so the solvent is intrinsically far less capable of protonating or deprotonating solutes than water is, and acid–base leveling occurs over a correspondingly different (and generally shifted, less acidic) window of intrinsic solute strength. The second organizing idea is the solvent-system definition of acids and bases, applied to ammonia exactly as it is applied to water. An ammono-acid is any species that increases the concentration of the solvent's characteristic cation, NH₄⁺, when dissolved; an ammono-base is any species that increases the concentration of the characteristic anion, NH₂⁻ (amide). Ammonium salts (NH₄Cl, NH₄NO₃, (NH₄)₂SO₄) are therefore ammono-acids in liquid ammonia — they behave analogously to how HCl, HNO₃, and H₂SO₄ behave in water — while alkali metal amides (NaNH₂, KNH₂) are ammono-bases, playing the role of NaOH and KOH. Neutralization,

\text{NH}_4^+ + \text{NH}_2^- \rightarrow 2\,\text{NH}_3

is the exact ammonia analog of $\text{H}_3\text{O}^+ + \text{OH}^- \rightarrow 2\,\text{H}_2\text{O}$ , and it can be followed by conductimetric titration in liquid ammonia exactly as acid–base titrations are followed in water, with a sharp conductance minimum at the equivalence point. Third, and unique to ammonia among common solvents, is the dissolution of electropositive metals to give solvated electrons. When an alkali metal (or Ca, Sr, Ba, Eu, Yb) is added to liquid ammonia, it dissolves — without evolving hydrogen, without net redox chemistry with the solvent, and reversibly (the pure metal can be recovered by evaporating the ammonia) — to give a solution containing solvated metal cations and solvated electrons:

M(s) + (x+y)\,\text{NH}_3 \rightarrow [M(\text{NH}_3)_x]^+ + [e^-(\text{NH}_3)_y]

The solvated electron occupies a quasi-spherical cavity roughly 3–4 Å in radius, created by orientation of ammonia N–H dipoles (hydrogens pointing inward, toward the excess negative charge) around the trapped electron; the species is best described as an electron delocalized over a multi-molecule solvation cavity rather than "an electron and one ammonia molecule" — spectroscopically and structurally it resembles a very large, very diffuse F-center in an ionic solid, or the excess electron states studied in cluster ion spectroscopy. This solvated electron is responsible for the deep blue color (broad absorption tailing from the near-IR into the visible, with $\lambda_{\max}$ near 1500 nm — the visible-region absorption is simply the blue-edge tail of this band), the very high electrical conductivity of dilute solutions (mobility far exceeding that of any normal ion, because conduction proceeds partly by an electron-hopping mechanism reminiscent of the Grotthuss mechanism for H⁺ in water), and the strong reducing power that makes metal–ammonia solutions among the most useful one-electron reducing agents in synthetic and organic chemistry (the classic Birch reduction reagent).

Figure

Autoionization and the acid–base parallel between water and liquid ammonia

Figure

The solvated electron cavity in metal–ammonia solutions

The core ideas

The behavior of liquid ammonia as a solvent system rests on three coupled relationships: the autoionization equilibrium that defines the acid/base scale, the ammonolysis (solvolysis) equations that parallel aqueous hydrolysis, and the energetics of electron solvation that governs metal–ammonia solutions.

K_{\text{auto}} = [\text{NH}_4^+][\text{NH}_2^-] \approx 10^{-33} \ \text{at} -50\,^{\circ}\text{C}

where: concentrations are in mol L⁻¹ referenced to pure liquid ammonia as the reference state (activity of NH₃ itself is taken as unity), and the equilibrium constant is inferred primarily from conductivity extrapolations since direct spectroscopic detection of NH₂⁻ in neutral ammonia is not feasible at such low concentration.

Reading: this constant is the ammonia analogue of $K_w$ , and it plays exactly the same normalizing role: an "ammono-neutral" solution has $[\text{NH}_4^+]=[\text{NH}_2^-]=\sqrt{K_{\text{auto}}}\approx 3\times10^{-17}\,\text{M}$ , defining the zero point of the ammono-pH scale, $\text{pNH}_4 + \text{pNH}_2 \approx 33$ replacing the aqueous $\text{pH}+\text{pOH}=14$ .

\text{ammonolysis general form:}\quad \text{E–X} + \text{NH}_3 \rightarrow \text{E–NH}_2 + \text{HX}

where: E–X is a polar covalent bond (commonly E = nonmetal or metal halide, X = halide or another leaving group), and the product E–NH₂ is an amide, which may further react (with excess NH₃, heat, or catalysis) to imide (=NH) or nitride (≡N) products.

Reading: this is the structural mirror of aqueous hydrolysis, $\text{E–X} + \text{H}_2\text{O} \rightarrow \text{E–OH} + \text{HX}$ ; wherever hydrolysis gives an oxide/hydroxide ladder (M–OH → M=O), ammonolysis gives the amide/imide/nitride ladder (M–NH₂ → M=NH → M≡N), and the depth to which the ladder proceeds depends on temperature and on the thermodynamic stability of the higher-condensed nitrogen species, exactly as the oxide/hydroxide equilibrium depends on pH and temperature in water.

\Delta G_{\text{solv}}(e^-) = \Delta H_{\text{cavity}} - T\Delta S_{\text{cavity}} + \Delta H_{\text{polarization}}

where: $\Delta H_{\text{cavity}}$ is the (positive) energy cost of opening a cavity of the right size in the hydrogen-bonded ammonia network, $\Delta S_{\text{cavity}}$ is the entropy of localizing the solvent dipoles around the trapped charge, and $\Delta H_{\text{polarization}}$ is the (negative, stabilizing) electrostatic interaction of the electron with the oriented N–H dipoles.

Reading: the solvated electron is thermodynamically stable in liquid ammonia because the polarization stabilization from many oriented, weakly hydrogen-bonded NH₃ dipoles exceeds the ionization energy given up by the metal atom (minus lattice/atomization energy); water cannot support a long-lived analogous species at ordinary concentrations because the hydrated electron reacts essentially instantaneously with water itself ( $e^-(aq) + \text{H}_2\text{O} \rightarrow \text{H} \cdot + \text{OH}^-$ occurs on a nanosecond–microsecond timescale, whereas $e^-(am) + \text{NH}_3$ is very slow at low temperature, requiring catalysis by transition-metal impurities to proceed at a useful rate).

Derivation AWhy the metal–ammonia dissolution is thermodynamically favorable — a Born–Haber-style cycle

Goal. Show, using an approximate thermochemical cycle, why an alkali metal such as sodium dissolves spontaneously in liquid ammonia to give solvated cation plus solvated electron, whereas the same process in the gas phase (bare ionization with no solvation) is strongly unfavorable.

Step 1 — write the overall process as a cycle. The dissolution

\text{Na}(s) \rightarrow \text{Na}^+(am) + e^-(am)

can be broken into steps whose energetics are separately known or estimable: (i) atomization of the solid metal, $\text{Na}(s) \rightarrow \text{Na}(g)$ , enthalpy $\Delta H_{\text{at}} = +107\ \text{kJ mol}^{-1}$ ; (ii) gas-phase ionization, $\text{Na}(g) \rightarrow \text{Na}^+(g) + e^-(g)$ , enthalpy equal to the first ionization energy, $I_1 = +496\ \text{kJ mol}^{-1}$ ; (iii) solvation of the bare cation, $\text{Na}^+(g) \rightarrow \text{Na}^+(am)$ , enthalpy $\Delta H_{\text{solv}}(\text{Na}^+) \approx -410\ \text{kJ mol}^{-1}$ (a value comparable in magnitude to, though somewhat smaller than, the aqueous hydration enthalpy of Na⁺, reflecting ammonia's smaller dipole moment and weaker short-range electrostatics); and (iv) solvation of the free electron, $e^-(g) \rightarrow e^-(am)$ , enthalpy $\Delta H_{\text{solv}}(e^-) \approx -190\ \text{kJ mol}^{-1}$ (estimated from spectroscopic and thermochemical studies of metal–ammonia solutions; it is this last step that has no counterpart at all in aqueous thermochemistry under ordinary conditions, since $e^-(aq)$ decomposes rather than persisting as a stable reference state).

Step 2 — sum the cycle.

\Delta H_{\text{overall}} = \Delta H_{\text{at}} + I_1 + \Delta H_{\text{solv}}(\text{Na}^+) + \Delta H_{\text{solv}}(e^-)

\Delta H_{\text{overall}} = (+107) + (+496) + (-410) + (-190)\ \text{kJ mol}^{-1}

\Delta H_{\text{overall}} \approx +3\ \text{kJ mol}^{-1}

Step 3 — interpret the near-zero result. The overall enthalpy for producing solvated Na⁺ and solvated e⁻ separately is close to thermoneutral — small compared with any single term in the cycle, meaning the coupled stabilization of both cation and electron in the ammonia cavity structure is what makes dissolution occur, not any one contribution dominating. Because dissolution also increases translational entropy substantially (one mole of ordered solid becomes two moles of independently mobile solvated species, each with its own solvent shell), $\Delta S_{\text{overall}} > 0$ and therefore

\Delta G_{\text{overall}} = \Delta H_{\text{overall}} - T\Delta S_{\text{overall}} < 0

at ambient and sub-ambient temperatures, driving spontaneous dissolution even though the enthalpy alone is roughly balanced. This is qualitatively different from the aqueous case, where the corresponding hypothetical process is strongly endothermic because water lacks any mechanism analogous to step (iv) — instead of solvating a free electron, water is reduced by it, releasing H₂ and heat, which is why sodium in water reacts violently rather than dissolving reversibly.

Step 4 — connect to observed behavior. Because $\Delta H_{\text{overall}}$ is small and near zero, the process is easily reversible: evaporating the ammonia from a metal–ammonia solution regenerates the pure metal quantitatively (this is in fact a standard preparative method for very pure alkali metal samples and for compounds like the alkalides and electrides that trap $M^+$ and $e^-$ , or $M^-$ , in crystalline solvates). The reversibility is direct experimental evidence that no irreversible chemical reaction with the solvent has occurred — the metal is simply ionized and solvated, not oxidized by the solvent, at least at low concentration and low temperature.

Go deeper · Derivation B · 2 minAmmonolysis of a covalent halide — the amide/imide/nitride ladder as ammonia's analog of the hydroxide/oxide ladder

Derivation BAmmonolysis of a covalent halide — the amide/imide/nitride ladder as ammonia's analog of the hydroxide/oxide ladder

Goal. Use PCl₃ (chosen as a representative covalent nonmetal halide, structurally and electronically analogous to species whose aqueous hydrolysis is well known) to show how ammonolysis proceeds by nucleophilic substitution at phosphorus and can be driven stepwise to higher-condensed nitrogen products, in exact parallel with the stepwise hydrolysis of P–Cl bonds to P–OH and eventually P=O species.

Step 1 — first substitution: NH₃ attacks P, Cl⁻ (as NH₄Cl) leaves. The nitrogen lone pair of ammonia is nucleophilic and attacks the electrophilic phosphorus center of PCl₃, displacing chloride, which is immediately protonated by a second ammonia molecule to form ammonium chloride:

\text{PCl}_3 + 6\,\text{NH}_3 \rightarrow \text{P(NH}_2)_3 + 3\,\text{NH}_4\text{Cl}

Mechanistically this is an associative substitution at a trivalent, pyramidal phosphorus(III) center — nitrogen's lone pair forms the new P–N σ bond while a P–Cl bond breaks heterolytically, with the departing chloride solvated and stabilized by protonation from bulk ammonia (the ammonia here is playing two simultaneous roles: nucleophile and Brønsted base/proton source, exactly as water plays both roles in hydrolysis of PCl₃ to P(OH)₃/H₃PO₃).

Step 2 — recognize the aqueous parallel exactly. The corresponding hydrolysis is

\text{PCl}_3 + 3\,\text{H}_2\text{O} \rightarrow \text{P(OH)}_3 + 3\,\text{HCl}

followed by rapid tautomerization of P(OH)₃ to the thermodynamically preferred H-phosphonate form, HP(O)(OH)₂. The ammonolysis product P(NH₂)₃ (triaminophosphine) is the direct nitrogen analog of P(OH)₃, and it is similarly prone to further condensation.

Step 3 — condensation to imide- and nitride-bridged species (the "ladder"). Just as P–OH groups on adjacent centers can condense with loss of water to form P–O–P linkages (polyphosphates, phosphate glasses), P–NH₂ groups can condense thermally with loss of ammonia to form P–NH–P (imide-bridged) and eventually P–N=P or cyclic/polymeric P–N frameworks approaching the nitride limit:

2\,\text{P(NH}_2)_3 \xrightarrow{\Delta,\ -\text{NH}_3} (\text{H}_2\text{N})_2\text{P–NH–P(NH}_2)_2 \xrightarrow{\Delta,\ -\text{NH}_3} \cdots \rightarrow (\text{PN})_n \text{ polymeric/cyclic phosphazene-type frameworks}

This is the ammonia-world counterpart of oxide condensation ladders such as orthophosphate → pyrophosphate → metaphosphate → P₂O₅, with –NH₂ replacing –OH, bridging –NH– replacing bridging –O–, and terminal ≡N (nitride) replacing terminal =O (oxide) at the fully condensed limit.

Step 4 — generalize and state the scope condition. The general ammonolysis ladder for an element E is

\text{E–X}_n \xrightarrow{\text{NH}_3} \text{E(NH}_2)_n \xrightarrow{-\text{NH}_3,\ \Delta} \text{E(NH)}_{n/2} \xrightarrow{-\text{NH}_3,\ \Delta} \text{E}_3\text{N}_n\ (\text{nitride limit, schematic})

exactly mirroring

\text{E–X}_n \xrightarrow{\text{H}_2\text{O}} \text{E(OH)}_n \xrightarrow{-\text{H}_2\text{O},\ \Delta} \text{EO}_{n/2} \xrightarrow{} \text{oxide, further condensation}

How far any real system proceeds down the ladder depends on the thermodynamic stability of the higher condensed nitrogen phase relative to the amide, on kinetic accessibility (activation barriers to NH₃ loss, which generally require appreciable heating), and on whether excess ammonia (favoring the less-condensed, NH₃-rich side by Le Chatelier's principle) or forcing dry/hot conditions (favoring condensation) are used. Well known examples that make it all the way to a genuine nitride include the ammonolysis of many metal halides at high temperature (e.g., TiCl₄ → TiN-type products under forcing thermal ammonolysis, and B–Cl or Si–Cl bonds ammonolyzing toward BN and Si₃N₄-related networks), while milder aqueous-analogous stops at the amide or imide stage are common for main-group nonmetal halides at low temperature, exactly as many metal halides hydrolyze only as far as the hydroxide at low temperature and require calcination to reach the oxide.

Worked examples

Example 1 — Foundational: classify species as ammono-acid, ammono-base, or ammono-neutral, and write the neutralization. Consider dissolving ammonium nitrate, NH₄NO₃, and potassium amide, KNH₂, together in liquid ammonia at −40 °C, in a 1 : 1 mole ratio, and predict the products. NH₄NO₃ dissociates in the ionizing solvent to give NH₄⁺(am) and NO₃⁻(am); since it raises the ammonium ion concentration above the ammono-neutral value ( $\sqrt{K_{\text{auto}}}\approx 3\times10^{-17}\,\text{M}$ at the relevant temperature), it is classified as an ammono-acid — the exact role HNO₃ plays in water. KNH₂ dissociates to K⁺(am) and NH₂⁻(am); since it raises the amide ion concentration, it is an ammono-base, playing the role KOH plays in water. Mixed in a 1 : 1 ratio, the acid–base neutralization

\text{NH}_4^+ + \text{NH}_2^- \rightarrow 2\,\text{NH}_3

consumes both characteristic ions, leaving the spectator ions K⁺ and NO₃⁻ in solution as the ammono-analog of a "neutral salt" — the overall metathesis is

\text{NH}_4\text{NO}_3 + \text{KNH}_2 \rightarrow \text{KNO}_3 + 2\,\text{NH}_3

which is the structural mirror of the aqueous $\text{HNO}_3 + \text{KOH} \rightarrow \text{KNO}_3 + \text{H}_2\text{O}$ . As a limiting-case check: if instead of a 1 : 1 ratio one used a large excess of KNH₂, the resulting solution would remain strongly ammono-basic (excess NH₂⁻ present), exactly as excess KOH leaves an aqueous solution strongly basic — the analogy holds at any stoichiometric ratio, not just at the equivalence point.

Example 2 — Applied: predicting solubility behavior via metathesis (double decomposition) in liquid ammonia. Silver nitrate and potassium chloride are both freely soluble in liquid ammonia individually. Predict what happens when solutions of the two are mixed, and contrast with the aqueous case. In water, mixing AgNO₃(aq) and KCl(aq) precipitates AgCl (Ksp ≈ 1.8 × 10⁻¹⁰), leaving K⁺ and NO₃⁻ in solution — the textbook metathesis/double-decomposition reaction. In liquid ammonia the same metathesis

\text{AgNO}_3 + \text{KCl} \rightarrow \text{AgCl}\downarrow + \text{KNO}_3

also occurs, and again AgCl precipitates, because silver ion's affinity for chloride (a hard–soft mismatch resolved by strong Ag–Cl covalent character and very low lattice solubility) outcompetes solvation by ammonia even though Ag⁺ itself is an excellent ammonia-solvated ion (forming $[\text{Ag(NH}_3)_2]^+$ readily, which is in fact why AgCl — insoluble in water — redissolves in aqueous ammonia as the diammine complex; but in the non-aqueous solvent itself, with no competing bulk water, the halide lattice energy of AgCl dominates and precipitation still occurs). The qualitative solubility "rules" are not identical between the two solvents (many salts insoluble in water are soluble in ammonia and vice versa, because ammonia solvates cations via strong N-donor coordination but solvates anions weakly, lacking a good hydrogen-bond-donor capability comparable to water's), but the mechanism and driving force — precipitation of the least-soluble combination, driven by lattice energy exceeding solvation energy — is exactly parallel. As the explicit limiting case: were both AgCl and KNO₃ appreciably soluble in ammonia (which in fact some analogous ammonia-soluble halide/nitrate pairs are), no precipitate would form and the ions would simply remain scrambled in solution — the metathesis "rule" is only ever a solubility-driven special case, not a universal law, in either solvent.

Example 3 — Exam/competition level: quantitative estimate of the concentration at which a metal–ammonia solution becomes metallic, using a Mott-transition–style estimate. Sodium–ammonia solutions undergo a well-documented transition from a non-metallic, localized-electron regime (dilute blue solutions, paramagnetic, low conductivity per electron) to a metallic, delocalized regime (concentrated bronze solutions, Pauli-paramagnetic, conductivity comparable to a liquid metal) as the metal mole fraction rises through a range centered near 4 mol % metal (roughly the concentration at which the mean electron–electron/cavity spacing becomes comparable to the effective Bohr radius of the solvated electron). Estimate this transition concentration from a simple Mott-criterion argument and compare with experiment. The Mott criterion for delocalization in a system of dilute, hydrogen-atom-like centers of effective Bohr radius $a^*$ states that the metal–insulator transition occurs near the electron density $n_c$ satisfying

n_c^{1/3}\,a^* \approx 0.25

Take the effective Bohr radius of the solvated electron in ammonia as roughly $a^* \approx 3.0\ \text{Å}$ (estimated from the optical absorption energy of the solvated electron band, treating it as a particle-in-a-spherical-box or hydrogenic-like state; this value is broadly consistent with independent estimates from the observed near-IR absorption maximum near 1500 nm). Then

n_c^{1/3} \approx \frac{0.25}{3.0\ \text{Å}} = 0.0833\ \text{Å}^{-1}

n_c \approx (0.0833)^3\ \text{Å}^{-3} = 5.8\times10^{-4}\ \text{Å}^{-3} = 5.8\times10^{20}\ \text{cm}^{-3}

Converting to molarity,

n_c = 5.8\times10^{20}\ \text{cm}^{-3} \times \frac{1\ \text{mol}}{6.022\times10^{23}} \times \frac{1000\ \text{cm}^3}{\text{L}} \approx 0.96\ \text{mol L}^{-1}

Pure liquid ammonia has a molarity of approximately

[\text{NH}_3] = \frac{\rho}{M} = \frac{0.68\ \text{g mL}^{-1}}{17.0\ \text{g mol}^{-1}} \times 1000\ \text{mL L}^{-1} \approx 40\ \text{mol L}^{-1}

so the estimated critical mole fraction of dissolved electrons (≈ metal) is

x_c \approx \frac{0.96}{40} \approx 0.024,\ \text{i.e. about}\ 2.4\ \text{mol \%}

This simple hydrogenic Mott estimate lands within roughly a factor of 1.5–2 of the experimentally observed crossover (commonly cited near 4 mol % metal, corresponding to the Mott density where electron wavefunctions in neighboring cavities begin to overlap appreciably), which is the expected level of agreement for a one-parameter order-of-magnitude estimate of this kind — the physical picture (delocalization once cavity wavefunctions overlap) is confirmed even though the specific numerical value depends sensitively on the assumed $a^*$ . As a limiting-case check: if $a^*$ were instead taken as small as a bare, uncompensated electron's de Broglie scale (sub-angstrom), $n_c$ would be predicted absurdly high (well above the density of pure ammonia itself), confirming that the transition can only be physically sensible for a solvated, spatially extended electron state of several-angstrom radius — precisely the kind of state that ammonia (and no comparably H-bonded solvent like water) is able to stabilize as a genuine, long-lived species.

Limiting cases, scope & common traps

Situation	Aqueous analogy	What actually happens in NH₃(l)	Why
Alkali metal added to solvent	Violent reaction with H₂O, H₂ evolved	Dissolves reversibly, deep blue solution, no H₂ (at low T, short time)	NH₃ stabilizes free e⁻ in a solvation cavity; H₂O reduces the electron irreversibly to H• + OH⁻
NH₄⁺ salts	Analogous to strong/weak acids in water	Ammono-acids; NH₄Cl, NH₄NO₃ act as "HCl", "HNO₃"	NH₄⁺ is the solvent's characteristic cation
Amide salts (NaNH₂)	Analogous to hydroxides	Ammono-bases, strong; used industrially in ammonia and organic synthesis	NH₂⁻ is the solvent's characteristic anion
Covalent halide + solvent	Hydrolysis to hydroxide/oxide	Ammonolysis to amide/imide/nitride	Same nucleophilic substitution mechanism, N lone pair replacing O lone pair
Precipitation metathesis	Standard double-decomposition, aqueous $K_{sp}$ rules	Same driving force, different solubility order	Ammonia solvates cations (soft, polarizable) well, anions poorly; opposite bias to water
Dilute metal solution (< 0.01 M)	No aqueous analog	Blue, paramagnetic, localized $e^-$ in separate cavities	Low density of solvated-electron centers, no wavefunction overlap
Concentrated metal solution (> 3–4 mol %)	No aqueous analog	Bronze, metallic conduction, Pauli-paramagnetic	Cavities overlap; delocalized conduction band forms (Mott transition)
Self-ionization constant	$K_w \approx 10^{-14}$	$K_{\text{auto}} \approx 10^{-33}$	Far weaker intrinsic proton self-transfer; ammonia is both a much weaker acid and a much weaker Brønsted base per molecule than water in the pure liquid, though it is an excellent Lewis base

Common trapCommon misconception 1

"Ammonia is simply a weaker version of water — same chemistry, milder." This is wrong in a specific and important way: ammonia's lower K_auto makes it a poorer auto-ionizing solvent for classical acid–base chemistry, but its distinctive electronic structure (exposed N lone pair, higher-lying antibonding orbitals, weaker O–H-style hydrogen bonding) makes it uniquely capable of stabilizing solvated electrons and forming extensive ammine coordination chemistry that water cannot match. "Weaker" and "different" are not the same claim; treating ammonia as merely an attenuated water leads students to wrongly predict, for instance, that alkali metals should react violently with NH₃(l) the way they do with H₂O, when in fact the opposite (reversible dissolution) is observed.

The chemist behind it

Origins · 1 minEdward C. Franklin and how this was discovered

Edward C. Franklin

1862–1937 · American chemist, Stanford University

Franklin, working largely at Stanford in the first two decades of the twentieth century (with his long-time collaborator Charles A. Kraus contributing much of the physical/electrochemical side of metal–ammonia solution chemistry), was the principal architect of the systematic "ammonia system" of acids, bases, and salts — the explicit generalization of aqueous acid–base and hydrolysis concepts to liquid ammonia as solvent. He introduced the terminology "ammono-acid," "ammono-base," and "ammonolysis" by direct analogy with the corresponding aqueous terms, argued (correctly, and ahead of most contemporaries) that the acid–base concept should be defined relative to the solvent's own self-ionization rather than tied exclusively to hydrogen and oxygen, and carried out extensive preparative work isolating and characterizing amides, imides, and nitrides made by ammonolysis of metal and metalloid halides. His framework is the direct historical root of the general solvent-system concept later formalized more broadly (and extended to other ionizing, non-aqueous, and even non-protonic solvents).

History

- 1809 — Davy first reports that potassium dissolves in liquid ammonia gas to give a blue coloration, though the systematic chemistry is not developed for nearly a century.

- 1897–1900s — Franklin, working with Kraus at Johns Hopkins and later Stanford, develops the systematic amide/ammono-acid framework and coins the term "ammonolysis."

- 1908 — Kraus reports that concentrated alkali metal–ammonia solutions display metallic electrical conductivity, establishing the metal–insulator transition as an experimental fact.

- 1920s–1930s — Franklin's monograph "The Nitrogen System of Compounds" systematizes ammono-acid/base/salt classification and ammonolysis reactions across the periodic table.

- 1950s–1970s — ESR, optical, and magnetic susceptibility studies (Ogg, Kaplan and Kittel, and subsequent workers) establish the modern solvated-electron/cavity picture and quantify the Mott transition in concentrated metal–ammonia solutions.

Why doesn't liquid ammonia react violently with alkali metals the way water does?

Water reacts with sodium because the resulting solvated electron is thermodynamically and kinetically unstable in water: it rapidly reduces a water molecule to give H• (which combines to H₂) and OH⁻, releasing substantial heat and driving the reaction to completion within microseconds. In liquid ammonia the analogous reduction of solvent, $e^-(am) + \text{NH}_3 \rightarrow \text{H}\cdot + \text{NH}_2^-$ , is thermodynamically much less favorable and, critically, extremely slow in the absence of catalytic impurities (transition-metal ions catalyze it strongly). This kinetic stability — not a fundamentally different thermodynamic driving force alone — is what allows the deep blue solvated-electron solution to persist for extended periods and be studied, titrated, and used synthetically, rather than decomposing on contact as it effectively does in water.

Is liquid ammonia a better or worse solvent than water for dissolving ionic salts?

Neither, universally — it depends on the ion. Ammonia has a lower dielectric constant (≈22 versus water's 78) and is a poor hydrogen-bond donor to anions, so it solvates small, hard anions (F⁻, OH⁻, and many oxoanions) much less effectively than water and correspondingly dissolves fewer simple ionic halides and hydroxides. But ammonia is an excellent σ-donor toward cations, especially soft, polarizable, or coordinatively unsaturated ones (Ag⁺, Cu⁺, many transition-metal ions, and NH₄⁺ itself), so salts and complexes built around such cations, plus many covalent and organic solutes, dissolve in liquid ammonia far more readily than in water. The "solubility rules" are genuinely different, not simply attenuated versions of the aqueous rules.

What exactly is an "ammono-acid," and how does it differ from a Brønsted acid in the usual aqueous sense?

An ammono-acid is defined purely in terms of the solvent's own autoionization: any solute that increases the concentration of NH₄⁺ (the solvent's characteristic cation) above the "ammono-neutral" value is an ammono-acid, exactly as an aqueous Brønsted acid increases [H₃O⁺]. Ammonium salts (NH₄Cl, NH₄NO₃) are the clearest ammono-acids because they dissociate directly to give NH₄⁺. This is not a fundamentally new kind of acidity — it is the same Brønsted proton-transfer concept, just referenced to the solvent's own conjugate-acid species (NH₄⁺) rather than to water's (H₃O⁺); the definitions coincide once you recognize both as instances of the general solvent-system concept.

Why does the solvated electron in ammonia absorb in the near-infrared/visible rather than the far infrared or ultraviolet?

The absorption reflects the energy gap between the electron's ground state in the solvation cavity and its first excited (or continuum) state, which is set by the size and depth of the cavity potential well created by the oriented ammonia dipoles. Because the cavity is a few angstroms across — comparable to, though larger than, a typical atomic radius — the energy-level spacing lands in the near-infrared to visible range (absorption maximum near 1500 nm, with a tail extending into the visible that produces the characteristic blue color), analogous to how quantum-confined systems (particle-in-a-box, quantum dots) show size-tunable optical gaps; a much smaller or much larger cavity would shift the absorption toward the ultraviolet or far infrared respectively.

Does liquid ammonia have a genuine analog of the hydronium ion, and if so why is it chemically less prominent?

Yes — NH₄⁺ is the exact structural and functional analog of H₃O⁺, formed by protonation of a solvent molecule by any stronger ammono-acid. It is less prominent in everyday chemical discussion mainly because ammonium salts are extremely familiar compounds studied for their own sake (fertilizers, buffers) independent of the solvent-system framework, and because ammonia's intrinsically low self-ionization constant means "ammono-pH" chemistry in the strict sense (fine acid–base titration in pure liquid ammonia) is a comparatively specialized subfield, practiced mainly in inorganic and physical-organic laboratories studying reactions that must avoid water altogether.

How does ammonolysis of a metal halide reach the nitride stage, and why does this matter industrially?

Ammonolysis proceeds by stepwise nucleophilic substitution: NH₃ displaces halide to give an amide (M–NH₂), which on heating condenses with loss of further NH₃ to give bridging imide (M–NH–M) and ultimately fully condensed nitride (M≡N or extended M–N network) linkages, in exact parallel with the hydroxide-to-oxide condensation ladder in aqueous hydrolysis chemistry. This matters industrially because high-performance ceramic nitrides — boron nitride, silicon nitride, aluminum nitride, widely used for their hardness, thermal conductivity, and electrical insulation properties — are frequently manufactured by exactly this route: ammonolysis of the corresponding halide or oxide precursor at elevated temperature, driving the condensation ladder all the way to the nitride endpoint.

Why is the metal–insulator (Mott) transition in metal–ammonia solutions considered important beyond inorganic chemistry specifically?

Concentrated metal–ammonia solutions are one of the cleanest experimental realizations of a continuously tunable metal–insulator transition in a disordered, liquid medium: by simply varying the metal concentration, the system can be driven smoothly from a localized, paramagnetic, non-conducting regime to a delocalized, Pauli-paramagnetic, metallically conducting regime, without changing crystal structure or introducing lattice disorder in the way a solid-state Mott transition would require. This makes the system a valuable testbed for theories of electron localization, percolation, and disorder-driven metal–insulator transitions that are otherwise studied in doped semiconductors, expanded liquid metals, and other strongly correlated electron systems.