Adduct Formation and Frustrated Lewis Pairs

Frustrated Lewis pair chemistry has moved from a laboratory curiosity to a genuine platform for metal-free catalysis with real industrial and environmental relevance. FLP-catalyzed hydrogenation of imines, enamines, and some polarized C=C/C=N substrates now provides routes to chiral amines and other reduced products without precious-metal catalysts, an attractive feature given the cost and toxicity concerns associated with platinum-group metals in fine chemical and pharmaceutical synthesis. Because many FLPs operate under comparatively mild H₂ pressures and temperatures, they are being explored as components of hydrogen-storage materials, where reversible FLP-type H₂ binding/release cycles offer an alternative to purely physical storage methods, provided the thermoneutrality challenge illustrated in Example 3 can be engineered by tuning substituents. FLPs also activate and functionalize CO₂, forming zwitterionic carboxylate/borate adducts in which the phosphine (or amine) attacks the electrophilic carbon of CO₂ while the borane captures one of the developing oxygen lone pairs — a reaction directly analogous in mechanism to the H₂ splitting derived above, substituting CO₂'s C=O π-system for H₂'s σ-bond as the substrate undergoing simultaneous nucleophilic and electrophilic attack. This chemistry is actively studied as part of broader carbon capture and utilization research, since main-group FLP systems can sometimes be tuned for milder regeneration conditions than metal-carbonate or amine-scrubbing systems. More broadly, the frustrated-pair concept has propagated well beyond phosphine-boranes: carbene-borane pairs, geometrically frustrated aluminum/nitrogen systems, and even frustrated radical pairs have all been reported, and the underlying steric-electronic logic developed here — comparing intrinsic bond strength against a steric penalty that scales steeply past a geometric threshold — recurs throughout inorganic and organometallic chemistry wherever chemists deliberately use bulky ligands to keep a reactive center "unquenched" for downstream reactivity. The same reasoning underlies bulky N-heterocyclic carbene and pincer-ligand design in transition-metal catalysis, and it sets up the natural next question tackled under Hard-Soft Acid-Base Applications in Synthesis: given that an FLP's two frontier orbitals remain available, which substrates will they prefer, and why does that preference track hardness/softness rather than simple sterics alone?

Ask the tutor: How does the choice of substituents on an FLP's boron center change its selectivity between H₂, CO₂, and alkynes as substrates?

Classical Lewis adduct formation is fundamentally an orbital-overlap event. A donor D has a filled orbital of appropriate symmetry (often an sp³ or sp² lone pair, occasionally a π-system) that overlaps with the empty acceptor orbital of A, most commonly an unhybridized p-orbital on a trigonal-planar boron, aluminum, or a σ orbital on a metal center. The result is a new two-electron, two-center D→A σ-bond, and the acceptor geometry usually relaxes from trigonal planar toward tetrahedral (sp²→sp³ at boron, for instance). The enthalpy released, ΔH_adduct, reflects both the intrinsic strength of the new bond and the energetic cost of any structural reorganization needed to reach the product geometry. Steric bulk enters twice. First, bulky substituents on the donor (like the tert-butyl groups of di-tert-butylphosphine) or acceptor (the ortho-fluorines and ipso ring bulk in tris(pentafluorophenyl)borane, B(C₆F₅)₃) physically increase the distance and torque needed to bring the reactive atoms into bonding contact — this is captured by ligand cone angles and A-values. Second, once an adduct nearly forms, the eclipsing/clashing interactions between substituents on D and A destabilize the pyramidalized product, effectively subtracting from the intrinsic bond energy. When this steric penalty exceeds the intrinsic D–A bond strength, no stable adduct exists: the system is "frustrated." Frustration does not quench reactivity — it redirects it. Both the lone pair on D and the empty orbital on A remain fully "charged up," waiting for something to react with. If a small molecule such as H₂, CO₂, SO₂, or an alkyne approaches the pocket between the two bulky partners, it experiences simultaneous* polarization: the acceptor's empty orbital withdraws electron density from one part of the substrate (often forming an incipient hydride or forming a new A–X bond), while the donor's lone pair pushes density into the remaining fragment (forming a new D–H⁺ or D–X bond). The net transformation looks like heterolytic cleavage of the substrate mediated cooperatively by two separate molecules (or two ends of one molecule) that could never have combined with each other.

The frontier-orbital picture is easiest to see in molecular orbital terms. In the unhindered pair, the donor HOMO (lone pair, roughly sp³ in character, energy ε_HOMO) and the acceptor LUMO (empty p_z orbital on boron, energy ε_LUMO) mix directly, and the bonding combination drops below ε_HOMO by an amount proportional to the square of their overlap integral divided by the energy gap — the familiar second-order perturbation result. Steric bulk does not change ε_HOMO or ε_LUMO very much (it is not primarily an electronic effect); it reduces the overlap integral S between them by forcing a longer D···A distance and a poorer angle of approach. Because the stabilization scales as S², even a modest overlap reduction from steric crowding can collapse the bonding stabilization by an order of magnitude, tipping the balance so that steric repulsion (which grows steeply, often faster than S² at short range due to Pauli/exchange repulsion) wins.

1. Adduct dissociation thermodynamics. For the general equilibrium D + A ⇌ D–A, the free energy of adduct formation controls whether the classical pathway dominates:

where:

• $\Delta H_{\text{adduct}}$ is the enthalpy of the new D–A σ-bond minus reorganization costs (pyramidalization at A, loss of donor conjugation, etc.),
• $\Delta S_{\text{adduct}}$ is typically strongly negative (two molecules become one, losing translational/rotational entropy),
• $T$ is temperature.

Reading: unhindered strong donor-acceptor pairs (Me₃N·BMe₃, ΔH ≈ −73 kJ/mol) have $\Delta H$ large enough in magnitude to overwhelm the entropic penalty at ordinary temperatures, so $\Delta G_{\text{adduct}} < 0$ and the adduct is thermodynamically stable. As bulk increases, $\Delta H_{\text{adduct}}$ becomes less negative (weaker net bonding after steric penalty) while $\Delta S_{\text{adduct}}$ stays roughly the same order, so past a threshold $\Delta G_{\text{adduct}}$ turns positive and no stable 1:1 adduct exists — the pair is frustrated.

2. Steric destabilization estimate (cone-angle overlap). A practical way to estimate whether frustration is geometrically forced is to compare the sum of the cone angles/A-values of the substituents to the space available at the bonding distance:

where:

• $\theta_D$, $\theta_A$ are effective cone half-angles (or summed A-values in kcal/mol as a steric proxy) of the donor and acceptor substituents,
• $\theta_{\text{crit}}$ is the critical combined angle beyond which substituents begin to clash at the normal D–A bonding distance,
• $k$ is an empirical force constant (units to make $E_{\text{steric}}$ an energy),
• the expression is zero (no penalty) when $\theta_D + \theta_A \le \theta_{\text{crit}}$.

Reading: this is a phenomenological quadratic "steric spring," analogous to a repulsive wall — mild crowding costs little, but past the critical overlap the penalty rises steeply (like van der Waals repulsion, which formally goes as $r^{-12}$ at short range). It rationalizes why small changes in substituent size (methyl → tert-butyl, or borane → B(C₆F₅)₃) can switch a system from "always forms an adduct" to "never forms an adduct" over a narrow structural range — precisely what is observed experimentally for phosphine/borane FLPs.

3. Heterolytic H₂ activation by an FLP. The signature reaction of frustrated Lewis pairs is splitting dihydrogen without a transition metal. The overall thermodynamics can be tracked through a Born–Haber–like cycle:

where:

• $D_{\text{H-H}}$ is the H–H bond dissociation energy (≈ 436 kJ/mol),
• $\Delta H_{\text{D-H}}$ is the strength of the new D–H (typically N–H or P–H) bond formed at the base,
• $\Delta H_{\text{A-H}}$ is the strength of the new A–H (typically B–H) bond formed at the acid,
• $E_{\text{ion-pair}}$ is the net electrostatic/solvation stabilization of the resulting [D–H]⁺[A–H]⁻ ion pair relative to two neutral radicals or a homolytic split.

Reading: unlike ordinary bond-breaking, this is a heterolytic process assisted by two separate binding events happening at once; even though $D_{\text{H-H}}$ is large, the sum of two new strong polar bonds (N–H ≈ 391 kJ/mol-type strength, B–H comparably strong) plus a favorable Coulombic ion-pair term can make the overall reaction exothermic and kinetically accessible at room temperature — which is exactly what is observed for pairs like tBu₃P/B(C₆F₅)₃.

Example 1 — Foundational: quantifying why Me₃N·BMe₃ forms but a bulkier analogue does not. Take the well-established gas-phase/solution data point that trimethylamine and trimethylborane form a stable 1:1 adduct with

Now compare a bulkier pair, 2,6-lutidine (2,6-dimethylpyridine) with BMe₃. Steric clash between the ortho-methyl groups of the pyridine and the methyls on boron destabilizes the adduct substantially relative to unhindered pyridine·BMe₃ (ΔH ≈ −71 kJ/mol for pyridine·BMe₃ itself). Experimentally, 2,6-lutidine·BMe₃ has a dissociation enthalpy reduced by roughly 40–50 kJ/mol relative to the pyridine case due almost entirely to the two additional ortho methyl clashes — consistent with an added steric penalty of order

per pair of ortho substituents in this specific system (an experimentally back-calculated number, not a universal constant). Plugging representative numbers into the enthalpy decomposition,

still favorable enough at room temperature for an isolable, if weaker, adduct — 2,6-lutidine·BMe₃ is indeed known but is markedly more thermally labile and more easily displaced than the pyridine analogue.

Limiting-case check: if the steric penalty is set to zero ($E_{\text{steric}}\to 0$, i.e., an unhindered base such as pyridine itself), the formula correctly reduces to $\Delta H_{\text{adduct}} = \Delta H_{\text{bond}} = -71$ kJ/mol, recovering the known strong, stable adduct — confirming the decomposition behaves sensibly at the no-strain limit.

Example 2 — Applied: estimating whether tBu₃P/B(C₆F₅)₃ is frustrated using the crossover-angle criterion. Tri-tert-butylphosphine has a very large Tolman cone angle, $\theta_D \approx 182°$, among the bulkiest common phosphines. Tris(pentafluorophenyl)borane, though nominally similar in size to BPh₃, is an unusually strong, sterically demanding Lewis acid with an effective cone angle (accounting for the three C₆F₅ rings) estimated around $\theta_A \approx 184°$ using comparable cone-angle conventions extended to boron acceptors. Their sum,

vastly exceeds a typical critical combined angle for direct P–B bond formation, generally estimated (by comparison with pairs that do form stable adducts, such as PMe₃/BF₃ with cone-angle sums well under 300°) at roughly

Since $\theta_D + \theta_A - \theta_{\text{crit}} \approx 70\text{–}95°$, a substantial positive excess remains even allowing generous uncertainty in these empirical angle conventions, and the quadratic steric penalty $E_{\text{steric}} = k(\theta_D+\theta_A-\theta_{\text{crit}})^2$ becomes large enough (empirically, tens to well over 100 kJ/mol for excesses of this size in related systems) to overcome any plausible intrinsic P→B bond strength (typically 30–50 kJ/mol for a phosphine-borane dative bond of this type). The prediction — no observable P–B adduct, i.e., a genuinely frustrated pair — matches the experimental finding that tBu₃P and B(C₆F₅)₃ remain uncombined in solution and react instead with H₂, forming [tBu₃PH]⁺[HB(C₆F₅)₃]⁻ readily at room temperature and modest H₂ pressure.

Limiting-case check: if instead the phosphine were replaced by unhindered PMe₃ ($\theta_D \approx 118°$), the combined angle with B(C₆F₅)₃ becomes

just above or comparable to $\theta_{\text{crit}}$, correctly predicting only marginal frustration — and indeed PMe₃ does form a classical (if somewhat labile) adduct with B(C₆F₅)₃, unlike the bulky tBu₃P case, confirming the criterion discriminates correctly between the two regimes.

Example 3 — Exam/competition level: energetics of an intramolecular ansa-FLP splitting H₂, with a full thermochemical cycle. Consider an ansa (tether-linked) phosphine-borane FLP, R₂P–CH₂–CH₂–B(C₆F₅)₂, geometrically forced by the short tether to keep P and B close (roughly 4.9 Å apart, well beyond bonding distance) but unable to reach true P–B bonding distance (~2.0 Å) without intolerable ring strain in the connecting ethylene bridge. Suppose this FLP reacts with H₂ (1 atm, 25 °C) to give the intramolecular zwitterion [R₂P(H)–CH₂–CH₂–B(H)(C₆F₅)₂], and you are given the following representative bond energies:

(these are representative magnitudes for P–H and B–H single-bond formation enthalpies in this electronic environment, not universal constants, since actual bond strengths depend sensitively on substituents). Using the thermochemical cycle,

(bond energies of the new bonds enter as negative — exothermic — contributions, while the H–H bond broken enters as a positive cost, following the convention that breaking bonds costs energy and forming bonds releases it):

a strongly exothermic result, consistent with the general experimental observation that FLP-mediated H₂ splitting is thermodynamically quite favorable once a suitable geometric pocket exists — the reaction proceeds readily at or below room temperature and 1 atm H₂ for many tether-linked and even some untethered intermolecular FLPs. Now examine the reverse process (H₂ release, relevant to FLP-catalyzed reversible hydrogenation and hydrogen-storage applications): since the overall $\Delta H_{\text{H}_2\text{-split}}$ is large and negative, the reverse dehydrogenation is correspondingly endothermic by the same 264 kJ/mol magnitude, explaining why practical FLP hydrogen-storage/-release cycles require carefully tuned, weaker pairs (smaller $|\Delta H_{\text{P-H}}|$ and $|\Delta H_{\text{B-H}}|$, often achieved using less fluorinated or bulkier boranes) so that the forward and reverse enthalpies are closer to thermoneutral, permitting practical reversible cycling rather than an essentially irreversible trap.

Limiting-case check: if the tether were shortened so much that true P–B bonding distance became accessible ($\theta_D+\theta_A$ effectively below $\theta_{\text{crit}}$ in the cone-angle language of Example 2), the molecule would simply form an intramolecular phosphine-borane adduct instead, and $\Delta H_{\text{H}_2\text{-split}}$ would become irrelevant because there would be no open pocket for H₂ to enter — correctly recovering the "non-frustrated" limit where classical adduct formation, not small-molecule activation, is the dominant pathway.

Stephan's early career combined a strong grounding in classical coordination and organometallic chemistry with a persistent interest in electron-deficient boron and phosphorus compounds. The 2006 discovery — that a simple, easily synthesized phosphine-borane split H₂ heterolytically and, remarkably, could later release it again on heating — was initially received with some surprise precisely because it violated the intuition that only transition metals with accessible d-orbitals could mediate H–H bond activation. Since then, the field has expanded rapidly, with contributions from groups worldwide (notably Gerhard Erker's extensive mechanistic and structural studies, and numerous computational chemists mapping the transition-state geometries of FLP-substrate encounter complexes) turning what began as a single curious observation into an entire subfield of metal-free catalysis.