On the Ambiphilic Reactivity of Geometrically Constrained Phosphorus(III) and Arsenic(III) Compounds: Insights into Their Interaction with Ionic Substrates

Abstract The ambiphilic nature of geometrically constrained Group 15 complexes bearing the N,N‐bis(3,5‐di‐tert‐butyl‐2‐phenolate)amide pincer ligand (ONO3−) is explored. Despite their differing reactivity towards nucleophilic substrates with polarised element–hydrogen bonds (e.g., NH3), both the phosphorus(III), P(ONO) (1 a), and arsenic(III), As(ONO) (1 b), compounds exhibit similar reactivity towards charged nucleophiles and electrophiles. Reactions of 1 a and 1 b with KOtBu or KNPh2 afford anionic complexes in which the nucleophilic anion associates with the pnictogen centre ([(tBuO)Pn(ONO)]− (Pn=P (2 a), As (2 b)) and [(Ph2N)Pn(ONO)]− (Pn=P (3 a), As (3 b)). Compound 2 a can subsequently be reacted with a proton source or benzylbromide to afford the phosphorus(V) compounds (tBuO)HP(ONO) (4 a) and (tBuO)BzP(ONO) (5 a), respectively, whereas analogous arsenic(V) compounds are inaccessible. Electrophilic substrates, such as HOTf and MeOTf, preferentially associate with the nitrogen atom of the ligand backbone of both 1 a and 1 b, giving rise to cationic species that can be rationalised as either ammonium salts or as amine‐stabilised phosphenium or arsenium complexes ([Pn{ON(H)O}]+ (Pn=P (6 a), As (6 b)) and [Pn{ON(Me)O}]+ (Pn=P (7 a), As (7 b)). Reaction of 1 a with an acid bearing a nucleophilic counteranion (such as HCl) gives rise to a phosphorus(V) compound HPCl(ONO) (8 a), whereas the analogous reaction with 1 b results in the addition of HCl across one of the As−O bonds to afford ClAs{(H)ONO} (8 b). Functionalisation at both the pnictogen centre and the ligand backbone is also possible by reaction of 7 a/7 b with KOtBu, which affords the neutral species (tBuO)Pn{ON(Me)O} (Pn=P (9 a), As (9 b)). The ambiphilic reactivity of these geometrically constrained complexes allows some insight into the mechanism of reactivity of 1 a towards small molecules, such as ammonia and water.


Introduction
Over the last decade significant advances have been made in the developmentofm ain-group species that are capable of activating small molecules. [1] Mostp rominent amongstt hese are alkyl(amino)c arbenes (AACs), [2] low oxidation state compounds of Groups13a nd 14, [3,4] and frustratedLewis pairs (FLPs). [5] Severalo ft he aforementioned compounds have also shown to be active in catalytic processes, such as the hydrogenation of imines,a lkenes and alkynes, [6] effectivelyh eralding an ew era in the chemistryo ft he p-block elements. It is also worth noting that the substrates cope of such compounds includes molecules such as ammonia,which hashistorically provendiffi-cult to activate using precious metalc atalysts due to the unfavourable coordination/activation equilibrium. Newr outes resultingi nt he activation of NÀHb onds in ammonia are particularly appealing, becauseo ft he dearth of transition-metal systems capable of effecting such at ransformation, [7][8][9][10][11][12][13][14][15][16][17][18][19] and the relevance of NÀHa ctivation to an umber of potentially important industrialp rocesses. [20] More recently,g eometrically constrained compounds based on phosphorus(III) have also received significant attention in this field. Studies by Arduengo, [21] Radosevich, [22][23][24] Kinjo, [25] and our own research group have demonstrated that the dis-tortedT -shaped phosphorus(III) compounds pictured in Figure1 are capable of reacting with polar substrates such as water, alcohols, ammonia anda mines. [26] In the vast majority of cases,t heser eactions result in af ormal oxidative addition of the substrate over the phosphorus(III) centre, although several theoretical studies have subsequently demonstrated that, in all likelihood, these processes involve the ligand backbone. [27,28] We recently reported ap hosphorus(III) compoundb earing the N,N-bis (3,5-di-tert-butyl-2-phenolate)amidel igand (P(ONO); 1a). [26] This speciesw as found to react with ammonia and water, activating the EÀHb onds in both substrates through af ormal oxidative addition to afford the corresponding phos-phorus(V) compounds. Under forcingc ondithereions (heating solid samples under ad ynamic vacuum) we were able to show that thesep rocesses are reversible. During the course of these studies, we observed that 1a exclusively reacts with nucleophilic specieswith protic EÀHb onds, whereas non-nucleophilic substrates, even those with polar EÀHb onds, such as phenylsilane, fail to react. These observations seemedt oi ndicate that the nucleophilic association of the substrate is necessary prior to any furtherr eactivity taking place. This promptedu st oe xplore the reactivity of 1a towards both nucleophilesa nd electrophiles in an effort to gain ab etter understanding of the reaction dynamics. These results, and analogouss tudies on the heavierarsenic(III) analogue (1b), are reported herein.

Results and Discussion
Structure and bonding considerations for 1a and 1b As previously reported, 1a exhibitsab ent geometry in the solid state with moderate pyramidalisation at both the phosphorusa nd nitrogen atoms (S angles(P) = 296.18; S angles(N) = 331.68), in contrast with planar compound A.T heoretical calculations at the density functional theory (DFT) level revealed that the pyramidal C s isomer (or electromorph to use the term coined by Arduengo) is the most stable, butr elatively closei ne nergy to the C 2v symmetric species (within4kJ mol À1 ). This energetic difference is within the error of the calculations, and it is worth noting that different computational analyses (varyingb asis sets and functionals) actually suggest that the planar structure is lower in energy than the C s isomer (albeit by similarly small values).T herefore, it is highly likelyt hat, in solution, there is ad ynamic, and concerted, pyramidal inversion at the phosphorusa nd nitrogen atoms resulting in aw ing-like "flapping" of the ligand backbone. The calculations also revealed that, regardlesso ft he symmetrya doptedb y1a,t here is an empty, energetically accessible orbitalt hat is largelyb ased on the phosphorus atom, which has anti-bonding characterw ith respect to the PÀNa nd PÀOb onds. It is also worth noting that the PÀNb ond in 1a is significantly polarised, with computed Hirshfeldc harges of 0.394 and À0.171 on the phosphorus and nitrogen atoms, respectively.T hese observations prompted us to explore the reactivity of 1a,and its heavierarsenic analogue (1b), towardsn ucleophiles in order to establish whether they associate with the phosphorus/arsenic atom.
The arsenic-containing species 1b can be preparedf ollowing as imilar synthetic methodology to that previously reported for its lighterc ongener.R eaction of the protonated ligand, N,N-bis (3,5-di-tert-butyl-2-phenol)amine, H 3 ONO, withA sCl 3 in the presence of three molar equivalents of triethylamine, affords 1b quantitatively, as evidenced by 1 Ha nd 13 C{ 1 H} NMR spectroscopy.T he 1 HNMR spectrum of 1b in C 6 D 6 reveals two equal intensity aromatic resonances at 8.39 and 7.51 ppm as well as two singlets at 1.71 and 1.46 ppm arising from the tertbutyl groups. Cooling of ac oncentrated pentane solutiona fforded bright red-orange crystalso ft he compound in good to high yields.
Reactions involving 1a were monitored by 31 PNMR spectroscopya nd show quantitative conversion to the desired products,w hich were observeda ss inglet resonances at 85.5 and 66.3 ppm for 2a and 3a,r espectively.T hese resonances are shifted upfield with respectt o1a (168.6 ppm), due to the enhanced electron density on the phosphorus centre.S imilarly the 1 HNMR spectra of both compounds displayt he requisite number of resonances for as ymmetrical N,N-bis (3,5-di-tertbutyl-2-phenolate)amide ligand backbonea nd for the nucleophilic substituents. Reactions involving 1b similarly give rise to products in which both of the 3,5-di-tert-butyl-2-phenolate arms of the ligand backbonea re equivalent as evidenced by 1 Ha nd 13 C{ 1 H} NMR spectroscopy.
The structures of all four novel anionic complexes were determined by single-crystal X-ray diffraction (Figures 3a nd 4). The complexesr eveal planar Pn(ONO)m oieties (mean deviation from plane for Pn1/O1/N1/O2:0 .0322 (2a), 0.0365 (2b)0 .0019 (3a)a nd 0.0150 (3b)) with the nucleophile orthogonal to the Pn(ONO)c ore. The planarity of the ligand backbonei sa lso evident in the sum of bond angles around the nitrogen atoms (359.4, 357.4, 359.7 and 360.08 for 2a, 2b, 3a and 3b,r espectively). These structures are consistent with lone-pair donation from the nucleophile into the lowest unoccupied molecular orbital( LUMO)o f1a and 2a.D FT calculations reveal that the most significant atomic orbital contribution to this orbital comes from the pnictogen atom po rbital that is perpendicular to the plane of the molecule( 53.75 and 49.97 %c ontributions for the phosphorus and arsenic p z orbitals for 1a and 2a,respectively;the z axis is defined as orthogonal to the plane of the molecule). TheL UMO of 1a and 2a are also notably p p -p p antibonding with respect to the PnÀN and PnÀOb onds.
Upon coordination of an anionic nucleophile there is asignificant elongation of the PnÀOb onds while the PnÀNb onds remain very similart ot hose of the parent compound (see Ta ble 1f or ac omparison of bond metric data for all complexes). In each of the adducts, the elongationo ft he PnÀO bonds is not uniform, butr ather more pronounced for one of the two aryloxide functionalities of the N,N-bis(3,5-di-tert-butyl-2-phenolate)amide ligand.T his significant weakening of the PnÀOb onds strongly suggeststhat on coordination of an ucleophile, the aryloxide functionalities are susceptible to electrophilic attack. This was probed by reacting 2a and 2b with ap yridinium trifluoromethanesulfonatea nd benzylb romide (BzBr).

Reactivity of 1a and 1b towardselectrophiles
The aforementioned resultsp rompted us to explore the reactivity of 1a and 1b towards electrophilic substrates in an effort to establish whether they too associate with the heavier pnictogen atom centre,o rw hether they attack the more electronegative atoms of the ligand backbone. The highest occupied molecular orbital (HOMO) for the pyramidal C s isomer of 1a has as ignificant contribution from the nitrogen atomico rbitals, while the computed Hirshfeld charges reveal significant polarisation of the PÀNa nd PÀOb onds, with negative charge accumulating equallyo ver the three atoms of the ligand backbone.
In at ypical reaction, one molar equivalent of EOTf (E = H, Me;O Tf = trifluoromethanesulfonate) was added to as olution of either 1a or 1b.T he reactions werem onitored by NMR spectroscopy and reveal quantitative conversion to [Pn{ON(H)O}][OTf] (Pn = P( 6a), As (6b)) and [Pn{ON(Me)O}] [OTf] (Pn = P( 7a), As (7b)) after heating or sonicating the mixtures (see Experimental Sectionf or full details). [31] The 31 PNMR spectra of the reactions involving 1a reveal broad resonances, with evidence of weak or non-existent PÀHc oupling,a t1 55.5 ( 2 J P-H = 12 Hz) and 149.4 ppm for 6a and 7a,r espectively (cf. 168.6 ppm for 1a), indicating that the electrophilic groupsd o not associate directly with the phosphorus(III) centre. The 1 HNMR spectrao fa ll four compounds are consistent with two equivalent aryloxide functionalities, which suggest functionalisation at the amide nitrogen atom. The 1 HNMR resonances for the proton-and methyl-group-functionalised nitrogen atoms were observed at 14.27 and 3.72 ppm for 6a and 7a,r espectively.Similarly,r eactions involving 1b also reveal clean conversion to the products and the 1 HNMR resonances of the electrophiles associated with the nitrogena tom were observed at 11.64 and3.41ppm for 6b and 7b,r espectively.
The structures of all four novel cationic complexes were determined by single-crystal X-ray diffraction (Figures 6a nd 7).
Interestingly,a ll four species exhibit close contacts between the pnictogen(III) centre andt rifluoromethanesulfonate anions (6a:2 .790 (2) (1) )i ndicating as ignificant degree of positive charge accumulating on the pnictogen centres on functionalisationo f the ligand backbone.T his is corroborated by DFTcalculations which showalargeH irshfeld charges on the pnictogen atoms relative to nitrogen (6a:0 .455 and À0.012, 6b:0 .598 and À0.024, 7a:0 .446 and 0.015, 7b:0 .586 and 0.004, for the heavierp nictogen and nitrogen atoms, respectively). Thus,a ll four complexes can be thought of as base-stabilised phosphenium or arsenium ions. This bonding formulation has previously been proposed for relatedp hosphorus-containing compounds. [32][33][34] It is interesting to note that while 1a reacts with HOTf to afford ap hosphorus(III) compound with ap rotonated ligand backbone( 6a), the analogous reaction with an acidt hat has am ore nucleophilic counter-anion, such as HCl, affords the phosphorus(V) species HPCl(ONO) (8a). This differencei nr eactivity suggests that mechanistically,asufficientlys trongn ucleophile is required to afford the formal phosphorus(V) oxidative addition product. Reaction of 6a with KOtBu, affords 4a,s uggesting that nucleophilic association of the anionic À OtBu moiety with the phosphorus centre induces proton migration from the ligand backbone. Similarly,r eaction of 6a with tetradodecylammonium chloride cleanly affords 8a and the corresponding ammonium trifluoromethane sulfonate salt. Conversely, reactions of 8a with one equivalent of trimethylsilyl trifluoromethane sulfonate show evidence for the formationo f 6a,a lthough full conversion was not observed at room temperature with such stoichiometric loadings.
In the aforementioned studies we have establishedt he relative inaccessibility of the arsenic(V) oxidation state for 1b.T his is borne out in the reactivity of 1b towards HCl. Whereas 1a reacts with HCl to afford 8a,t he reactiono f1b with one molar equivalent of HClr esults in the addition of the acid across one of the AsÀOb onds resulting in the arsenic(III) compound AsCl{(H)ONO} (8b). This was evident from the 1 HNMR spectrao ft hese reactionm ixtures which reveal the presence of two inequivalent arms of N,N-bis(3,5-di-tert-butyl-2-phenolate)amide pincerl igand on protonation.
The structure of 8b reveals at rigonal pyramidal geometry about the arsenic(III)c entre (S angles(As) = 284.88; S angles(N) = 355.68) with ar elativelyo btuse N1-As1-Cl1 angle due to the steric repulsion of the chloride with the free arm of the N,N-bis(3,5-ditert-butyl-2-phenolate)amide ligand.T he AsÀNa nd AsÀOd istances, 1.839(2) and 1.783 (2) ,respectively,are notably shorter than those observed for 1b (AsÀN: 1.862(3) andA s ÀO: 1.933(4) ), presumably due to the relaxation of steric strain and the loss of AsÀNm ultiple bond character on bending the ligand backbone. It is worth noting that the AsÀNb ond in 8b is still very short and indicative of some multiple bond character.
As mentioned previously,r eactions of 6a with KOtBu result in the association of the tert-butoxide anion with the phosphorus(III) centre andm igration of the ligand proton to afford the phosphorus(V) product 4a (which can also be accessed by af ormal oxidative addition of HOtBu to 1a). The facility with which the protono ft he ligand backbonem igrates prompted us to carry out related studies with the methylateds pecies 7a and 7b.W eh ypothesised that methyl migration would not occur andt hat, consequently novel complexes in which both the nitrogen atom of the N,N-bis(3,5-di-tert-butyl-2-phenolate)amide ligand, and the heavierp nictogen centers could be functionalised.
In at ypical reaction, 7a or 7b were treated with one molar equivalent of potassium tert-butoxide. These reactions were found to quantitively afford novel complexesb earing as ymmetricall igand environment and an additional tert-butoxide functionality (as pictured in Scheme 3). Reactions involving 7a reveal an upfield shift in resonance in the 31 PNMR spectra from 149.4 to 142.8 ppm. The association of the À OtBu moiety with the phosphorus centre is evident from the appearance of ab road resonance in the 1 HNMR spectrum at 1.57 ppm. This resonance is accompanied by two aromatic resonances (7.40 and 7.31 ppm), two resonances arising from the ligand tertbutoxide groups andadoublet due to the methyl group of the ligand backbone.C omparable spectroscopic data were recordedf or the arsenic-containing analogue, 9b,a lthough this sample could not be isolated as ac ompositionally pure compound due to its relative instability andt endency to decompose.

Conclusion
We have explored the reactivity of the geometrically constrained Group 15 complexes P(ONO) (1a)a nd As(ONO)( 1b)t owards ionic nucleophiles and electrophiles. These studies show that anionic nucleophilesr eadily associate with the pnictogen(III)c entres in both complexes,s uggesting that such an associationm ay play an important role in the mechanism for the bond activation of NH 3 and H 2 Ob y 1a.O ur studies also reveal that while phosphorus(V) compounds are readily accessible by sequential reactions involving 1a,t he corresponding arsenic(V) compounds cannot be synthesised from the heavier analogue 1b.
Reactions involving charged electrophilic substrates give rise to pnictogen(III) compounds in which the electrophile associates with the nitrogen atom of the ligand backbone. Interestingly,when the electrophile in question is aproton, it will associate with the nitrogen atom of the ligand backboneo nly in the presence of aw eakly coordinating counteranion( such as OTf À ). When am ore nucleophilicc ounter-anion is employed (such as Cl À )t hese reactions result in the generation of ap hosphorus(V) compound by protonm igration from the ligand backbonet ot he phosphorus centre. As with previouss tudies, the analogous arsenic(V) compound was found to be inaccessible.
These studies have helped us probe possible pathways by which 1a is able to activate nucleophilic substrates with polarised EÀHb onds. Studies are currently on-going with regard to elucidating am echanism for such formal oxidative addition reactions.