Trimerization and cyclization of reactive P-functionalities confined within OCO pincers

In order to stabilize a 10–P–3 species with C2v symmetry and two lone pairs on the central phosphorus atom, a specialized ligand is required. Using an NCN pincer, previous efforts to enforce this planarized geometry at P resulted in the formation of a Cs-symmetric, 10π-electron benzazaphosphole that existed as a dynamic “bell-clapper” in solution. Here, OCO pincers 1 and 2 were synthesized, operating under the hypothesis that the more electron-withdrawing oxygen donors would better stabilize the 3-center, 4-electron O–P–O bond of the 10–P–3 target and the sp3-hybridized benzylic carbon atoms would prevent the formation of aromatic P-heterocycles. However, subjecting 1 to a metalation/phosphination/reduction sequence afforded cyclotriphosphane 3, resulting from trimerization of the P(i) center unbound by its oxygen donors. Pincer 2 featuring four benzylic CF3 groups was expected to strengthen the O–P–O bond of the target, but after metal–halogen exchange and quenching with PCl3, unexpected cyclization with loss of CH3Cl was observed to give monochlorinated 5. Treatment of 5 with (p-CH3)C6H4MgBr generated crystalline P-(p-Tol) derivative 6, which was characterized by NMR spectroscopy, elemental analysis, and X-ray crystallography. The complex 19F NMR spectra of 5 and 6 observed experimentally, were reproduced by simulations with MestreNova.


Introduction
Transition metal (TM) catalysis 1 has revolutionized the chemical industry, enabling the conversion of cheap feedstocks into valuable products for pharmaceuticals, polymers, and other specialty chemicals. 2 However, some of the most commonly employed metals like Ru, 3 Rh, 4 Ir, 5 Pd, 6 and Pt 7 are scarce, while the supporting ligands are oen phosphines, 8 N-heterocyclic carbenes (NHCs), 9 and/or other elaborate scaffolds including chiral diols, 10 functionalized cyclopentadienyl ligands (ansa-metallocenes), 11 and rare and/or non-naturally occurring amines. 12 These metal-ligand platforms are both expensive and toxic, resulting in a delicate balance between the benets of chemical synthesis and its harmful effects to the environment and human health. An environmentally friendly and sustainable TM alternative would be to use a more benign and earth-abundant Main Group (MG) element such as phosphorus as the active center. Yet, unlike TMs that have closely spaced HOMO-LUMO gaps, MG compounds feature orbitals that are far apart energetically, which limits their ability to engage in TM-type reactivity like oxidative addition, insertion, and reductive elimination. 13 Fortunately, by distorting phosphines away from their classic three-fold symmetry, their frontier orbitals can become energetically accessible. 14 For example, C s -symmetric phosphorus triamide A can promote oxidative addition of alcohols and amines. [15][16][17] Related C 2v -symmetric B, 18 describable by numerous resonance structures including B 0 and B 00 due to extensive conjugation within the ONO ligand 19 will oxidatively add H 2 from H 3 N-BH 3 and transfer that hydrogen equivalent to azobenzene in a catalytic fashion, producing hydrazines (Fig. 1). 20 Resonance structure B 00 is a T-shaped 10-P-3 species, 21 which contains a frontier orbital environment remarkably similar to a C 2v -symmetric, d 8 ML 3 TM complex like the Ir(PCP) pincer fragment with a low-lying s acceptor orbital and a higher energy lone pair with p-symmetry, 22 capable of engaging in backbonding to an incoming substrate (Fig. 2). In the case of Ir(PCP) pincers, 23 this results in the oxidative addition of dihydrogen, 24 alkanes, 25 and ammonia. 26 However, B does not add H 2 because the highenergy p-lone pair necessary for backbonding is delocalized into the ligand scaffold. 27 Therefore, we speculated that replacing the tridentate ONO ligand with a traditional pincer featuring a central aryl donor would prevent p-delocalization, rendering the p lone pair available for backbonding to small molecules.
In 2010, 28 the Dostál group demonstrated that heavier, C 2vsymmetric 10-Bi-3 species C (Scheme 1, inset) could be stabilized within an NCN pincer and similar analogues could perform oxidative addition of the weak bonds of diphenyldichalcogenides 29 (ex. PhS-SPh, S-S bond ¼ 55 kcal mol À1 ). 30 This led us to attempt to stabilize a 10-P-3 species within a related NCN scaffold with the hope that substrates with stronger bonds, like those present in dihydrogen (H-H bond ¼ 104 kcal mol À1 ), 31 could be broken at P. Yet, unlike the Bi analogue, reduction of the P(III) intermediate did not afford the desired 10-P-3 species, but rather a C s -symmetric 10p-electron benzazaphosphole with a tethered imine arm, which existed as a dynamic "bell-clapper" in solution (Scheme 1). 32 In order to access the targeted 10-P-3 species, we hypothesized that strengthening the 3-center, 4-electron bond between the axial donors and phosphorus (indicated in red, Chart 1, le) could be accomplished using more electron-withdrawing oxygen atoms. 19 Additionally, sp 3 -hybridized benzylic carbons would be employed on the pincer arms to prevent the formation of aromatic P-heterocycles. 33 Guided by these ligand design  Scheme 1 Synthesis of a benzazaphosphole "bell-clapper" with 10-Bi-3 species C shown in the inset.
principles, brominated OCO pincers 1 (ref. 34) and 2 were synthesized (Chart 1, right), and we report here on the unexpected trimerization and cyclization chemistry encountered when installing the P-functionality within these scaffolds.
Monobrominated OCO pincer 1 was readily identied by 1 H NMR spectroscopy with its downeld shied benzylic signals (CDCl 3 : d 4.93; relative to D, CDCl 3 : d 4.55), integrating in a 4 : 2 : 1 ratio with the diagnostic doublet/triplet pattern of the central aryl protons. The proligand was further characterized by 13 C{ 1 H} NMR spectroscopy and elemental analysis, and its structure was unequivocally conrmed by X-ray crystallography (Fig. 3).
Based on literature precedent with heavier pnictogens, 28,36 it was anticipated that the bromide in 1 could be substituted by a PCl 2 unit via a lithium-halogen exchange/phosphination sequence. Reduction of the intermediate dichlorophosphine would then produce the desired 10-P-3 species. However, treatment of 1 with BuLi, followed by quenching with PCl 3 and reduction 37 with PMe 3 did not afford the target, but rather cyclotriphosphane 3 in 56% yield (Scheme 3). The 31 P{ 1 H} NMR spectrum of 3 (inset) contained upeld shied resonances at À116 and À144 ppm with a J PP ¼ 186 Hz, consistent with a solution structure in which two P atoms are spectroscopically distinct from a third. 38 This NMR signature is in line with other (PR) 3 species like (PIs) 3 (Is ¼ 2,4,6-tri-isopropylbenzene) and (PMes) 3 , 38 but not diphosphenes like Mes*P]PMes* (Mes* ¼ 2,4,6-tri-tert-butylbenzene), which feature bona de P]P double bonds (2.034(2)Å), 39 downeld shied 31  The structure of cyclotriphosphane 3 was further corroborated by 1 H and 13 C{ 1 H} NMR spectroscopy. Specically, this "2-Down, 1-Up"-type (referring to the organic substituents on P) structure was readily apparent as two distinct aryl O-Mes singlets were observed in the 1 H NMR spectrum (CDCl 3 ) at 6.71 and 6.62 ppm in an 8 : 4 ratio with all the remaining resonances paired (although some broadened) in a similar 2 : 1 fashion. In addition, the 13 C{ 1 H} NMR spectrum displayed two separate benzylic signals and four methyl signals, all consistent with the assignment of 3. Ultimately, the structure of 3, as proposed, was established by X-ray crystallography (Fig. 4). Structural characterization of cyclotriphosphanes is rare, 43 but the bond lengths and angles of 3 are quite similar to [P(t-Bu)] 3 and Scheme 2 Synthesis of 1. [PCH(SiMe 3 ) 2 ] 3 . 44,45 In particular, the P-P bonds measure 2.217(2), 2.194(2), and 2.237(2)Å, respectively with PPP angles of 60.95 (7), 59.02 (7), and 60.03(7) deg, indicative of P-P single bonds and unhybridized P-centers conned into a small ring system. 43 The P-C bonds in 3 (avg ¼ 1.847Å) are slightly shorter than observed with [P(t-Bu)] 3 and [PCH(SiMe 3 ) 2 ] 3 , which may be due to the presence of an sp 2 -hybridized 46 C-substituent with considerably less bulk and more exibility than the t-Bu and CH(SiMe 3 ) 2 groups. In fact, in comparison with cyclotriphosphanes [P(t-Bu)] 3 , [PCH(SiMe 3 ) 2 ] 3 , (PIs) 3 , and (PMes) 3 , it is somewhat surprising that 3 adopts a related structure because normally, as steric bulk decreases, the size of the oligomeric fragment increases (note: (PMes) 3 versus (PPh) 5 ); however, here, many close contacts (under 4Å) 47 between the aryl rings of the OMes units may impart added stability to the cyclotriphosphane structure.
Regardless, the structure of 3 suggested that the axial 3center, 4-electron bond of the targeted 10-P-3 species was still the weak point. In fact, cyclotriphosphane 3 is formally the result of trimerization 48 of an OCO-supported P(I) intermediate aer the O-donors rotated away from the electron-decient phosphorus center. In order to elucidate any mechanistic details, the phosphination and reduction steps (shown in Scheme 3) were monitored by 31 P{ 1 H} NMR spectroscopy, but did not reveal the generation of RP]PMe 3 (R ¼ OCO pincer), a proposed precursor to the formation of cyclotriphosphanes such as (PIs) 3 , nor did a solution of 3, PMe 3 , and benzaldehyde produce any phosphaalkenes. 48 Synthesis of 2 and cyclization of PCl 2 -functionalized 4 In order to further strengthen the 3-center, 4-electron O-P-O bond of the potential 10-P-3 species, the electronegativity of the O-donors was increased using benzylic CF 3 groups. To this end, diol E, 49 now accessible in a single step 50 from dimethyl 2-bromoisophthalate and TMSCF 3 and previously used to stabilize numerous hypervalent MG species 51 including 10-Br-3 (ref. 52) and 12-I-5, 49 was selected as the building block to brominated OCO pincer 2. Aer isolation of E in multi-gram quantities, 53 dimethylation with MeI/K 2 CO 3 in DMF resulted in the isolation of 2 in 78% yield (Scheme 4).
The 1 H NMR spectrum (CDCl 3 ) featured a prominent singlet for the OMe groups at 3.52 ppm, which integrated in a 6 : 2 : 1 ratio with the remaining central aryl protons, while the 19 F NMR spectrum contained a single resonance at À66.9 ppm, all consistent with the expected C 2v symmetry of 2. Additional characterization by 13 C{ 1 H} NMR spectroscopy, elemental analysis, and Xray crystallography conrmed its structure (Fig. 5).
Cognizant that lithium-halogen exchange using BuLi in the presence of uorine atoms may be complicated by LiF formation 54 and that magnesium-halogen exchange is faster with electron-withdrawing substrates and more functional group tolerant, 55 2 was exposed to i-PrMgCl$LiCl, 56 resulting in smooth in situ conversion to the Grignard reagent, which was subsequently quenched with a precooled solution (À35 C) of PCl 3 in THF (Scheme 5).
The reaction mixture was then analyzed by 31 P{ 1 H} NMR spectroscopy (THF), revealing the presence of some unreacted PCl 3 (218 ppm), (i-Pr)PCl 2 (202 ppm), and an unidentied product (171 ppm). The 19 F NMR spectrum (C 6 D 6 ) displayed four complex multiplet CF 3 signals at À68.9, À70.4, À73.4, and À76.3 ppm, demonstrating that the unidentied product no longer contained C 2v symmetry and was not PCl 2 -substituted 4. Instead, we suspected an intramolecular reaction occurred between the highly electrophilic PCl 2 functionality and one of the O-donors, resulting in cyclization and the loss of chloride, which subsequently dealkylated 57 the O-Me unit affording monochlorinated 5. Using MestreNova, 58 simulated 19 F NMR signals of 5 that closely matched the experimental spectrum were generated (Fig. 6), revealing the presence of both second order and long range coupling between the diastereotopic CF 3 groups, the heterocyclic P atom, and the aryl protons (see ESI † for details). Experimentally, the cyclization to 5 was conrmed by synthesizing a more crystalline derivative via nucleophilic substitution. Specically, a solution of 5 in THF at 0 C was treated with (p-CH 3 )C 6 H 4 MgBr, leading to the isolation of 6 as large off-white crystals in 38% yield (from 2, Scheme 5, above; see Fig. 9 for picture of crystals). The 31 P{ 1 H} NMR spectrum (C 6 D 6 ) of 6 exhibited an apparent septet at 129.7 ppm ( 4 J PF ¼ 6 Hz), while its 19 F NMR spectrum, like 5, displayed four distinct resonances for the diastereotopic CF 3 groups at À69.5, À69.6, À74.0, and À76.5 ppm. These complex 19 F NMR signals could also be reproduced with MestreNova (Fig. 7, see ESI † for details).  addition, two distinct methyl signals (OMe, 3.11 ppm and p-Me, 1.92 ppm) were observed. The 13 C{ 1 H} NMR spectrum also highlighted the inequivalency of the CF 3 groups with four overlapping signals (see ESI † for zoomed in NMR spectrum): two quartets (J CF $ 290 Hz) combined with two quartets of doublets (J CF $ 290 Hz and J CP $ 9 or 2 Hz, respectively) in the 122-123 ppm range, while the two quaternary carbons C(CF 3 ) 2 resonated as a septet (80.8 ppm, J CF ¼ 28 Hz) and a septet of doublets (89.5 ppm, J CF ¼ 31.5 Hz, J CP ¼ 16.5 Hz, Fig. 8).
Ultimately, although disordered across a pseudo mirror plane, X-ray crystallography established the structure of 6 with its bulk purity veried by elemental analysis (Fig. 9).   Unfortunately though, like 1, uorinated OCO pincer 2 also failed to deliver access to the 10-P-3 target. Here, the increased electron-withdrawing power of four benzylic CF 3 groups resulted in, aer a metal-halogen exchange/phosphination sequence, highly reactive dichlorophosphine 4, which via an unexpected cyclization 57 to 5, partially disassembled the pincer framework before reduction could be attempted.

Conclusions and future directions
Undoubtedly, the hypothesis (vide supra) that the formation of 10-P-3 species over benzazaphosphole "bell-clappers" would be favored by switching from NCN pincers to more electron withdrawing OCO pincers with sp 3 -hybridized benzylic carbon atoms was overly simplistic. As the phrase goes, "hindsight is 20/20" and here, the biggest oversight was expecting a P-center to adopt a high-energy planarized geometry when lower barrier pathways to phosphorus maintaining its preferred pyramidal geometry exist. For example, 10-P-3 species like B 00 (Fig. 1) and the target (Chart 1, le) can be considered "internally-solvated phosphinidenes" 19 that are supported by a 3-center, 4-electron O-P-O bond. However, B 00 features a conjugated ONO ligand that locks the O-donors into place, while C Ar -C sp 3 or C sp 3 -O bond rotations within OCO pincer 1 expose the reactive P(I) center, resulting in oligomerization chemistry, 43,48,59 a known route by which phosphorus preserves its pyramidal structure and is exemplied by the formation of trimer 3. To prevent this, we aimed to strengthen the 3-center, 4-electron bond using more electron-withdrawing O-donors, but neglected how the enhanced electrophilicity of PCl 2 -substituted 4 can make the ligand framework susceptible to nucleophilic attack, in this case, cyclization to 5 (and its functionalization to 6). Guided by these lessons, a pincer framework that is rigid featuring electron-withdrawing O-donors, but lacking sp 3 -hybridized 57 carbon atoms prone to nucleophilic attack may provide the stabilization necessary to conne a P-center into a planarized 10-P-3 arrangement and will be investigated.

General experimental details
Unless otherwise specied, all reactions were performed under an atmosphere of nitrogen in an MBraun or Vacuum Atmospheres glovebox or using standard Schlenk techniques. All glassware was dried overnight in an oven at 140 C prior to use. Solvents used in the glove box were purchased directly from chemical suppliers (Aldrich or Acros), pumped directly into the glove box, and stored over oven-activated 4 or 5Å molecular sieves (Aldrich). Solvents used outside the glove box were purged with N 2 for 30 min and stored over molecular sieves. TMSCF 3 was dried by cryogenic transfer. 1 H, 13  All other chemicals were used as received, unless otherwise noted.

Synthesis of 1
In the glovebox, 2,4,6-trimethylphenol (3.00 g, 22.03 mmol, 2.5 equiv.) was dissolved in THF (100 mL), and a 1.6 M solution of n-BuLi in hexanes (13.0 mL, 20.8 mmol, 2.3 equiv.) was added dropwise at room temperature. The reaction mixture was stirred for 15 min. Subsequently, tribromide D (3.086 g, 9.0 mmol) was added to the stirred solution, and the reaction mixture was transferred to a Schlenk bomb equipped with a Teon screw cap and heated at 100 C for 1 day. The volatiles were then removed under reduced pressure, and the crude residue was extracted with a toluene : hexane mixture (1 : 1, 3 Â 15 mL). The combined extracts were ltered through a Celite plug, and the ltrate was concentrated under vacuum then puried by column chromatography (silica gel, toluene : hexane 1 : 1, R f ¼ 0.63). The title product (2.64 g, 5.822 mmol) was obtained as white plates in 65% yield. X-Ray quality crystals were obtained by slow evaporation from a concentrated solution of 1 in n-hexanes.

Synthesis of 2
Diol E (3.61 g, 7.38 mmol) and DMF (180 mL) were combined in a Schlenk ask with a stir bar. The solution was purged with N 2 on the Schlenk line for 30 min, then K 2 CO 3 (5.10 g, 36.9 mmol, 5 equiv.) was added with stirring under positive N 2 pressure. Next, MeI (2.30 mL, 36.9 mmol, 5 equiv.) was injected via syringe, and the reaction mixture was stirred overnight, then quenched with NH 4 Cl (aq) and extracted with toluene (3 Â 50 mL). The combined organic layers were washed with water, brine, dried over MgSO 4 , ltered, and the ltrate concentrated under vacuum, resulting in pale yellow solid 2 (2.97 g, 5.74 mmol, 78%). Recrystallization from acetonitrile (8 mL) at 4 C afforded crystals suitable for X-ray diffraction.
Anal. Calcd for C 14

Synthesis of 3
OCO-supported aryl bromide 1 (1.134 g, 2.50 mmol) was loaded into a Schlenk ask, dissolved in Et 2 O (60 mL), taken outside of the glovebox, and cooled to À78 C. A 1.6 M solution of n-BuLi in hexanes (1.7 mL, 2.72 mmol, 1.1 equiv.) was injected (under N 2 ), and the reaction mixture was stirred at À78 C for 5 min. A second Schlenk ask containing a solution of PCl 3 (550 mg, 4.01 mmol, 1.6 equiv.) in Et 2 O (5 mL) was transferred via cannula to the cooled reaction mixture. The reaction mixture was stirred at À78 C for 10 min and then at room temperature (RT) for 1 h, leading to the precipitation of a white solid. The volatiles were removed under reduced pressure, the Schlenk ask was brought back into the glovebox, the residue was triturated with Et 2 O (10 mL), and the volatiles were again removed. The residue was dissolved in THF (20 mL), and PMe 3 was added (485 mg, 6.375 mmol, 2.6 equiv.). The reaction mixture was stirred at RT for 1 day. The organic volatiles were removed under reduced pressure, and the residue was extracted with toluene (3 Â 20 mL). The combined extracts were ltered through a Celite plug, the ltrate was concentrated to dryness, triturated with n-pentane (10 mL), and again concentrated to dryness under reduced pressure. The resulting residue was triturated with acetonitrile (10 mL) and stirred at RT for 1 h, generating a white precipitate that was collected by ltration and dried (565 mg, 0.466 mmol, 56%). X-Ray quality crystals of 3 were obtained by recrystallization from a solution of hot acetonitrile. Anal

Synthesis of 6
Fluorinated aryl bromide 2 (500 mg, 0.967 mmol) was dissolved in 6 mL of THF in a vial with a stir bar inside the glovebox and (i-Pr)MgCl$LiCl was added dropwise via syringe (0.82 mL, 1.064 mmol, 1.1 equiv., 1.3 M in THF), resulting in a homogeneous, yellow reaction mixture. The reaction mixture was stirred for 1 h at room temperature (RT) then directly ltered into 2 mL of a pre-cooled solution (À35 C, 1 h) of PCl 3 (146 mg, 1.064 mmol, 1.1 equiv.) in THF. The solution was warmed to RT for 1 h then analyzed by 31 P{ 1 H} NMR spectroscopy, revealing the presence of unreacted PCl 3 (218 ppm), i-PrPCl 2 (202 ppm) and the chlorophosphine (170 ppm). The "intermediate" product mixture was then concentrated under vacuum to remove the volatile and unwanted phosphorus byproducts (PCl 3 and i-PrPCl 2 ), leaving a pale yellow residue that was dissolved in 6 mL of THF and transferred to a Schlenk bomb tted with a screw-top Teon cap. The bomb was taken outside of the glovebox, cooled to 0 C, and (p-CH 3 )C 6 H 4 MgBr (1.06 mL, 1.064 mmol, 1.1 equiv., 1.0 M in THF) was injected via syringe under positive N 2 pressure, affording a light orange reaction mixture, which was subsequently warmed to RT. The Schlenk bomb was resealed (under positive N 2 pressure), brought back inside the glove box, and an aliquot was analyzed by 31 P{ 1 H} NMR spectroscopy displaying a prominent signal at 129 ppm with slight impurities at À2 and À8 ppm. The entire reaction mixture was concentrated under vacuum, extracted with pentane (3 Â 50 mL), and ltered through a Kimwipe plug. The ltrate was concentrated under vacuum, dissolved in ether (2 mL) and cooled to À35 C overnight, resulting in large white/ colorless blocks suitable for X-ray diffraction (202 mg, 0.371 mmol, 38% yield).
Anal. Calcd for C 20  Experimental NMR spectra, simulated 19 F NMR spectra using MestreNova, and X-ray crystallography See the ESI † for details.

Conflicts of interest
There are no conicts to declare.