Regioselective N-versus P-Deprotonation of Aminophosphane Tungsten(0) Complexes

: 1,2-Bifunctional ligands are rare, in general, which holds especially for those with a P-N linkage. Herein, we report on the synthesis of P-tert -butyl substituted aminophosphane W(CO) 5 complexes 3a – f (a: R = R’ = H; b: R = H, R = Me; c: R = H, R’ = ally; d: R = H, R’ = i -Pr; e: R = H, R’ = t -Bu; f: R = R’ = Me) obtained via formal N-H insertion reactions of Li/Cl phosphinidenoid complex 2 into NH bonds of ammonia and different amines. The 1,2-bifunctionality of 3b was addressed in targeted regioselective deprotonation reactions leading to amidophosphane complexes M-4b or M/N(H)Me phosphinidenoid complexes M-5b , respectively (M = Li, K). Remarkable was the observation that reactions of M-4b and M-5b with MeI as the electrophile resulted in the formation of the same product 7b . The constitution of all the compounds has been established by means of NMR and IR spectroscopy and mass spectrometry. Two possible reaction pathways were studied in detail using high-level DFT calculations.


Introduction
The first mention of aminophosphanes I (Figure 1) bearing no P-H functionality dates to 1957, when Harris reported the synthesis of (CF 3 ) 2 P-NH 2 via the reaction of bis(trifluoromethyl)chlorophosphane with two equivalents of ammonia in the gas phase [1].Following this, Burg [2] and Smith [3] reported early examples using the aminolysis or treatment of chlorophosphanes with sodium amides.Transition metal complexes II with, e.g., iron(0), nickel(0) or molybdenum(0), were first reported in 1971 [4,5].The synthesis of aminophosphanes III, which have a P-H functionality, remained unsuccessful for many years and may be also related to the firm belief that α-elimination under the formation of an amine H-NR 2 and a phosphinidene R-P [6] could occur.Nevertheless, the corresponding complexes of secondary aminophosphanes IV have been known since 1975 [6] and have been frequently reported thereafter [7,8].Two examples of IV were obtained from a phosphanido iron(0) complex [9].In 1977, Niecke synthesized the first P-H containing derivative III, (Me 3 Si) 2 NP(H)-NHSiMe 3 [10], and then the first primary aminophosphane (Me 3 Si) 2 N-PH 2 [11], thus paving the way for further research in various directions.The strategy to use N-H insertion reactions to gain access to aminophosphane complexes of type II was first used in 1982 by Mathey [7] and, more recently, by Lammertsma [9].The alternative use of Li/Cl phosphinidenoid complexes in the formation of aminophosphane complexes of type IV was first reported by us in 2012, using the formal N-H insertion [12,13].The broader scope of this E-H insertion chemistry was reported more recently [14][15][16][17] including an overview of formal N-H insertion of Li/X phosphinidenoid complexes [18], adding first examples of V, but focusing on sterically demanding substituents, such as triphenylmethyl and bis(trimethylsilyl)methyl, together with W(CO) 5 and Fe(CO) 4 metal fragments [12][13][14][15][16][17][18][19].Studies on the chemistry of 1,1 -bifunctional aminophosphane complexes V included reactions targeting the P-N bond with hydrogen halides [7] and the P-H bond via deprotonation [12][13][14][15][16][17][18].The latter has resulted in new M/NR' 2 phosphinidenoid complexes if the P-H deprotonation takes place.In contrast, the new 1,2-bifunctionality, i.e., the presence of a P-H and N-H bond, has allowed for initial investigations concerning N-functionalization and/or P/N bifunctional nucleophilicity, thus enabling to form threemembered P-heterocyclic ligands, if sterically demanding P-substituents were used [16].
Herein, 1,2-bifunctional aminophosphane complexes V were synthesized, bearing the sterically less demanding tert-butyl group [20], and used to address the quest for selective deprotonation and N-vs.P-functionalization.The latter part was also studied in great detail using DFT calculations.

General experimental details.
The syntheses of all compounds were performed under an argon atmosphere using Schlenk techniques and dry solvents.Tetrahydrofuran (THF), diethyl ether, petroleum ether (PE) and n-pentane were dried over sodium wire/benzophenone, dichloromethane over calcium hydride and toluene over sodium and further purified by subsequent distillation.All NMR spectra were recorded on a Bruker AV I 300 (300.1 MHz for 1 H, 75.5 MHz for 13 C; 121.5 MHz for 31 P), Bruker AV I 400 (400.1 MHz for 1H, 100.6 MHz for 13 C; 162.0 MHz for 31 P) and Bruker AV III HD Prodigy 500 (500.2MHz for 1 H, 125.8 MHz for 13 C and 202.5 MHz for 31 P) spectrometers at 25 • C. The 1 H and 13 C NMR spectra were referenced to the residual proton resonances and the 13 C NMR signals of the deuterated solvents and 31 P NMR spectra were referenced to 85% H 3 PO 4 as external standards, respectively.Elemental analyses were carried out on a Vario EL gas chromatograph.Mass spectrometric data were collected on a micrOTOF-Q Bruker Daltonik TOF mass spectrometer (ESI, ACPI) or a MAT 90 Thermo Finnigan sector instrument (EI).IR spectra of all compounds were recorded on a Thermo Nicolet 380 FT-IR spectrometer with an attenuated total reflection (ATR) attachment or a Bruker Alpha Diamond ATR FTIR spectrometer.
Complex 1 was prepared according to the literature [21].

Synthesis and Characterization of P-tert-Bu Substituted Aminophosphane Complexes
According to Tolman's cone angle concept extrapolated from phosphine ligands to the organyl R substituent attached to P in computed (PBEh-3c level) R-PCl 2 -W(CO) 5 compounds 1, the estimated cone angle for the trityl group (Θ = 178.4• ) reveals a much higher steric protection at P than the t-Bu group (Θ = 143.5 • ) [34,35].The addition of t-BuLi to a THF solution of dichloro(tert-butyl)phosphane complex 1 in the presence of 12-crown-4 at −100 • C generated the Li/Cl phosphinidenoid complex 2 [34,35], which reacted with ammonia and different amines (Scheme 1).Formal N-H insertion reactions led to the formation of 1,1 -bifuntional aminophosphane W(CO) 5 complexes 3a-f (Scheme 1, Table 1).Highly selective reactions were observed in case of derivatives 3a-d, which were isolated as yellow to red oils in moderate to good yields (Table 1).In the case of 3e and 3f, the desired products could be detected only by 31 P NMR; hence, 3f was only partly characterized in an inseparable mixture containing 88% of 3f.
Surprisingly, the reactions of 2 with other primary and secondary amines, such as ethyl amine, aniline or diethyl amine, did not yield the desired products but an inseparable mixture of compounds, which were not further characterized.In comparison to, e.g., the CPh 3 -substituent, the tert-butyl substituent exerts a reduced steric shielding onto the phosphorus centre according to Tolman's cone angle concept [20].This makes this finding particularly astonishing, as a lesser steric shielding onto the phosphorus centre should allow for these reactions to happen, especially as the same reactions had proceeded selectively for the P-CPh 3 substituted analogue of complex 2 [16].
The 31 P NMR resonance signals of complexes 3a-f were observed in the range of 37 to 91 ppm and are highfield-shifted compared to the starting material 1.The analogous P-CPh 3 substituted complexes are 5-10 ppm highfield-shifted [15,16], indicating that the tert-butyl substituent leads to a slightly lower shielding of the P nucleus.Two substituents at the nitrogen centre shift the resonance to the downfield region.The observed 1 J W,P and 1 J P,H coupling constants of 3a-f are of similar magnitude, as expected from related derivatives [12,13,[15][16][17].
The FT-IR spectra of 3a displays the asymmetric and symmetric NH 2 vibration modes at 3476 cm −1 and 3384 cm −1 , while the bending vibration mode is found at 1556 cm −1 .Complexes 3b-d show their N-H bond vibration modes between 3436 cm −1 and 3421 cm −1 .An unscaled value of 3479 cm −1 was computed for 3b at the working level of theory (see Computational Details).The values for the stretching vibrations are slightly higher than for known complexes, e.g., the P-CPh 3 substituted derivatives, which display values at 3400 cm −1 to 3350 cm −1 .The absorption bands due to P-H vibration modes are measured at about 2290 cm −1 with only slight variations in all complexes 3a-f.

Regioselective Deprotonation of 1,2-Bifunctional Aminophosphane Complexes
Compared to P-CPh 3 or P-CH(SiMe 3 ) 2 derivatives, the tert-Bu substituent seemed to have a clear influence on the outcome of otherwise feasible reactions.Therefore, the quest of regioselective deprotonation of 1,2-bifunctional aminophosphane complexes 3a-f emerged.Previous deprotonation studies of the related, but sterically more demanding, 1,2-bifunctional complexes have shown that one equivalent of base is sufficient, in the presence of a crown ether, to achieve full conversion to the M/N(H)R' phosphinidenoid complex [12,13,[15][16][17].To study the system in hand, the following three bases were tested with respect to the deprotonation of 3a-c: KHMDS, which was used before [15][16][17], MeLi and n-BuLi, varying in basicity and nature of the counter ion.In particular, the Li cation can be expected to display stronger P-Li and/or N-Li interactions, compared to interactions with the significantly larger K cation.
In contrast to previous studies, the reaction of 3a-c with one equivalent of base in THF did not lead to any of the desired products M-4a-c or M-5a-c (Scheme 2).Instead, mixtures of 3a, 3b or 3c, respectively, and several unknown phosphorus compounds were formed.This finding was independent from other reaction conditions, such as temperature and concentration.
Further investigations of 3b in THF showed that the use of two equivalents of any of the bases mentioned above leads to the N-H deprotonation product M-4b (Scheme 2).In the row of bases, KHMDS provided the worst and n-BuLi the best result concerning selectivity, which could be tentatively attributed to the highest basicity of the latter.The reaction outcome was temperature independent, and the reaction could also be conducted at room temperature.
To check the effect of a more separated counter cation, two equivalents of a base and two equivalents of the corresponding crown ether were added to a THF solution of 3b at −80 • C, which resulted in P-H deprotonation, instead, yielding the mono-metalated aminophosphane complex M-5b selectively (Scheme 2).The same behaviour in regioselective deprotonation of 1,2-bifunctional complexes could be confirmed by NMR spectroscopy for complexes 3a and 3c.The 31 P resonances of the amide complexes M-4b were found to be 20 ppm highfieldshifted compared to 3b and showed significantly decreased 1 J W,P and 1 J P,H coupling constant magnitudes.This trend is in accordance with the observations made earlier for the P-CPh 3 substituted derivative [16].As shown by the calculations (vide supra), N-H deprotonation significantly strengthens the PN bond, which weakens the other phosphorus bonds and decreases the respective J values (4b : MBO PN 1.66; 3b: MBO PN 1.08).The significant decrease in the 1 J P,C coupling constant to 20.8 Hz in Li-4b from 28.3 Hz in 3b is in line with these findings.
Li-4b could be isolated as a yellow solid, which was very sensitive towards air and moisture.It decomposed over hours in solution and in the solid state, even under inert atmosphere, reforming the starting material, but also leading to two unknown phosphorus compounds (δ( 31 P)/ppm = 71.5 ( 1 J W,P = 233 Hz, 1 J P,H = 338 Hz), 89.8 (no detectable Table 1 J W,P , 1 J P,H = 278 Hz)).The 13 C{ 1 H} NMR spectrum of Li-4b showed a 10 to 20 ppm downfield shift of the CO resonance signals, which fits with an increased electron density in the PN bond.In comparison to 3b, all 1 H NMR resonance signals of Li-4b were approximately 0.4 ppm downfield-shifted.Remarkably, the 1 H NMR spectrum also revealed the presence of one THF molecule in the complex, which remained after washing and drying.Therefore, it seems to be tightly bound to the lithium cation (vide infra) and to be necessary for the stabilization (calculated Li-O distance in Li(thf)-4b of 1.91 Å, with MBO Li-O 0.56).In the FT-IR spectrum of Li-4b, the N-H vibration mode was not present and, hence, confirmed the proposed composition.
The 31 P resonance signals of M-5b are 20 ppm downfield-shifted in comparison to 3b, in general, which is in accordance to previously observed M/N(H)R phosphinidenoid complexes [15][16][17].In addition, the 1 J W,P coupling constant has a characteristically small value [12][13][14][15][16][17][18][19].Presumably related to the decomposition of crown ethers in basic media at ambient temperatures, M-5b turned out to be thermally unstable and could not be further characterized, in contrast to previous studies, which even allowed for the isolation of a K/N(H)R phosphinidenoid complex [16,17], but not for the isolation of the corresponding amide complex.Upon warming up, M-5b decomposed, providing first the amide complex M-4b and, afterwards, a mixture of four not further characterized main products, displaying signals in the 31 P NMR spectrum at 41.9 ppm (no J W,P , no 1 J P,H ), 86.1 ppm ( 1 J W,P = 226 Hz), 91.5 ppm ( 1 J W,P = 230 Hz, 1 J P,H = 276 Hz) and 99.8 ppm ( 1 J W,P = 228 Hz, 1 J P,H = 283 Hz) (see ESI).
However, the presence of Li-5b in solution could be further proven by a followup reaction with MeI as an electrophile and giving rise to the expected P-methylated complex 7b, possessing a resonance at 70.7 ppm ( 1 J W,P = 252.0Hz) (Scheme 3).Due to its similarity to the aminophosphane complex 3b, the analytical data do not differ too much, except for those corresponding to the missing PH unit, such as the vibration mode in the FT-IR spectrum.
Interestingly, also the reaction of Li-4b with MeI yielded the same product 7b, and not the expected N-methylation product 3f.For 7b to form from both compounds Li-4b and Li-5b, a proton transfer has to occur in the former, which will be discussed below.The 31 P NMR spectroscopic reaction monitoring of the formation of 7b showed an intermediate that displayed a 31 P resonance signal at 55.0 ppm without a 1 J P,H coupling and a 1 J W,P coupling of 196.5 ppm, being somehow similar to values of the amide complexes M-4b.Based on these observations, the structure of this intermediate can be tentatively assigned to the P-methylated amide complex 6b (cf.Scheme 3).Intermediate 6b and compound 7b are initially formed in a ratio of 86:14 (at room temperature), while after 4.5 h, it changed to 23:77, with solely 7b remaining in the end (see ESI).The existence of a dianionic complex, bearing neither a P-H nor a N-H function, was excluded based on the 31 P NMR parameter.Scheme 3. Reactions of P-amido complex Li-4b and phosphinidenoid complex Li-5b with MeI as the electrophile, yielding complex 7b as the final product in both cases.

Theoretical Investigations on the Mechanism
To further unveil the very surprising formation of 7b in both methylation reactions and particularly from Li-4b (Scheme 4), quantum chemical calculations were performed.As stated beforehand, initial N-coordination of the solvated MeLi reagent favours Ndeprotonation of 3b to afford complexes Li(thf)-4b or Li(thf) 2 -4b, used here merely as simplified models of Li-coordinated amide salts.Noteworthy is the higher stability of the former, bearing only one THF molecule bound, despite featuring a less coordinatively saturated Li + cation, in agreement with the experimental observations (vide supra).).However, we assume that the tentatively proposed PH group-lacking intermediate 6b is indeed Li(thf) 2 -5b N , whose formation could be alternatively explained not via intra-but intermolecular trans-protonation (not computed) from Li(thf) 2 -4b.

Scheme 2 .
Scheme 2. Regioselective deprotonation of 1,2-bifunctional aminophosphane complex 3b.Table 2 contains selected 31 P NMR data of the Li/K-4b and Li/K-5b in THF solution.The comparison of the two metalated derivatives M-4b and M-5b revealed that their experimentally observed resonances were almost invariant, i.e., there was no clear cation dependency.The values are very close to the computed (GIAO/TPSS-D3/def2-QZVP//CPCM thf /TPSS-D3/def2-TZVP ecp ) values for the respective naked anions, which most likely parallels the occurrence of solvent and/or crown ether separated ion pairs in solution.

Table 1 .
Selected NMR data in THF and yields of synthesized aminophosphane complexes 3a

Table 2 .
Selected experimental and calculated NMR data (THF) of species that might result in the deprotonation processes.amounting to 100.66 and 19.95 kcal/mol, respectively), which would explain the destabilization of the P-H bond in 4b − with an MBO P-H of 0.84, in comparison with 3f MBO P-H of 0.95.The latter may further enhance the formation of 5b − from 4b − and, therefore, 7b in the end.Coordination to the K cation shows little influence on the P-N/P-H bonding situation in 4b − with MBO P-N 1.50 and MBO P-H 0.88. (