Catalytic hydrophosphination of alkynes using structurally diverse sodium diphenylphosphide donor complexes

Homogeneous catalysis relies heavily on the use of precious transition metal catalysts. Here, Whitelaw et al. describe the use of a series of alternative, structurally well-deﬁned compounds based on the earth-abundant metal sodium, which can catalyze hydrophosphination reactions of alkynes leading to alkenes. Currently, there is a drive to develop the organoelement chemistry of sodium, the most abundant alkali metal on earth, as an alternative to that of rarer lithium, with the prime focus on sustainability. Organo-lithium compounds have been essential to the success of synthetic chemistry for more than a century, although their implementation has been essentially conﬁned to stoichiometric synthesis. Here, we report on synthetic, structural, catalytic, mechanistic, and theoretical studies of a series of sodium diphenylphosphides, having unique structures deﬁned by the Lewis base donor D solvating the Lewis acidic sodium cation. These donor complexes are explored as hydrophosphination catalysts on reacting Ph 2 P–H with a range of alkynes and prove to be generally effective under ambient conditions, especially when n = 1 in [{Ph 2 PNa(D) x } n ]. Density functional theory (DFT) studies have shed light on the possible mechanisms of these catalytic cycles and how they relate to the E, Z, or a isomer formed. sodium bis(2,6-diisopropylphenyl)phosphide, 48 with each sodium atom being solvated by two THF units, and two Na–P bonds (lengths of 2.9246(8) A˚ and 3.0178(8) A˚), both considerably longer than the 2.825(2) A˚ observed in 1 . When [Ph 2 PNa.PMDETA] is searched, two hits show; 48,54 one monomeric compound with the previously mentioned bis(2,6-diisopropylphenyl)phosphide anion, and another involving a monomeric phos-phinimide complex. These have Na–P bond lengths of 2.8745(12) A˚ and 2.9619(11) A˚ respectively, this time notably shorter than the 2.9499(17) A˚ and 3.036(2) A˚ in 2. Finally, searching for [Ph 2 PNa.O 5 ] brings up one hit, 55 namely a sodium bis(2-oxy-3,5-di- tert -butyl)tert-butylphosphide polymer with one glyme molecule solvating each sodium cation, and an extra glyme molecule bridging between sodium cations. This gives a Na–P bond length of 2.8757(14) A˚, which is in good agreement with the 2.8742(8) A˚ found in 3. This paucity of structural hits also underlines the novelty of the new compounds 1–4. compounds, each existing in a different aggregation state owing to the speciﬁc Lewis base donor ligand employed. We then compared and contrasted their ability to act as catalysts for the Ph 2 P–H hydrophosphination of a range of alkenes and alkynes. It was observed that lowering the aggregation mmol; 0.007 g), the substrate precursor (0.6 mmol), and HPPh 2 (0.5 mmol; 0.085 mL). This reaction mixture was contained in a sealed J. Young’s tap NMR tube and the reaction was regularly monitored by 1 H and 31 P NMR until complete consumption of the HPPh 2 had occurred, determined by integration against the adamantane standard. Using catalysts 1 and 2 , the reactions were all carried out at 110 (cid:3) C unless otherwise spec-iﬁed.


INTRODUCTION
Phosphorus-containing compounds are invaluable due to their extensive applications in areas such as agriculture, pharmaceuticals, and organometallic catalysis. 5 Fertilizers, 6 antibiotics, 6 and flame retardants 7 are just three examples of the diverse roles that phosphorus compounds have that benefit the everyday lives of humankind. It follows, then, that hydrophosphination, the process of inserting a phosphorus-hydrogen bond into an unsaturated carbon-carbon or carbon-heteroatom bond, is of significant interest in modern research. Much current focus is on catalytic hydrophosphination as it provides a pathway to higher product yields with lower reaction times and less-demanding conditions. The first metal-catalyzed hydrophosphination was reported over 30 years ago in a seminal communication by Pringle and Smith,8 who used a Pt(0) complex to catalyze phosphine addition to acrylonitrile. In the intervening years, significant advances have been made using a plethora of catalyst types. [9][10][11][12][13][14][15][16][17][18] Reports of ''catalyst-free'' systems have also emerged. 19,20 With scientists shifting more of their research toward sustainability issues and the development of green chemistry, it would be desirable if the less-abundant and generally more toxic precious transition metal catalysts could be replaced with more-abundant and less-toxic main-group alternatives. Toward that goal, progress has been made, with, for example, reports of Ca, Sr, Ba, Al, and Ge-based catalysts, [21][22][23][24] while our own group has contributed the heterobimetallic lithium aluminate catalyst i Bu 3 AlPPh 2 Li(THF) 3 . 25 Organoalkali-metal compounds have long been one of the most widely used classes of reagents in the history of synthetic chemistry, with organosodium and organopotassium compounds first reported in 1858 by Wanklyn,26 while organolithium compounds, by far the most utilized organoalkali-metal reagents, were introduced by Schlenk and Holtz in 1917. 27 The numerous applications of s-block complexes have been the subject of many review articles, highlighting the importance of these compounds. [28][29][30][31][32][33][34] Nearly all of these applications have been within the confines of stoichiometric chemistry. On the other hand, d-block metals have dominated the catalysis world for decades. With advantages of high earth abundance (especially sodium and potassium), low cost, environmental friendliness, and relatively low toxicity compared with their d-block counterparts, alkali-metal-based compounds could become prize assets within the catalytic world moving forward, which is why the ''ignition'' has now been pressed and the development of alkali-metal homogeneous catalysis is starting to accelerate. Here, our interests lie in developing organosodium-based catalysts for hydrophosphination. The use of alkali metals in hydrophosphination chemistry was achieved as early as 1966 by Aguiar and Archibald 35 when they studied the addition of lithium diphenylphosphide to phenylacetylene in stoichiometric reactions. Related work was reported in 1970 by Issleib, Boehme, and Rockstroh when they reacted lithium and potassium diphenylphosphides stoichiometrically with acetylenes. 36 Since then, hydrophosphination has been advanced to the catalytic regime with other alkali-metal catalysts, such as KO t Bu [37][38][39][40] and KHMDS (HMDS = 1, 1, 1, 3, 3, 3-hexamethyldisilazide). 41 The study of sodium compounds for carrying out various chemical transformations is an area currently undergoing a resurgence. 30,[42][43][44][45][46][47] In our aforementioned research on employing a heterobimetallic lithium aluminate compound for catalytic hydrophosphination of various alkynes, alkenes, and carbodiimides, 25 monometallic lithium diphenylphosphide was also screened for comparison and, although it gave slightly lower yields than the aluminate i Bu 3 AlPPh 2 Li(THF) 3 , it was still catalytically competent. Given that sodium is both more abundant than lithium and its compounds are generally more reactive than lithium compounds, it left us pondering whether sodium could be a more effective and more sustainable replacement for its lighter congener in catalytic hydrophosphination processes. To address this question, we have therefore synthesized and structurally characterized four monometallic sodium diphenylphosphide complexes and tested their catalytic competence in the hydrophosphination of benchmark alkenes and alkynes.

Synthesis and characterization
Although sodium diphenylphosphide has been synthesized previously via metalation of the parent secondary phosphine Ph 2 PH by benzyl sodium, 48 transmetalation of LiPPh 2 with NaO t Bu, 49 or by direct metalation using sodium dispersion, 50,51 we found it convenient to make it by a simple deprotonative metalation reaction between n-butylsodium 52 and Ph 2 PH in a 1:1 stoichiometric ratio in hexane solution. This produced a bright yellow suspension, to which a stoichiometric quantity of a polydentate Lewis base donor ligand was added, either PMDETA (N,N,N 0 ,N 00 ,N 00pentamethyldiethylenetriamine), 15-crown-5, or 2,2,2-cryptand, while monodentate tetrahydrofuran (THF) was added in excess. Each of these solutions deposited crystals of a suitable quality for X-ray crystallographic determination, which estab- lowering aggregation state with increasing denticity of the Lewis base donor ligand, with the THF-solvate a polymer, the PMDETA-solvate a dimer, the 15-crown-5-solvate a monomer, and the highest dentate 2,2,2-cryptand sequestering the sodium cation to generate a charge-separated ion pair. Figure 1 shows the asymmetric units of 1-4; while Figure 2 gives more extended views of the aggregated structures of 1 and 2.
As can be seen from Tables S1 and S2 in the supplemental information, there are significant differences in the metrics between the different complexes 1-3. Structural data for compound 4 can also be found in the supplemental information (Tables S3 and S4) and has been omitted from the above tables due to the complexity of the compound and the lack of any Na-P contact. The sodium cations in each compound have different coordination numbers and adopt different geometries; it is 4-coordinate in 1, adopting a tetrahedral geometry (mean bond angle of 109.5 ), 5-coordinate in 2, adopting a distorted trigonal bipyramidal geometry where P2 and N2 occupy the axial positions (bond angle of 170.12 [6]) and P1, N1, and N3 lie equatorially (bond angles ranging from 110.05 [14] to 125.47 [11]), and is 6-coordinate in 3 with a distorted trigonal prismatic geometry having a mean bond angle of 89.16 around the edges of the coordination sphere and 127.3 diagonally across it. The geometry of the phosphorus atoms change across the set of compounds also, from 4-coordinate distorted tetrahedral in 1 and 2 (mean bond angles of 116.8 and 106.2 respectively) to 3-coordinate distorted trigonal pyramidal in 3 with a mean bond angle of 102. sodium bis(2,6-diisopropylphenyl)phosphide, 48 with each sodium atom being solvated by two THF units, and two Na-P bonds (lengths of 2.9246(8) Å and 3.0178(8) Å ), both considerably longer than the 2.825(2) Å observed in 1. When [Ph 2 PNa.PMDETA] is searched, two hits show; 48,54 one monomeric compound with the previously mentioned bis(2,6-diisopropylphenyl)phosphide anion, and another involving a monomeric phosphinimide complex. These have Na-P bond lengths of 2.8745(12) Å and 2.9619(11) Å respectively, this time notably shorter than the 2.9499(17) Å and 3.036(2) Å in 2. Finally, searching for [Ph 2 PNa.O 5 ] brings up one hit, 55 namely a sodium bis(2-oxy-3,5-di-tertbutyl)tert-butylphosphide polymer with one glyme molecule solvating each sodium cation, and an extra glyme molecule bridging between sodium cations. This gives a Na-P bond length of 2.8757(14) Å , which is in good agreement with the 2.8742(8) Å found in 3. This paucity of structural hits also underlines the novelty of the new compounds 1-4.

Catalytic hydrophosphination studies
Next, we moved on to quantifying the comparative effectiveness of the solvated sodium phosphides for catalytic hydrophosphination of the model alkyne, phenylacetylene PhChCH (Table 1). Since our previous hydrophosphination work using the heterobimetallic catalyst i Bu 3 AlPPh 2 Li(THF) 3 required a 10 mol % catalyst loading at 110 C, 25 we viewed this as a logical starting point for our new aluminum-free phosphides. However, simple visual inspections suggested that compounds 3 and 4 were reacting significantly faster than the 30-min measurement mark, so we therefore lowered the temperature for these reactions to room temperature. Pleasingly, both compounds still produced quantitative conversion to the hydrophosphinated alkene products in 30 min. The conversions in Table 1 were calculated by 1 H NMR spectroscopy using adamantane as the internal standard and are based on the consumption of diphenylphosphine starting material. Entries 1, 2, 5, and 6 were therefore taken as the optimized conditions for each catalyst. The catalysts were each tested on 11 different substrates and the full set of results are collated in Figure 3. The stereoselectivities obtained in each case are also noted. Phosphide 1 was shown to be the least efficient catalyst as it required the most demanding conditions and produced the lowest conversions. Notably, however, it was still able to convert 4-cyanophenylacetylene, 3-phenyl-1-propyne, and methyl acrylate in just 30 min at room temperature in good to excellent conversions (entries e, f, and k). Understandably, the more challenging aliphatic substrates Figure 2. Part of the polymeric structure of 1 and dimeric structure of 2 with key atom labeling H atoms have been removed for clarity. PMDETA ligands in 2 have been set to 60% transparency for clarity. Here and elsewhere, for compound 1, the prime symbol '' 0 indicates the symmetry operation x + 1/2, 1 À y, z; and for compound 2, it indicates 1 À x, y, 3/2 À z.

OPEN ACCESS
4-phenyl-1-butyne and 1-octyne required 3 h at 110 C in order to obtain moderate to good conversions (entries g and h), whereas the most difficult 3-hexyne produced only a 10% conversion after a full 24 h at 110 C (entry i).
Phosphide 2 performed marginally better than 1 and notably was more efficient when electron-withdrawing substituents were present, quantitatively converting 4-trifluoromethyl phenylacetylene and 4-cyanophenylacetylene in 30 min at room temperature (entries d and e); it also only required these conditions for the more activated methyl acrylate (entry k). Longer times were necessary to successfully convert 3-phenyl-1-propyne, 4-phenyl-1-butyne and 1-octyne, producing good to moderate conversions in each case (entries f, g, and h). Interestingly, 3 h were also required to convert styrene using 2 (entry j). Finally, phosphide 2 secured a slightly better result with 3-hexyne than 1, but the conversion was still poor (entry i). Moving to monomeric 3, this proved significantly more reactive than the previous two phosphides and quantitatively converted all but three of the substrates under the standard conditions, with 4-phenyl-1-butyne producing a low conversion, and 1-octyne requiring 3 h at 110 C (entries g and h). A good result was obtained with the obstinate 3-hexyne, albeit still requiring harsh conditions (24 h at 110 C, entry i). Compound 4 proved to be the most efficient catalyst of the set with the non-activated aliphatic alkynes the only substrates that required deviation from standard mild conditions. Notwithstanding, 1-octyne required 3 h to secure a good conversion, and 4 was the only compound to successfully hydrophosphinate 1-octyne at room temperature (entry h), demonstrating its heightened reactivity, and while 3-hexyne did still require 24 h at 110 C, 4 was the only catalyst to obtain quantitative conversion of this substrate (entry i). In order to examine whether or not the alkali metal is necessary, we attempted the hydrophosphination of phenylacetylene, catalyzed by the ammonium analogue [Bu 4 N] + [Ph 2 P] -. After 3 h at 110 C there was only a 13% conversion observed, and this did not change when left for a further 72 h at room temperature. This result demonstrates the necessity of the alkali metal in our work. Significant work by Wipff and co-workers 56 on the molecular dynamic simulations of 2,2,2-cryptand encapsulating a series of alkali-metal cations suggest that some of the sodium surface is available for reactivity, which may explain in part the high reactivity of compound 4.
In order to obtain selected yields of products as well as conversions, we repeated four of the above catalytic reactions and each reaction was quenched with an

OPEN ACCESS
EtOAc/water mixture before being run through a column packed with silica and using a 1:19 EtOAc/hexane eluant mixture. This allowed us to obtain yield by mass for four of our substrates, with compound 3 being employed as the catalyst in each case.
It should be noted that although compound 4 produced better catalytic results, 3 was employed as the model catalyst throughout these further studies due to its higher stability, making it easier to isolate, store, and use from a consistent batch. Displayed in Table 2, the results show a general decrease from the conversions but the isolated yields are still good with the lowest one at 73%. The corresponding turnover numbers (TONs) and turnover frequencies (TOFs) have been determined using NMR spectroscopy and the results are presented in the supplemental information (Table S5). In 2019, Schmidt et al. 57 reported the use of the N,N-dimethylbenzylamine-derived lanthanum catalyst, La(Dmba) 3 for the hydrophosphination of various alkynes. Applying a 5 mol % catalyst loading, they required 12 h at room temperature to obtain an 87% yield of hydrophosphinated phenylacetylene and 48 h for a 65% yield of hydrophosphinated diphenylacetylene, whereas our compounds 3 and 4 required 30 min at room temperature for 99% conversions of each, and 3 obtained 86% and 73% yields of phenylacetylene and diphenylacetylene respectively. Another example is that of Westerhausen 1 in 2008, where he hydrophosphinated diphenylacetylene using a calcium catalyst, and although this did achieve quantitative conversion, it required a 20 mol % catalyst loading and 12 h at 75 C.
While our new methodology is not stereoselective, it is interesting to note the ratio of isomers obtained in each case. With some exceptions, the stereoselectivity appeared to be more dependent on the substrate rather than the catalyst, with each substrate producing similar stereoselectivities irrespective of the catalyst involved, suggesting similar mechanistic pathways. For example, the presence of an electron-donating group favored production of the Z isomer, whereas electron-withdrawing groups favored production of the E isomer. There were also a few cases where significant quantities of the a isomer were observed, where the two nonhydrogen substituents are attached to the same carbon atom.
We first wanted to investigate the effect of changing temperature on the stereoselectivity of the reaction. Using 3 as an example, E:Z ratios of 45:55 and 35:65 were found for phenylacetylene and diphenylacetylene, respectively. When these 3 respectively. Furthermore, when the reaction was carried out with phenylacetylene at the lower temperature of À20 C, a 20:80 E:Z ratio was observed (see supplemental information). These findings are consistent with the Z isomer being the kinetically favored product and the E isomer the thermodynamically favored product.
Next, we sought to shed light on the mechanism behind this isomerization process, as it could potentially lead us to achieve better stereoselectivity in the future. We first investigated the effect of temperature on the stereoselectivity (Table 3) and used four of the previously investigated substrates, encapsulating terminal and internal alkynes, as well as alkynes containing both electron-donating and electron-withdrawing groups. The crown ether monomer 3 was utilized as the catalyst throughout this part of the study. While the first two substrates produced promising results with nearly quantitative stereoselectivity to E upon heating, entries 3 and 4 demonstrate the inherent complexity of this process, as the electronics of the substituent groups on these alkyne substrates appear to have a clear and notable impact on the isomerization, showing no change upon heating.
Schmidt et al. 57 also discussed a potential isomerization mechanism with their aforementioned La(Dmba) 3 catalyst and proposed that the substrates were temporarily undergoing double hydrophosphination. This would produce an alkane intermediate, allowing free rotation of the carbon-carbon bond, before b-phosphide elimination re-formed the alkene product in the more favorable E conformation (Figure 4).
To add merit to this hypothesis, they demonstrated that reducing the quantity of phosphine in the reaction selectively produced the Z isomer, whereas alternatively increasing the quantity of phosphine selectively produced the E isomer. To investigate whether or not our system could be undergoing a similar process, we attempted the same reaction. However, no difference was observed in our system when the phosphine concentration was increased or reduced, implying that our system proceeds by a different mechanism.
In 1966, Aguiar and Archibald 35 published work showing the stoichiometric and stereoselective addition of phosphines to alkynes via lithium diphenylphosphide. They showed that, by carrying out their reactions in THF and adding a primary amine to the mixture (e.g., n-butylamine), they witnessed E selectivity. In contrast, when Article they added a secondary amine (e.g., N-methylaniline), there was a stereoselective switch to the Z isomer. Interestingly, they also noted that this was not the case on changing their polar metal phosphide source to sodium from lithium. Regardless, we decided to attempt this method with our sodium system, and the results are displayed in Table 4.
While Aguiar and Archibald carried out their process stoichiometrically, we opted to attempt this catalytically and began by using [NaPPh 2 ] ꝏ in d 8 -THF to replicate the conditions used by Aguiar and Archibald (entries 1-3) and found that the stereoselectivity does change when a sodium catalyst is used in the presence of either a primary or secondary amine. We then moved to using one of our own catalysts for this purpose (entries 4-7) and found that the choice of solvent was significant, observing no change in the stereoselectivity of hydrophosphination of diphenylacetylene when the reaction was run in d 8 -toluene. That notwithstanding, we were able to partially control the stereoselectivity of diphenylacetylene using 3 in d 8 -THF, obtaining an E:Z ratio of 97:3 when a primary amine was added, and reversing this to 30:70 on changing to a secondary amine additive. When we attempted this method on 4 0 -methylphenylacetylene, a substrate that has been shown to be Z selective previously, there was unfortunately no notable difference when a primary or secondary amine was added to the reaction (entries 8-10).
There have been other reports of main-group metal-catalyzed hydrophosphination over the past few years; for example, Webster et al. reported their use of [GeCl {N(SiMe 3 ) 2 } 3 ] as a pre-catalyst for hydrophosphination in 2020 and showcased the good performance of this catalyst as it only required a 5 mol % catalyst loading and had a wide-ranging substrate scope. 24 However, one of the main limitations of their system was the longer reaction time with internal alkynes, with durations as long as 96 h in one case. This was shown not to be a problem in our sodium work. Our catalysts, for example, were all able to quantitatively convert diphenylacetylene to the respective hydrophosphinated product either in 1 h at 110 C (1 and 2) or in 30 min at room temperature (3 and 4), whereas the germanium catalyst reported by Webster et al. required 48 h at room temperature to obtain an 81% conversion. The reported germanium catalyst also struggled with stereoselectivity, something we were also familiar with. Two years prior to this, the Webster group also demonstrated the ability of KHMDS to carry out catalytic double hydrophosphination. 41 A 10 mol % catalyst loading was required, along with longer reaction times than were required for our system (6 h). It is worth noting that we have not observed double hydrophosphination in any of our work.
In an early seminal piece of work, Knochel and Bunlaksananusorn introduced the use of KO t Bu for hydrophosphination in 2002 39 and disclosed the ability of this easy-toobtain and cheap starting material as a catalyst. Their work required a 20 mol % loading of the catalyst though, and the product yields spanned the range 63%-88%, notably lower than what was generally observed with our catalysts. There were also no alkynes tested in Knochel's study, only activated alkene substrates. Manners et al. also carried out work with KO t Bu, albeit for the purpose of homoand heterodehydrocoupling of phosphines, 40 but it is worth noting that reported intermediates in their work are hydrophosphination products. While they do carry out these reactions with only a 10 mol % loading of their potassium catalyst, the process requires harsh conditions (e.g., 130 C in THF for 16 h). Intrigued by these publications, we attempted the hydrophosphination of phenylacetylene and 3-hexyne using

OPEN ACCESS
KO t Bu as the pre-catalyst; this resulted in quantitative conversion of phenylacetylene in 2 h at 110 C (E:Z 35:65) and an 8% conversion of 3-hexyne after 29 h at 110 C. When a catalytic quantity of 18-crown-6 was added to this reaction, a 78% conversion of 3-hexyne was obtained after 47 h at 110 C. While these results do reinforce the conclusion that KO t Bu can be successfully used as a pre-catalyst for the hydrophosphination of alkynes, these preliminary results show that it is not as efficient as compounds 3 and 4 described in this work.

DFT studies
In an attempt to gain insight into the mechanism behind our sodium-catalyzed hydrophosphinations, and the formation of each isomer, we turned our attention to density functional theory (DFT) studies. For calculational simplicity, we employed the diphenylphosphine-analogue of compound 3 as the catalyst in these studies, diphenylphosphine as the phosphorus reagent, and 1-propyne as the benchmark substrate, to investigate the catalytic process, carried out in toluene solvent. All calculations were performed using the Gaussian 16 3 program with the M06-L 4 function and def2tzvp 2 basis set using an ultrafine integral with the toluene solvent modeled via the CPCM (conductor-like polarizable continuum model) method. 58 Reaction pathways were confirmed by following the intrinsic reaction coordinates (IRC) 2,59 in both directions from the optimized transition states; followed by optimization and frequency calculations of the resulting minima to yield reaction complexes and intermediates. Figure 5 shows competitive energy profiles for the a, E, and Z isomers, from which it can be seen that the Z isomer has a lower initial energy barrier (DG* AC À > TS1 = 26.8 kcal/mol) from the active sodium phosphide catalyst (AC), therefore showing it to be the kinetic product consistent with our reasoning from the experimental observations with the E isomer (DG* AC À > TS1 = 33.0 kcal/ mol) and a-isomer (DG* AC À > TS1 = 38.8 kcal/mol) tending toward inaccessible, and indeed the a-isomer is not observed experimentally for this reaction. However, when the first intermediate of the Z isomer is formed (Int1), the relative barrier height for the forward (DG* Int1 À > TS2 = 17.6 kcal/mol) is higher than the backward reaction (DG* Int1 À > TS1 = 13.3 kcal/mol) and as such the reverse reaction to form the reaction complex (RC) can provide new starting material for the E isomer, as the energies and molecular orientations of the E/Z RC are similar. This isomerization pathway is also feasible for the E-isomer intermediate; however, it is much less accessible than the Z isomer due to the forward reaction (DG* Int1 À > TS2 = 13.7 kcal/mol) being favorable for the E isomer over the backward reaction (DG* Int1 À > TS1 = 20.7 kcal/mol), suggesting that, once the E isomer intermediate is formed via the higher-energy transition state, the reaction continues to form the thermodynamic product. This does not provide an easy route to the a isomer as the energy of that RC is significantly higher and the reactant is orientated differently with respect to the catalyst (visualizations of the 3 RCs are shown in the supplemental information, Figure S150). The intermediate structure is the only structure that can be isomerized in these systems, as following the second transition state (TS2) the exergonic nature of the reaction shows the protonation step to be irreversible. The E/Z relative energy barriers are achievable at the experimental conditions. Moreover, the subtle variations in the barrier height of the forward and backward barriers for the Z isomer is suggestive of the role that the substituents of the alkyne substrate can play in determining the stereoselectivity. That is, small changes in the substrate can result in significant relative ordering of the forward and backward barrier heights and consequently the ratio of isomers generated. From these computational data, we were able to propose a plausible general mechanism by which this catalytic hydrophosphination proceeds, for the formation of the a, E, and Z isomers. This mechanism of formation of the E and Z isomers is displayed in Figure 6 and the mechanism for the a isomer can be found in Fig-ure S149 of the supplemental information.
The cycle begins with an initial deprotonation step in order to form the AC, followed by coordination of the catalyst to the alkyne substrate, forming the RC. The test reaction with the NBu 4 + salt as stated above bears testimony to the nature of the RC, where the roles of both the Na + and the phosphide anion are important. There is then a complexation step, the rate-determining step, which leads to transition state 1 (TS1). This then undergoes formal insertion, producing an alkene intermediate (Int1) before a second equivalent of phosphine approaches and coordinates to the alkene, resulting in intermediate 2 (Int2). A second complexation step is observed to generate transition state 2 (TS2), which is followed by a formal protonation step to replace the sodium atom with a proton and form the product complex (PC) where the newly regenerated sodium phosphide catalyst is still coordinated to the phosphorus atom of the alkene. Finally, this undergoes a dissociation step to release the product and fully regenerate the catalyst.
To summarize, in this work we have reported the synthesis and structural characterization of four new sodium diphenylphosphide compounds, each existing in a different aggregation state owing to the specific Lewis base donor ligand employed. We then compared and contrasted their ability to act as catalysts for the Ph 2 P-H hydrophosphination of a range of alkenes and alkynes. It was observed that lowering the aggregation ll OPEN ACCESS state of the compounds increases catalytic performance, with the monomeric 15-crown-5 solvate (3) and the charge-separated ion pair, where 2,2,2-cryptand is the donor ligand (4) performing exceptionally well under ambient conditions. We also noted the stereoselectivity for each substrate and attempts thus far to obtain a completely stereoselective process have been unsuccessful. DFT studies have been used to propose possible mechanisms for the hydrophosphination of alkynes that provide insight into the complicated pattern of the E/Z isomer ratios obtained in these catalytic reactions that suggests a high dependency on the particular substrate studied. Collectively, this work advances the development of earth-abundant homogeneous catalysts for fundamentally important chemical transformations, and further work on improving the design of alkali-metal catalysts is ongoing.

EXPERIMENTAL PROCEDURES
Resource availability Lead contact Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Robert Mulvey (r.e.mulvey@strath.ac.uk). Materials availability Hexane, THF, and toluene were dried by heating to reflux over sodium benzophenone ketyl and then distilled under nitrogen prior to use. D 8 -toluene and d 8 -THF were degassed by freeze-pump-thaw methods and stored over activated 4 Å molecular sieves prior to use. All reagents were purchased from commercial sources and used as received, unless stated otherwise. PMDETA was distilled and stored over activated 4 Å molecular sieves prior to use. n BuNa was synthesized according to literature procedures and stored as a white powder in the glovebox freezer. 52  1 All structures were solved and then refined to convergence against F 2 using all independent reflections by the full-matrix least-squares method using SHELXL-2018 implemented within WINGX or Olex2. 2,3,4 Crystals of 1 were cut from fragile thin sheets of material. These were found to shatter at low temperature and so a collection temperature of 230 K was used. Even at this higher temperature, the samples were observed to bend upon mounting. These material properties resulted in non-ideal diffraction behavior and a relatively low-quality final model. This model does, however, confirm the atomic connectivity of this compound. In all structures, non-hydrogen atoms were refined using anisotropic thermal parameters. Two THF ligands of 1 and a C 2 H 4 section of the crown ether in 3 were treated as disordered. All of these groups were modeled with atoms split over two sites and with appropriate restraints and constraints applied to ensure normal geometries and displacement behaviors. Selected geometric parameters for 1-4 are given in Tables S1-S4 To a suspension of n BuNa (0.16 g; 2 mmol) in hexane (10 mL) was added HPPh 2 (0.34 mL; 2 mmol) and this was stirred at room temperature for 1 h. To the yellow ll OPEN ACCESS suspension was added 2,2,2-cryptand (0.752 g; 2 mmol) and this formed a red suspension, which was allowed to stir for a further hour at room temperature before THF (5 mL) was added with heating from a heat gun, resulting in formation of a red solution. This was allowed to cool to room temperature and, once cool, a crop of red needle-like crystals were obtained. These were isolated by filtration (0.2519 g; 0.43 mmol; 21.5% yield). 1  General catalytic procedure A 10 mol % loading of the desired catalyst (1-4) was added to 0.5 mL of d 8 -toluene solution containing adamantane as an internal standard (0.05 mmol; 0.007 g), the substrate precursor (0.6 mmol), and HPPh 2 (0.5 mmol; 0.085 mL). This reaction mixture was contained in a sealed J. Young's tap NMR tube and the reaction was regularly monitored by 1 H and 31 P NMR until complete consumption of the HPPh 2 had occurred, determined by integration against the adamantane standard. Using catalysts 1 and 2, the reactions were all carried out at 110 C unless otherwise specified. Reactions using catalysts 3 and 4 were carried out at room temperature unless otherwise specified. For isomerization studies by temperature, the reaction mixture was monitored by 1 H and 31 P NMR at room temperature for 30 min before being heated to 110 C for 1 h and measured again. For the low-temperature study, the catalyst was added to the NMR tube at À20 C and monitored by 1 H and 31 P NMR at this temperature before being allowed to warm up to room temperature. For isomerization studies by amine additive, a stoichiometric quantity of either N-methylaniline (0.055 mL; 0.5 mmol) or n-butylamine (0.05 mL; 0.5 mmol) was added to the mixture before addition of the catalyst. These reactions were all carried out in d 8 -THF, unless otherwise specified, and at room temperature.