Article Cite This: Acc. Chem. Res. XXXX, XXX, XXX−XXX
pubs.acs.org/accounts
Triamidoamine-Supported Zirconium Compounds in Main Group Bond-Formation Catalysis Rory Waterman*
Downloaded via NOTTINGHAM TRENT UNIV on August 6, 2019 at 20:12:06 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Department of Chemistry, University of Vermont, Burlington, Vermont 05405, United States CONSPECTUS: The rationale to pursue long-term study of any system must be sound. Quick discoveries and emergent fields are more than temptations. They remind us to ask what are we gaining through continued study of any system. For triamidoaminesupported zirconium, there has been a great deal gained with yet more ahead. Initial study of the system taught much that is applied to catalysis. Cyclometalation of a trimethylsilyl substituent of the ancillary ligand, abbreviated (N3N) when not metalated for simplicity, via C−H bond activation is facile and highly reversible. It has allowed for the synthesis of a range of Zr−E bonds, which are of fundamental interest. More germane, cyclometalation has emerged as our primary product liberation step in catalysis. Cyclometalation also appears to be a catalyst resting state, despite how cyclometalation is a known deactivation step for many a compound in other circumstances. Catalysis with triamidoamine-supported zirconium has been rich. Rather than summarizing the breadth of reactions, a more detailed report on the dehydrocoupling of phosphines and hydrophosphination is provided. Both reactions demonstrate the outward impact that the study of (N3N)Zr-based catalysis has afforded. Dehydrocoupling catalysis, or bond formation via loss of hydrogen, is particular to 3p and heavier main group elements. The reaction has been important in the formation of E−E and E−E′ bonds in the main group for molecular species and materials. While study of this reaction at (N3N)Zr compounds provides key insights into mechanism, discoveries in the area of P−P and Si−Si bond formation with (N3N)Zr derivatives as catalysts have greater reach than merely the synthesis of main group element containing products. For example, that work has informed design principles for the identification of catalysts that transfer lowvalent fragments. The successful application of these principles was evident in the discovery of a catalyst that transfers phosphinidene (“PR”) to unsaturated substrates. Hydrophosphination exhibits perfect atom economy in the formation of P−C bonds. The reaction can proceed without a catalyst, but the purpose of a catalyst is enhanced reactivity and selectivity. Nevertheless, significant challenges in this reaction remain. In particular, (N3N)Zr compounds have demonstrated high activity in hydrophosphination and readily utilize unactivated unsaturated organic molecules, challenging substrates for any heterofunctionalization reaction. This activity has led to not only impressive metrics in the catalysis but access to previously untouched substrates and formation of unique products. The particular properties of the (N3N)Zr system that engage in this reactivity may influence other heterofunctionalization reactions. The recently discovered photocatalytic hydrophosphination with (N3N)ZrPRR′ compounds already appears to be general rather than unique and may drive additional bond formation catalysis among early transition-metal compounds.
■
substantiated the terrific success of Schrock2 and Scheer4 among others to prepare triamidoamine and other fivecoordinate compounds with triple bonding in an axial coordination site. Group 4 metal triamidoamine chemistry, however, did not lay dormant in Verkade’s absence. Scott led a deeper investigation of the fundamental zirconium organometallic chemistry.5,6 One of the critical advances that has fueled our work was the identification of cyclometalation of a trialkylsilyl substituent. Scott nobly sought to isolate the intermediate zirconium hydride in his reactions that were proposed to precede metalation. While those efforts failed to yield an isolable product, they gave substantial insight into the broader category of triamidoaminesupported zirconium compounds.5,6 In particular, efforts to branch out beyond trialkylsilyl derivatives was particularly
EARLY TRANSITION-METAL TRIAMIDOAMINE COMPOUNDS Verkade reported the earliest group 4 compounds supported by triamidoamine ligands as CVD precursors.1 Indeed, the family of (N3N)MX (N3N = N(CH2CH2NSiMe3)33−, M = group 4 metal, X = anionic ligand) compounds are almost surprisingly volatile with modest heating under reduced pressure. While Verkade’s legacy in compounds with these “tren” ligands is substantial, there was no additional work on these early metal systems. That absence was no coincidence. Emergent work from Schrock in triamidoamine-supported group 5 metal compounds gave Verkade pause,2 and he retreated from those systems to avoid competition.3 Schrock’s (ongoing) work with group 5 metals provided some of the foundation of our understanding of d0 metals with these ligands.2 Rigorous characterization of a range of compounds led Schrock to propose an orbital configuration based on canonical trigonal bipyramidal molecular geometry. That proposal © XXXX American Chemical Society
Received: May 30, 2019
A
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
despite the steric bulk imparted by any substituent on a triamidoamine ligand. Structural data and computational evidence, described below, disproved this thinking. Being incorrect in this hypothesis was a surprising boon. Rather than providing an accessible orbital, the (N3N)Zr-phosphido derivatives have long, reactive Zr−P bonds, which are the result of no significant ligand-to-metal πbonding contribution from the phosphido ligand.10 This was an excellent way to be incorrect. A primarily σbonding interaction is rare for group 15 elements bonded to electropositive transition metals, and more importantly, a σ-only Zr−P bond should be much more reactive than one with a πbonding component. The latter supposition has borne out, and the Zr−P bond of these derivatives has proven quite reactive.
informative. The sum of these efforts laid the groundwork for later successful investigations from Scott’s group in enantioselective early metal catalysis (among other areas).7 While this terse recount does not fully address the field at our entry point (2006), it highlights a few of the influential discoveries and researchers.
■
WHY TRIAMIDOAMINE-SUPPORTED ZIRCONIUM? The appeal of triamidoamine-supported zirconium chemistry for us arose from several general considerations and one original, albeit incorrect, hypothesis about the reaction chemistry at these compounds. Zirconium is an attractive element for exploratory catalysis. It has a relatively high natural abundance compared to several 3d metals including cobalt and nickel, has low toxicity, and is relatively inexpensive.8,9 Organometallic chemistry of zirconium has historically been dominated by zirconocene derivatives for both their fundamental importance in polymerization catalysis and the tremendously rich and interesting reactivity that those compounds exhibit. Finally, the ligand precursor, like cyclopentadiene, is fairly inexpensive, and preparation of substituted derivatives is straightforward. From Schrock’s description of the electronic structure of triamidoamine-metal compounds2 and the structural data from Verkade on (N3N)ZrNMe2,1 it was anticipated that there would be a ligand-to-metal π-bonding contribution from a terminal phosphido ligand. We (really, just me) incorrectly hypothesized that the orbital arrangement of the triamidoamine zirconium derivatives would provide more facile orbital access to a phosphido ligand as compared to zirconocene derivatives (Chart 1). This thinking came from the fact that a phosphido
■
SYNTHESIS AND MOLECULAR STRUCTURE OF PRECURSORS Verkade reported the original synthesis of (N3N)ZrNMe2.1 From that compound, we derived a straightforward preparation of (N3N)ZrCl, which availed various products from salt metathesis reactions. It quickly became evident that the cyclometalated derivative [κ5-N,N,N,N,C(Me3SiNCH2CH2)2NCH2CH2NSiMe2CH2]Zr (1) was a more important precursor than the chloride compound. Our preparation of 1 mirrored that developed by Scott for related compounds (Scheme 1).6 Our observation is that any polar E−H bond will react with the silaazametalacyclobutane to give the Zr−E product with restoration of a trimethylsilyl substituent (Scheme 2).10 This reaction proved more general than salt metathesis with (N3N)ZrCl in which the relative electronegativities of the participants as well as reduction potential can influence the outcome of the reaction. For example, reaction of (N3N)ZrCl with LiPHPh gives a complex mixture of which (N3N)ZrPHPh is one minor component. Given that 1 is the product of Zr(CH2Ph)4 and N(CH2CH2NHSiMe3)3, the overall efficiency of a preparation of (N3N)ZrPHPh from 1 is greater than other routes.10 Ring-opening of 1 is a facile process for polar E−H bonds, but there are limitations. For example, size matters. We have identified that otherwise viable E−H bonds (e.g., P−H or As− H) will fail to produce the (N3N)ZrX product in part or completely when the substrate is sterically encumbered.10−12 In one telling instance, the reaction of 1 with dmpAsH2 (dmp = 2,6-dimesitylphenyl) produces an equilibrium quantity of (N3N)ZrAsHdmp that is detectable but not isolable (Scheme 3).11 Cyclometalation is an outward indicator that the (N3N)ZrX compound is unstable, but steric factors are not the only determinant. We have been unable to isolate a silyl derivative, and given how similar the steric profile of phenylsilane and phenylphosphine are, this appears to be the result of thermodynamics in which such reactions favor cyclometalation
Chart 1. (left) Comparison of Frontier Orbitals between Idealized Zirconocene (C2v) and Triamidoamine Zirconium Compound (C3v) in Support Orbital Accessibility for a πBonded Ligand and (right) Key Triamidoamine Zirconium Derivatives in This Account
ligand would have an adjacent ligand in the cleft of the zirconocene that would hinder access from one side. For (N3N) Zr derivatives, orbital access from either direction was the same Scheme 1. Preparation of Compound 1
B
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research Scheme 2. Examples of the Direct Preparation of (N3N)Zr Derivatives from the Common Precursor 1
reactions (vide infra). Prior investigators, notably each Schrock and Scott,5,6,14 have observed intramolecular C−H bond activation and cyclometalation for group 4 metals with triamidoamine ligands. That reactivity arose from efforts to isolate the respective hydride derivatives or in reactions with hydrogen. None of the family of triamidoamine group 4 hydrides are stable, but the titanium derivative has been observed in solution with a chemical shift of δ 8.29.14 We have observed (N3N)ZrH in solution under hydrogen pressure,15 and that chemical shift, δ 10.59, compares favorably to that observed by Schrock and co-workers. The instability of these hydride derivatives is surprising. Not only have we calculated stable gas phase ground state structures of (N3N)ZrH,10 but other zirconium hydride derivatives with amido-based ligands are isolable.16 On first glance, one may anticipate that the formation of hydrogen gas represents a driving force in this reactivity. That may be true, but (N3N)ZrMe is an isolable compound that will eliminate methane but only upon heating.12,17 Thermodynamic arguments are always useful, and given the similarity of the bond dissociation energies of H−H and H−CH3, the relative strength of the Zr−E bond must play a critical role. In a slightly vexing turn, computational analysis reveals that the Zr−H bond is stronger than the Zr−C bond of the respective compounds. (This is only slightly vexing in that we have reasonable, albeit empirical, handle on the situation.).10 At this point, one might retreat to an argument for kinetic facilitation by improved overlap with an s orbital, which may also indicate a reason for Zr−Si instability. Further discussion ceases to be fully evidence
Scheme 3. Equilibrium Formation of (N3N)ZrAsHdmp
and formation of 1 and free silane (Scheme 4).13 The interplay of steric and electronic effects in this regime are subtle. For Scheme 4. Equilibrium Formation of a Silyl Derivative (N3N)ZrSiH2Ph Prior to Decomposition
example, the benzyl derivative, (N3N)ZrCH2Ph, is observable though unstable under ambient conditions, despite (N3N)ZrnBu being isolable,10 a difference that we attribute to steric factors. Regardless of the apparent stability, cyclometalation of a trimethylsilyl substituent of (N3N)Zr derivatives is rapid and ubiquitous in the chemistry of these compounds. Moreover, we continue to harbor the poorly tested hypothesis that this cyclometalation is a critical feature in successful catalytic
Scheme 5. Exchange of Deuterium from Phenylsilane-d3 to the Trimethylsilyl Substituents of 1 That Illustrates Formation of an (N3N)ZrX (e.g., (N3N)ZrSiH2Ph) Derivative That May Not Be Otherwise Observable
C
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
structure. Second, an interaction between the pseudoaxial amine nitrogen and the zirconium is ever present, though weak. Third, all structurally characterized examples of compounds with πbasic 3p and heavier elements exhibit no signs of ligand-to-metal π donation. The apparent consequence of the strong π-donation from the amido arms of the triamidoamine ligand is diminished if not nonexistent ligand-to-metal π-donation for heavier πbases. This conclusion has been upheld in computational analysis of these systems, which have also revealed highly ionic bonding.10
based, but the takeaway is that fundamental questions remain despite the apparent simplicity of these compounds. The reversibility of this cyclometalation has been a useful tool. We have observed nearly complete H/D exchange of the trimethylsilyl substituents of 1 under D2.15 This is consistent with Scott’s observations with −SitBuMe2 derivatives and harkens to early H/D exchange reported by Andersen on actinide compounds supported by bis(trimethylsilyl)amido ligands.6,18 Leveraging H/D exchange has allowed us to establish ring opening events at 1, even if a particular (N3N)ZrX compound is unobservable. For example, reaction of PhSiH3 with 1 fails to provide an observable silyl compound prior to decomposition, but treatment of 1 with PhSiD3 results in formation of 1-dn.13 This reaction indicates that the metallacycle of 1 activates an Si−D bond but is followed by rapid ring closure and restoration of phenylsilane (Scheme 5). The relative stability of C−D versus Si−D bonds doubtlessly drives this process, and the rapid ring-opening/ring-closing results in sequential deuterium transfer. Because there is very little dehydrocoupling of phenylsilane by 1, the H/D exchange at 1 with phenylsilane-d3 must be primarily from of Si−D activation,13 making deuteration a powerful mechanistic probe, particularly of unobservable intermediates. The rapid ring-opening/ring-closing reactions at 1 can present some challenges as well. Efforts to measure a kinetic isotope effect in the dehydrocoupling of phosphines were stymied by rapid H/D exchange between phenylphosphine and phenylphosphine-d2, reactivity that was facile at ambient temperature when productive catalysis required elevated temperatures.12 That challenge and solution are explored in detail below. Other observations suggested that ring opening of 1 without formation of a stable (N3N)ZrX can be productive. For example, dehydrocoupling reactions with MesPH2 (Mes = 2,4,6trimethylphenyl) provide a scant but reliable conversion to (MesPH)2 with hydrogen loss, but (N3N)ZrPHMes was not observed.12 Prior work and our observations confirm that 1 is an excellent precursor to a broad family of (N3N)NZrX derivatives with X ligands derived from carbon or group 15 and 16 elements (Scheme 2).10 Their identification and characterization was straightforward, but most are staunchly air sensitive. For example, samples of (N3N)ZrMe for X-ray analysis would decompose under dry Paratone-N oil with surprisingly vigorous methane evolution. Beside the apparent instability of heavier group 14 elements, alkoxide ligands can be problematic. Even with careful addition, many O−H donors will completely remove the triamidoamine ligand.10 This observation would appear to be a synthetic challenge because isolated (N3N)ZrOR compounds are themselves stable, even amenable to sublimation.19 These observations indicate that the ability of E−H substrates to open the metallacycle of 1 is dependent more on bond polarity rather than acidity, which can be a complication in the system. We have reported the molecular structures of many (N3N)Zr derivatives among those already known, and there are three commonalities. First, all three amido arms of the triamidoamine ligand are planar, indicative of π-bonding. For symmetry reasons, the three amido substituents are unable to simultaneously engage in ligand-to-metal π-bonding, which demonstrates that the structural features (planar nitrogen atoms and short Zr−N bond length) represent an average for the three arms of the triamidoamine ligand in a highly delocalized
■
CATALYSIS Our main application of these triamidoamine zirconium compounds has been catalysts for bond-formation reactions with main group substrates. Many transformations have been mediated or catalyzed by (N3N)Zr derivatives, but for the purpose of making general conclusions about this system and how it can impact other catalysts more broadly, the discussion will be limited to dehydrocoupling and hydrophosphination catalysis. Safety note: Many phosphines represent potential hazards for both acute and chronic exposure. Some phosphines, especially those in this Account, can have suf f iciently low oxidation potentials to be pyrophoric. Appropriate controls are critical for safe handling.
■
DEHYDROCOUPLING We initiated studies of the (N3N)ZrX family intending these for dehydrocoupling of phosphines. Triamidoamine-supported zirconium is a good catalyst for this reaction, comparing favorably with known catalysts in terms of activity.12,20 What was remarkable about the (N3N)Zr system was selectivity. Dehydrocoupling of phenylphosphine, for example, provided solely 1,2-diphenyldiphosphine, and extended reaction times or more forcing conditions gave rise to the expected phosphacycles (eq 1).12 At that time, the substrate scope was the deepest for any reported catalysts, though studies of substrate scope have not been commonplace in phosphine dehydrocoupling catalysis.20 5 mol 1
RPH 2 ⎯⎯⎯⎯⎯⎯→ RHPPHR −H 2
(1)
The successful phosphine dehydrocoupling catalysis provided opportunity for a more detailed mechanistic investigation. That study was initiated under the hypothesis that these zirconium phosphido compounds were reacting with substrate via σ-bond metathesis.21 Indeed, kinetic studies as well as activation parameters were consistent with σ-bond metathesis.12 It was, however, challenging to obtain kinetic isotope effect data. Phenylphopshine-d2 reacted rapidly at ambient temperature with 1 or (N3N)ZrPHPh to give mixtures of PhPD2, PhPHD, and PhPH2 with formation of 1-dn. This frustrating development suggested the importance of cyclometalation (N3N)Zr derivatives and gave a qualitative indication that relative rate of cyclometalation was quite high. A kinetic isotope effect was sought nevertheless, and an internal competition experiment was designed to circumvent competitive H/D exchange. In that reaction, 1 was treated with phenylphosphine-d1 and the distribution of 1,2-diphenyldiphosphine-dn derivatives was measured for the first equivalent of substrate consumed. Examining the reaction at one equivalent consumption was meant to circumvent extensive disproportionation of the phenylphosphine-d1 substrate to PhPH2 and PhPD2. For D
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
dehydrocoupling catalyst. In these reactions, phenylsilane was converted to Ph(NMe2)SiH2 by (N3N)ZrNMe2 and formed 1. As we had observed, 1 reacts with Ph(NMe2)SiH2 to afford (N3N)ZrNMe2. In the presence of excess phenylsilane, however, (PhSiH2)2 is formed with low molecular mass oligosilanes via apparent Si−H bond insertion. These observations and stoichiometric test reactions indicated α-silylene elimination as the critical step. Such α elimination reactions are rare, and catalytic examples had been reported for heavier main group elements such as tin, antimony, and arsenic with d0 metal catalysts.11,25−27 Observation of α-silylene elimination, our own prior work in α elimination, and the literature reports, led to the formation of three design properties for identifying potential α elimination catalysts. 1. E−H bond activation. Perhaps a trite requirement, this narrows the field of potential catalysts. 2. Relative M−E and M−H/E′ bond energies. Silyl compounds of (N3N)Zr were unstable with respect to silane loss and formation of 1 but a π-donor promoted elimination. The relative Zr−Si, Zr−H, and Zr−N bond energies were critical in the observed chemistry. 3. Low reactivity with unsaturated substrates. Reported group 4 α elimination catalysts react rapidly and irreversibly with unsaturated substrates28 limiting α elimination to dehydrocoupling and undercutting the possibility of group transfer catalysis. These design principals led us to Cp(CO)2FeMe, which demonstrated α-phosphinidene elimination and catalytic phosphinidene transfer to unsaturated substrates.29 There are still limitations to that catalysis as compared to the success for stoichiometric phosphinidene transfer reactions,30,31 but it and the related work by Layfield with iron and cobalt compounds32 demonstrate that understanding of these early metal catalyzed dehydrocoupling reactions can have substantial reach in catalysis.
statistical reasons, loss of phenylphosphine-d1 would give suppression of the measured KIE. Therefore, the observed value of ca. 3.5 represented a lower limit of sorts. The magnitude of the value, even with associated error, was consistent with other data and strongly implicated σ-bond metathesis in P−P bond formation. The hypothesis that phosphine dehydrocoupling is based on σ-bond metathesis was predicated on much broader work for group 4 and related d0 metals,21 yet other studies that implicated phosphorus in σ-bond metathesis were absent from the literature at that time. The relative acidity of P−H bonds in comparison to other E−H bonds that had been involved in σbond metathesis does, however, leave the possibility of proton transfer as an important step in early metal reactions with phosphines. Further evidence for σ-bond metathesis came from early heterodehydrocoupling (or “cross-dehydrocoupling,” for those who live in Iowa) experiments with phosphines and organosilanes or -germanes. Those reactions were highly selective for the P−E products with no formation of competitive P−P or E− E products.22 These observations and (N3N)ZrPRR′ as the sole intermediates buttressed the notion of dehydrocoupling catalysis based on σ-bond metathesis steps (Scheme 6). While Scheme 6. Proposed Mechanism for the Heterodehydrocoupling of Phosphines with Silanes or Germanes Catalyzed by 1
■
HYDROPHOSPHINATION As we initiated study of dehydrocoupling, we also probed the stoichiometric reactivity of the Zr−P bond. One reaction that was facile was 1,2-insertion reactions of polar substrates such as ketones and nitriles (Scheme 7).33 An extension to catalysis Scheme 7. Example Stoichiometric Insertion Reactions at Zirconium Phosphido Derivatives
we have seen couplings between other sets of elements that have been reported and some not,23 an intriguing pairing was phenyland cyclohexylphosphine, which afforded a nonstatistical excess of the unsymmetrical product, PhHP−PHCy.12 We became enchanted with the notion of selective phosphine heterodehydrocoupling with 1, but a more general exploration of that transformation failed to yield substantial selectivity in such a reaction.24 In developing our mechanistic proposal for P−Si/Ge heterodehydrocoupling, we hypothesized that (N3N)ZrPRR′ derivatives were more stable than (N3N)ZrSiR3, not knowing how true that was. All subsequent efforts to prepare silyl (i.e., (N3N)ZrSiHxR3−x) derivatives failed with compounds decomposing,13 which was surprising given the relative stability of other (N3N)Zr derivatives. The exception was reaction of 1 with Ph(NMe2)SiH2, which resulted in formation of (N3N)ZrNMe2. That reaction indicated to us that a silylene fragment may be lost, which led to the use of (N3N)ZrNMe2 as a silane
appeared logical. The diphenylphosphido derivative, (N3N)ZrPPh2, had been structurally characterized and appeared to be a natural candidate for catalysts with the ubiquitous hydrophosphination substrate, diphenylphosphine.34 An exploration of potential substrates was discouraging, to say the least. Potential substrates with heteroatoms underwent insertion but failed to liberate product. Michael acceptors, internal alkynes, E
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
catalysis and were only revisited by happenstance. It was hypothesized that more efficient hydrogen separation in dehydrocoupling (vide supra) would accelerate phosphine dehydrocoupling, and we considered trapping evolved H2 via hydrogenation as an option. That hypothesis is, of course, predicated on the dehydrocoupling catalyst being an effective hydrogenation catalyst with molecular hydrogen. We now know that (N3N)Zr derivatives are not effective hydrogenation catalysts with H2, rather they appear to be effective as transfer hydrogenation type catalysts with amine-boranes.23 That bit of knowledge might have stopped us, but as this is an Account, a reader may correctly suppose that our hydrogenation understanding came later, and sometimes knowing all the answers in advance is not ideal. We tested this hypothesis by adding alkenes to dehydrocoupling reactions, and our penchant for primary phosphines led to those being tested first. The first discovery we made was that no alkene was sacrificed in these reactions. More important, alkene was consumed to make secondary and tertiary phosphine products via hydrophosphination. We were able to develop straightforward conditions to select for each the secondary ortertiary phosphine product in high isolated yields (Scheme 6).37 It would seem reasonable that steric factors would sufficiently slow the second P−H activation step to give the tertiary product. However, should the phosphorus act as a nucleophile, a secondary phosphine may be a better substrate. Rather than needing judicious conditions, we were fortunate to have identified a system with the appropriate bias to select for secondary phosphine products, which are potential substrates in the preparation of privileged tertiary phosphine ligands among other value-added products. Beyond the excellent selectivity, it was most remarkable that unactivated alkene substrates, absent from the intermolecular hydrophosphination literature, had reactivity (Scheme 9).38 At the same time, we observed remarkably good reactivity with substrates that were utilized though still challenging (e.g., internal alkenes). This initial work was suggestive that electrophilic metal centers promote this higher activity in hydrophosphination catalysis, and a cadre of leaders and emergent leaders in the field have demonstrated that notion
styrene derivatives, and simple alkenes all failed to give productive catalysis. That recount begs the question, why hammer on a bad catalyst? The answer is simple: It did not start as a bad catalyst. As an article of convenience, phenylacetylene was one of the first, if not the first, substrates tested, and it worked (eq 2).
Unfortunately, conversions with terminal alkynes were nevertheless modest and did not compete with literature reports.35,36As a substrate scope was developed, two observations were made. First, the catalysis appeared to be limited to terminal alkynes, and second, alkynyl compounds, (N3N)ZrCCR, were routinely observed. That observation prompted reconsideration of the working hypothesis that catalysis involved substrate insertion. Very little definitive insight appeared likely from a protracted kinetic analysis, thus only a more precise measure of Keq between (N3N)ZrPPh2 and (N3N)ZrCCPh was sought. Measurement of the individual equilibrium constants of the left and right reactions in Scheme 8 was undertaken. No reaction between stoichiometric (N3N)ZrC CPh and diphenylphosphine was observed at ambient temperature. However, reaction of (N3N)ZrCCPh and diphneylphopshine-d1 was revealing, affording diphenylphosphine and deuterium incorporation at the trimethylsilyl substituents (Scheme 8, middle).33 This reaction was an excellent indication that cyclometalation is perhaps the single fastest reaction step for any transformation involving (N3N)Zr compounds, and it also demonstrated a mechanism for X ligand lability at (N3N)ZrX derivatives that we have consistently observed.13 Reaction of (N 3 N)ZrPPh 2 and phenylacetylene gave primarily (N3N)ZrCCPh, but some vinyl phosphine product was also observed, suggesting an insertion-based mechanism of catalysis (Scheme 8, bottom). More important than the actual reactions, the work afforded a critical understanding of the system relevant to a range of other reactions. Due to limited reactivity in that initial effort at catalysis, (N3N)Zr derivatives lay dormant in hydrophosphination
Scheme 8. Equilibrium between the Phosphido and Alkynyl Derivatives with Outcomes for Attempts to Measure Keq Starting from the Right (middle) and Left (bottom) Sides of the Equation
F
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
of isolated samples of 2 and related phosphido derivatives.45 The origin of the color and the transition responsible for the catalytic activity is an n → d Zr−P MLCT band based on TDDFT calculations and further spectroscopic characterization.45 As a result of this new understanding, we began to revisit catalytic hydrophosphination reactions with 1 under irradiation. As anticipated, reactivity was substantially enhanced under direct irradiation, which resulted in reduced reaction times. This MLCT band of 2 is centered at approximately 364 nm,46 which prompted exploration of wavelength. Irradiation with sources around this wavelength gave yet greater reactivity. Yet greater reactivity was seen under irradiation at 253.7 nm, the wavelength of a high intensity commercial photoreactor and near to a higher energy charge-transfer band of 2.45 Because Kasha’s Rule is obeyed (vide infra), we sampled a variety of wavelengths from different sources. Sources with the greatest photon density at wavelengths nearest to substantial transitions provided greatest activity, but it was also equally facile to use highly efficient (fluorescent or LED) sources with substantial improvement in catalytic activity, cutting reaction times by as much as two orders of magnitude.45 Armed with these data, we established substantially faster catalysis under photolysis than our prior conditions and returned the (N3N)Zr family back to the realm of fast hydrophosphination catalysts for styrene derivatives. More importantly, irradiation provided the basis for substantially increased reactivity and scope among unactivated alkene substrates as well as phosphines.45 This enhanced reactivity enabled new synthesis. For example, we were able to directly couple an unconjugated diene with a primary phosphine via two sequential hydrophosphination events to directly prepare a phosphorus heterocycle from commercial reagents (eq 4). Ring-
Scheme 9. Differentiation of Hydrophosphination for Secondary or Tertiary Products Using 1a
a
All reactions proceed with high if not quantitative conversions.
with a variety of electrophilic metal compounds as catalysts for similar substrates, many with more impressive turnover metrics.39,40 Unactivated alkenes, though, have remained our private intermolecular substrates for hydrophosphination catalysis, but we would welcome others to that sandbox. We expanded on this catalysis, looking at air-stable primary phosphines with an eye toward substrates with imaging and detection applications.41 Despite the advances in leveraging electronic effects to impart oxidative stability to primary phosphines, some of these substrates are relatively large. Despite their steric profile, the substrates were reasonably effective.42,43 While interesting, it is more relevant to this story to focus on work with alkyne substrates. Initial reactions with internal alkynes, to avoid the C−H bond activation of terminal alkynes that plagued early studies, gave the vinyl phosphines but at an unexpectedly sluggish pace. Under more forcing conditions, we were able to also observe the double hydrophosphination38 (two P−H equivalent additions) to these alkynes to afford the secondary diphosphinoethane products (eq 3).34 The continued
importance of chelating phosphine ligands cannot be overstated, and the “double hydrophosphination” that prepares these ligands has only been known since Nakazawa’s report in 2012.44 Nakazawa’s iron catalysts and the two other reported systems all exclusively functionalize terminal alkynes with secondary phosphines, making the zirconium catalyst a key compliment, leveraging internal alkynes and primary phosphines. While running control experiments, it became apparent that there was a light enhancement if not dependence in the catalytic hydrophosphination reactions with alkyne substrates. Dependence on wavelength was not fully explored at that time, but irradiation with a commercial visible-light source (i.e., LED bulb) gave improved conversions and allowed for decreased reaction temperatures. The net result was a system that showed good activity for the preparation of vinyl secondary phosphines from a range of terminal alkyne substrates as well as the formation of diphosphinoethane products under extended reaction times.34 The light dependence was intriguing, and we sought to better understand the behavior to develop more efficient catalysis. The UV−vis spectrum of the terminal phosphido compound, (N3N)ZrPHPh (2) revealed several transitions with very little absorbance in the visible, which is consistent with the faint color
closure via intermolecular hydrophosphination has been known for some time47 and was recently revisited by Webster.48 The initial intermolecular step is the greater challenge being an unactivated alkene, and this example demonstrates the substantial potential for new syntheses via highly active hydrophosphination catalysis.In the time since, these observations appear to be quite broad. Our early investigations indicate that the triamidoamine ligand is not required for zirconium photocatalysis and that photoactivation of π-basic ligands other than phosphido at d0 metals may be a general phenomenon. These are areas under current investigation.
■
FUTURE AND CONCLUSIONS “It is difficult to make predictions, especially about the future.”49 Thirteen years later, it is surprising that we are still learning from the same metal compound. Our continued interest stems from (N3N)Zr staunchly refusing to work by incremental improvement, instead demonstrating significant new and unique reactivity. The triamidoamine ligand is more than a spectator in this chemistry. Amido π-donation is the critical feature in discouraging π-donation from the X ligand. In this Account, the impact of primarily σ-donor phosphido ligands was explored in G
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
(7) Knight, P. D.; Scott, P. Predetermination of Chirality at Octahedral Centres with Tetradentate Ligands: Prospects for Enantioselective Catalysis. Coord. Chem. Rev. 2003, 242, 125−143. (8) Cox, P. A. The Elements: Their Origin, Abundance, and Distribution; Oxford University Press: Oxford, 1989. The relatively high abundance of zirconium does not necessarily make ZrCl4 less expensive per mole than some less abundant 3d metals. Processing and market ultimately dictate price more than abundance. Nevertheless, we sleep well at night knowing there is ZrO2 in the earth’s crust. (9) NIH National Library of Medicine. https://toxnet.nlm.nih.gov/ cgi-bin/sis/search/a?dbs+hsdb:@term+@DOCNO+7347 (accessed March 16 2019). (10) Roering, A. J.; Maddox, A. F.; Elrod, L. T.; Chan, S. M.; Ghebreab, M. B.; Donovan, K. L.; Davidson, J. J.; Hughes, R. P.; Shalumova, T.; MacMillan, S. N.; Tanski, J. M.; Waterman, R. General Preparation of (N3N)Zrx (N3N = N(CH2CH2NIiMe3)33-) Complexes from a Hydride Surrogate. Organometallics 2009, 28, 573−581. (11) Roering, A. J.; Davidson, J. J.; MacMillan, S. N.; Tanski, J. M.; Waterman, R. Mechanistic Variety in Zirconium-Catalyzed BondForming Reaction of Arsines. Dalton Trans 2008, 4488−4498. (12) Waterman, R. Selective Dehydrocoupling of Phosphines by Triamidoamine Zirconium Catalysts. Organometallics 2007, 26, 2492− 2494. (13) Erickson, K. A.; Cibuzar, M. P.; Mucha, N. T.; Waterman, R. Catalytic N-Si Coupling as a Vehicle for Silane Dehydrocoupling Via ASilylene Elimination. Dalton Trans 2018, 47, 2138−2142. (14) Cummins, C. C.; Schrock, R. R.; Davis, W. M. Synthesis of Vanadium and Titanium Complexes of the Type Rm[(Me3SiNCH2CH2)3n] (R = Cl, Alkyl) and the Structure of ClV[(Me3SiNCH2CH2)3n]. Organometallics 1992, 11, 1452−1454. (15) Leshinski, S. E.; Wheaton, C. A.; Sun, H.; Roering, A. J.; Tanski, J. M.; Fox, D. J.; Hayes, P. G.; Waterman, R. Triamidoamine-Supported Zirconium: Hydrogen Activation, Lewis Acidity, and Rac-Lactide Polymerization. RSC Adv. 2016, 6, 70581−70585. (16) Turculet, L.; Tilley, T. D. Synthesis and Reactivity of D0 Alkyl, Silyl, and Hydride Complexes of Titanium and Zirconium Featuring an Aryl-Substituted Tripodal Triamido Ligand Derived from cis, cis-1,3,5Triaminocyclohexane. Organometallics 2004, 23, 1542−1553. (17) MacMillan, S. N.; Tanski, J. M.; Waterman, R. {N,N-Bis[2(Trimethylsilylamino)Ethyl]-N’-(Trimethylsilyl)Ethane-1,2Diaminato(3-)-κ4n}Methylzirconium(IV). Acta Crystallogr., Sect. E: Struct. Rep. Online 2008, 64, m477. (18) Simpson, S. J.; Turner, H. W.; Andersen, R. A. Preparation and Hydrogen-Deuterium Exchange of Alkyl and Hydride Bis(Trimethylsilyl)Amido Derivatives of the Actinide Elements. Inorg. Chem. 1981, 20, 2991−2995. (19) Menger, F. M.; Sorrells, J. L. Chronology of a Difficult Synthesis. J. Chem. Educ. 2009, 86, 859. (20) Waterman, R. Dehydrogenative Bond-Forming Catalysis Involving Phosphines: Updated through 2010. Curr. Org. Chem. 2012, 16, 1313−1331. (21) Waterman, R. σ-Bond Metathesis: A 30-Year Retrospective. Organometallics 2013, 32, 7249−7263. (22) Roering, A. J.; MacMillan, S. N.; Tanski, J. M.; Waterman, R. Zirconium-Catalyzed Heterodehydrocoupling of Primary Phosphines with Silanes and Germanes. Inorg. Chem. 2007, 46, 6855−6857. (23) Erickson, K. A.; Stelmach, J. P. W.; Mucha, N. T.; Waterman, R. Zirconium-Catalyzed Amine Borane Dehydrocoupling and Transfer Hydrogenation. Organometallics 2015, 34, 4693−4699. (24) Ghebreab, M. B.; Costanza, S.; Waterman, R. Selectivity Effects in Zirconium-Catalyzed Heterodehydrocoupling Reactions of Phosphines. Phosphorus, Sulfur Silicon Relat. Elem. 2016, 191, 668−670. (25) Neale, N. R.; Tilley, T. D. A New Mechanism for MetalCatalyzed Stannane Dehydrocoupling Based on α-H-Elimination in a Hafnium Hydrostannyl Complex. J. Am. Chem. Soc. 2002, 124, 3802− 3803. (26) Waterman, R. Mechanisms of Metal-Catalyzed Dehydrocoupling Reactions. Chem. Soc. Rev. 2013, 42, 5629.
detail, but the effects would easily extend throughout the p block. Additionally, cyclometalation, the kiss of death for many another ligand’s reactivity (with exceptions50), is a boon in this system where azasilylametalacyclobutane 1 is “home base” for so much reactivity at (N3N)Zr compounds. However, this is not a narrow story of a special ancillary ligand because the results are not unique to triamidoamine. Our exploration of dehydrocoupling has wide bearing on catalytic synthesis involving low-valent main group fragments, an area long acknowledged for greatly improved synthetic efficiency. This is evidenced by the development of design principals in the search for new α elimination catalysts. Likewise, hydrophosphination catalysis with (N3N)Zr derivatives has influenced other systems and, most recently, yielded what may be general photocatalysis for d0 metals with π-basic ligands. While the catalysis with (N3N)Zr is broader than described here, there is impact of this system beyond those specific examples.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Rory Waterman: 0000-0001-8761-8759 Notes
The author declares no competing financial interest. Biography Rory Waterman loves making molecules and was fortunate to learn how with some of the greats, like the late Gregory Hillhouse and T. Don Tilley. He rose through the ranks at the University of Vermont since his initial appointment there in 2006 and spends his time chasing down problems in inorganic and organometallic chemistry, catalysis, materials science, K−20 professional development, and curricular development.
■
ACKNOWLEDGMENTS This work has been primarily funded through long-standing, generous support from the US National Science Foundation and most recently CHE-1565658, with contributions to various aspects from the Petroleum Research Fund and Research Corporation for Science Advancement as well. The hard work and dedication of students (high school, undergraduate, and graduate) and postdocs has made this a rich and complete story. Personally, I have been blessed to collaborate with these individuals but more so to witness their tremendous growth.
■
REFERENCES
(1) Duan, Z.; Naiini, A. A.; Lee, J.-H.; Verkade, J. G. Novel Volatile Azatranes of Group 4 Metals. Inorg. Chem. 1995, 34, 5477−5482. (2) Schrock, R. R. Transition Metal Complexes That Contain a Triamidoamine Ligand. Acc. Chem. Res. 1997, 30, 9−16. (3) Verkade, J. G., Iowa State University, Personal Communication, 2008. (4) Johnson, B. P.; Balázs, G.; Scheer, M. Low-Coordinate E1 Ligand Complexes of Group 15 Elementsa Developing Area. Coord. Chem. Rev. 2006, 250, 1178−1195. (5) Morton, C.; Gillespie, K. M.; Sanders, C. J.; Scott, P. Complexes of Zirconium with Aryl Substituted Triamidoamines: Molecular Structures of Amide and Alkyl Derivatives. J. Organomet. Chem. 2000, 606, 141−146. (6) Morton, C.; Munslow, I. J.; Sanders, C. J.; Alcock, N. W.; Scott, P. Triamidoamine Chemistry of Zirconium. Organometallics 1999, 18, 4608−4613. H
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research (27) Waterman, R.; Tilley, T. D. Catalytic Antimony-Antimony Bond Formation through Stibinidene Elimination from Zirconocene and Hafnocene Complexes. Angew. Chem., Int. Ed. 2006, 45, 2926−2929. (28) Leshinski, S.; Shalumova, T.; Tanski, J. M.; Waterman, R. Insertion Reactions Involving a Triamidoamine-Supported Zirconium Complex. Dalton Trans 2010, 39, 9073−9078. (29) Pagano, J. K.; Ackley, B. J.; Waterman, R. Evidence for IronCatalyzed A-Phosphinidene Elimination with Phenylphosphine. Chem. - Eur. J. 2018, 24, 2554−2557. (30) Mathey, F. Developing the Chemistry of Monovalent Phosphorus. Dalton Trans 2007, 1861−1868. (31) Waterman, R. Metal-Phosphido and -Phosphinidene Complexes in P-E Bond-Forming Reactions. Dalton Trans 2009, 18−26. (32) Pal, K.; Hemming, O. B.; Day, B. M.; Pugh, T.; Evans, D. J.; Layfield, R. A. Iron- and Cobalt-Catalyzed Synthesis of Carbene Phosphinidenes. Angew. Chem., Int. Ed. 2016, 55, 1690−1693. (33) Roering, A. J.; Leshinski, S. E.; Chan, S. M.; Shalumova, T.; MacMillan, S. N.; Tanski, J. M.; Waterman, R. Insertion Reactions and Catalytic Hydrophosphination by Triamidoamine-Supported Zirconium Complexes. Organometallics 2010, 29, 2557−2565. (34) Bange, C. A.; Waterman, R. Zirconium-Catalyzed Intermolecular Double Hydrophosphination of Alkynes with a Primary Phosphine. ACS Catal. 2016, 6, 6413−6416. (35) Crimmin, M. R.; Barrett, A. G. M.; Hill, M. S.; Hitchcock, P. B.; Procopiou, P. A. Calcium-Catalyzed Intermolecular Hydrophosphination. Organometallics 2007, 26, 2953−2956. (36) Crimmin, M. R.; Barrett, A. G. M.; Hill, M. S.; Hitchcock, P. B.; Procopiou, P. A. Heavier Group 2 Element Catalyzed Hydrophosphination of Carbodiimides. Organometallics 2008, 27, 497−499. (37) Ghebreab, M. B.; Bange, C. A.; Waterman, R. Intermolecular Zirconium-Catalyzed Hydrophosphination of Alkenes and Dienes with Primary Phosphines. J. Am. Chem. Soc. 2014, 136, 9240−9243. (38) Bange, C. A.; Waterman, R. Challenges in Catalytic Hydrophosphination. Chem. - Eur. J. 2016, 22, 12598−12605. (39) Selikhov, A. N.; Plankin, G. S.; Cherkasov, A. V.; Shavyrin, A. S.; Louyriac, E.; Maron, L.; Trifonov, A. A. Thermally Stable Ln(Ii) and Ca(Ii) Bis(Benzhydryl) Complexes: Excellent Precatalysts for Intermolecular Hydrophosphination of C−C Multiple Bonds. Inorg. Chem. 2019, 58, 5325−5334. (40) Basalov, I. V.; Liu, B.; Roisnel, T.; Cherkasov, A. V.; Fukin, G. K.; Carpentier, J.-F.; Sarazin, Y.; Trifonov, A. A. Amino Ether−Phenolato Precatalysts of Divalent Rare Earths and Alkaline Earths for the Single and Double Hydrophosphination of Activated Alkenes. Organometallics 2016, 35, 3261−3271. (41) Fleming, J. T.; Higham, L. J. Primary Phosphine Chemistry. Coord. Chem. Rev. 2015, 297−298, 127−145. (42) Bange, C. A.; Ghebreab, M. B.; Ficks, A.; Mucha, N. T.; Higham, L.; Waterman, R. Zirconium-Catalyzed Intermolecular Hydrophosphination Using a Chiral, Air-Stable Primary Phosphine. Dalton Trans 2016, 45, 1863−1867. (43) Bange, C. A.; Mucha, N. T.; Cousins, M. E.; Gehsmann, A. C.; Singer, A.; Truax, T.; Higham, L. J.; Waterman, R. Zirconium-Catalyzed Alkene Hydrophosphination and Dehydrocoupling with an Air-Stable, Fluorescent Primary Phosphine. Inorganics 2016, 4, 26. (44) Kamitani, M.; Itazaki, M.; Tamiya, C.; Nakazawa, H. Regioselective Double Hydrophosphination of Terminal Arylacetylenes Catalyzed by an Iron Complex. J. Am. Chem. Soc. 2012, 134, 11932−11935. (45) Bange, C. A.; Conger, M. A.; Novas, B. T.; Young, E. R.; Liptak, M. D.; Waterman, R. Light-Driven, Zirconium-Catalyzed Hydrophosphination with Primary Phosphines. ACS Catal. 2018, 8, 6230− 6238. (46) This is conveniently near the λmax of commercial compact fluorescent “blacklight” bulbs. The first such bulb we purchased was called a Party Light, and reactions run under blacklight irradiation became known as “party conditions.” (47) Douglass, M. R.; Marks, T. J. Organolanthanide-Catalyzed Intramolecular Hydrophosphination/Cyclization of Phosphinoalkenes and Phosphinoalkynes. J. Am. Chem. Soc. 2000, 122, 1824−1825.
(48) Espinal-Viguri, M.; King, A. K.; Lowe, J. P.; Mahon, M. F.; Webster, R. L. Hydrophosphination of Unactivated Alkenes and Alkynes Using Iron(Ii): Catalysis and Mechanistic Insight. ACS Catal. 2016, 6, 7892−7897. (49) Attributed to Danish origin and used by Niels Bohr among others: Shapiro, F. R. The Yale Book of Quotations; Yale University Press: New Haven, CT, 2006; p 92. (50) Wicker, B. F.; Scott, J.; Fout, A. R.; Pink, M.; Mindiola, D. J. Atom-Economical Route to Substituted Pyridines Via a Scandium Imide. Organometallics 2011, 30, 2453−2456.
I
DOI: 10.1021/acs.accounts.9b00284 Acc. Chem. Res. XXXX, XXX, XXX−XXX
