High hydride count rhodium octahedra, [Rh6(PR3) 6H12][BArF4]2: Synthesis, structures, and reversible hydrogen uptake under mild conditions

Published on Web 04/18/2006

High Hydride Count Rhodium Octahedra,

[Rh₆(PR₃)₆H₁₂][BAr^F] : Synthesis, Structures, and Reversible

4 2

Hydrogen Uptake under Mild Conditions

Simon K. Brayshaw,^†Michael J. Ingleson,^†Jennifer C. Green,^‡J. Scott McIndoe,^§

Paul R. Raithby,^†Gabriele Kociok-Ko¨hn,^†and Andrew S. Weller*^,†

Contribution from the Departments of Chemistry, UniVersity of Bath, Bath BA2 7AY, U.K.,

UniVersity of Oxford, Oxford OX1 3QR, U.K., and UniVersity of Victoria,

Victoria, BC V8W3V6, Canada

Received January 28, 2006; E-mail: a.s.weller@bath.ac.uk

i

Abstract: A new class of transition metal cluster is described, [Rh₆(PR₃)₆H₁₂][BAr^F

]

4 2

(R ) Pr (1a), Cy

(2a); BAr^F₄) [B{C₆H₃(CF₃)₂}₄]^-). These clusters are unique in that they have structures exactly like those

of early transition metal clusters with edge-bridging π-donor ligands rather than the structures expected for

late transition metal clusters with π-acceptor ligands. The solid-state structures of 1a and 2a have been

determined, and the 12 hydride ligands bridge each Rh-Rh edge of a regular octahedron. Pulsed gradient

spin-echo NMR experiments show that the clusters remain intact in solution, having calculated

hydrodynamic radii of 9.5(3) Å for 1a and 10.7(2) Å for 2a, and the formulation of 1a and 2a was

unambiguously confirmed by ESI mass spectrometry. Both 1a and 2a take up two molecules of H₂to

afford the cluster species [Rh₆(PⁱPr₃)₆H₁₆][BAr^F₄]₂(1b) and [Rh₆(PCy₃)₆H₁₆][BAr^F₄]₂(2b), respectively, as

characterized by NMR spectroscopy, ESI-MS, and, for 2b, X-ray crystallography using the [1-H-CB₁₁Me₁₁]^-

1

salt. The hydride ligands were not located by X-ray crystallography, but H NMR spectroscopy showed a

15:1 ratio of hydride ligands, suggesting an interstitial hydride ligand. Addition of H₂is reversible: placing

1b and 2b under vacuum regenerates 1a and 2a. DFT calculations on [Rh₆(PH₃)₆H_x]²⁺(x ) 12, 16) support

the structural assignments and also show a molecular orbital structure that has 20 orbitals involved with

cluster bonding. Cluster formation has been monitored by ³¹P{¹H} and ¹H NMR spectroscopy, and

mechanisms involving heterolytic H₂cleavage and elimination of [HPⁱPr₃]⁺or the formation of trimetallic

intermediates are discussed.

Scheme 1. Representative Examples of Octahedral Transition

Metal Clusters with π-Donor and π-Acceptor Ligands

Introduction

The chemistry of transition metal clusters occupies two

“classical” areas.¹The first encompasses clusters with metal

atoms in low to medium oxidation states combined with

π-donating ligands,²exemplified by halide edge-bridged clusters

such as [Nb₆Cl₁₈]K₄, Nb₆Cl₁₄(PⁿPr₃)₄(A, Scheme 1), or Zr₆-

3

4

Cl₁₄H₄(PEt₃)₄and by the face-bridged chalcogenides [W₆S₈-

(PEt₃)₆] (B)⁵or [Re₆Se₈I₆]^4-,⁶which are molecular models for

Chevrel phases.⁷The second grouping is that of the later

transition metals in low oxidation states with π-acceptor ligands

such as Rh₆(CO)₁₆(C) or [Ru₆(µ-C)(CO)₁₆]^2-.⁸Aside from the

fascinating structural and bonding aspects of these compounds,

the driving force for the study of such cluster systems is their

potential to bridge the areas of heterogeneous and homogeneous

catalysis^1,9,10and their unique spatial, electronic, and magnetic

properties.¹¹

The electronic descriptions for both classes of cluster are well

understood, especially for octahedral species where symmetry

allows for a concise account of the molecular orbital struc-

tures.^12,13The octahedral late transition metals with π-accepting

ligands (C) generally have cluster valence electron counts (cve)

of 86 electrons,¹⁴although exceptions do exist.¹⁵For clusters

with π-donor edge-bridging ligands (A), electron counts of 74-

78 cve have been noted, with 76 cve being usual, while the

^†University of Bath.

^‡University of Oxford.

^§University of Victoria.

(1) Metal clusters in chemistry; Braunstein, P., Oro, L. A., Raithby, P. R.,

Eds.; Wiley-VCH: Weinheim, 1999.

(2) Early transition metal clusters with π-donor ligands; Chisholm, M. H.,

Ed.; VCH: New York, 1995. Gray, T. G. Coord. Chem. ReV. 2003, 243,

213-235.

(3) Klendworth, D. D.; Walton, R. A. Inorg. Chem. 1981, 20, 1151-1155.

(4) Chen, L.; Cotton, F. A.; Wojtczak, W. A. Inorg. Chem. 1996, 35, 2988-

2994.

(5) Saito, T. AdV. Inorg. Chem. 1997, 44, 45-91.

(6) Long, J. R.; McCarty, L. S.; Holm, R. H. J. Am. Chem. Soc. 1996, 118,

4603-4616.

(7) Chevrel, R.; Sergent, M.; Prigent, J. J. Solid State Chem. 1971, 3, 515-

519.

9

J. AM. CHEM. SOC. 2006, 128, 6247-6263

6247

A R T I C L E S

Brayshaw et al.

Scheme 2. Examples of Cluster Buildup under Kinetic Conditions

faced-bridged chalcogenides (B) span a larger range of cluster

electron counts, from 80 cve in Mo₆(µ-S)₈(PEt₃)₆to 98 cve in

Co₆(µ-S)₈(PEt₃)₆.¹⁶

W₆H₅(CⁱPr)(OⁱPr)₁₂,^22,23and reaction of H₂with [(^tBuO)Cu]₄

in the presence of P(NMe₂)₃gives the octahedral copper

polyhydride Cu₆(P(NMe₂)₃)₆H₆(Scheme 2). An alternative

24

The synthesis of transition metal cluster species from mono-

or dinuclear starting materials often favors the isolation of what

might be considered the thermodynamic product, by using

moderate to high temperature and pressure, or sometimes

both.^2,12,17This invariably leads to products that are electroni-

cally and coordinatively saturated, although notable recent

examples exist of electronically unsaturated clusters formed from

thermolysis reactions when bulky phosphines are used, such as

Pt₃Re₂(CO)₆(P^tBu₃)₃.¹⁸Rarer are routes that offer up some

degree of kinetic control to cluster formation from mono- or

dinuclear precursors, and, relevant to this paper, to our

knowledge many of these kinetic routes often result in cluster

species with attached hydride ligands, although non-hydride-

containing clusters can also be synthesized.¹⁹Hydrogenolysis

of precursor complexes and elimination of non-coordinating

alkane or alcohol is a common synthetic method. For example,

reaction of Pt(PR₃)(η²-H₂CdCH₂)₂precursors with H₂gives

platinum hydrido-phosphine clusters such as Pt₄(PⁱPr₂Ph)₄H₈,^20,21

while addition of H₂to (ⁱPrO)₂(^tBu)WtW(ⁱPrO)₂(^tBu) affords

method is to reduce a mononuclear transition metal halide with

a hydride source such as HSnBu₃or H₃SiPh to afford com-

4,25

pounds such as Zr₆(PR₃)₄Cl₁₄H₄

and [(η⁵-C₅Me₄SiMe₃)-

LuH₂]₆,^26,27respectively. Kinetic routes to related metallaborane

cluster species through reduction of mononuclear cyclopenta-

dienyl halides have also been developed,²⁸and hydrido clusters

can also be synthesized under kinetic conditions from condensa-

tion of bimetallic hydride complexes with monometallic

fragments.^29-31

We report in this paper the synthesis under mild conditions

of a new class of octahedral cluster, [Rh₆(PR₃)₆H₁₂]²⁺. Such

clusters bridge the gap between early and later transition metals

inasmuch as they are composed of late transition metal

fragments {Rh(PR₃)} but have molecular geometries and

(19) Adams, R. D.; Captain, B.; Fu, W.; Smith, M. D. J. Am. Chem. Soc. 2002,

124, 5628-5629. Adams, R. D.; Captain, B.; Fu, W.; Hall, M. B.; Manson,

J.; Smith, M. D.; Webster, C. E. J. Am. Chem. Soc. 2004, 126, 5253-

5267. Stone, F. G. A. Pure Appl. Chem. 1986, 58, 529-536. Stone, F. G.

A. Angew. Chem., Int. Ed. Engl. 1984, 23, 89-99.

(20) Gregson, D.; Howard, J. A. K.; Murray, M.; Spencer, J. L. J. Chem. Soc.,

Chem. Commun. 1981, 716-717.

(8) Dyson, P. J.; McIndoe, J. S. Transition metal carbonyl cluster chemistry;

Gordon and Breach Science Publishers: Amsterdam, 2000.

(9) Dumestre, F.; Chaudret, B.; Amiens, C.; Renaud, P.; Fejes, P. Science 2004,

303, 821-823.

(21) Frost, P. W.; Howard, J. A. K.; Spencer, J. L.; Turner, D. G.; Gregson, D.

J. Chem. Soc., Chem. Commun. 1981, 1104-1106.

(22) Chisholm, M. H.; Folting, K.; Kramer, K. S.; Streib, W. E. J. Am. Chem.

Soc. 1997, 119, 5528-5539.

(10) Thomas, J. M.; Johnson, B. F. G.; Raja, R.; Sankar, G.; Midgley, P. A.

Acc. Chem. Res. 2003, 36, 20-30.

(23) Chisholm, M. H.; Kramer, K. S. Chem. Commun. 1996, 1331-1332.

(24) Lemmen, T. H.; Folting, K.; Huffman, J. C.; Caulton, K. G. J. Am. Chem.

Soc. 1985, 107, 7774-7775.

(11) Beltran, L. M. C.; Long, J. R. Acc. Chem. Res. 2005, 38, 325-334. Gabriel,

J. C. P.; Boubekeur, K.; Uriel, S.; Batail, P. Chem. ReV. 2001, 101, 2037-

2066.

(25) Chen, L. F.; Cotton, F. A. Inorg. Chim. Acta 1997, 257, 105-120.

(26) Tardif, O.; Nishiura, M.; Hou, Z. M. Organometallics 2003, 22, 1171-

1173.

(12) Mingos, D. M. P.; Wales, D. J. Introduction to cluster chemistry; Prentice

Hall: London, 1990.

(13) Chisholm, M. H.; Clark, D. L.; Hampdensmith, M. J.; Hoffman, D. H.

Angew. Chem., Int. Ed. Engl. 1989, 28, 432-444.

(14) Mingos, D. M. P. J. Chem. Soc., Dalton Trans. 1974, 133-138.

(15) Paquette, M. S.; Dahl, L. F. J. Am. Chem. Soc. 1980, 102, 6621-6623.

(16) Lin, Z. Y.; Williams, I. D. Polyhedron 1996, 15, 3277-3287. Lin, Z. Y.;

Fan, M. F. In Structural and Electronic Paradigms in Cluster Chemistry;

Springer-Verlag: Berlin, 1997; Vol. 87.

(27) Cui, D. M.; Tardif, O.; Hou, Z. M. J. Am. Chem. Soc. 2004, 126, 1312-

1313. Luo, Y.; Baldamus, J.; Tardif, O.; Hou, Z. M. Organometallics 2005,

24, 4362-4366.

(28) Fehlner, T. P. Organometallics 2000, 19, 2643-2651.

(29) Rhodes, L. F.; Huffman, J. C.; Caulton, K. G. J. Am. Chem. Soc. 1983,

105, 5137-5138.

(30) Suzuki, H.; Kakigano, T.; Tada, K.; Igarashi, M.; Matsubara, K.; Inagaki,

A.; Oshima, M.; Takao, T. Bull. Chem. Soc. Jpn. 2005, 78, 67-87.

(31) Vieille-Petit, L.; Tschan, M. J. L.; Suss-Fink, G.; Laurenczy, G.; Hagen,

C. M.; Finke, R. G.; Geneste, F.; Moinet, C. Polyhedron 2005, 24, 1961-

1967.

(17) The chemistry of metal cluster complexes; Shriver, D. F., Kaesz, H. D.,

Adams, R. D., Eds.; VCH: New York, 1990.

(18) Adams, R. D.; Captain, B. Angew. Chem., Int. Ed. 2005, 44, 2531-2533.

9

6248 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

Scheme 3. Synthesis of the 12-Hydride Clusters 1a and 2a

electronic structures that are more closely related to those of

the early octahedral transition metal clusters. Not only do these

clusters have unprecedented structures for a late transition metal

cluster species but they also have an extraordinarily large

number of hydride ligands associated with them and thus are

potential models for nanoparticle Rh clusters used in arene

hydrogenations,^10,32hydrogen storage materials,^33-35and hy-

drogen spillover systems.³⁶We demonstrate their relevance to

these areas by describing the facile and reversible dihydrogen

uptake by these clusters under very mild conditions of room

temperature and pressure. The structures of these new cluster

species have been probed by a combination of X-ray crystal-

lography, NMR spectroscopy, electrospray ionization mass

spectrometry, and DFT calculations, and we also report on

studies to elucidate the mechanism of cluster formation from

mononuclear rhodium precursor complexes. Aspects of this

work have been communicated previously.^37,38

anion [1-closo-HCB₁₁Me₁₁]^{- 40}afforded clusters in effectively

the same yield as for [BAr^F]^-, but [PF₆]^-or [BF₄]^-resulted

4

in no cluster formation, principally because these latter two

anions are not compatible with the formation of the intermediate

dihydrogen complexes (Scheme 3). Compound 1a is sparingly

soluble in fluorobenzene (marginally better in CH₂Cl₂), while

2a shows better solubility in either solvent. They are both

insoluble in pentane and toluene.

The solid-state structures of the triisopropylphosphine cluster

1a and the tricyclohexylphosphine cluster 2a are shown in

Figure 1. Table 1 gives data collection and refinement details,

and Table 2 presents a comparison of the key structural metrics

of these clusters. Full listings of bond lengths and angles are

given in the Supporting Information. 1a has a crystallographi-

cally imposed centrosymmetric structure in the solid state, while

2a does not, but both are effectively regular Rh₆octahedra. Each

rhodium vertex is coordinated with a single phosphine, which

is directed approximately radial to the cluster core, with the

cross-cluster Rh‚‚‚RhsP angles, e.g. Rh(2′)‚‚‚Rh(2)sP(2) in

1a, being only a few degrees away from 180° (average 176.6-

(1) and 177.8(1)° for 1a and 2a, respectively). The Rh-Rh

distances are only slightly larger for 2a than measured for 1a

(averages for 1a, 2.720 Å; for 2a, 2.735 Å), and there is only

a slight increase in cross-cluster P‚‚‚P distances, showing that

the bulkier tricyclohexylphosphine ligands (cone angle for PCy₃

is 169° vs 159° for PⁱPr₃⁴¹) have only a small effect on the

cluster geometry. The deviation of Rh-Rh bonding distances,

δ(Rh-Rh), is essentially the same for both and is small (∼0.03

Å), as is deviation in cross-cluster Rh-Rh distances. DiSalvo

Results and Discussion

Synthesis and Solid-State Structure of [Rh₆(PR₃)₆H₁₂]-

[BAr^F] (R ) ⁱPr, Cy). We have recently reported that addition

4 2

of H₂(∼4 atm initially, 298 K) to CH₂Cl₂or fluorobenzene

i

solutions of [Rh(PR₃)₂(nbd)][BAr^F] (R ) Pr, Cy; BAr^F

)

4

[B{C₆H₃(CF₃)₂}₄]^-, nbd ) norbornadiene) initially results in

the formation of the almost colorless dihydrogen/dihydride

complexes [Rh(PR₃)₂(H)₂(η²-H₂)_x][BAr^F] (x ) 1 or 2).³⁹Gentle

4

heating (313-323 K) of these solutions results in a gradual

(32) Dyson, P. J. Dalton Trans. 2003, 2964-2974. Hagen, C. M.; Vieille-Petit,

L.; Laurenczy, G.; Suss-Fink, G.; Finke, R. G. Organometallics 2005, 24,

1819-1831. Hagen, C. M.; Widegren, J. A.; Maitlis, P. M.; Finke, R. G.

J. Am. Chem. Soc. 2005, 127, 4423-4432.

darkening to very dark red over 4 days and the formation of

the dicationic octahedral cluster species [Rh₆(PR₃)₆H₁₂][BAr^F] ,

4 2

isolated in moderate yield as analytically pure salts on recrys-

tallization (1a, R ) ⁱPr, 31%; 2a, R ) Cy, 40% yield, based on

rhodium). Recrystallization from solutions containing an excess

of 1-hexene ensures the isolation of the 12-hydride species and

not clusters with a higher hydride count (16 hydrides), which

can form readily on exposure of 1a or 2a to H₂present in the

reaction vessel (vide infra). The mechanism of cluster formation

and hydrogen uptake experiments will be discussed in due

course, and we first present the complete characterization of

clusters 1a and 2a by X-ray crystallography, NMR spectroscopy,

and mass spectrometry. Using CH₂Cl₂as the reaction solvent

rather than fluorobenzene resulted in lower yields of cluster,

(33) Schlapbach, L.; Zuttel, A. Nature 2001, 414, 353-358.

(34) Rowsell, J. L. C.; Eckert, J.; Yaghi, O. M. J. Am. Chem. Soc. 2005, 127,

14904-14910. Rowsell, J. L. C.; Spencer, E. C.; Eckert, J.; Howard, J. A.

K.; Yaghi, O. M. Science 2005, 309, 1350-1354. Rowsell, J. L. C.; Yaghi,

O. M. Angew. Chem., Int. Ed. 2005, 44, 4670-4679.

(35) Dinca, M.; Long, J. R. J. Am. Chem. Soc. 2005, 127, 9376-9377. Kaye,

S. S.; Long, J. R. J. Am. Chem. Soc. 2005, 127, 6506-6507.

(36) Gao, H. R.; Angelici, R. J. J. Am. Chem. Soc. 1997, 119, 6937-6938.

Conner, W. C.; Falconer, J. L. Chem. ReV. 1995, 95, 759-788. Lachawiec,

A. J.; Qi, G. S.; Yang, R. T. Langmuir 2005, 21, 11418-11424.

(37) Brayshaw, S. K.; Ingleson, M. J.; Green, J. C.; Raithby, P. R.; Kociok-

Kohn, G.; McIndoe, J. S.; Weller, A. S. Angew. Chem., Int. Ed. 2005, 44,

6875-6878.

(38) Ingleson, M. J.; Mahon, M. F.; Raithby, P. R.; Weller, A. S. J. Am. Chem.

Soc. 2004, 126, 4784-4785.

(39) Ingleson, M. J.; Brayshaw, S. K.; Mahon, M. F.; Ruggiero, G. D.; Weller,

A. S. Inorg. Chem. 2005, 44, 3162-3171.

(40) King, B. T.; Janousek, Z.; Gruner, B.; Trammell, M.; Noll, B. C.; Michl,

J. J. Am. Chem. Soc. 1996, 118, 3313-3314. Clarke, A. J.; Ingleson, M.

J.; Kociok-Kohn, G.; Mahon, M. F.; Patmore, N. J.; Rourke, J. P.; Ruggiero,

G. D.; Weller, A. S. J. Am. Chem. Soc. 2004, 126, 1503-1517.

(41) Tolman, C. A. Chem. ReV. 1977, 77, 313-348.

and thus the latter is preferred. Anions other than [BAr^F]^-gave

4

mixed results. Use of the “greasy ball” weakly coordinating

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6249

A R T I C L E S

Brayshaw et al.

location of hydride ligands in transition metal clusters by X-ray

crystallography is often problematic, given the low scattering

associated with the M-H bond, but, as will be reported later,

NMR, mass spectral, and microanalysis data all unequivocally

support the formulation as 12-hydride clusters for both 1a and

2a. Comparison of the Rh-H distances in 1a suggests that,

within error, six of these hydrides are more semi-bridging than

bridging in nature, although the difference between the two

modes is rather subtle, especially as measured by X-ray

crystallography. DFT calculations (vide infra) also suggest a

mixture of bridging and semi-bridging hydrides, with four

octahedral edges bridged symmetrically and the remaining eight

bridged unsymmetrically, supporting such a structural assign-

ment. Of course, a neutron diffraction study would help pin

down the positioning of the hydrides, as it has for cluster hydride

ligands previously,^44-47and attempts are currently underway

to produce crystals of a size suitable for such an experiment.

Transition metal clusters with hydride ligands are ubiqui-

tous.^1,8,12,17However, those with more than four hydride ligands,

or a hydride:metal ratio of >1, are significantly less well

documented. Well-characterized examples of clusters with a high

(>4) hydride count include Pt₅H₈(P^tBu₂Ph)₅,²⁰[Pt₄H₇(P^tBu₃)₄]-

[BF₄],⁴⁸Pt₂Os₇(CO)₂₃H₈,⁴⁹[Zr₆Cl₁₈H₅][PPh₄]₃,⁴⁷Cu₆(PR₆)₆H₆,²⁴

Pd₂₈(PtPMe₃)(PtPPh₃)₁₂(CO)₂₇H₁₂,⁵⁰[{Ru(η⁶-C₆H₆)}₄H₆]Cl₂,⁵¹

Os₅Pd(CO)₁₆H₆,⁵²(Cp*Ru)₃H₅,³⁰Os₇Rh₃(CO)₂₃H₁₁,⁵³OsPt₃-

Re₂(CO)₆(P^tBu₃)₃H₆,¹⁸[(η⁵-C₅Me₄SiMe₃)LuH₂]₆,²⁶[Re₄Cu₂H₁₆-

(PMe₂Ph)₈][PF₆],²⁹and [Ir₃(L)₃(L′)₃H₇][PF₆]₂(L ) PCy₃, L′ )

py;⁵⁴L ) L′ ) PHOX⁵⁵). In particular, the edge-bridging

arrangement of the hydride ligands in 1a and 2a is reminiscent

of that observed in octahedral W₆(CⁱPr)(OⁱPr)₁₂H₅,²²while the

simple stoichiometry of {Rh(PR₃)H₂} is similar to that of the

hydrido-platinum monophosphine clusters such as Pt₄(PⁱPr₂-

Ph)₄H₈(Scheme 2).²¹There is also a marked structural similarity

between the arrangement of hydride ligands and the positioning

of edge-bridging chlorides in the early transition metal clusters

25

(Figure 1), such as [NbCl₁₈]^{2- 3}or Zr₆Cl₁₂H₂(PEt₃)₆and

especially Zr₆Cl₁₂(PMe₂Ph)₆.⁵⁶This structural motif is unknown

in the chemistry of rhodium clusters, which is dominated by

complexes with carbonyl ligands and {Rh(CO)₃} fragments,

although a small number of non-carbonyl bi- and trimetallic

hydrido-phosphine species with bridging hydride ligands have

been reported, such as [Rh{P(OR)₃}₂H]₃

Pr₃)₄]⁺.⁵⁹

Figure 1. Solid-state structure of the dications in 1a (top) and 2a (bottom).

Thermal ellipsoids are shown at the 50% probability level. The counterions

are not shown. Apart from the hydride ligands, all other hydrogen atoms

have been omitted for clarity.

57,58

and [Rh₂H₅(Pⁱ-

(44) Bau, R.; Ho, N. N.; Schneider, J. J.; Mason, S. A.; McIntyre, G. J. Inorg.

Chem. 2004, 43, 555-558.

and co-workers have noted similar small changes to the cluster

core on going from PEt₃to the bulkier PCy₃in the octahedral

face-bridged sufide clusters W₆S₈(PR₃)₆.⁴²The direct Rh-Rh

distances in 1a and 2a are also comparable to those found in

Rh₆(CO)₁₆(average 2.750 Å).⁴³

For both 1a and 2a, the 12 hydride ligands were located in

the final difference map, and each Rh-Rh edge in the respective

octahedra is spanned by a bridging hydride ligand, forming a

cuboctahedron surrounding the metal core (Figure 2 shows 1a).

Three independently synthesized samples of 1a gave similar

high-quality data and are all in agreement with the gross

positioning of the hydride ligands over Rh-Rh edges. The

(45) Jackson, P. F.; Johnson, B. F. G.; Lewis, J.; Raithby, P. R.; McPartlin, M.;

Nelson, W. J. H.; Rouse, K. D.; Allibon, J.; Mason, S. A. J. Chem. Soc.,

Chem. Commun. 1980, 295-297.

(46) Hart, D. W.; Teller, R. G.; Wei, C. Y.; Bau, R.; Longoni, G.; Campanella,

S.; Chini, P.; Koetzle, T. F. J. Am. Chem. Soc. 1981, 103, 1458-1466.

(47) Chen, L. F.; Cotton, F. A.; Klooster, W. T.; Koetzle, T. F. J. Am. Chem.

Soc. 1997, 119, 12175-12183.

(48) Goodfellow, R. J.; Hamon, E. M.; Howard, J. A. K.; Spencer, J. L.; Turner,

D. G. J. Chem. Soc., Chem. Commun. 1984, 1604-1606.

(49) Adams, R. D.; Pompeo, M. P.; Wu, W. G. Inorg. Chem. 1991, 30, 2899-

2905.

(50) Bemis, J. M.; Dahl, L. F. J. Am. Chem. Soc. 1997, 119, 4545-4546.

(51) Suss-Fink, G.; Plasseraud, L.; Maisse-Francois, A.; Stoeckli-Evans, H.;

Berke, H.; Fox, T.; Gautier, R.; Saillard, J. Y. J. Organomet. Chem. 2000,

609, 196-203.

(52) Chan, S.; Wong, W. T. J. Chem. Soc., Dalton Trans. 1995, 3987-3994.

(53) Lau, J. P. K.; Wong, W. T. Dalton Trans. 2005, 2579-2587.

(54) Chodosh, D. F.; Crabtree, R. H.; Felkin, H.; Morehouse, S.; Morris, G. E.

Inorg. Chem. 1982, 21, 1307-1311.

(42) Jin, S.; Venkataraman, D.; DiSalvo, F. J. Inorg. Chem. 2000, 39, 2747-

2757.

(43) Farrar, D. H.; Grachova, E. V.; Lough, A.; Patirana, C.; Poe, A. J.; Tunik,

S. P. J. Chem. Soc., Dalton Trans. 2001, 2015-2019.

(55) Smidt, S. P.; Pfaltz, A.; Martinez-Viviente, E.; Pregosin, P. S.; Albinati,

A. Organometallics 2003, 22, 1000-1009.

(56) Cotton, F. A.; Kibala, P. A.; Roth, W. J. J. Am. Chem. Soc. 1988, 110,

298-300.

9

6250 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

Table 1. Crystal and Structure Refinement Data for the New Cluster Compounds

compound

(C₇H₈) (CH₂Cl₂)_2.63

1a[BAr^F

]

4

‚

2

(C₆H₅F)

2a[BAr^F

]

4

‚

2b[HCB₁₁Me₁₁] ‚CH₂Cl₂

2

empirical formula

formula weight

temperature, K

wavelength, Å

crystal system

space group

a, Å

b, Å

c, Å

C

₁₂₄H₁₆₆B₂F₄₉P₆Rh₆

C

_181.63H_247.25B₂F₄₈P₆Rh₆Cl_5.25

C₁₃₃H₂₇₁B₂₂Cl₂P₆Rh₆

2982.50

150(2)

0.71073

orthorhombic

Pca2₁

17.6720(2)

28.0280(3)

31.0780(5)

90

3412.47

150(2)

0.71073

triclinic

4353.56

150(2)

0.71073

monoclinic

C2/c

69.5388(4)

20.4464(1)

27.8999(2

90

100.821(1)

90

38963.2(4)

8

1.484

P1h

13.3466(1)

17.3989(2)

17.8425(1)

73.208(4)

68.263(4)

88.452(3)

3669.66(5)

1

R, °

â, °

90

γ, °

volume/Å³

15393.3(3)

4

1.287

0.769

6292

Z

density (calcd), mg/m³

absorption coefficient, mm^-1

F(000)

1.544

0.830

1721

0.712

17826

crystal size, mm

θ range for data collection, °

reflections collected

independent reflections

data/restraints/parameters

largest diff peak and hole, e/Å³

final R^a,bindices [I > 2σ(I)]

0.35 × 0.15 × 0.10

3.08 to 28.31

68058

0.40 × 0.40 × 0.10

3.54 to 30.07

286379

0.25 × 0.13 × 0.08

3.55 to 27.10

64581

17997 [R(int) ) 0.0433]

17997/0/980

0.788 and -0.775

R₁) 0.0319

wR₂) 0.0739

R₁) 0.0398

wR₂) 0.0770

1.025

56646 [R(int) ) 0.0601]

56646/12/2347

2.533 and -1.157

R₁) 0.0509

wR₂) 0.1188

R₁) 0.1050

wR₂) 0.1452

1.022

27469 [R(int) ) 0.0614]

27469/1/1535

1.136 and -1.106

R₁) 0.0561

wR₂) 0.1105

R₁) 0.0940

wR₂) 0.1216

1.028

R^a,bindices (all data)

goodness-of-fit on F^{2 c}

2

^aR₁) ∑||F_o| - |F_c||/∑|F_o|. ^bwR₂) {∑[w(F_o- F_c)²]/∑[w(F_o²⁾²]}^1/2

. .

^cGOF ) S ) {∑[w(F_o- F_c)²]/(n - p)}^1/2

Table 2. Selected Interatomic Distances (Å) for the Crystallographically and Theoretically (ADF, TZVP) Determined 12- and 16-Hydride

Clusters

2

+

2+

1a[BAr^F

]

2a[BAr^F

]

2b[HCB₁₁Me₁₁]

2

[Rh₆(PH₃)₆H₁₂

calculated

]

[Rh₆(PH₃)₆H₁₆

calculated

]

4

2

4

2

H

₁₂-PⁱPr₃

H₁₂-PCy₃

H₁₆-PCy₃

Rh-Rh mean, Å

Rh-Rh, Å

2.720

2.700(1)-2.733(1)

0.033

3.825(1)-3.863(1)

0.038

8.313(1)-8.351(1)

0.038

2.246(1)-2.247(1)

0.001

2.246

176.2(1)-177.2(1)

1.0

2.735

2.719(1)-2.743(1)

0.024

3.858(1)-3.869(1)

0.011

8.379-8.392

0.013

2.260(1)-2.264(1)

0.004

2.262

177.3(1)-178.3(1)

1.0

2.815

2.720(1)-3.132(1)

0.412

3.840(1)-4.063(1)

0.223

8.131(2)-8.519(2)

0.388

2.235(2)-2.271(2)

0.036

2.252

166.8(1)-177.1(1)

10.3

2.708

2.667-2.747

0.080

3.795-3.866

0.071

8.219-8.266

0.047

2.203-2.220

0.017

2.212

172.5-176.5

4.0

2.789

2.642-3.017

0.375

3.718-4.124

0.406

8.016-8.385

0.369

2.209-2.227

0.018

2.218

153.2-174.2

21.0

δ(Rh-Rh), Å^a

Rh‚‚‚Rh cross-cluster, Å

δ(Rh‚‚‚Rh), Å

P‚‚‚P cross-cluster, Å

δ(P‚‚‚P), Å^a

Rh-P, Å

δ(Rh-P), Å

Rh-P mean, Å

Rh‚‚‚Rh-P, deg

δ(Rh‚‚‚Rh-P), deg

^aMaximum deviation in distances. ^bCalculated assuming a regular octahedron using the mean Rh-Rh distance.

The cluster cores in both 1a and 2a are surrounded by an

alkyl sheath formed by the phosphine ligands, presenting a

nonpolar outer layer, shielding the inner {Rh₆P₆H₁₂}²⁺core

structure of these clusters, this protective cover may well be

the reason why they can be isolated.

Solution NMR and Mass Spectrometric Characterization

of 1a and 2a. Complex 1a shows lower solubility in CH₂Cl₂

solution than 2a, which is readily soluble (>40 mg/cm³).

Although 1a will dissolve in CH₂Cl₂(∼10 mg/cm³) on

sonication for 10 min, overnight it tends to recrystallize, even

from CH₂Cl₂solutions. Nevertheless, the solubility is more than

adequate for the acquisition of good room-temperature solution

NMR data. Both complexes are stable in CD₂Cl₂solutions for

weeks under rigorous exclusion of air and moisture. The ³¹P-

{¹H} NMR spectrum of 1a shows a single, sharp peak centered

at δ 106.2 ppm which shows coupling to ¹⁰³Rh [J(RhP) ) 104

Hz]. For 2a, a broadened doublet is observed at δ 91.9 ppm

i

(Figure 3). For 1a, the Pr₃P ligands appear to pack so as to

allow more appreciable gaps in the alkyl sheath, whereas in 2a

the cyclohexyl groups shroud the core more effectively. It is

remarkable that, especially for the tricyclohexyl phopshines in

2a, enough space can be found to accommodate six ligands,

and the interdigitation of the cyclohexyl groups clearly occurs

effectively. Similar packing of cyclohexyl groups has been

observed in the closely related cluster W₆S₈(PCy₃)₆.⁴²The two

clusters have appreciable van der Waals diameters of 1.6 nm

(1a) and 1.8 nm (2a). With reference to the unusual electronic

(57) Brown, R. K.; Williams, J. M.; Sivak, A. J.; Muetterties, E. L. Inorg. Chem.

1980, 19, 370-374.

(58) Sivak, A. J.; Muetterties, E. L. J. Am. Chem. Soc. 1979, 101, 4878-4887.

(59) Wolf, J.; Nurnberg, O.; Schafer, M.; Werner, H. Z. Anorg. Allg. Chem.

1994, 620, 1157-1162.

[J(RhP) ) 103 Hz]. In its ¹H NMR spectrum, 1a shows signals

i

due to [BAr^F]^-and Pr₃groups in the correct ratio for a

4

dicationic cluster. The high-field region displays a single, rather

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6251

A R T I C L E S

Brayshaw et al.

Figure 4. Variable temperature ¹H NMR spectrum of 2a between 248 and

353 K, showing the high-field region (1,2-C₆H₄F₂solution).

at low temperature. At temperatures higher than 320 K, complex

1a starts to decompose (C₆H₄F₂solution). The room-temperature

i

¹³C{¹H} NMR spectrum shows only two signals for the Pr

groups, which demonstrates that the phosphine ligands rotate

(or at least librate) around their Rh-P axes at room temperature.

A low-temperature spectrum was not possible to acquire due

to the low solubility of 1a.

Figure 2. Solid-state structure of cationic cluster 1a showing the cluster

core (phosphine alkyl groups have been omitted for clarity). Inset shows

selected Rh-H bond distances (Å).

1

The room-temperature H NMR spectrum of 2a, unlike that

of 1a, displays two broad high-field peaks that consistently

integrate in the ratio 9 H:3 H with respect to the other signals

in the spectrum, at δ -27.0 ppm (fwhm ) 230 Hz) and δ -28.9

ppm (fwhm ) 330 Hz), respectively. The signals are invariant

in ratio on slow cooling to 200 K, suggesting that they are not

due to two species in dynamic equilibrium. They do sharpen

significantly at the lower temperature and (as for 1a) shift to

lower field (δ -21.7 and -25.2 ppm, respectively at 248 K).

At higher temperatures these signals coalesce and then become

a single peak (at 353 K) shifted upfield to δ -36.8 ppm, which

integrates to 12.2 H (Figure 4). The ³¹P{¹H} NMR spectra show

one phosphorus peak over all temperature ranges that also

sharpens on cooling and shifts appreciably upfield by ∆δ -12

ppm. Although there are a number of possible reasons for the

observation of two signals at low temperature (e.g., an alternative

trigonal-prismatic structure, or a mixture of bridging and

terminal hydrides), we favor the simple explanation of restricted

rotation of the bulky tricyclohexyl phosphine groups at or below

room temperature. This contrasts with triisopropyl phosphine

ligands in 1a. This restricted rotation leads to hydride ligands

that either lie directly under a cyclohexyl group or between two

cyclohexyl groups, which results in different bridging hydride

environments (Figure 5). Close inspection of the solid-state

structure of 2a reveals three such environments, in the ratio

6:3:3, consistent with the 9:3 pattern observed in solution by

¹H NMR spectroscopy. The observation of different hydride

environments also shows that the hydrides are not fluxional over

the surface of the cluster at low temperature. A ¹³C{¹H} NMR

spectrum of 2a at 250 K (see Supporting Information) shows

15 cyclohexyl environments (some representing coincident

peaks), consistent with restricted rotation of the phosphine

ligand. These signals broaden on warming, but decomposition

at higher temperatures (350 K) prevented the acquisition of a

Figure 3. Space-filling diagrams of the cationic portions of 1a (left) and

2a (right). Atoms are shown at their respective van der Waals radii. The

approximate diameter of these molecules in the solid-state is 1.6 and 1.8

nm, respectively.

broad resonance at δ -25.44 ppm (fwhm ) 46 Hz) that

i

integrates to 12 H relative to both the Pr protons and the

[BAr^F]^-aromatic protons. Correct integration of hydride signals

4

can often be problematic due to slow relaxation, and long delays

between pulses (5 s) and a large number of scans were used to

ensure that the true integral value was obtained. Average integral

values over at least three samples were also taken, with

individual integrals between each sample deviating by less than

(0.5 H. No fine structure to this hydride signal was observed,

and decoupling ³¹P did not narrow the signal significantly or

1

reveal any coupling to ¹⁰³Rh, which was unfortunate as H-

{³¹P} NMR spectra are potentially useful in confirming the

number of rhodium atoms with which each hydride environment

couples.⁵⁸Cooling to 200 K broadens the hydride peak

significantly and shifts it downfield by ca. 5 ppm to δ -19.0

ppm. This might be indicative of a fluxional process, possibly

restricted rotation of the phosphine (see later), but the low

solubility of 1a could also contribute to a poorly resolved peak

9

6252 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

Figure 5. Stick representation of the cluster core in 2a and the two postulated fluxional processes occurring in 1a/2a. Cyclohexyl groups apart from the

ipso-carbon have been removed to more clearly demonstrate the alignment of the tricyclohexylphosphine ligands.

meaningful ¹³C{¹H} NMR spectrum at the high temperature

limit. A freely rotating ligand set would be expected to show

only six separate environments for the PCy₃ligands in the ¹³C-

{¹H} NMR spectrum, while a static ligand set would give 18.

The gradual upfield shift of the hydride signal on warming from

200 to 353 K suggests a rapid, temperature-dependent equilib-

rium between two Rh-H environments (e.g., a bridging vs

terminal tautomerism), one of which is favored at high tem-

perature. Involvement of solvent at higher temperatures is

discounted, as the same spectral changes are observed in CD₂-

Cl₂, fluorobenzene, or 1,2-difluorobenzene solutions.

At 200 K, the longitudinal relaxation time of the hydrogen

ligands on the clusters (T₁)^60,61is measured as 320 ms for 1a

and 890 ms for each resonance in 2a, firmly in the region

Figure 6. Plot of the observed intensity changes (I/I₀) vs arbitrary units

proportional to the square of the gradient amplitude for PGSE measurements

associated with hydride and not dihydrogen ligands, consistent

with both the X-ray structures and the structure calculated by

DFT (vide infra). For comparison, the bridging hydrides in Ir₄H₄-

(µ-H)₄(CO)₄(PPh₃)₄have a T₁of 690 to 290 ms,⁶²while in

[Ru₄H₆(C₆H₆)₄]²⁺the bridging hydrides have T₁of between 190

and 310 ms.⁵¹This latter complex is also suggested to have a

cluster dihydrogen ligand on the basis of a short (T₁) 34 ms)

relaxation time at 153 K.

on 1a and 2a (CD₂Cl₂solutions, 60 mM solutions). The lines represent

linear least-squares fits to the experimental data.

expected to have a significantly smaller hydrodynamic radii for

the cation (ca. 7 Å or less⁶³).

Crucially, with regard to the solution characterization of 1a

and 2a, that the clusters remain intact in solution and have 12

hydride ligands is unequivocally shown by electrospray ioniza-

tion mass spectrometry (ESI-MS⁶⁵), which shows only the intact

clusters and no other charged species.⁶⁶ESI-MS on CH₂Cl₂

solutions of 1a and 2a under very soft conditions (cone voltage

of 10 V, source and desolvation gas temperatures both 30 °C)

The structure of the clusters in solution has also been probed

using pulsed gradient spin-echo (PGSE) NMR experiments.

Such experiments have been used to measure the size of

molecules in solution by their calculated hydrodynamic radii.^63,64

A PGSE experiment on 1a in CD₂Cl₂solution afforded a

hydrodynamic radius (r_H) of 9.5(3) Å for the dication. As

expected for the tricyclohexyl phosphine cluster 2a, a slightly

larger r_Hvalue is calculated, 10.7(2) Å (Figure 6). These values

are both slightly larger than those measured in the solid state

are in full agreement with the formulation [Rh₆(PR₆)₆H₁₂]²⁺

,

with a single molecular species observed at m/z 795.1 (calcd

795.2) for 1a and 1155.5 (calcd 1155.7) for 2a, which have

experimentally determined isotope patterns that are an excellent

match with theory (Figure 7).

(8.0 and 9.0 Å, respectively). The [BAr^F]^-anions in each (r_H

4

) 7.8(2) Å) are also slightly larger in solution than found in

the solid state (7.2 Å). These increases in calculated radii are

suggestive of ion-pairing between the anions and dication in

solution, a not-uncommon feature of ion-pairs in CH₂Cl₂

solution.⁶³What these values confirm is that, in solution, the

clusters are intact and do not dissociate appreciably into

mononuclear solvent-stabilized fragments, e.g., [Rh(PR₃)(H)₂-

Reversible Dihydrogen Uptake by the Clusters [Rh₆H₁₂-

(PR₃)₆]²⁺. The reversible uptake of hydrogen by molecular or

extended materials (such as nanoclusters) is an area that is

attracting significant current interest. Much of this interest has

centered on the role that such species play in hydrogenation

reactions (especially arene hydrogenation), in particular the

nature of the actual species in catalysis (nanocluster versus

molecular cluster)^1,9,10,32and potential hydrogen storage

applications.^33-35Given the high hydride content of the clusters

1a and 2a, we reasoned that they may act as well-defined

molecular models for the attachment of hydrogen on a metal

(solvent)_x][BAr^F], on the NMR time scale, as these would be

4

(60) Crabtree, R. H.; Lavin, M. J. Chem. Soc., Chem. Commun. 1985, 1661-

1662. Crabtree, R. H. Acc. Chem. Res. 1990, 23, 95.

(61) Kubas, G. J. Metal Dihydrogen and σ-Bond Complexes; Kluwer Academic/

Plenum Publishers: New York, 2001.

(62) Garlaschelli, L.; Greco, F.; Peli, G.; Manassero, M.; Sansoni, M.; Gobetto,

R.; Salassa, L.; Della Pergola, R. Eur. J. Inorg. Chem. 2003, 2108-2112.

(63) Pregosin, P. S.; Kumar, P. G. A.; Fernandez, I. Chem. ReV. 2005, 105,

2977-2998.

(64) Pregosin, P. S.; Martinez-Viviente, E.; Kumar, P. G. A. Dalton Trans. 2003,

4007-4014.

(65) Henderson, W.; McIndoe, J. S. Mass Spectrometry of Inorganic and

Organometallic Compounds; John Wiley and Sons: Chichester, 2005.

(66) Butcher, C. P. G.; Dyson, P. J.; Johnson, B. F. G.; Khimyak, T.; McIndoe,

J. S. Chem. Eur. J. 2003, 9, 944-950.

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6253

A R T I C L E S

Brayshaw et al.

Figure 8. ESI-MS of 1a and 2a after addition of H₂to afford 1b (top) and

2b (bottom).

Figure 7. ESI-MS of 1a (top) and 2a (bottom) in CH₂Cl₂solution. The

inset shows the theoretically determined isotope patterns overlaid with

experiment. Theoretical isotope patterns were calculated using Matthew

Monroe’s Molecular Weight Calculator (http://jjorg.chem.unc.edu/personal/

monroe/mwtwin.html).

hydrogen uptake occurs under mild conditions of temperature

and pressure, happening at room temperature in CD₂Cl₂solution

under 1 atm of H₂over ∼10 min, and effectively instantaneously

at room temperature under 4 atm of H₂. The uptake of hydrogen

also occurs in the solid state by placing crystalline material under

4 atm of H₂. In this instance this solid-state reaction is, not

unexpectedly, slower (16 h) and is also dependent on the size

of the crystals. Pressurizing 1a to 100 bar of H₂in CD₂Cl₂

solution afforded only 1b, showing that hydrogen uptake stops

at two molecules of H₂. The ESI-MS spectra of 1b and 2b (CH₂-

Cl₂solutions under the mild conditions used previously)

unambiguously show the uptake of two molecules of H₂(an

increase of m/z 2 ) 4 Da for a 2+ cation to m/z 797.2 and

1157.5, respectively), displaying “flagpole spectra” with excel-

lent isotope distribution match to the calculated patterns (Figure

8). Under the conditions used, the potential intermediate 14-

hydride species was not observed.

surface and/or hydrogen storage devices, and consequentally

we investigated their reactivity with regard to dihydrogen uptake/

loss.

Even though 1a and 2a have an exceptionally large number

of hydride ligands, placing either cluster under a dynamic

vacuum (10^-3Torr, 5 days, room temperature) did not result in

loss of H₂, as measured by NMR spectroscopy and ESI-MS. In

fact, the 12 hydride ligands are so firmly placed on the clusters

that 1a and 2a can be routinely recrystallized from solutions

containing an excess (>100-fold) of 1-hexene, and addition of

an excess (>100-fold) of the excellent hydrogen acceptor tert-

butylethylene (tbe) does not remove any of the 12 hydride

ligands. This was at first surprising, but in light of the electronic

structure of the clusters (vide infra) that affords a very stable

76 cve count, perhaps it should not be. Although H₂is not lost

either under vacuum or by addition of a hydrogen acceptor,

addition of H₂is facile and results in the immediate uptake of

1

The room-temperature H NMR spectra (CD₂Cl₂solutions)

of 1b and 2b show resonances due to [BAr^F]^-anions and the

4

alkyl phosphines in the appropriate ratio. In the high-field region,

two hydride signals are observed in the ratio 15:1 with respect

to both the anion and phosphine resonances. One of these is

two molecules of H₂to afford the octahedral clusters [Rh₆-

i

(PR₃)₆H₁₆][BAr^F] (Scheme 4, R ) Pr (1b), Cy (2b)). This

4 2

Scheme 4. Reversible Dihydrogen Uptake

9

6254 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

On cooling of 1b and 2b to 200 K (CD₂Cl₂), their hydride

resonances become broader and resolve into many separate

signals with an overall integral lower than 16 H, both being

nearer 12 H, showing that the hydrides are mobile over the

surface of the cluster at room temperature. The broad linewidths

of some of these signals indicate that any fluxional processes

occurring are not completely frozen out at 200 K, as does the

lower-than expected integral value. T₁measurements at 200 K

indicate that all the obserVed signals are best assigned to hydride

ligands (T₁> 230 ms), although the presence of two unresolved

dihydrogen ligands in the baseline cannot be discounted. Higher

temperatures (up to 350 K, 1,2-difluorobenzene solvent) result

in the apparent coalescence of the two peaks, possibly suggesting

exchange, although as the integral 1 H peak is very broad, it

could simply be unresolved in the baseline. At this temperature,

partial decomposition of the cluster to unidentified material is

also observed. Under a H₂atmosphere (1 atm) at 298 K,

dissolved H₂is observed as a sharp resonance at δ 4.6 ppm,

demonstrating that exchange between free and bound H₂is slow

on the NMR time scale. Slow exchange with free H₂does occur,

however, as placing 1b or 2b under a D₂atmosphere (∼4 atm)

results in complete H/D exchange for the hydride ligands in 72

h and the appearance of HD as a characteristic 1:1:1 triplet at

δ 4.55 ppm [J(DH) ) 43 Hz], along with H₂. The observation

of HD and free H₂suggests coordinated dihydrogen on the

cluster surface, resulting from H/D exchange via a dihydrogen

intermediate.^51,61To our knowledge, the only cluster identified

as containing a dihydrogen ligand (as identified by T₁measure-

1

Figure 9. High-field region of the room-temperature H NMR spectra of

2a/2b. Integrals are quoted relative to the alkyl phosphine protons.

relatively sharp, while the other is much broader, e.g., for 1b δ

-21.46 ppm (15 H, fwhm 40 Hz) and δ -29.81 ppm (1 H,

fwhm 200 Hz). This pattern of integrals is consistent over a

number of independently synthesized, crystalline samples. Figure

9 shows the high-field region for the pair 2a/2b.

This 15:1 ratio suggests a structure in which 15 hydrides are

mutually exchanging on the NMR time scale and one is not.

Mutual exchange of the 15 hydrides is necessary, as it is not

possible to arrange 15 hydrides around an octahedron sym-

metrically. To our minds, the only plausible structure that

accounts for this observation is one that invokes an interstitial

hydride with 15 hydride ligands decorating the outside of the

Rh₆octahedron. Interstitial hydrides are well-known in transition

metal clusters and have been characterized by NMR spectro-

scopy and, in some cases, neutron diffraction techniques.^45,46,67

The chemical shift of interstitial hydrides can vary over a

considerable range, for example δ +16.4 ppm for [HRu₆-

(CO)₁₈]^{- 68}to δ -26.9 ppm for [HNi₁₂(CO)₂₁]^-,⁶⁹and it has

been suggested that hydrides that are located in the center of a

regular octahedron show the lowest-field chemical shifts, while

those more asymmetrically orientated have high-field chemical

shifts.⁷⁰For 2b, an X-ray study (vide infra) shows a rather

distorted octahedron, and thus a chemical shift observed for the

unique hydride around δ -29 ppm is not unreasonable. DFT

studies also indicate that a structure with 15 hydride ligands on

the surface and one interstitial is significantly more stable than

one with 16 hydrides on the surface, and these studies also show

that this central hydride sits rather asymmetrically inside the

Rh₆core. Of course, in the absence of a neutron diffraction

study, the assignment of an interstitial hydride must remain

plausible but tentative, especially as examples exist of instances

where interstitial hydrides have been originally proposed but

later proven to be, in fact, located on the cluster surface.⁷¹The

room-temperature ³¹P{¹H} NMR spectra for both 1b and 2b

show broad environments, which for 1b resolves into a doublet

J(RhP) ) 140 Hz. Within error, the hydrodynamic radii of 1b

and 2b as calculated using PGSE in CD₂Cl₂solutions are the

same as for 1a/2a [9.5(3) and 10.3(5) Å,³respectively].

ments and DFT calculations) is [Ru₄H₆(C₆H₆)₄]^{2+ 51}

although

,

bimetallic complexes with dihydrogen ligands are known,^61,72

and dihydrogen complexes have been suggested to be possible

intermediates in clusters that catalyze H₂-D₂equilibration to

form HD.^73,74Interestingly, after 3 days under D₂at 313 K, for

1b some deuterium is also incorporated into the isopropyl groups

on the phosphines (as shown by ²H NMR and ESI-MS),

suggesting that these ligands do not play a completely innocent

role in the clusters. We have not yet pursued the mechanism of

H/D exchange, but it is likely that it involves reversible oxidative

addition of an alkyl C-H bond to the cluster core.

Despite repeated attempts, the crystalline material of 1b

afforded X-ray diffraction patterns that indicated high levels of

disorder. Crystals suitable for X-ray diffraction for 2b where

grown using the alternative anion [1-H-CB₁₁Me₁₁]^{- 40}from CH₂-

Cl₂/pentane solution under 1 atm of H₂. Crystals of the [BAr^F]^-

4

salt of 2b were of marginal quality, and the structure obtained

is not reported here. Nevertheless, the resulting cluster core

structural metrics are essentially the same as for that with the

[1-H-CB₁₁Me₁₁]^-anion, demonstrating that the anions do not

(68) Eady, C. R.; Jackson, P. F.; Johnson, B. F. G.; Lewis, J.; Malatesta, M. C.;

McPartlin, M.; Nelson, W. J. H. J. Chem. Soc., Dalton Trans. 1980, 383-

392.

(69) Eguchi, T.; Heaton, B. T.; Harding, R.; Miyagi, K.; Longoni, G.; Nahring,

J.; Nakamura, N.; Nakayama, H.; Pakkanen, T. A.; Pursiainen, J.; Smith,

A. K. J. Chem. Soc., Dalton Trans. 1996, 625-630.

(67) Bau, R.; Drabnis, M. H. Inorg. Chim. Acta 1997, 259, 27-50. Bau, R.;

Drabnis, M. H.; Garlaschelli, L.; Klooster, W. T.; Xie, Z. W.; Koetzle, T.

F.; Martinengo, S. Science 1997, 275, 1099-1102. Adams, R. D.; Babin,

J. E.; Tanner, J. T. Organometallics 1988, 7, 2027-2033. Hart, D. W.;

Teller, R. G.; Wei, C. Y.; Bau, R.; Longoni, G.; Campanella, S.; Chini, P.;

Koetzle, T. F. Angew. Chem., Int. Ed. Engl. 1979, 18, 80-81. Eady, C.

R.; Johnson, B. F. G.; Lewis, J.; Malatesta, M. C.; Machin, P.; McPartlin,

M. J. Chem. Soc., Chem. Commun. 1976, 945-946. Broach, R. W.; Dahl,

L. F.; Longoni, G.; Chini, P.; Schulz, A. J.; Williams, J. M. AdV. Chem.

Ser. 1978, 167, 93.

(70) Eguchi, T.; Heaton, B. T. J. Chem. Soc., Dalton Trans. 1999, 3523-3530.

(71) Bashall, A.; Gade, L. H.; Lewis, J.; Johnson, B. F. G.; McIntyre, G. J.;

McPartlin, M. Angew. Chem., Int. Ed. Engl. 1991, 30, 1164-1167.

(72) Sola, E.; Bakhmutov, V. I.; Torres, F.; Elduque, A.; Lopez, J. A.; Lahoz,

F. J.; Werner, H.; Oro, L. A. Organometallics 1998, 17, 683-696.

(73) Aubart, M. A.; Chandler, B. D.; Gould, R. A. T.; Krogstad, D. A.;

Schoondergang, M. F. J.; Pignolet, L. H. Inorg. Chem. 1994, 33, 3724-

3734.

(74) Aime, S.; Dastru, W.; Gobetto, R.; Krause, J.; Matas, L.; Viale, A.

Organometallics 1996, 15, 4967-4970.

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6255

A R T I C L E S

Brayshaw et al.

Figure 11. Stacked plot of ESI-MS of 1b on addition of the hydrogen

acceptor tert-butylethene (∼100-fold excess) in CH₂Cl₂solution. The bottom

spectrum was obtained 5 min after addition of tbe and the top after 24 h.

The 16-hydride complex 2a slowly loses H₂in the solid state

under an argon atmosphere to give 1a, giving a ratio of 1:0.7

for 2a:1a by ¹H NMR spectroscopy after 1 week. By contrast,

after 1 week 2b is unchanged as shown by NMR and ESI-MS.

Under vacuum (5 × 10^-3Torr), H₂loss is faster. For the ⁱPr₃P

cluster, 1b, complete H₂loss is considerably more rapid (12 h)

than for the PCy₃congener 2b (∼1 week). This increased

resistance to H₂loss for 2b reflects the increased kinetic

stabilization toward H₂loss provided by the bulky cyclohexyl

ligands, as electronically both 1b and 2b would be expected to

be very similar. This process is completely reversible for both

clusters, and repressurizing with H₂regenerates quantitatively

(NMR, ESI-MS) the H₁₆species. Loss of two hydrogen

molecules is also effected by addition of hydrogen acceptors

such as tert-butylethene or 1-hexene, as monitored by NMR

and ESI-MS (Figure 11). The loss of dihydrogen from 1b or

2b is not at odds with their putative interstitial hydride structure,

as interstitial hydrides in octahedral complexes have been noted

to be mobile between the center of the octahedron and the cluster

surface. For example, protonation of [Ru₆(µ₆-H)(CO)₁₈]^-affords

H₂Ru₆(CO)₁₈, which has two bridging hydrides,⁶⁸while the

interstitial hydride in [Co₆(µ₆-H)(CO)₁₅]^-undergoes rapid

migration from the inside to the outside of the cluster.^46,69,70

Figure 10. Solid-state structure of the dication of 2b[HCB₁₁Me₁₁]₂.

Thermal ellipsoids are shown at the 50% probability level. The counterions

are not shown. The hydride ligands around the cluster core were not located

reliably. All other hydrogen atoms have been omitted for clarity. See

Supporting Information for labeling scheme.

have a pronounced effect on the cluster geometry. Figure 10

shows the solid-state structure of 2b[1-H-CB₁₁Me₁₁]₂. Selected

structural comparisons are given in Table 2.

The cluster core in 2b describes an approximate octahedron,

but compared to the H₁₂-hydride counterparts 1a and 2a it is

significantly distorted. Notably, the Rh-Rh distances span a

wider range, δ(Rh-Rh) being an order of magnitude larger for

the H₁₆species than the H₁₂clusters. The cross-cluster Rh‚‚‚

Rh and P‚‚‚P distances are also larger and span a similarly wide

range. The cross-cluster Rh‚‚‚RhsP angles also span a much

wider range for 2b than for 2a, e.g., 178.3(1)-177.7(1)° for 2a

and 177.1(1)-166.8(1)° for 2b. The hydride ligands around the

{Rh₆} core were not located with any certainty. As discussed,

the location of hydrogen atoms near metal centers is always

potentially problematic, and coupled with the fact that the

hydrides in 2b will be distributed asymmetrically in the solid

state, potentially leading to partial occupancy of sites, this means

that the definitive location of the hydrides using X-ray crystal-

lography would always be difficult. Nevertheless, the combina-

tion of NMR and ESI-MS data unambiguously points to clusters

with 16 hydride ligands. Moreover, the structural comparisons

between the 12-hydride clusters, 1a/2a, and the 16-hydride

cluster, 2b, indicate the accommodation of four extra hydride

ligands in the latter by significantly larger Rh-Rh distances

and deviations, larger cross-cluster Rh‚‚‚Rh distances, and

greater deviations from linearity for Rh‚‚‚RhsP angles. These

structural variations are underscored by DFT calculations on

both 12- and 16-hydride structures (vide infra) that, for

comparison, are shown in Table 2. Even though the phosphine

PH₃was used in the DFT model instead of the bulky PⁱPr₃or

PCy₃ligands, all the changes observed experimentally on

addition of dihydrogen are reproduced satisfactorily in the

calculated structures.

Although complexes with hydride ligands are common in

cluster chemistry, those that can reVersibly take up and release

one molecule of H₂are few.^73,75,76Of these species that

reversibly take up H₂, only one approaches the high hydride

count of 1b/2b, [Pt₄H₇(P^tBu₃)₄][BF₄].⁴⁸This cluster reversibly

loses four molecules of H₂on exposure to air, to afford a cluster

with no hydrides (Scheme 5), or on exposure to ethene, to give

a cluster with one hydride ligand. Pt₃Re₂(CO)₆(P^tBu₃)₃has been

reported to take up a relatively large amount of H₂(3 equiv),

but the process is not reported to be reversible for this cluster.^18,76

(75) Safarowic, F. J.; Bierdeman, D. J.; Keister, J. B. J. Am. Chem. Soc. 1996,

118, 11805-11812. Farrugia, L. J.; Green, M.; Hankey, D. R.; Orpen, A.

G.; Stone, F. G. A. J. Chem. Soc., Chem. Commun. 1983, 310-312. Arif,

A. M.; Bright, T. A.; Jones, R. A.; Nunn, C. M. J. Am. Chem. Soc. 1988,

110, 6894-6895.

(76) Adams, R. D.; Captain, B.; Smith, M. D. Angew. Chem., Int. Ed. 2006, 45,

1109-1112.

9

6256 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

Scheme 5. Reversible H₂Uptake in [Pt₄H₇(P^tBu₃)₄][BF₄]

The cluster species [Rh₆(PR₃)₆H₁₆][BAr^F] are thus unique in

precision afforded by an X-ray diffraction experiment. Addition

of two molecules of H₂to give [Rh₆(PH₃)₆H₁₆]²⁺has been

modeled in two ways: by 16 hydrides on the surface of the

polyhedron, and by 15 on the surface and 1 central (interstitial).

The interstitial model comes out as being significantly more

stable by 145 kJ mol^-1, in line with the experimental NMR

data that show a 15:1 ratio of hydride ligands. Of the 15 hydrides

on the outside of the octahedron, four of them (two pairs) have

short H-H distances, indicative of dihydrogen ligands, d(H-

H) ) 1.061 and 0.965 Å. Similar structures, implicating a

mixture of dihydrogen and hydride ligands, have been calculated

for Pd₅H₆clusters.⁷⁷The presence of dihydrogen ligands is

consistent with the experimental observations that the cluster

undergoes H/D exchange with D_2(g)to produce HD_(g)and that

the low-temperature, high-field ¹H NMR spectra of both 1b and

2b show a total integral of 12 H, the four missing protons

4 2

that H₂uptake and loss is facile and the “low” hydride species

that result still have an exceptionally high number of hydrides.

To our knowledge, the complexes 1b and 2b are without

precedent in cluster chemistry, combining (i) the greatest number

of hydride ligands so far reported on a discrete metal cluster,

(ii) a hydride:metal ratio of greater than 2:1 (Pd₂₈(PtPMe₃)-

(PtPPh₃)₁₂(CO)₂₇H₁₂⁵⁰has 12-hydride ligands but has 41 metals

55

in its core, while [Ir₃(PHOX)₃H₇][PF₆]₂has a high hydride:

metal ratio of 7:3 but only seven hydride ligands^54,55), and (iii)

reversible hydrogen loss/uptake of two molecules of H₂.

DFT Calculations on the Model Cluster Pairs [Rh₆-

(PH₃)₆H₁₂]²⁺and [Rh₆(PH₃)₆H₁₆]²⁺. DFT calculations on the

cluster pair [Rh₆(PH₃)₆H₁₂]²⁺and [Rh₆(PH₃)₆H₁₆]²⁺have been

carried out at the ADF/TZVP level, with the alkyl phosphine

groups substituted by PH₃for computational expediency. Full

details of these calculations, the molecular orbital description

of the bonding, and the hydrogen uptake will appear in a

dedicated full paper. Here we briefly outline the main conclu-

sions and highlight how these calculations mirror the experi-

mentally determined results and support the assignment of the

hydride count for the cluster pairs 1a/1b and 2a/2b. We also

comment briefly on the molecular orbital structure of the 12-

hydride species.

Figure 12 shows the calculated structures for [Rh₆(PH₃)₆H₁₂]²⁺

and [Rh₆(PH₃)₆H₁₆]²⁺, and Table 2 presents selected structural

metrics for the pair, alongside the experimentally determined

structures from single-crystal X-ray diffraction. Although the

substitution of bulky PⁱPr₃or PCy₃for the model phosphine

PH₃might be expected to have a significant effect on the cluster

geometry, that there is a surprisingly good correlation between

theory and experiment shows that, although bulky, the phosphine

ligands exert less influence on the cluster core than might be

expected. Dealing with the {Rh₆P₆} cluster cores first, all the

observed structural trends on going from H₁₂to H₁₆are mirrored

in the DFT-calculated structures (Table 2). Specifically, on

addition of two molecules of H₂, the octahedron becomes

significantly less regular, Rh-Rh distances get significantly

larger, as do their range of distances, the cross-cluster Rh‚‚‚Rh

distances become larger, and the Rh‚‚‚RhsP angles deviate

more from linear in the 16-hydride cluster. The hydride ligands

in calculated [Rh₆(PH₃)₆H₁₂]²⁺describe two distinct bonding

motifs: four symmetrically bridging around a central octahedral

plane, and the remaining eight that can be considered semi-

bridging/terminal. The underlying electronic reasons for this will

be discussed in a more detailed contribution, but pleasingly,

this arrangement is partially reflected in the crystal structure of

1a, which shows six hydrides to be semi-bridging within the

Figure 12. DFT-calculated structures of [Rh₆(PH₃)₆H₁₂]²⁺(top) and [Rh₆-

(PH₃)₆H₁₆]²⁺(bottom).

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6257

A R T I C L E S

Brayshaw et al.

Figure 13. Partial molecular orbital diagram for [Rh₆(PH₃)₆H₁₂]²⁺under the constraint of O_hsymmetry, showing the orbitals involved with cluster bonding.

Diagrammatic representations of the HOMO and LUMO orbitals are also shown.

presumably broad and not observed. Also consistent with the

¹H NMR data is that the interstitial hydride is not located in

the geometric center of the cluster, with the internal Rh-H

distances varying between 2.082 and 1.875 Å in the model. The

dissociation energy of the two H₂molecules from the H₁₆cluster,

ing cluster orbitals (Figure 13). Of these, 12 may be regarded

as built from the 12 three-center, two-electron Rh-H-Rh bonds

(1a_1g, 1e_g, 1t_1u, 1t_2u, and 2t_2gin symmetry); these are equivalent

to the Nb-Cl-Nb symmetric bonding orbitals. The other 20

are principally metal-localized and correspond to the 8 Nb-

Nb bonding orbitals of the niobium cluster (2a_1g, 1t_2g, 3t_1u, and

1a_2uin symmetry) and 12 metal-localized orbitals (2t_2u, 3t_2g,

1e_u1t_1g, and 1a_2gin symmetry), which are numerically equal

(though not symmetrically equivalent) to the 12 antisymmetric

Nb-Cl-Nb bridging bonds; the π-donor interaction of the 12

Cl’s with the Nb₆core is replaced in the Rh₆H₁₂cluster by metal-

localized electrons. The remaining 6 orbitals that complete a

total orbital count of 38 (and consequently an electron count of

76) are associated with Rh-P bonding (2t_1u, 2e_g, and 3a_1g).

Replacing the chlorides in [Nb₆Cl₁₈]^4-with hydrides and

phosphines reduces the overall electron count by 24, but moving

from a group 5 metal to a group 9 metal results in a net gain of

24 electrons to the cluster. Thus, the low ligand electron count

of [Rh₆(PH₃)₆H₁₂]²⁺is exactly balanced by the move from Nb

to Rh and the subsequent use of 12 additional metal-localized

orbitals (and 24 electrons). Although with regard to octahedral

late transition metal clusters composed of {ML₃} fragments such

as Rh₆(CO)₁₆1a and 2a count less by 10 electrons, they are in

fact electron-precise for a cluster constructed from {ML₅}

i.e., the SCF energy difference between [Rh₆H₁₂(PH₃)₆]²⁺

+

2H₂and [Rh₆H₁₆(PH₃)₆]²⁺, is calculated as 151 kJ mol^-1in favor

of the [Rh₆H₁₆(PH₃)₆]²⁺cluster.

In terms of simple electron book-keeping, the 12 hydride

clusters 1a and 2a have 76 cve, this being 10 cve less than that

found for late transition metal octahedral clusters (86 cve) such

as Rh₆(CO)₁₆. However, the electron count is exactly like that

of the early transition metal clusters such as [Nb₆Cl₁₈]^4-and is

similar to that of the structurally closely related zirconium-

phosphine clusters Zr₆Cl₁₂(PMe₂Ph)₆(72 cve)⁵⁶and Zr₆Cl₁₄H₄-

(PMe₃)₄(74 cve).⁴Dicationic hexanuclear clusters [Au(PR₃)]₆

2+

of 76 cve are also known, and these can adopt either octahedral⁷⁸

or edge-shared bitetrahedral structures.⁷⁹The structural similarity

between the early transition metal clusters with edge-bridged

halides (Scheme 1) and 1a/2a, with 12 edge-bridged hydrides,

is remarkable and underscores that both sets of clusters are built

from {ML₅} fragments, e.g., {ClNbCl₄} and {(PR₃)RhH₄}. The

molecular orbital description of clusters such as [Nb₆Cl₁₈]^4-is

well understood,¹⁶and cluster bonding arises from 8 cluster

bonding orbitals. By comparison, calculations on [Rh₆(PH₃)₆-

H₁₂]²⁺, with the Rh₆P₆H₁₂core constrained to octahedral

symmetry for simplicity, revealed 20 bonding and 12 nonbond-

fragments such as [Nb₆Cl₁₈]^4-

.

The principal difference in the orbital structure when the

symmetry is lowered from octahedral to that of the optimized

calculated structure is that two of the three Rh-H-Rh 2t_2g

orbitals are considerably stabilized by the shift of eight

hydrogens from central bridging to semi-bridging positions as

a consequence of improved overlap between the H 1s orbitals

(77) Moc, J.; Musaev, D. G.; Morokuma, K. J. Phys. Chem. A 2003, 107, 4929-

4939.

(78) Bellon, P.; Manasser, M.; Sansoni, M. J. Chem. Soc., Dalton Trans. 1973,

2423-2427.

(79) Briant, C. E.; Hall, K. P.; Mingos, D. M. P.; Wheeler, A. C. J. Chem. Soc.,

Dalton Trans. 1986, 687-692.

9

6258 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

Figure 14. Time versus relative ³¹P integral plot for the synthesis of cluster 1b. 2, “Rh(PⁱPr₃)₂” species at δ(³¹P) 64-58 ppm, see text; (, [HPⁱPr₃][BAr^F₄];

b, 1b[BAr^F₄]₂; /, [RhH₂(PⁱPr₃)(η⁶-C₆H₆)][BAr^F₄]; 9, [RhH₂(PⁱPr₃)(η⁶-C₆H₅F)][BAr^F₄].

and the Rh 4d orbitals. Mixing between a 3e_gorbital and the

1a_2gorbital localizes a pair of electrons in the same plane as

the symmetrically bridging hydrogens.

at δ -5.8 ppm in the ¹H NMR spectrum at high field.³⁹

Following the reaction by NMR spectroscopy during gentle

heating at 313 K over 5 days revealed a number of intermediate

species, along with the final product 1b that forms from addition

of H₂to 1a. Figure 14 shows a graph of integrated ³¹P

resonances against time for this experiment.

The HOMO-LUMO gap in [Rh₆(PH₃)₆H₁₂]²⁺as calculated

under O_hsymmetry is relatively small, ∼0.3 eV, and the LUMO

is of e_gdegeneracy and effectively nonbonding. These two

observations suggests that addition of four electrons should be

facile. These four electrons can come from two molecules of

H₂, and addition results in the 16-hydride, 80 cve species [Rh₆-

(PH₃)₆H₁₆]²⁺having a much larger HOMO-LUMO gap of 1.3

eV, which disfavors the addition of further hydrogen, as is

observed experimentally. A similar analysis of frontier molecular

orbitals has previously been used to account for the facile uptake

of three molecules of H₂by the electronically unsaturated cluster

Pt₃Re₂(CO)₆(P^tBu₃)₃, to give Pt₃Re₂(CO)₆(P^tBu₃)₃H₆with no

major structural change.¹⁸

There are three new species observed after 2 h of reaction in

the ³¹P{¹H} NMR spectrum. The first is [HPⁱPr₃][BAr^F],⁸⁰

4

identified by a distinctive resonance at δ 46 ppm which splits

into a doublet, J(PH) ) 336 Hz, in the ³¹P NMR spectrum.

Using D₂in the reactions rather than H₂afforded [DPⁱPr₃]-

[BAr^F] and, ultimately 1b-d₁₆. The second new species is a

4

very small amount of cluster 1b. Both of these species continue

to increase in concentration throughout the experiment. The third

new species is spectroscopically identified as the fluorobenzene

complex [RhH₂(PⁱPr₃)(η⁶-C₆H₅F)][BAr^F] (3). Complex 3 has

4

The full molecular orbital structure description, discussion

of the facile uptake of H₂, and electron counting for the clusters

will be dealt with in detail in a future full paper. Importantly

for this paper, which deals with the synthesis and structures of

these new clusters, the structural descriptors of the H₁₂and H₁₆

complexes are reproduced very well in the DFT calculations,

which give confidence to the assignment of the solid-state

structures of the cluster pair 1a/1b as well as 2b.

been independently synthesized by addition of H₂to [(binor-

S)Rh(PⁱPr₃)][BAr^F] (a source of a reactive {Rh(PR₃)}⁺frag-

4

ment)⁸¹in fluorobenzene solution (Scheme 6), and the NMR

data match exactly those observed in the reaction mixture.

After 3 h the concentration of complex 3 drops, to be replaced

by a species identified as the benzene complex [RhH₂(PⁱPr₃)-

(η⁶-C₆H₆)][BAr^F] (4).⁸²Concomitant with the appearance of

4

1

4, free benzene also appears in the H NMR spectrum (δ 7.21

Mechanism of Cluster Formation. As discussed in the

Introduction, clusters that form under kinetic conditions are

relatively unusual, and the attractive time scales (5 days) and

moderate temperatures involved (313-323 K) in the formation

of 1a and 2a afford a rare opportunity to probe a cluster self-

assembly process using NMR spectroscopy. Addition of H₂(∼4

ppm) as well as traces of fluorocyclohexane (C₆H₁₁F, identified

by a characteristic d of t of t at δ 4.40 ppm⁸³). The generation

of benzene most likely comes from hydrido-defluorination of

fluorobenzene by traces of colloidal Rh, while hydrogenation

(80) Torres, F.; Sola, E.; Martin, M.; Lopez, J. A.; Lahoz, F. J.; Oro, L. A. J.

Am. Chem. Soc. 1999, 121, 10632-10633.

atm) to a fluorobenzene solution of [Rh(PⁱPr₃)₂(nbd)][BAr^F]

4

(81) Brayshaw, S. K.; Green, J. C.; Sceats, E.; Kociok-Kohn, G.; Weller, A. S.

Angew. Chem., Int. Ed. 2006, 45, 452-456.

in a sealed J. Youngs tube initially results in the formation of

the almost colorless dihydrogen/dihydride complexes [Rh-

(82) Canepa, G.; Brandt, C. D.; Ilg, K.; Wolf, J.; Werner, H. Chem. Eur. J.

2003, 9, 2502-2515.

(PⁱPr₃)₂(H)₂(η²-H₂)_x][BAr^F] (x ) 1 or 2), as shown by a broad

4

(83) Lambert, J. B.; Greifenstein, L. G. J. Am. Chem. Soc. 1974, 96, 5120-

5124.

³¹P{¹H} NMR resonance at δ 63 ppm and a very broad signal

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6259

A R T I C L E S

Brayshaw et al.

Scheme 6. Synthesis of [RhH₂(PⁱPr₃)(η⁶-C₆H₅X)][BAr^F

]

4

Scheme 7. Observed Species in the Cluster Buildup Reaction and Suggested Mechanistic Routes (Dashed Arrows)^a

-

^a[BAr^F

]

4

anions not shown.

of fluorobenzene to form fluorocyclohexane is also catalyzed

by colloidal Rh.⁸⁴The benzene thus produced readily displaces

the weakly bound fluorobenzene in 3 to give 4. After about 22

h, complex 4 starts to disappear, so by the end of the monitoring

experiment the concentrations of 3 and 4 are both zero. During

this time, the peak at δ 63 ppm in the ³¹P{¹H} NMR spectrum

gradually moves to being a doublet at δ 58.2 ppm [J(RhH) )

110 Hz] and reaches what appears to be an equilibrium

concentration of approximately half the initial concentration of

starting material. We assign this peak at δ 58.2 ppm to [Rh-

the only observed species being [Rh(PⁱPr₃)₂(H)₂(η²-H₂)₂][BAr^F].

4

Removal of H₂under vacuum after the initial hydrogenation of

[Rh(nbd)(PⁱPr₃)₂][BAr^F] stopped the reaction and afforded the

4

previously spectroscopically characterized complex [Rh(PⁱPr₃)₂-

(H)₂(L)₂][BAr^F] (L ) solvent or agostic interaction).³⁹Use of

4

less bulky phosphine ligands, such as PEt₃(cone angle 132° ⁴¹),

starting from [Rh(nbd)(PEt₃)₂][BAr^F], cleanly affords the Rh-

4

(I) species [Rh(η⁶-C₆H₅F)(PEt₃)₂][BAr^F] (5), which has a

4

coordinated fluorobenzene ligand and no hydride ligands (see

Supporting Information for full characterization). On addition

of H₂no cluster is formed under 4 atm of H₂on heating to 323

K, with 5 remaining unchanged. This suggests that phosphines

with sufficient steric bulk to force the trans arrangement

(PⁱPr₃)₂(H)₂(η²-H₂)][BAr^F], formed from loss of one H₂from

4

[Rh(PⁱPr₃)₂(H)₂(η²-H₂)₂][BAr^F], as has been noted to happen

4

when the H₂pressure drops from ∼4 atm to ∼1-2 atm.³⁹The

final products of the reaction after 5 days are 1b, [HPⁱPr₃]-

[BAr^F], and [Rh(PⁱPr₃)₂(H)₂(η²-H₂)][BAr^F]. Repeating the

observed in [Rh(PⁱPr₃)₂(H)₂(η²-H₂)₂][BAr^F] are required for

4

cluster formation.

4

experiment under 100 bar H₂did not afford the cluster, with

Scheme 7 outlines the species observed by NMR spectro-

scopy. That cluster formation does not occur under very high

pressures of H₂, but also requires some H₂to be available, shows

(84) Yang, H.; Gao, H.; Angelici, R. J. Organometallics 1999, 18, 2285-2287.

Blum, J.; Rosenfeld, A.; Gelman, F.; Schumann, H.; Avnir, D. J. Mol.

Catal. A 1999, 146, 117-122.

that the bis-dihydrogen complex [Rh(PⁱPr₃)₂(H)₂(η²-H₂)₂][BAr^F]

4

9

6260 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

Scheme 8. Comparison of Possible Mechanisms Leading to

that once a significant amount of protonated phosphine has built

up in the reaction, cluster formation will cease. This observation

is consistent with the apparent equilibrium concentration of [Rh-

Cluster Formation^a

(PⁱPr₃)₂(H)₂(η²-H₂)][BAr^F] observed at the end of the reaction,

4

and the idea that repressurizing the reaction with H₂does not

afford more cluster. This inhibition likely comes from a back

reaction (protonation and coordination of phosphine) between

^aThe number of hydrogen atoms (H_x) is arbitrary in each step.

[HPⁱPr₃][BAr^F] and a mono-phosphine intermediate involved

4

is not an intermediate, but instead a complex with only one H₂

ligand might be important mechanistically. To form a cluster,

one phosphine must be lost from such an intermediate, and we

suggest three possible mechanistic pathways for this. In pathway

A, simple loss of PⁱPr₃from the intermediate dihydrogen

complex generates a low-coordinate cationic {Rh(PⁱPr₃)H_x}⁺

fragment that is stabilized by fluorobenzene to give 3 (and

subsequently 4). Although these species are observed by NMR

spectroscopy, self-assembly of six cationic fragments such as

3 would be unlikely due to charge repulsion. Consistent with

this, cluster was not observed to form when pure samples of 3

or 4 were heated at 313 K under H₂. Complexes 3 and 4, at

least in pure form, thus represent mechanistic dead-ends in

cluster formation.

in cluster buildup.

Although the order of all these possible events discussed, and

the actual species involved, needs further clarification before a

complete mechanistic picture of the cluster self-assembly can

be determined, that the clusters are reproducibly isolated in a

good yield (∼30-40%) and are of excellent purity shows that

whatever self-assembly mechanisms are operating are experi-

mentally robust. This is an important consideration if the

chemistry of these cluster species is to be developed.

Conclusions

The synthesis and characterization of octahedral rhodium

cluster complexes under conditions that favor kinetically

controlled cluster products has been described: [Rh₆(PR₃)₆H₁₂]-

[BAr^F] (PR₃) PⁱPr₃, PCy₃; BAr^F) [B{C₆H₃(CF₃)₂}₄]^-).

The second pathway (B) involves elimination of [HPⁱPr₃]⁺

(as observed in the reaction), possibly through heterolytic H₂

cleavage of coordinated dihydrogen.⁸⁵We have previously

discussed this possibility,³⁹and examples where such a process

is suggested to occur in other cationic dihydrogen complexes

exist.⁸⁶This would afford a reactive neutral monophopshine

{Rh(PⁱPr₃)H_x} fragment that could then self-assemble to give a

neutral cluster. Subsequent oxidation, or more likely protonation

by either [HPⁱPr₃]⁺(pK_a) 9.0) or HF formed from the hydrido-

defluorination of fluorobenzene in the reaction mixture, would

afford the observed dicationic cluster (eq 1, Scheme 8). We

have yet to test whether a low-coordinate neutral {Rh(PR₃)H_x}

fragment is a viable intermediate in cluster buildup because

suitable synthetic routes to such monophosphine complexes have

not yet been developed. A third alternative is the combination

of four neutral and two cationic fragments, pathway C. This

would afford a cluster directly. One would expect that a neutral

{Rh(PR₃)H_x} fragment would dimerize at the very least by

hydride bridging,⁵⁸and thus cluster condensation could result

from addition of “[Rh(PR₃)H_x]₂” to [RhH₂(PⁱPr₃)(η⁶-C₆H₅X)]-

4 2

4

These clusters are unique in that they have structures that are

exactly like those of early transition metal clusters that have

edge-bridging π-donor ligands rather than those expected for a

late transition metal. These clusters reversibly take up two

molecules of dihydrogen to afford clusters that have 16

hydrogens surrounding the metal core. This uptake of H₂is a

consequence of two low-lying unoccupied molecular orbitals

that readily accept two electron pairs (or two mole equivalents

of H₂). The reversible uptake of dihydrogen is of clear relevance

to applications in hydrogen storage, and the drivers behind the

reversible storage of large amounts of H₂(>6% by mass for

mobile applications) have been described extensively and will

not be repeated here.³³The current best molecular materials are

those based around metal-organic frameworks (MOFs),³⁴which

can reach storage efficiencies of 2.5% at 77 K.⁸⁸Based on the

cluster dication core, complex 1a can store a modest ∼0.25%

of H₂at room temperature for the 12-hydride to 16-hydride

process, and in terms of the volume of H₂uptake, 1a can

reversibly take up ∼20 times its volume of H₂. Clearly, they

do not make suitable candidates for hydrogen storage for mobile

applications. However, their well-defined molecular nature that

can be probed by NMR, X-ray crystallography, and mass

spectrometry as well as DFT calculations means that they make

excellent models for reversible hydrogen attachment on a metal

surface. Moreover, the ease of filling and emptying the frontier

molecular orbitals with electrons (in this case from H₂) without

significant structural change suggests that these clusters should

present interesting electrochemistry and magnetic behavior,

especially when linked to the reversible uptake of H₂and the

extremely high hydride count of the clusters. These aspects,

and ways of increasing the hydrogen storage capacity of these

clusters by “unlocking” the remaining 12 hydride ligands that

surround the metal core, must be future goals associated with

this work.

[BAr^F] (X ) F (3), H (4)) to afford a cationic trimetallic species

4

that could then itself dimerize (eq 2, Scheme 8). A related self-

assembly process is suggested to occur in the formation of

W₆(CⁱPr)(OⁱPr)₁₂H₅from hydrogenolysis of W₂(ⁱBu)₂(OⁱPr)₄,

where reactive species such as “W₂H₂(OⁱPr)₄” are implicated

to combine to form “W₄H₄(OⁱPr)₄” and subsequently associate

with W₂(ⁱBu)₂(OⁱPr)₄.²²Related to the loss of [HPR₃]⁺as a

driver to cluster formation, triflic acid-promoted loss of phos-

phine has been reported to encourage the formation of [Pt₆(µ-

P^tBu₂)₄(CO)₆][OTf]₂from the trimetallic precursor Pt₃(µ-

P^tBu₂)₃H(CO)₂.⁸⁷

Repeating the reaction in the presence of a 2-fold excess of

[HPⁱPr₃][BAr^F] completely inhibited cluster formation, showing

4

(85) Kubas, G. AdV. Inorg. Chem. 2004, 56.

(86) Chin, B.; Lough, A. J.; Morris, R. H.; Schweitzer, C. T.; Dagostino, C.

Inorg. Chem. 1994, 33, 6278-6288. Yi, C. S.; Lee, D. W.; He, Z.;

Rheingold, A. L.; Lam, K.-C.; Concolino, T. E. Organometallics 2000,

19, 2909.

(87) de Biani, F. F.; Lenco, A.; Laschi, F.; Leoni, P.; Marchetti, F.; Marchetti,

L.; Mealli, C.; Zanello, P. J. Am. Chem. Soc. 2005, 127, 3076-3089.

(88) Chen, B. L.; Ockwig, N. W.; Millward, A. R.; Contreras, D. S.; Yaghi, O.

M. Angew. Chem., Int. Ed. 2005, 44, 4745-4749.

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6261

A R T I C L E S

Brayshaw et al.

[Rh₆(PⁱPr₃)₆H₁₂][BAr^F]₂(1a). In a typical experiment, a solution

Experimental Section

4

of [Rh(nbd)(PⁱPr₃)₂] (200 mg, 0.145 mmol) in fluorobenzene (10 mL)

in a J. Young tube was hydrogenated (ca. 4 atm of H₂) by freezing a

solution to 77 K, placing the tube under 1 atm of H₂, sealing the tube,

and warming to room temperature (298/77 ∼ 4) and stirred at 40 °C

under ca. 4 atm of H₂for 5 days. Pentane (50 cm³) was added, and the

mixture was crystallized at -18 °C overnight. The material that resulted

was crystallized in the same manner an additional two times. The sample

was dissolved in fluorobenzene, and 1-hexene was added to ensure

General. All manipulations were performed under an inert atmo-

sphere of argon, using standard Schlenk line and glovebox techniques.

Glassware was dried in an oven at 130 °C overnight and flamed with

a blow torch, under vacuum, three times before use. C₆H₅F was distilled

from CaH₂. CH₂Cl₂, pentane, and hexane were purified using an

MBraun solvent purification system. CD₂Cl₂was distilled under vacuum

from CaH₂. [(Cy₃P)₂Rh(nbd)][1-H-closo-CB₁₁Me₁₁], [(Cy₃P)₂Rh(nbd)]-

[BAr^F], and [(ⁱPr₃P)₂Rh(nbd)][BAr^F] were prepared by the published

4

that only [Rh₆(PⁱPr₃)₆H₁₂][BAr^F]₂(and none of the 16-hydride species)

4

route³⁹using Ag[closo-1-H-CB₁₁Me₁₁],⁴⁰K[BAr^F],⁸⁹and ⁱPr₃P or PCy₃.

4

was present. After 24 h, the solution was layered with hexanes, and

[Rh(binor-S)(PⁱPr₃)][BAr^F] was synthesized by the published proce-

4

slow diffusion gave [Rh₆(PⁱPr₃)₆H₁₂][BAr^F]₂as a deep red crystalline

4

dure.⁸¹

solid (25 mg, 31%).

1

NMR Spectroscopy. H and ³¹P NMR spectra were recorded on a

¹H NMR (400 MHz, CD₂Cl₂): δ 7.71 (m, 16H, BAr^F), 7.55 (m,

4

Bruker Avance 400 MHz FT-NMR spectrometer. Residual protio

8H, BAr^F), 2.32 (m, 18H, PCH), 1.23 (m, 108H, CH₃), -25.44 (s,

solvent was used as reference for ¹H NMR spectra (CD₂Cl₂: δ ) 5.33).

4

fwhm ) 46 Hz, 12H, Rh-H). ³¹P NMR (162 MHz, CD₂Cl₂): δ 106.2

1

³¹P NMR spectra were referenced against 85% H₃PO₄(external). H

1

[d, J(RhP) ) 104 Hz]. Selected low-temperature H NMR data: at

NMR spectra in fluorobenzene were referenced to the solvent signal,

the position of which was determined using an external sample of TMS.

For samples run in protio-fluorobenzene (reaction monitoring experi-

ments), shimming was performed by optimization of line shapes

obtained by visual inspection of a real-time Fourier-transformed FID.

Coupling constants are quoted in hertz. ¹H NMR spectra were recorded

using long delays between pulses (>5 s) to avoid saturation. T₁

measurements were made using the standard inversion-recovery-delay

method (180°-τ-90°). PGSE diffusion data were collected at 294 K

in 5-mm NMR tubes using the previously outlined method,⁶³measuring

integrated resonances of the PR₃groups in the ¹H NMR spectrum. The

diffusion coefficient was calculated from the slope of the regression

line of the plot {ln(I/I₀) vs G²}.

250 K, δ -25.44 (s, fwhm 55 Hz); at 200 K, δ -19.03 (s, fwhm 76

Hz, T₁) 320 ms). ¹³C NMR (162 MHz, CD₂Cl₂): δ 29.92 (br, fwhm

30 Hz, PCH), δ 20.53 (s, CH₃). ESI-MS (CH₂Cl₂): calcd for

[P₆Rh₆C₅₄H₁₃₈]

²⁺, 795.2; obsd 795.1. Elemental analysis: C₁₁₈H₁₆₂F₄₈B₂P₆-

Rh₆‚C₆H₅F requires C 43.63, H 4.93; found C 43.48, H 4.86.

[Rh₆(PⁱPr₃)₆H₁₆][BAr^F]₂(1b). A solution of 1a (5 mg) in dichlo-

4

romethane (0.5 cm³) was placed under a hydrogen atmosphere (4 atm)

for 2 min in a J. Young NMR tube by freezing a solution to 77 K,

placing the tube under 1 atm of H₂, sealing the tube, and warming to

room temperature (298/77 ∼ 4). Alternatively, placing the sample under

1 atm of H₂effects conversion in 10 min. Conversion to [Rh₆(PⁱPr₃)₆H₁₆]-

[BAr^F]₂(1b) was quantitative by NMR spectroscopy. Although

4

crystalline material was readily available by recrystallization from

fluorobenzene/pentane under a H₂atmosphere (as checked by ESI-MS

and NMR), the crystals that resulted always gave diffraction patterns

that were indicative of long-range disorder in the solid state. Other

solvent/anion combinations did not produce better quality crystals.

Alternatively, finely crystalline or finely powdered 1a (∼5 mg) was

treated with H₂(∼4 atm) in a J. Young tube for 10 h, resulting in

quantitative conversion (by ESI-MS and NMR) to 1b.

Mass Spectrometry. ESI-MS data were collected on a Waters

Micromass Q-Tof micromass spectrometer in positive-ion mode.

Samples were infused by means of a syringe pump at 5 µL min^-1

.

Solutions were made up in an inert-atmosphere glovebox using dry

CH₂Cl₂. Capillary voltage was set to 2900 V. To minimize fragmenta-

tion of the parent ion, the cone voltage was set to 10 V, and source

and desolvation gas temperatures were both set to 30 °C. Theoretical

isotope patterns were calculated using Matthew Monroe’s Molecular

Weight Calculator (http://jjorg.chem.unc.edu/personal/monroe/mwtwin.

html).

¹H NMR (400 MHz, CD₂Cl₂, 298 K under 1 atm of H₂; free H₂

observed at δ ∼4.6 as a sharp singlet): δ 7.71 (m, 16H, BAr^F), 7.55

4

(m, 8H, BAr^F), 2.27 (m, 18H, PCH), 1.24 [dd, J(PH) ) 15.4 Hz, J(HH)

Computational Details. Calculations were carried out using the ADF

code version 2004.1.⁹⁰Vosko, Wilk, and Nusair’s local functional was

employed with the Becke 88 and the Perdew 86 nonlocal exchange

and correlation gradient corrections. The basis sets used were uncon-

tracted triple-ú Slater-type orbitals (STOs). Hydrogen and phosphorus

were given extra polarization functions (2p on H and 4d on P).⁹¹The

cores of atoms were frozen, P up to the 2p level and Rh up to 4p.

Scalar ZORA relativistic corrections were used. Calculations for [Rh₆-

(PH₃)₆H₁₆]²⁺with 16 hydrides on the octahedral Rh₆surface result in

a final energy of -168.18 eV. Putting one hydride ligand interstitial

lowers this energy to -169.68 eV, a difference of 1.5 eV (145 kJ

mol^-1). The dissociation energy of the two H₂molecules from the H₁₆

4

) 7.1 Hz, 108H, CH₃], -21.46 (s, fwhm 40 Hz, 15H, Rh-H), -29.81

(br s, fwhm 200 Hz, 1H, Rh-H). ³¹P{¹H} NMR (162 MHz, CD₂Cl₂):

1

δ 109.2 [d, J(RhP) ) 140 Hz]. Selected low-temperature H NMR

data: at 200 K, δ -16.5 to -25.9 (at least eight signals, all signals

show T₁≈ 330 ms), -29.3 (T₁) 430 ms), total integral over hydride

region ) 12.9 relative to ⁱPr groups; the peak at δ -29.3 (assigned as

the interstitial hydride in the 298 K spectrum) integrates to ∼1H in

both room-temperature and low-temperature spectra. ESI-MS (CH₂-

Cl₂): calcd for [P₆Rh₆C₅₄H₁₄₂]

²⁺, 797.2; obsd, 797.3. Elemental analysis

was not obtained due to loss of H₂under vacuum. Characterization by

NMR and ESI-MS is unequivocal.

cluster, i.e., the SCF energy difference between [Rh₆H₁₂(PH₃)₆]²⁺

+

Synthesis of [Rh₆(PCy₃)₆H₁₂][BAr^F]₂(2a). In a typical experiment,

4

2H₂and [Rh₆H₁₆(PH₃)₆]²⁺, is calculated as 151 kJ mol^-1, with the 16-

hydride cluster being the more stable. In addition to full optimizations

of both structures, [Rh₆H₁₂(PH₃)₆]²⁺was also optimized with the

Rh₆H₁₂P₆core constrained to octahedral symmetry. Coordinates for the

calculated structures have been published as Supporting Information

previously.³⁷

a solution of [Rh(nbd)(PCy₃)₂][BAr^F] (500 mg, 0.31 mmol) in

4

fluorobenzene (20 mL) in a J. Young tube was hydrogenated (ca. 4

atm of H₂) by freezing a solution to 77 K, placing the tube under 1

atm of H₂, sealing the tube, and warming to room temperature (298/77

∼ 4 atm) and held at 50 °C under ca. 4 atm of H₂for 5 days. Pentane

(80 cm³) was added, and the mixture was crystallized at 5 °C overnight.

The residue was dissolved in dichloromethane (5 mL), 1-hexene (0.5

cm³) was added, and the mixture was heated to 40 °C for 40 h to effect

conversion to the 12-hydride species. The solvent was removed in

vacuo, and the dark material was repeatedly crystallized from dichlo-

(89) Buschmann, W. E.; Miller, J. S. Inorg. Synth. 2002, 33, 85.

(90) Velde, G. T.; Bickelhaupt, F. M.; Gisbergen, S. J. A. V.; Guerra, C. F.;

Baerends, E. J.; Snijders, J. G. J. Comput. Chem. 2001, 22, 931. Guerra,

C. F.; Snijders, J. G.; Velde, G. T.; Baerends, E. J. Theor. Chem. Acc.

1998, 99, 391. ADF, 2004.0 ed.; SCM, Theoretical Chemistry, Vrije

Universiteit: Amsterdam, The Netherlands, 2004 (http://www.scm.com).

(91) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200-1211.

Perdew, J. P. Phys. ReV. B 1986, 33, 8822-8824. Becke, A. D. Phys. ReV.

A 1988, 38, 3098-3100.

romethane/toluene until negligible colorless material ([HPCy₃][BAr^F])

4

cocrystallized. The solid was dissolved in dichloromethane, and slow

diffusion of toluene into the solution gave [Rh₆(PCy₃)₆H₁₂][BAr^F]₂as

4

9

6262 J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006

High Hydride Count Rhodium Octahedra

A R T I C L E S

a deep red solid (84 mg, 40%). NMR and ESI data of the resulting

crystals were unequivocal in the characterization as the 12-hydride

species.

) 23 Hz], indicating a hydride count of two (the small ¹⁹F coupling is

not resolved in this experiment).

[Rh(PⁱPr₃)(H₂)(η⁶-C₆H₆)][BAr^F] (4). A solution of [(binor-S)Rh-

4

¹H NMR (400 MHz, CD₂Cl₂, 298 K): δ 7.71 (m, 16H, BAr^F), 7.55

(PⁱPr₃)][BAr^F] (5 mg) and benzene (50 µL) in CD₂Cl₂(0.5 mL) was

4

(m, 8H, BAr^F), 0.8-2.5 (m, 198H, PCy₃), -27.20 (br s, fwhm 230

4

placed under 4 atm of dihydrogen in a J. Young NMR tube. The solution

was shaken and allowed to stand for 15 min, resulting in quantitative

conversion to 5, as determined by in situ ¹H and ³¹P NMR spectroscopy.

The NMR data for the cation are essentially the same as those reported

Hz, 9H, Rh-H), -28.9 (br s, fwhm 330 Hz, 3H, Rh-H). Selected

variable-temperature ¹H NMR data: at 330 K, δ -21.7 (12H); at 250

K, δ -21.7 (9H), -25.2 (3H); at 200 K, δ -19.52 (9H, T₁) 890 ms),

-21.94 (3 H, T₁) 890 ms). ³¹P{¹H} (162 MHz, CD₂Cl₂, 298 K): δ

91.9 [d br, J(RhP) ) 102Hz]; at 200 K, δ 88.2 [d, J(RhP) ) 105 Hz];

at 348 K, 100.4 (vbr s, fwhm 200 Hz]. ESI-MS (CH₂Cl₂): calcd for

82

for the [PF₆]^-salt.

Selected ¹H NMR (400 MHz, CD₂Cl₂): δ 6.77 (s, 6H, C₆H₆), 1.94-

2.07 (m, 3H, PCH), 1.12 [dd, J(HH) ) 16.0 Hz, J(PH) ) 7.1 Hz, 18H,

CH₃], -14.54 [dd, J(PH) ) 27.7 Hz, J(RhH) ) 23.6 Hz, 2H, Rh-H].

³¹P{¹H} NMR (162 MHz, CD₂Cl₂): δ 96.3 [d, J(RhP) ) 142 Hz].

[P₆Rh₆C₁₀₈H₂₁₀]

²⁺, 1155.5; obsd, 1155.7. Elemental analysis: C₁₇₃H₂₃₄B₂-

F₄₈P₆Rh₆‚1.5CH₂Cl₂requires C 50.02, H 5.73; found C 49.95, H 5.49.

Solvent of crystallization (CH₂Cl₂) is found in the solid-state structure.

X-ray Crystallography. Intensity data for 1a, 2a, and 2b were

collected at 150 K on a Nonius KappaCCD diffractometer equipped

with a low-temperature device, using graphite-monochromated Mo KR

radiation (λ ) 0.71073 Å). Data were processed using the supplied

Nonius software. For 2a, a symmetry-related (multiscan) absorption

correction was employed. Crystal parameters and details on data

collection, solution, and refinement for the complexes are provided in

Table 1. Structure solution, followed by full-matrix least squares

refinement, was performed using the WinGX-1.70 suite of programs

throughout. For 1a and 2a, all 12 hydrogen atoms were found in

bridging positions around the Rh₆core and could be refined freely,

whereas in 2b only 5 out of 16 hydrogen atoms could be found and

refined reliably in the final stages of the refinement. In 1a, a few of

Synthesis of [Rh₆(PCy₃)₆H₁₆][BAr^F]₂(2b). A solution of [Rh₆-

4

(PCy₃)₆H₁₂][BAr^F]₂(5 mg) in CD₂Cl₂(0.5 mL) was shaken under a

4

hydrogen atmosphere (4 atm) for 2 min in a J. Young NMR tube.

Alternatively, placing a solution of 2a under 1 atm of H₂effects

conversion in 10 min. Conversion to [Rh₆(PCy₃)₆H₁₆][BAr^F]₂(2b) was

4

quantitative by NMR spectroscopy. The solution was layered with

pentane, and slow diffusion gave [Rh₆(PCy₃)₆H₁₆][BAr^F]₂as a deep

4

red solid (4.5 mg, 90%). The crystalline material that resulted gave a

poor-quality refinement (R₁> 9%). Other solvent combinations did

not produce better quality crystals. Use of the weakly coordinating anion

[1-H-closo-CB₁₁Me₁₁]^-in place of [BAr^F]^-(starting from [(Cy₃P)₂-

4

Rh(nbd)][1-H-closo-CB₁₁Me₁₁]) gave reasonable quality crystals (R₁

< 5%). The structures of both cluster cores are similar, and the hydrogen

atoms could not be reliably located in either. Alternatively, finely

crystalline or finely powdered 2a (∼5 mg) was treated with H₂(∼4

atm) for 10 h in a J. Young tube, resulting in quantitative conversion

(by ESI-MS and NMR) to 2b.

the CF₃groups of the [BAr^F]^-anion show rotational disorder in the

4

ratio 40:60. Compound 1a co-crystallizes with one molecule of

fluorobenzene solvent, which is located around a center of inversion

and shows symmetry-related disorder of the F atom. In 2a, co-

crystallization occurs with one toluene molecule and dichloromethane

solvent that refines best to a partial occupancy of 2.75. All the solvent

molecules show some kind of disorder, and most of them are only partly

occupied within the lattice. The toluene molecules share their site with

CH₂Cl₂molecules and therefore had to be restrained and refined

isotropically. Crystallographic data files have been deposited with the

Cambridge Crystallographic Data Service (CCDC, 12 Union Rd.,

Cambridge CB2 1EZ, UK; tel. (+44) 1223-336-408; fax (+44) 1223-

336-033, E-mail deposit@ccdc.cam.ac.uk) under the following codes:

1a, 603489; 2a, 603490; and 2b, 275849.

¹H NMR (400 MHz, CD₂Cl₂, 298 K under 1 atm of H₂; free H₂is

observed at δ ∼4.6): δ 7.71 (m, 16H, BAr^F), 7.55 (m, 8H, BAr^F),

4

1.0-2.4 (m, 198H, PCy₃), -21.59 (br s, fwhm 190 Hz, 15H, Rh-H),

-28.37 (br s, fwhm 390 Hz, 1H, Rh-H). Selected low-temperature

¹H NMR data: at 200 K, δ -26.71 (T₁) 940 ms), others (all ∼1100

ms); total integral over hydride region ) 12.3 H relative to Cy groups.

³¹P{¹H} NMR (162 MHz, CD₂Cl₂, 298 K): δ 107.4 (br s, fwhm 1100

Hz, 2P), 89.0 (br s, fwhm 900 Hz, 4 P). ESI-MS (CH₂Cl₂): calcd for

[P₆Rh₆C₁₀₈H₂₁₄]

²⁺, 1157.5; obsd, 1157.8. Elemental analysis was not

obtained due to loss of H₂under vacuum. Characterization by NMR

and ESI-MS is unequivocal.

Acknowledgment. We thank the EPSRC, under grants GR/

R36824/01 and GR/T10169, and the Royal Society for support.

Prof. Paul Dyson and Dr. Gabor Laurenczy (EPFL, Switzerland)

are thanked for acquiring the high-pressure NMR data.

[Rh(H)₂(C₆H₅F)(PⁱPr₃)][BAr^F]₂(3). A solution of [(binor-S)Rh-

4

(PⁱPr₃)][BAr^F] (5 mg) and fluorobenzene (0.5 mL) was placed under

4

4 atm of of dihydrogen (see 1a for details). The solution was shaken

and allowed to stand for 15 min, resulting in quantitative conversion

to 3, as determined by in situ ¹H and ³¹P NMR spectroscopy. Compound

3 could not be isolated in solid form, and so characterization rests on

spectroscopic data alone.

¹H NMR (400.1 MHz, C₆H₅F): δ 6.32 (m, 2H, C₆H₅F), 6.18 (m,

2H, C₆H₅F), 5.82 (m, 1H, C₆H₅F), 1.54 [apparent octet, J(HH) ≈ J(PH)

≈ 7.5 Hz, 3H, PCH], 0.74 [dd, J(PH) ) 16.2 Hz, J(HH) ) 7.1 Hz,

18H, CH₃], -14.98 [ddd, J(RhH) ) 29.1 Hz, J(PH) ) 23.2 Hz, J(FH)

) 1.6 Hz, 2H, Rh-H]. ³¹P NMR (162.0 MHz, C₆H₅F): δ 95.45 [dd,

J(RhP) ) 141 Hz, J(FP) ) 1.7 Hz]. ¹⁹F NMR (376.5 MHz, C₆H₅F):

δ -109.8 (br s). In ³¹P{¹H-selective} experiments, only the isopropyl

alkyl phosphine protons (δ 2.3-1.2 ppm) lead to a ³¹P{¹H-selective}

NMR spectrum that is a doublet of triplets [J(RhP) ) 141 Hz, J(HP)

Supporting Information Available: X-ray crystallographic

data for 1a, 2a, and 2b; low-temperature (250 K) ¹³C{¹H} NMR

spectrum of 2a; room-temperature (298 K) ¹³C{¹H} NMR

spectrum of 1a; ¹H and ³¹P{¹H selective} NMR data for 3; ³¹P-

{¹H} NMR spectra over 5 days for the cluster formation

reaction; synthesis, NMR, and X-ray crystallographic data for

5. CIF files for all the crystal structures reported in this paper.

This material is available free of charge via the Internet at

http://pubs.acs.org.

JA0604663

9

J. AM. CHEM. SOC. VOL. 128, NO. 18, 2006 6263

Article Doi

High hydride count rhodium octahedra, [Rh6(PR3) 6H12][BArF4]2: Synthesis, structures, and reversible hydrogen uptake under mild conditions

DOI: 10.1021/ja0604663

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja0604663

Full text of DOI:10.1021/ja0604663

Products guided by the article

R&D Labs maybe for 890054-64-5

Relevant to this article

Hot Product